Effectiveness Problem Gambling Brief Telephone Interventions

EFFECTIVENESS OF PROBLEM GAMBLING BRIEF
TELEPHONE INTERVENTIONS: A RANDOMISED

CONTROLLED TRIAL
Provider Number: 467589
Contract Number: 326673/00 and 326673/01
FINAL REPORT
13 December 2012
Prepared for:
Ministry of Health
PO Box 5013
WELLINGTON
Authors:
Professor Max Abbott
Dr Maria Bellringer
Associate Professor Alain Vandal
Professor David Hodgins
Katie Palmer Du Preez
Dr Jason Landon
Dr Sean Sullivan
Professor Valery Feigin
ACKNOWLEDGEMENTS
This report has been prepared by the Gambling and Addictions Research Centre, National
Institute for Public Health and Mental Health Research, School of Public Health and
Psychosocial Studies, Faculty of Health and Environmental Sciences, Auckland University of
Technology, Private Bag 92006, Auckland 1142, New Zealand.
The authors are highly appreciative of, and would like to thank, Gambling Helpline Ltd (now
part of Lifeline Auckland) which was the partner organisation for this research. Gambling
Helpline staff embraced this project with enthusiasm and recruited all participants to the
study, providing the interventions with fidelity and integrity. Without Gambling Helpline’s
commitment to the study, this research would not have been possible.
Thanks are also due to Professor Philip Schluter for advice in the developmental stages of the
project, Dr Justin Pulford for project management of the pilot phase, and to Lore Le Pabic and
Nick Garrett for additional data analyses.
Grateful acknowledgement is made of all the gamblers who participated in this study, the
majority of whom continued to participate throughout the follow-up interviews.
1
Effectiveness of problem gambling brief telephone interventions: A randomised controlled trial
Provider No: 467589, Contract No: 326673/00 and 326673/01
Gambling and Addictions Research Centre, Auckland University of Technology
Final Report, 13 December 2012
CONTENTS
EXECUTIVE SUMMARY.......................................................................................................9
1. BACKGROUND.......................................................................................................13
2. LITERATURE REVIEW...........................................................................................15
3. RESEARCH METHODOLOGY...............................................................................21
3.1 Ethics approval......................................................................................................21
3.2 Trial design............................................................................................................21
3.3 Participants............................................................................................................21
3.3.1 Eligibility criteria 21
3.3.2 Setting and location 22
3.4 Interventions..........................................................................................................22
3.4.1 Group 1: Helpline standard care (TAU) 23
3.4.2 Group 2: Single brief motivational interview (MI) 23
3.4.3 Group 3: MI plus self-help workbook (MI+W) 23
3.4.4 Group 4: MI+W plus four follow-up motivational booster sessions (MI+W+B) 23
3.4.5 Treatment integrity and fidelity 24
3.5 Outcome measures.................................................................................................25
3.5.1 Initial assessment 25
3.5.2 Follow-up assessments 27
3.5.3 Collateral assessments 27
3.6 Sample size............................................................................................................28
3.7 Randomisation.......................................................................................................28
3.8 Blinding.................................................................................................................29
3.9 Trial hypotheses and statistical methods................................................................29
3.9.1 Study hypotheses 29
3.9.2 Study endpoints 30
3.9.3 Analysis sets 31
3.9.4 Statistical methods 31
4. RESULTS..................................................................................................................39
4.1 Participants............................................................................................................39
4.1.1. Participant flow and study sample 39
4.1.2. Recruitment dates 40
4.2 Descriptive statistics..............................................................................................42
4.2.1. Number of participants 42
4.2.2. Socio-demographic characteristics 43
4.2.3. Gambling characterisation 43
4.2.4. Treatment assistance, goal and prospects 44
4.2.5. Co-existing issues 44
4.2.6. Primary efficacy outcomes 45
4.2.7. Secondary efficacy outcomes 47
4.2.8. Treatment engagement 51
4.3 Primary analyses ITT data set................................................................................53
4.4 Subgroup analyses ITT data set.............................................................................56
4.4.1. Gender subgroups 56
4.4.2. Ethnicity subgroups 57
4.4.3. Gambling mode 59
4.4.4. Baseline PGSI score 59
4.4.5. Baseline Kessler-10 score 61
4.4.6. Baseline AUDIT-C score 64
2
4.4.7. Baseline gambling goal 65
4.4.8. Baseline belief in treatment success 68
4.4.9. Goal achievement 70
4.5 Primary analyses PP data set..................................................................................71
4.6 Secondary analyses ITT data set............................................................................74
4.6.1 Primary variables 74
4.6.2 PGSI 77
4.6.3 Motivation to overcome gambling problem 79
4.6.4 Control over gambling 79
4.6.5 Psychological distress, alcohol abuse/dependence, drug abuse, quality of life and
deprivation index 80
4.6.6 Mental disorders mood module 80
4.6.7 Tobacco use 80
4.6.8 Treatment for co-existing issues 80
4.6.9 Gambling impacts 81
4.6.10 Legal problems 81
4.6.11 Workbook and other formal treatment engagement 81
4.7 Collateral assessments...........................................................................................84
4.8 Treatment integrity and fidelity.............................................................................86
5. DISCUSSION............................................................................................................88
5.1. Limitations............................................................................................................88
5.1.1. Differential attrition and imbalance 88
5.1.2. Multiplicity 88
5.1.3. Ascertainment bias 88
5.1.4. Selection bias 89
5.1.5. Post-intervention baseline assessment 89
5.2. Discussion and interpretation.................................................................................89
5.2.1 Scene setting 89
5.2.2 Study interpretation 91
6. OTHER INFORMATION..........................................................................................96
6.1 Registration...........................................................................................................96
6.2 Protocol.................................................................................................................96
6.3 Funding.................................................................................................................96
7. REFERENCES..........................................................................................................97
APPENDIX 1 Ethical approval.............................................................................................103
APPENDIX 2 Trial hypotheses notation...............................................................................105
APPENDIX 3 Summary table of analyses............................................................................108
APPENDIX 4 Tables - Descriptive statistics.........................................................................116
Table 4.1: Socio-demographics.............................................................................................116
Table 4.2: Area of residence..................................................................................................118
Table 4.3: Gambling characterisation....................................................................................121
Table 4.4: Treatment prospects.............................................................................................123
Table 4.5: Primary efficacy - gambling, money lost and gambling cessation/improvement..124
Table 4.6: Secondary efficacy outcomes - PGSI...................................................................125
Table 4.7: Secondary efficacy outcomes - Control over gambling........................................126
3
Table 4.8: Secondary efficacy outcomes - Co-existing issues...............................................127
Table 4.9: Secondary efficacy outcomes - Gambling impacts...............................................131
Table 4.10: Secondary efficacy outcomes - Goal setting and motivation..............................134
Table 4.11: Secondary efficacy outcomes - Workbook reception and use.............................135
Table 4.12: Secondary efficacy outcomes - Treatment service assistance.............................137
Table 4.13: Secondary efficacy outcomes - Treatment service assistance.............................138
APPENDIX 5 Tables - Subgroup analyses...........................................................................139
Table 5.1: TAU vs. MI Days Gambled, Money Lost by gender............................................139
Table 5.2: TAU vs. MI Gambling-quit or improved by gender.............................................139
Table 5.3: Hypotheses B and C - Days Gambled, Money Lost, PGSI - females...................139
Table 5.4: Hypotheses B and C - Gambling-quit or improved by gender..............................140
Table 5.5: TAU vs. MI Days Gambled, Money Lost by ethnicity.........................................140
Table 5.6: TAU vs. MI Gambling-quit or improved by ethnicity..........................................141
Table 5.7: Hypotheses B and C - Days Gambled, Money Lost, PGSI - by ethnicity.............142
Table 5.8: Hypotheses B and C - Gambling-quit or improved by ethnicity...........................144
Table 5.9: TAU vs. MI Days Gambled, Money Lost by gambling mode..............................145
Table 5.10: TAU vs. MI Gambling-quit or improved by gambling mode.............................145
Table 5.11: Hypotheses B and C - Days Gambled, Money Lost, PGSI by gambling mode. .145
Table 5.12: Hypotheses B and C - Gambling-quit or improved by gambling mode..............146
Table 5.13: TAU vs. MI Days Gambled, Money Lost by dichotomised baseline PGSI score
..............................................................................................................................................147
Table 5.14: TAU vs. MI Gambling-quit or improved by dichotomised baseline PGSI score 147
Table 5.15: Hypotheses B and C - Days Gambled, Money Lost, PGSI by baseline PGSI ≤ 17
..............................................................................................................................................148
Table 5.16: Hypotheses B and C - Gambling-quit or improved by baseline PGSI ≤ 17........148
Table 5.17: TAU vs. MI Days Gambled, Money Lost by baseline Kessler-10 score.............149
Table 5.18: TAU vs. MI Gambling-quit or improved by baseline Kessler-10 score..............149
Table 5.19: Hypotheses B and C - Days Gambled, Money Lost, PGSI by baseline K-10 ≤ 30
..............................................................................................................................................149
Table 5.20: Hypotheses B and C - Gambling-quit or improved by baseline K-10 ≤ 30........150
Table 5.21: TAU vs. MI Days Gambled, Money Lost by baseline AUDIT-C score..............150
Table 5.22: TAU vs. MI Gambling-quit or improved by baseline AUDIT-C score...............150
Table 5.23: Hypotheses B and C - Days Gambled, Money Lost, PGSI by baseline AUDIT-C
..............................................................................................................................................151
Table 5.24: Hypotheses B and C - Gambling-quit or improved by high risk AUDIT-C score
..............................................................................................................................................152
Table 5.25: Hypotheses B and C - Days Gambled, Money Lost, PGSI by baseline quit
gambling goal.......................................................................................................................153
4
Table 5.26: Hypotheses B and C - Gambling-quit or improved by quit gambling goal.........153
Table 5.27: Hypotheses B and C - Days Gambled, Money Lost, PGSI by belief in treatment
success..................................................................................................................................154
Table 5.28: Hypotheses B and C - Gambling-quit or improved by high belief treatment
success..................................................................................................................................155
Table 5.29: Hypotheses B and C - Goal met in past 3-months..............................................156
APPENDIX 6 Tables - Secondary analyses..........................................................................157
Table 6.1: Hypotheses B and C - Motivation to overcome gambling....................................157
Table 6.2: Hypotheses B and C - Kessler-10, AUDIT-C, DAST, WHOQoL and NZDI........158
Table 6.3: Hypotheses B and C - PRIME-MD......................................................................159
Table 6.4: Hypotheses B and C - Tobacco use......................................................................160
Table 6.5: Hypotheses B and C - Treatment for co-existing issues.......................................161
Table 6.6: Hypotheses B and C - Gambling impacts.............................................................162
Table 6.7: Hypotheses B and C - Legal problems.................................................................163
Table 6.8: Hypotheses B and C - Other formal service engagement.....................................163
APPENDIX 7 Tables - Call timings......................................................................................164
Table 7.1: Intervention delivery timing (minutes).................................................................164
Table 7.2: Follow-up assessment timing (days)....................................................................165
Table 7.3: Booster call timing (days)....................................................................................166
5
LIST OF TABLES
Table 1: Number of booster calls received..............................................................................41

Table 2: ITT data set at each time point..................................................................................42
Table 3: PP data set at each time point....................................................................................42
Table 4: Current and past treatment........................................................................................44
Table 5: Treatment goal..........................................................................................................44
Table 6: Co-existing issues.....................................................................................................45
Table 7: MI vs. TAU Days Gambled, Money Lost..................................................................53
Table 8: MI vs. TAU time-averaged Gambling-quit or improved...........................................53
Table 9: Hypotheses B and C - Days Gambled, Money Lost, PGSI........................................54
Table 10: Hypotheses B and C - Gambling-quit or improved.................................................55
Table 11: Number of participants by gender...........................................................................56
Table 12: Hypotheses B and C - days gambled, money lost, PGSI - males.............................57
Table 13: Number of participants by gender...........................................................................57
Table 14: Hypothesis C - Money Lost gambling at 12-months by ethnicity...........................58
Table 15: Number of participants by gambling mode.............................................................59
Table 16: Number of participants by dichotomised PGSI baseline score................................60
Table 17: Hypotheses B and C - Days Gambled, Money Lost, PGSI by baseline PGSI >17. .61
Table 18: Hypotheses B and C - Gambling-quit or improved by baseline PGSI > 17.............61
Table 19: Number of participants by dichotomised Kessler-10 baseline score........................62
Table 20: Hypotheses B and C - Days gambled, Money lost, PGSI by baseline K-10 score >
30............................................................................................................................................63
Table 21: Hypotheses B and C - Gambling-quit or improved by Kessler-10 score > 30.........64
Table 22: Number of participants by dichotomised AUDIT-C baseline score.........................64
Table 23: Hypotheses B and C - gambling-quit or improved by low risk AUDIT-C score......65
Table 24: Number of participants by dichotomised baseline gambling goal...........................65
Table 25: TAU vs. MI days gambled, money lost by dichotomised baseline gambling goal...66
Table 26: TAU vs. MI Gambling-quit or improved by dichotomised baseline gambling goal 66
Table 27: Hypotheses B and C - Days gambled, Money lost, PGSI by baseline control
gambling goal.........................................................................................................................67
Table 28: Hypotheses B and C - gambling-quit or improved by control gambling goal.........68
Table 29: Number of participants by dichotomised belief in treatment success......................68
Table 30: TAU vs. MI days gambled, money lost by belief in treatment success....................69
Table 31: TAU vs. MI Gambling-quit or improved by dichotomised belief in treatment
success....................................................................................................................................69
Table 32: Hypotheses B and C - Gambling-quit or improved by low belief treatment success
................................................................................................................................................70
Table 33: TAU vs. MI Days Gambled, Money Lost................................................................71
Table 34: TAU vs. MI time-averaged Gambling-quit or improved.........................................71
Table 35: Hypotheses B and C - Days Gambled, Money Lost, PGSI......................................72
Table 36: Hypotheses B and C - Gambling-quit or improved.................................................73
Table 37: Hypotheses C* and D: Days Gambled, Money Lost and PGSI...............................75
Table 38: Hypothesis E - Days Gambled, Money Lost, control over gambling and PGSI......76
Table 39: Hypotheses C*, D and E - Gambling-quit or improved and goal met.....................77
Table 40: TAU vs. MI PGSI....................................................................................................77
Table 41: Hypotheses B and C - PGSI....................................................................................78
Table 42: Hypotheses B and C - dichotomised PGSI..............................................................79
Table 43: Hypotheses B and C - Control over gambling.........................................................80
Table 44: Number of participants reporting receiving workbook............................................81
Table 45: Hypotheses F and G - Workbook engagement, time-averaged and at 3-months......83
Table 46: Gambler and collateral reports of gambling............................................................84
6
Table 47: Gambler and collateral reports of gambling by collateral confidence ‘fairly’ or
‘extremely’.............................................................................................................................84
Table 48: Gambler and collateral reports of gambling by collateral confidence ‘not at all’ or
‘somewhat’.............................................................................................................................85
Table 49: Reliability...............................................................................................................87
7
LIST OF FIGURES
Figure 1: Participant flow.......................................................................................................40

Figure 2: Median Days Gambled per month...........................................................................45
Figure 3: Median Money Lost per day....................................................................................46
Figure 4: Percentage Gambling-quit or improved...................................................................46
Figure 5: Median PGSI score, past 3-month time frame.........................................................47
Figure 6: Control over gambling behaviour............................................................................48
Figure 7: Median Kessler-10 score.........................................................................................48
Figure 8: Median AUDIT-C score...........................................................................................49
Figure 9: Percentage goal met in past three months................................................................50
Figure 10: Treatment adherence..............................................................................................86
Figure 11: Mean elements.......................................................................................................86
8
1. EXECUTIVE SUMMARY
Background
Problem gambling and wider gambling-related harms constitute a significant health and social
issue. A variety of policies and services have been developed with the intent of preventing
and reducing problem gambling and related harms. In New Zealand this includes, among
other measures, a national gambling helpline and face-to-face counselling services. It is not
known how effective these services are, generally or for particular client groups. The
evidence base is weak, locally and internationally, a consequence of both the relatively small
number of studies conducted and the poor quality of most of them including clinically
heterogeneous and statistically underpowered studies. The present statistically powered study
used internationally accepted methods and outcome measures and was designed to assess the
effectiveness of three brief telephone interventions relative to standard gambling helpline
treatment and each other. Two of these brief interventions had been examined previously in
North American efficacy trials involving volunteers recruited via the mass media. They are
among only three forms of psychological intervention that can be considered, on the basis of
research to date, to be ‘possibly efficacious’ in the treatment of problem gambling. However,
it was not known how readily these brief interventions could be integrated into the day-to-day
operations of an existing service or how effective they are when delivered by community-
based practitioners and evaluated by researchers independent of the person or team that
developed them.
Since two of the interventions had been evaluated previously in efficacy trials that included
wait-list controls, a further objective of the present study was to see how helpline standard
care outcomes compared with those from these treatments. This would provide an indication
of the extent to which current helpline practice reduces gambling problems beyond what
would occur if the clients had wanted but not received helpline engagement until a later time.
Given that the helpline does not have a waiting list it would not be ethical to include a control
group of this type in evaluating the service.
A number of studies in the gambling and wider addictions field have found that even very
brief interventions can be effective in reducing problems though the sustainability of the
intervention effects remain unclear. For this reason, one of the interventions included in the
present study involved just a single motivational interview, to enable comparison with
standard treatment and the other interventions that, while brief, were somewhat more
intensive.
A further objective of the present study was to identify subgroups of clients who do better
with different types and intensities of intervention. This is important in terms of matching
clients to interventions that are more effective for them and developing stepped-care models
that are cost-effective in reaching larger numbers of problem gamblers, including the majority
who do not currently access care. Very little is known about this important topic in relation to
problem gambling, in large part because clinical trials to date have not included sufficient
numbers of participants to examine subgroup differences in treatment response. The present
study was designed to be sufficiently large to address this issue.
Methodology
9
The study was designed as a single-site Randomised Controlled Trial (RCT). The inclusion
criteria were: minimum age of 18 years; perception of having a gambling problem; and
willingness to read a short workbook (to ensure reading ability), have calls recorded, provide
follow-up data on gambling, and provide the name of collateral/s. Present or past
involvement in treatment or mutual help groups for gambling or other mental health problems
was documented and did not preclude participation. Callers were excluded from the trial if
they were considered by the counsellor to be actively psychotic, or they required immediate
crisis or police intervention because they posed a serious risk to themselves or others.
Four hundred and sixty-two first-time helpline callers who met eligibility criteria were
randomly assigned to four groups on a 1:1:1:1 ratio using a computer-generated block
randomisation procedure. The block size was 20, allocating participants to one of the four
treatment groups. Random assignment continued until there was a minimum of 110
participants in each group. The trial had 70% power to significantly detect a one-day
difference in mean days gambled between treatment groups (after accounting for time
changes), a $20/day difference in dollars gambled between treatment groups and a quit or
improved gambling rate difference of 0.13.
The treatments were: (1) Helpline standard care (TAU) 1, (2) Single motivational interview
(MI), (3) Single motivational interview plus cognitive-behavioural self-help workbook
(MI+W) and, (4) Single motivational interview plus workbook plus four follow-up
motivational telephone interviews (MI+W+B). Callers could choose their own treatment goal
(quit some or all forms of gambling, or control their gambling). The primary outcome
measures were self-reports of days gambled, money lost gambling and treatment goal success.
Secondary outcome measures included problem gambling severity, control over gambling,
gambling impacts, psychiatric comorbidity, general psychological distress and quality of life.
Initial assessments were conducted by helpline counsellors prior to participants receiving a
randomly allocated intervention. Further information 2 was collected by research staff, blind
to treatment allocation, within seven days after the telephone intervention and the primary and
secondary outcome measures were generally repeated at three, six and 12 months post-
intervention. Collateral information3, from one or more persons nominated by callers, was
obtained at three and 12 months. Intention To Treat and Per Protocol analyses were used.
The primary hypotheses are:

1. All four groups will evidence significant reduction in gambling
2. The Motivational Interview (MI) group will show similar improvement to Helpline
standard care (TAU)
3. The Motivational Interview plus Workbook group (MI+W) and the Motivational
Interview plus Workbook plus Booster group (MI+W+B) will show greater
improvement than the MI and TAU groups
4. The MI+W+B group will show greater improvement than the other three groups at
the 12-month follow-up.
The trial was registered with the Australian New Zealand Clinical Trials Registry (registration
number ACTRN12609000560291). The study was approved by the Multi-region Ethics
Committee (reference number MEC/09/04/043, 3 June 2009).
1
Brief screening, problem identification and referral to face-to-face problem gambling counselling
services or other services and websites and/or suggestions for self-care. Motivational interviewing
aspects were excluded.
2
More detailed gambling/problem gambling history, the mood module of the Primary Care Evaluation
of Mental Disorders, and the New Zealand Index of Socio-economic Deprivation for Individuals.
3
Collaterals were asked about the participant’s involvement with gambling over the last month, and the
confidence they had in the accuracy of their reports.
10
Results
All Helpline counsellors involved in the trial were successfully trained to reliably and
consistently deliver motivational interviews, the standard helpline intervention and follow-up
booster sessions. Following training, the new counselling approach and other trial procedures
became integrated into the operations of the helpline service. The great majority of the 462
callers recruited into the trial (N = 451) received the applicable, randomised intervention,
although only a minority of MI+W+B participants received all four booster sessions (N = 39,
34%). Overall trial retention was 81%, 74% and 64% at three-, six- and 12-months
respectively, with participant retention varying slightly across the four interventions.
Interview duration did not differ across the intervention groups and there was no significant
differential loss to follow-up between the study groups or overall.
With respect to treatment outcome, participants in all four intervention groups evidenced
statistically and clinically significant, sustained improvement on the three primary measures
self-reports of days gambled, money lost gambling and treatment goal success). This applied
when performance was time-averaged across the duration of the trial and when assessed at 12
months. Substantial improvement was also found for problem gambling severity and other
measures including self-ratings of control over gambling, gambling impacts on work, social
life, family and home and health, psychological distress, major and minor depression and
quality of life. Little or no change was evident with respect to alcohol misuse and tobacco
use.
As hypothesised, there were no significant outcome differences between the MI and TAU
interventions. Contrary to expectation, participants in the more intensive MI+W and
MI+W+B interventions did not have better outcomes on the primary outcome measures than
those who received MI and TAU. Although there were no significant primary outcome
differences between participants in each of the treatment groups overall, differences were
found for a number of subgroups. Usually these differences were evident for only one or a
few outcome measures. In most cases MI+W+B participants had significantly better
outcomes than their counterparts receiving MI alone. MI participants with lower levels of
belief in their success in achieving their treatment goal did worse on one outcome measure
than those in TAU. In this case those in the more intensive MI+W+B condition had better
outcomes than their MI counterparts. Participants who, at the baseline assessment, had more
serious gambling problems or whose goal was to control/reduce their gambling rather than
quit gambling had better outcomes in the MI+W+B group than in the TAU and MI groups.
Similarly, participants in the MI+W+B group with higher levels of psychological disorder and
lower alcohol misuse levels had better outcomes in relation to money lost gambling and/or
having quit or improved control over gambling, compared with their counterparts in the MI
group. The only finding related to ethnicity was that Maori in the MI+W+B group showed
greater improvement in money lost gambling (i.e. lost less money on average) at the 12-
month assessment than Maori in the MI group.
Discussion
This study demonstrated that brief motivational and cognitive-behavioural interventions can
be readily integrated into the everyday operations of an existing problem gambling helpline.
Prior to the present study, efficacy had only been demonstrated in trials involving volunteers,
recruited via advertising and conducted by research teams led by the person who developed
these interventions. The present study demonstrated that these interventions are also
effective, producing sustained statistically and clinically significant outcomes in people
11
seeking help from an existing treatment service. Given the nature of both the setting and the
study population it is likely that these brief interventions would perform well in other services
that provide clinical assistance to problem gamblers. This may well extend to face-to-face
and internet delivery services. TAU participants did as well as those in the two interventions
(MI+W and MI+W+B) that had previously been shown to produce significantly better
outcomes than wait-list controls. This suggests that standard helpline treatment would also
perform well relative to wait-list controls. The finding that participants receiving a single
motivational interview did as well as those receiving the more intensive TAU, MI+W and
MI+W+B interventions regarding the primary outcome measures is consistent with a growing
body of treatment literature in the gambling and wider addictions field that indicates that
‘more’ is not necessarily better than ‘less’.
The study design did not allow determination of the various therapy components that
contributed to the significant positive outcomes, across a range of gambling and other
measures, or provide a clear indication of why similar outcomes were generally achieved in
the four intervention groups. Identification of the major ingredients of effective gambling
treatment remains an important object for further investigation. The finding that particular
subgroups of participants, including those with different treatment goals and problem severity,
did significantly better with some interventions than with others is of particular note. While
further research and replication is required prior to reaching firm conclusions, the study
provides an indication of client groups that may do significantly better with particular types
and intensities of intervention. It is also possible that further differences in treatment
response will be found over time, for participants overall in each of the treatments and/or for
subgroups within these treatments. It is anticipated that a further follow-up assessment will
be conducted to assess longer term impacts. The inclusion of cost-benefit analysis in future
studies would assist in making decisions regarding the incorporation of these and other
evidence-based interventions into existing services and their optimal application to different
client groups.
12
2. BACKGROUND
Problem gambling is a significant public health issue, contributing to a broad spectrum of

morbidity and harm to individuals, families and communities. Maori, Pacific people and
populations in areas of high deprivation are disproportionately impacted. The Ministry of
Health accords high priority to the prevention and reduction of gambling-related harm and
funds intervention services including the gambling helpline and face-to-face counselling. It is
not known how effective these services are, in general, or for particular groups. A weak
evidence base internationally further impedes service improvement. Only three forms of
psychological intervention (cognitive treatments, cognitive behavioural treatments, and brief
motivational plus self-help interventions) can be considered ‘possibly efficacious’ (Ladouceur
et al., 2001; Echeburua et al., 1996; Hodgins et al., 2001; Hodgins et al., 2004; Petry et al.,
2006; Petry et al., 2008). The brief intervention approach involving a motivational interview
and self-help workbook appears to produce outcomes comparable to more intensive therapies.
However, none of these interventions has been demonstrated to be effective when conducted
in every day clinical or community settings.
Reflecting on their examination of the gambling intervention literature, Westphal and Abbott
(2006) concluded: “In sum gamblers respond to several possibly efficacious treatments …
with the majority benefiting, at least in the short term, when conducted by the original
investigators. There is no evidence that the beneficial effects occur when the treatments are
performed by other investigators or community based clinicians” (p.131). This is clearly a
very serious shortcoming. These authors identified barriers to the development of evidence-
based treatments including low sample size, heterogeneous samples, lack of protocol driven
treatments, single site clinical trials, lack of replication of studies by independent
investigators and high rates of non-specific treatment response. They examined other fields
that have overcome many of these barriers by, among other things, developing close
collaborations between treatment providers and investigators and conducting multi-site
studies. The trial detailed in the present report was designed to address most of the fore-
mentioned deficiencies.
In addition to the foregoing limitations, the small sample size of studies and lack of outcome
and effectiveness research means that little is known about individual characteristics
associated with success in different treatments. This information would enable clients to be
matched to particular interventions and treatment outcomes improved.
The reviews suggest that, for most problem gamblers, short-term and less intense (‘minimal’
or ‘brief’) interventions might be as effective as longer, more intensive therapies. Such
approaches, typically including brief motivational interviews and/or self-help workbooks,
have been shown to be effective with a variety of problems including alcohol and substance
misuse. Meta-analyses comparing self-help workbooks and no treatment controls or
therapist-directed interventions indicate that workbooks are more effective than no treatment
controls and as effective as the same programmes administered by therapists (Gould & Clum,
1993). While it appears that workbooks are generally effective, it remains unclear which
particular types of intervention are most beneficial to which type of individual (Babor, 1994).
In the alcohol field, however, there are indications that brief interventions are particularly
effective, and highly cost effective, for people with less serious forms of disorder (Bertholet et
al., 2005).
The current trial progressed the evaluation of brief motivational interviews and self-help
workbooks from efficacy testing with community volunteers to an assessment of effectiveness
with a representative sample of problem gamblers who sought information and help for
13
gambling from a telephone helpline. In addition to assessing the effectiveness of a ‘probably
efficacious’ treatment in a real life clinical setting, the current trial assessed whether or not the
addition of follow-up telephone booster sessions enhanced treatment outcomes, both
generally and for particular client groups. The current trial also included a ‘dismantling’
component via the addition of a motivational interview only condition, to contribute to our
understanding of how therapy works. From the previous Hodgins et al. randomised
controlled trials (2001; 2004) it was not known whether it was the motivational interview or
the combination of motivational interview and cognitive-behavioural workbook that was
responsible for the treatment effect.
The current trial assessed whether or not the interview alone could produce results
comparable to those of the original combination, as well as enable comparison of the
motivational interview with the other treatment conditions. Petry and colleagues’ (2008)
findings suggest that ‘more’ is not necessarily ‘better’.
The main purpose of the current trial was to examine the effectiveness of three ‘experimental’
brief telephone interventions relative to standard helpline treatment 4 (the control group), and
to compare their performance relative to each other. However, it also provided important
information about client uptake, choice, nature of, and evaluation of currently available
treatments. The control group for this trial is of particular note in this regard. Assessment of
this group documented the nature of ‘standard’ care currently accessed by helpline callers
during and following their initial helpline contact. This condition can be regarded as an
uncontrolled outcome study (but not an efficacy or effectiveness study) in its own right.
The current study is a definitive randomised controlled trial involving four groups with
repeated measures (pre-treatment, three months, six months and 12 months) enabling
investigation of independent and some interaction effects of the different interventions.
The four groups are:

 Group 1: Helpline standard care (control group; ‘Treatment as Usual’)
 Group 2: Single brief motivational interview
 Group 3: Single brief motivational interview plus self-help workbook
 Group 4: Single brief motivational interview plus self-help workbook plus four
follow-up motivational booster sessions.
The primary hypotheses were:

5. All four groups will evidence significant reduction in gambling
6. The Motivational Interview (MI) group will show similar improvement to Treatment
as Usual (TAU)
7. The Motivational Interview plus Workbook group (MI+W) and the Motivational
Interview plus Workbook plus Booster (MI+W+B) group will show greater
improvement than the MI and TAU groups
8. The MI+W+B group will show greater improvement than the other three groups at
the 12-month follow-up
In December 2008, the Gambling and Addictions Research Centre at Auckland University of
Technology was commissioned by the Ministry of Health to conduct the research project
National problem gambling intervention effectiveness which is reported in this document
titled Effectiveness of problem gambling brief telephone interventions: A randomised
controlled trial.
4
Includes brief screening, reflective listening to clients’ concerns, referral to face-to-face problem
gambling counselling services, and/or suggestions for self-care.
14
3. LITERATURE REVIEW
Gambling has been a major growth industry during the last 20 years. Increased availability of
some forms of gambling, particularly electronic gaming machines (EGMs) and casino table
games, have been associated with a rise in gambling-related problems. In a number of
jurisdictions, 15% to 30% of regular EGM participants experience gambling problems
(Abbott, 2006). Pathological gambling has long been included in psychiatric classification
manuals (Abbott & Volberg, 2006). The public health significance of the wider spectrum of
gambling-related harms experienced by individuals, families and communities has also
received recognition with the Ministry of Health (2005, 2007a, 2010) developing policies and
strategies to prevent and reduce gambling-related harms.
Approximately one to two percent of adult New Zealanders are estimated to be problem
gamblers, with about twice as many experiencing less serious problems (Abbott & Volberg,
2000; Ministry of Health, 2006, 2009). The Ministry of Health estimates that the effects of
problem gambling result in a loss of 3,300 to 10,600 years of ‘quality of life’ in this country
per year, or $330 million to $1.06 billion per annum (Ministry of Health, 2004). Prevalence is
particularly high for Maori and Pacific people (four to six times higher than for
European/Pakeha). Other risk factors include being aged 25 to 34 years, residence in lower
socio-economic areas, lower educational attainment, having paid employment and living
alone.
General population and clinical studies indicate significant comorbidity, with elevated rates of
numerous mental health and physical disorders (Abbott, Williams, & Volberg, 2004a; Petry &
Weinstock, 2007). The 2006/07 New Zealand Health Survey found that when compared to
people with no gambling problems, problem gamblers were 3.73 times more likely to be a
current smoker and 5.20 times more likely to be engaging in hazardous drinking behaviour
(Ministry of Health, 2009). The nature of relationships between gambling and comorbid
behaviours and conditions (e.g. temporal sequence and causality) are not well understood
since few prospective studies have been conducted (Abbott & Clarke, 2007). However,
problematic alcohol consumption has been found to predict more persistent gambling
problems (Abbott, Williams, & Volberg, 2004b; Hodgins & Holub, 2007).
Many other impacts have been identified including impaired quality of life for gamblers,
suicide and financial, legal, family and social problems (Abbott, Volberg, Bellringer, & Reith,
2004; Grinols, 2007). The Australian Productivity Commission concluded that, on average,
seven other people were affected by each problem gambler’s behaviour and reported
“enormous” family pressures with relationship breakdown and domestic violence
(Productivity Commission, 1999). In addition to the financial costs borne by problem
gamblers and their families, the Australian Productivity Commission estimated that each
problem gambler costs society between A$10,000 and A$30,000 (Productivity Commission,
2010). Problem gamblers commit high rates of gambling-related crime. National prison
surveys in New Zealand found 15% of male and 26% of female prisoners reported having
committed a crime to gamble or pay gambling debts (Abbott & McKenna, 2005; Abbott,
McKenna, & Giles, 2005).
From the foregoing it is evident that problem gambling is a significant health issue, both
directly and through its negative impacts in various other domains. These impacts (harms)
fall most heavily on Maori and Pacific people, and populations in high deprivation areas
(Ministry of Health, 2006, 2008). Consequently, measures to prevent and reduce problem
gambling are highly relevant to the goal of reducing health inequalities. Maintaining and
developing accessible, responsive and effective interventions is one of the objectives
15
specified in the Ministry of Health’s strategic plan to reduce gambling harm (Ministry of
Health, 2010).
New Zealand was one of the first jurisdictions to introduce services for problem gamblers
(from 1992 onwards). In 2009/10 the Ministry of Health allocated over $11 million to
intervention services (Ministry of Health, 2010). The gambling helpline had 2,122 new
clients in 2011, of which 1,242 were problem gamblers and 600 were people affected by
others’ gambling (Gambling Helpline, 2012). Including brief interventions, the total number
of gambling clients assisted face-to-face over a similar period (July 2010 to June 2011) was
7,594 (Ministry of Health, 2012). The gambling helpline provides information, screening,
brief intervention, referral and follow-up services. Ministry 2007-2010 purchasing plans
included further development of the gambling helpline and expansion of screening and early
intervention in other settings including primary care (Ministry of Health, 2007a). New
service specifications introduced in 2008 broadened the scope of intervention to include more
brief and public health modalities as well as full intervention. The Ministry moved to a 24-
hour helpline service in late 2008 which also began providing full intervention services,
ensuring access for people in areas without face-to-face services and for people who prefer a
telephone-based service (Ministry of Health, 2010).
Although substantial financial resources are allocated to problem gambling treatment in New
Zealand, it is not yet known how effective these services are or whether or not comparable
outcomes could be produced more effectively using different approaches. Internationally, a
variety of interventions have been developed (Abbott, Volberg, et al., 2004; Hodgins &
Holub, 2007). A meta-analysis of relevant studies concluded that, in general, psychological
interventions for problem gamblers are associated with favourable outcomes compared with
no treatment (Pallesen, Mitsem, Kvale, Johnsen, & Molde, 2005). However, a recent
comprehensive review of psychological interventions, conducted by Australian researchers at
the Problem Gambling Research and Treatment Centre, revealed large diversity in treatment
strategies, mode of delivery, materials used, location, dose and practitioner involvement in
treatment in the gambling field (Problem Gambling Research and Treatment Centre, 2011).
The authors commented that to date, despite widespread agreement that one is needed, there
is no standard taxonomy for describing the content of gambling interventions. In addition, the
psychological treatment outcome literature tends to be heavily compromised by
methodological limitations such as small sample size, high attrition and lack of intention to
treat analyses (Problem Gambling Research and Treatment Centre, 2011). Failure to include
comparative or control groups, randomly assign to treatment, or evaluate manualised
interventions has contributed to the relatively weak evidence base in comparison to the
substance addiction treatment fields. This recent review also shows little movement from the
findings of earlier reviews (Abbott, Volberg, et al., 2004; Hodgins & Holub, 2007; Petry,
2005; Toneatto & Ladouceur, 2003; Toneatto & Millar, 2004; Westphal & Abbott, 2006)
where it is evident that problem gambling treatment efficacy, effectiveness and outcome
studies are limited.
Some interventions, however, correspond to the “possibly efficacious” category (Chambless

& Ollendick, 2001) of at least one randomised controlled trial from one investigator group.
Psychological therapies in this category include a cognitive treatment (Ladoucer, et al., 2001),
cognitive behavioural treatments (Echeburúa, Báez, & Fernández-Montalvo, 1996; Gooding
& Tarrier, 2009; Petry, et al., 2006) and brief motivational and self-help interventions
(Hodgins, Currie, Currie, & Fick, 2009; Hodgins, Currie, & el-Guebaly, 2001; Hodgins,
Currie, el-Guebaly, & Peden, 2004; Petry, Weinstock, Ledgerwood, & Morasco, 2008; Petry,
Weinstock, Morasco, & Ledgerwood, 2009). While these and similar interventions are
apparently being used in every day clinical settings, few have been evaluated in effectiveness
or benchmarking studies (controlled studies of efficacious treatment) and just one has been
16
evaluated in a naturally occurring community treatment setting (Carlbring et al., 2010). There
are also few outcome studies (uncontrolled reports of treatment characteristics, number of
clients and client characteristics and outcomes).
Reflecting on their examination of the gambling intervention literature, Westphal and Abbott
(2006) concluded: “In sum gamblers respond to several possibly efficacious treatments…
with the majority benefitting, at least in the short term, when conducted by the original
investigators. There is no evidence that the beneficial effects occur when the treatments are
performed by other investigators or community based clinicians” (p.131). This is clearly a
very serious shortcoming. To date, one study has examined the effectiveness of either
motivational interviewing (MI) or cognitive behavioural group therapy (CBGT) with
participants recruited and treated through an outpatient dependency clinic. Carlbring et al.
(2010) found both face-to-face MI and CBGT produced significant within-group effects on
problem gambling screening scores up to 12-month follow up. However, because wait-listed
controls were treated prior to follow-up, between-group comparisons at the various time
points were not possible.
A recent Cochrane review of psychological therapies for pathological and problem gambling
examined 14 randomised controlled trials using CBT, motivational interviewing therapy,
integrative therapy or other psychological therapy. The authors of the review concluded that
CBT is efficacious in reducing problematic gambling behaviour and other related symptoms
in the short-term with sustainability of treatment effects remaining unknown. They also noted
that there was preliminary evidence for motivational interviewing therapy reducing gambling
behaviour though there was less evidence for reduction in other problem gambling symptoms.
The evidence was too scant to evaluate integrative and other psychological therapies
(Cowlishaw, Merkouris, Dowling, Anderson, Jackson, & Thomas, 2012).
Barriers to the development of evidence-based treatments identified by Westphal and Abbott

(2006) and Cowlishaw et al. (2012) remain including low sample size leading to low
statistical power, heterogeneous samples, lack of protocol driven treatments, missing or
skewed data, single site clinical trials, lack of replication of studies by independent
investigators and high rates of non-specific treatment response. Westphal and Abbott (2006)
examined other fields that have overcome many of these barriers by, among other things,
developing close collaborations between treatment providers and investigators and conducting
multi-site studies. In addition to the foregoing limitations, the small sample size of studies
and lack of outcome and effectiveness research means that little is known about individual
characteristics associated with success in different treatments. This information would enable
clients to be matched to particular interventions and treatment outcomes improved.
The reviews suggest that, for most problem gamblers, short-term and less intense (‘minimal’
or ‘brief’) interventions might be as effective as longer, more intensive therapies. Such
approaches, typically including brief motivational interviews and/or self-help workbooks,
have been shown to be effective with a variety of problems including alcohol and substance
misuse. Meta-analyses comparing self-help workbooks and therapist-directed interventions
have suggested that workbooks are as effective as the same programmes administered by
therapists (Gould & Clum, 1993). Conflicting results have been found in regard to
comparison of workbook (mainly CBT-style) interventions with wait-list control, prompting
recommendations for further Randomised Controlled Trials (RCT) of their effectiveness
(Problem Gambling Research and Treatment Centre, 2011). While it appears that workbooks
can be effective, it remains unclear which particular types of intervention are most beneficial
to which type of individual (Babor, 1994). In the alcohol field, however, there are indications
that brief interventions are particularly effective, and highly cost effective, for people with
less serious forms of disorder (Bertholet et al., 2005).
17
Brief motivational interviews and self-help tools are beginning to be examined specifically in
the gambling field. Seven RCTs have been conducted looking at the effectiveness of brief
motivational interviewing techniques. Three have found brief MI conducted face-to-face to
be as effective as cognitive treatments (Carlbring, et al., 2010; Larimer, et al., 2012; Petry, et
al., 2009) and two studies have involved telephone delivery of MI and the trial of a self-help
workbook (Hodgins, et al., 2009; Hodgins, et al., 2001; Hodgins, et al., 2004). One RCT has
examined the effectiveness of a particular brief self-help toolkit intervention in comparison to
wait-list control (LaBrie, et al., 2012). One study compared single-session face-to-face MI
with a control interview (Diskin & Hodgins, 2009).
Recent review of RCT research suggests that motivational interviewing is superior to wait-list
control in reducing gambling behaviour (Problem Gambling Research and Treatment Centre,
2011). In our view the most promising application of brief interventions to problem gambling
to date involves a short motivational telephone interview, followed by a self-help workbook
(Hodgins, et al., 2001; Hodgins, et al., 2004). As indicated earlier, this is one of three
psychological interventions that meet criteria for possible efficacy. Hodgins and colleagues
(2001) compared this intervention with receipt of the workbook alone, and a wait-list control
in a Randomised Controlled Trial. The interview in the efficacy study used motivational
enhancement therapy principles directed towards building commitment to change. The
workbook was based on a cognitive-behavioural model of problem gambling, relapse
prevention and the findings of research on problem gambling recovery processes.
Participants who received a motivational telephone interview and workbook in the mail, but
not those who received the workbook only, had significantly better outcomes at one-month
follow-up than participants in the wait-list control. Participants in the combined motivational
interview plus workbook group also gambled less frequently and spent less money gambling
at three and six months than those who only received the workbook. At three months, 42% of
the former group was abstinent compared with 19% of the latter. At six months their
respective outcomes were 33% and 22%. While there was no overall difference at 12 months,
motivational interview plus workbook participants with less severe gambling problems
maintained a therapeutic advantage.
Hodgins and colleagues (2004) followed up 67 participants 24 months after they had
completed the programme. While motivational interview plus workbook and workbook only
participants did not differ with respect to abstinence rates during the preceding six months,
those in the former group gambled less often, lost less money, had lower problem gambling
scores and were more often rated as having improved. Overall, more than three-quarters of
the total participants were rated as improved, over half scored below the cut-off for past year
pathological gambling and over a third reported six months of abstinence.
The foregoing indicates that brief interventions involving no face-to-face contact can have
clinically significant, enduring impact. Hodgins and colleagues (2001) recommended that
future studies examine the impact of treatment on other areas such as psychological distress
and family and social functioning. They also suggested that the addition of further
motivational interviewing ‘booster’ sessions might enhance outcomes.
Across a variety of mental disorders, motivational interviewing has been shown to improve
outcomes by enhancing treatment compliance (Arkowitz et al., 2007). A growing body of
literature supports the value of specifically targeting motivation to change as part of brief
interventions for gambling (Diskin & Hodgins, 2009; Hodgins, et al., 2004). A pilot study
with pathological gamblers (Wulfert, Blanchard, Freidenberg, & Martell, 2006) found that
treatment drop-out was significantly higher for treatment-as-usual than it was for a combined
18
motivational interview-cognitive behaviour therapy intervention. This study did not assess
whether or not greater compliance was associated with improved outcome. Diskin and
Hodgins (2009) found that a single session motivational interview conducted face-to-face had
larger impact on gambling outcomes than a non-motivational interview.
In contrast to the previous study, a recent trial of brief face-to-face interventions (Petry et al.,
2008) did not find that a motivational interviewing component enhanced outcomes for
problem gamblers recruited from substance abuse programmes and medical clinics.
Furthermore, this study found that relative to participants who received assessment only (no
treatment control), those who received 10 minutes of brief behavioural advice significantly
decreased their gambling behaviour at a six-week follow-up. Additionally, participants in that
group had clinically meaningful reductions in gambling at nine months. That study also
examined some participant characteristics in relation to outcome. Participants with less
severe gambling problems and fewer medical problems had better outcomes. Contrary to
expectation, comorbid substance misuse/dependence and psychological distress did not
influence outcome.
More recently, Hodgins (2009) compared a brief intervention (motivational interviewing plus
self-help workbook), to brief intervention and additional MI ‘booster sessions’ delivered on
six occasions over the follow-up period. This RCT included a six-week wait-list condition
and a workbook-only condition. As hypothesised by the authors, the brief intervention and
brief intervention with booster treatment participants reported less gambling at six weeks than
those assigned to the control groups. Intervention and intervention plus booster treatment
participants gambled significantly less often over the first six months of the follow-up than
workbook only participants. However, the workbook only participants were as likely to have
significantly reduced their losses over the year and to have not met criteria for pathological
gambling. Contrary to the hypothesis, participants in the brief booster treatment group
showed no greater improvement than brief treatment participants.
The foregoing studies raise interesting and important questions about the optimal length,
format and content of brief interventions. They also raise questions about which groups
respond best to which form or mix of interventions. It is unclear whether or not the addition
of follow-up booster sessions can enhance treatment outcomes, both generally and for
particular client groups.
This review highlights the need for definitive Randomised Controlled Trials that evaluate the
effectiveness of a well-developed and documented brief intervention for problem gambling
(Hodgins et al., 2001; 2004) and modifications to it with representative samples of problem
gamblers who seek information and help for gambling. Boosters may increase workbook use
and application and lead to improved outcomes, particularly at 12 months follow-up. It is
also possible that clients with more serious problems will do better in this condition.
From the previous Hodgins et al. (2001; 2004; 2009) RCTs it remains unclear whether it was
the motivational interview or the combination of motivational interview and cognitive-
behavioural workbook that was responsible for the treatment effect. Petry and colleagues’
(2008) and Hodgins and colleagues’ (2009) findings suggest that ‘more’ is not necessarily
‘better’. A recent study of patients in a hospital trauma centre with alcohol problems supports
assessing the value of motivational interview alone (Apodaca, Miller, Schermer, & Amrhein,
2007). While compromised by small sample size, that study found similar reductions in
drinking and related problems following brief assessment/interview and brief assessment/
interview plus a self-help workbook. It is possible that it is the motivational interview that is
the most important ingredient. If so, efficiencies would be made through removing
19
unnecessary intervention components (i.e. workbook, workbook plus booster follow-up calls),
at least for some clients.
Currently it is not known whether interventions provided in this country for problem
gamblers do better than natural or self-recovery, or non-specific ‘placebo’ effects associated
with seeking help and being assessed. Evaluating the effectiveness of a probably efficacious
intervention and extensions of it in a clinical setting will contribute to the understanding of
current gambling treatment provision in New Zealand while addressing a number of
deficiencies identified in the literature and enabling more robust conclusions to be reached
regarding treatment effectiveness in various populations.
20
4. RESEARCH METHODOLOGY
3.1 Ethics approval
The trial proposal was submitted to the Multi-Region Health and Disability Ethics Committee
which is a Health Research Council accredited human ethics committee. All participant
materials (i.e. survey questionnaires, information sheets and consent forms) and other relevant
documents were submitted to the Committee, which considers the ethical implications of
proposals for research projects with humans where participants are asked questions in relation
to their health.
The ethics approval for the trial was granted on 3 June 2009 (Appendix 1). The Ethics
Committee was kept apprised of any changes to the trial at the study progressed.
During the research the following measures were taken to protect the identity of the
participants:
 All participants were allocated a code by the research team to protect their identities
 No personal identifying information has been reported.
In addition:
 Participants were informed that participation in the research was voluntary and that
they could withdraw at any time, prior to data reporting.
3.2 Trial design
This was a single-site Randomised Controlled Trial (RCT) with gambler callers to the
gambling helpline randomly assigned to one of four parallel groups in a 1:1:1:1 ratio:
 Group 1: Helpline standard care (control group; ‘Treatment as Usual’ (TAU))
 Group 2: Single brief motivational interview (MI)
 Group 3: Single brief motivational interview plus self-help workbook (MI+W)
 Group 4: Single brief motivational interview plus self-help workbook plus four
follow-up motivational booster sessions (MI+W+B).
Participants were randomly assigned (computer generated) to the four groups until each group
contained a minimum of 110 participants (described in more detail in section 3.7).
3.3 Participants
3.3.1 Eligibility criteria
Participants were recruited from callers to the gambling helpline who sought information or
assistance for their own gambling problem.
The inclusion criteria were:

 Minimum age of 18 years
 Perception of having a gambling problem
 Willingness to:
21
o Read a short workbook (to ensure reading ability)
o Have calls recorded
o Provide follow-up data on gambling
o Provide the name of collateral/s.
Present or past involvement in treatment or mutual help groups for gambling or other mental
health problems was documented and did not preclude participation.
Callers were excluded from the trial if:

 They were considered by the counsellor to be actively psychotic
 They required immediate crisis or police intervention because they posed a serious
risk to themselves or others.
3.3.2 Setting and location
The study took place at the gambling helpline, Auckland, New Zealand in that the
interventions were delivered by trained gambling helpline counsellors. As the interventions
were delivered by telephone, participants were based throughout New Zealand. Recruitment
and delivery of interventions occurred from August 2009 to February 2011.
Follow-up assessment calls were made by telephone by trained university research assistants
from the North Shore Campus of Auckland University of Technology (AUT), Auckland, New
Zealand. Research assistants were blind to participants’ treatment group.
3.4 Interventions
All callers to the helpline initially received brief non-directive counselling to identify
presenting concern/s and establish rapport. If the caller met eligibility criteria they were
asked if they would like to participate in the study. Immediately after consenting to take part
in the study, participants underwent an initial baseline assessment (detailed in section 3.5) and
then received their randomly allocated intervention which was delivered by telephone by a
trained gambling helpline counsellor.
The counsellors were trained to deliver all four interventions, removing potential
contamination of the RCT design by therapist effects. The training included practice in
introducing the project, recruitment of participants, the initial assessment questions, treatment
option selection and delivery, and booster session delivery. The training also incorporated
how to use the protocol developed to facilitate consistency and integrity in the delivery of the
standard care interview (TAU) and specific motivational interviewing training. All training
included pilot interviews with volunteers that were digitally recorded and assessed for
compliance and consistency by Dr Sean Sullivan and Professor David Hodgins, both of whom
are very experienced in use of motivational interviewing techniques with problem gamblers.
The training included additional ad hoc sessions, particularly at the beginning of the trial, to
address any issues; this was an opportunity for counsellors to air any difficulties or to state
their confidence in the trial to others.
Dr Sullivan also trained the AUT researchers who conducted the follow-up assessments. The
training included identification of risk level of participants and how to safely intervene when
participants expressed symptoms of risk or suicidal ideation.
22
3.4.1 Group 1: Helpline standard care (TAU)
Group 1 participants received a protocolled version of the helpline’s standard care. This
included brief screening, listening to clients’ concerns (problem identification) and, in the
instance of first time callers or regular callers who were experiencing persistent difficulties,
referral to face-to-face problem gambling counselling services or other services and websites
and/or suggestions for self-care (e.g. controlling access to money, coping with gambling
urges, alternative activities to gambling, and goals around saving money). No motivational
interviewing aspects were included to differentiate this intervention from the three trial
interventions (Groups 2, 3 and 4) 5. The protocol was developed with staff to ensure it was
functional and similar to their normal practice. Additionally, participants were offered an
information pack (relevant information pamphlets, for example detailing venue self-exclusion
processes, or budgeting advice). In this respect it is similar to one of the control conditions in
the original Hodgins et al. (2001; 2004) efficacy study.
3.4.2 Group 2: Single brief motivational interview (MI)
Group 2 participants received a brief motivational interview, as used in the Hodgins et al.
(2001; 2004) study. The interview was structured to encourage the client to build a
commitment to change by emphasising the reasons why change is desirable. This approach
was shaped by five therapeutic guidelines, namely: (1) expression of empathy (acceptance of
individual and recognition that ambivalence about change is normal), (2) development of a
discrepancy between the individuals’ present behaviour and their goals and self-image,
(3) avoidance of argumentation and confrontation, (4) rolling with resistance (looking for
opportunities to reinforce accurate perceptions versus correcting misperceptions), and
(5) support of self-efficacy. Interviews ended with a summary of participants’ stated reasons
for changing and specific therapeutic goals.
3.4.3 Group 3: MI plus self-help workbook (MI+W)
Group 3 participants received a brief motivational interview, as for Group 2, combined with
the use of a workbook. Within 24 hours of the initial interview, participants were mailed a
self-help workbook6 adapted from the Hodgins et al. (2001; 2004) study, along with a written
summary of the clients’ stated reasons for changing and their specific goals. Changes to the
original workbook were minimal, reflecting differences in phrasing and common word usage
between Canada and New Zealand.
3.4.4 Group 4: MI+W plus four follow-up motivational booster

sessions (MI+W+B)
Group 4 participants received the same intervention as Group 3 and also received four follow-
up motivational booster sessions of 10 to 15 minutes duration at one week after the initial
interview and at one, three and six months. These booster sessions focused on motivation of,
5
To further differentiate this intervention from the three trial interventions, counsellors specifically did
not do any of the following with participants: send the Gambling Helpline workbook (which was
similar to the trial workbook); provide gambling screen feedback; ask about behaviour changes; ask
about commitment, motivation, confidence or likelihood of success; or offer additional telephone calls.
6
Becoming a Winner: Defeating Problem Gambling.
23
and reinforcement for, behaviour change through the use of the workbook. At each session,
progress was reviewed, motivation and commitment renewed, and new short-term goals
developed. If a participant could not be contacted for a particular booster session, that session
was missed and the next attempt at contact was at the next scheduled booster session.
3.4.5 Treatment integrity and fidelity
To assess how well the counsellors delivered each intervention and to ensure that there was no
cross-contamination between interventions (i.e. to assess treatment integrity and fidelity),
particularly between the Group 1 standard care (TAU) (which had no motivational
interviewing elements) and the three treatment groups (which were based on motivational
interviewing techniques), approximately 20% of telephone calls (including intervention
delivery) with participants were randomly digitally recorded. The recordings were
subsequently (usually within one month) listened to by Dr Sean Sullivan who is an
experienced psychologist with substantial knowledge of motivational interviewing
techniques. The recordings were coded based on the Motivational Interviewing Treatment
Integrity (MITI) scale (Moyers, Martin, Manuel, Hendrickson, & Miller, 2004) and tailored
for this trial to assess for shared processes between the TAU and other interventions, for
motivational interviewing aspects, and for TAU specific aspects. Following assessment of the
recordings, personal feedback and, where required, additional training was provided to
counsellors by Dr Sullivan. Approximately 33% of the recordings was also assessed by
Professor David Hodgins in order to assess reliability of the first assessment of the recordings.
Assessment of the recordings was based on three categories: Motivational interviewing

elements (eight or nine elements), TAU only elements (eight elements), and elements shared
across the treatment and control (TAU) interventions (five elements).
The motivational interviewing elements included:

 Reflective listening
 Potential benefits of not gambling
 Affirmation
 Offer feedback around Problem Gambling Severity Index score
 Summarise concerns, motivations
 Ask for commitment
 Rating of commitment and/or success
 Connect with workbook (MI+W and MI+W+B groups only)
 Ask about other change attempts (e.g., smoking).
TAU only elements included:

 Soft/hard referral to face-to-face service
 Information regarding other helplines/websites
 Offer of postal ‘information pack’
 Discussion of options for controlling access to money
 Discussion of options for coping with urge to gamble
 Discussion of options for keeping busy with alternative activities
 Advice to set saving goal
 Advice in other area (unspecified).
24
Shared elements included:
 Reasons for contacting gambling helpline/concerns about gambling
 Financial concerns
 Relationship problems
 Emotional difficulties
 Legal problems.
3.5 Outcome measures
Primary outcome measures were self-reports of:

 Days gambled
 Money lost gambling
 Treatment goal success.
Secondary outcome measures included control over gambling, gambling impacts, problem
gambling severity, psychiatric comorbidity and substance use, tobacco and psychotropic
medication use, general psychological distress and quality of life.
Collateral assessment (at three months and one year) from people nominated by participants
included participant’s gambling over the past month, observed changes and confidence in
accuracy of their (collateral) reports.
3.5.1 Initial assessment
The initial assessment was conducted with the participant by a helpline counsellor prior to the
participant receiving a randomly allocated intervention. Due to the length of the initial
assessment, some of the baseline initial assessment was conducted by an AUT researcher
within seven days of a participant receiving the intervention. This is detailed at the end of
this section.
Gambling/problem gambling history, impacts and past help-seeking

A brief gambling history was obtained including length of gambling problem; type/s of
gambling causing problems; number, nature and outcomes of past attempts to quit or reduce
gambling; and past treatment and mutual help involvement. The impacts of gambling on
financial status, employment, family and other relationships, criminal offending and general
health (adapted from Abbott & Volberg, 1992; Abbott, 2001b) were also assessed.
Problem gambling
The nine-item Problem Gambling Severity Index (PGSI) (Ferris & Wynne, 2001) was used to
measure severity of gambling problems. It was administered in both a past 12-month and a
past three-month time frame (reported as PGSI-12 and PGSI-3, respectively). The two-item
Lie-Bet screen was also administered to assess problem or non-problem status (Johnson et al.,
1997). Both the PGSI and the Lie-Bet screen have been validated against clinician-derived
DSM-IV pathological gambling diagnoses and other problem gambling measures including
the widely used South Oaks Gambling Screen/South Oaks Gambling Screen-Revised
(SOGS/SOGS-R) (Abbott & Volberg, 2006).
Comorbidity and substance use
25
A brief version (10-item scale) of the Drug Abuse Screening Test (DAST; Skinner, 1982) was
administered to assess drug abuse. The DAST has very good internal reliability in samples of
substance abusers and psychiatric patients and correlates strongly with a number of drug use
measures (Cocco & Carey, 1998).
To identify hazardous alcohol consumption or active alcohol use disorders (including alcohol
abuse or dependence) a brief version (AUDIT-C, three-item scale) of the Alcohol Use
Disorders Identification Test (AUDIT; Saunders et al., 1993) was administered. In a review
of research using the AUDIT and shortened versions, the AUDIT-C was reported as showing
promise in being time-efficient and accurate when compared with full AUDIT results (Reinert
& Allen, 2002).
Participants were also asked about lifetime and current tobacco use and any previous success
at quitting a problematic behaviour (i.e. smoking, alcohol, other drugs and other behaviour).
The Kessler-10 (K-10) questionnaire was included to provide a continuous measure of general
psychological distress that is responsive to change over time. The K-10 has been well
validated internationally. Its brevity and simple response format are attractive features. It
also produces a summary measure indicating probability of currently experiencing an anxiety
or depressive disorder (Kessler & Mroczek, 1994).
Quality of life
Quality of life was assessed by the WHOQoL-8, an eight item version of a widely used
measure. This short form has been used in a number of countries, is robust psychometrically,
and overall performance is strongly correlated with scores from the original WHOQoL
instrument (Schmidt, Muhlan & Power, 2005).
Treatment goal
Participants were asked whether their goal was to stop all forms of gambling, stop only
problematic forms of gambling, or to reduce their gambling.
Self-efficacy
A simple rating was employed to assess belief in likelihood of a participant achieving their
treatment goal (0 “not at all confident” to 10 “extremely confident”) in the next six months.
Motivation and perceived control over gambling

Treatment goal motivation was measured on the same 0 to 10-point scale (“not at all” to
“extremely”). Participant-rated sense of control over gambling was assessed using a similar
0 to 10-point scale (“no control” to “total control”).
Socio-demographics
Age, gender, ethnicity, marital status, highest educational level, employment status and area
of residence data were collected.
26
Within seven days of the initial assessment and delivery of the intervention, an AUT
researcher asked some follow-up baseline questions of each participant. These included:
Gambling/problem gambling history

A detailed timeline of types of gambling, frequency and money spent gambling over the past
two months was administered (based on Sobell & Sobell, 1992). Participants were provided
with memory cues such as recent holidays and news events to facilitate retrieval of this
information.
Comorbidity and substance use

The mood module of the Primary Care Evaluation of Mental Disorders (PRIME-MD, Spitzer
et al., 1994) was administered to provide diagnoses of major depressive disorder, dysthymia,
minor depressive disorder, and alcohol abuse/dependence. This is a structured interview
designed for primary care clinicians and researchers to diagnose these and other current
DSM-IV mental health disorders. It has been validated against the Structured Clinical
Interview for the DSM-IV (Spitzer et al., 1992) and has been administered by telephone and
shown to yield valid diagnoses (Spitzer et al., 1994; Kobak et al., 1997). The use of
psychotropic medication and history of manic episodes was assessed using questions
modified from the Gambling Impact Study (Gerstein et al., 1999).
Socio-demographics
The eight-item New Zealand Index of Socio-economic Deprivation for Individuals (Salmond,
2005) was administered.
3.5.2 Follow-up assessments
Participants were contacted by an AUT researcher to complete a follow-up assessment at

three, six and 12 months post-intervention. At each follow-up assessment, a timeline follow-
back interview was conducted to capture the number of days gambling during the follow-up
period and the amount of money lost on each occasion. Participants were asked whether they
had met their goal (‘not at all’, ‘partially’, ‘mostly’, ‘completely’) and what their present goal
and personal sense of control over their gambling were (0 ‘no control’ to 10 ‘total control’).
Participants were also asked whether they had received the ‘Becoming a Winner’ workbook
and if so, whether they had read it (‘not at all’, ‘some sections’, ‘completely’), whether they
had followed the procedures (‘not at all’, ‘to some extent’, ‘completely’), and whether they
had used the strategies (‘not at all’, ‘occasionally’, ‘regularly’). They were asked what they
had found to be most and least helpful from: (a) the initial interview, and (b) the workbook, in
reaching their goal during the follow-up period, and why.
At each assessment, participants were asked what other treatment or help, if any, they
received for their problem gambling during the follow-up period. These forms of treatment/
help were listed and, for each, they were asked how often the treatment or help was obtained
(number of occasions) and how helpful it was in reaching their goal (‘not at all’, ‘partially’,
‘mostly’, ‘completely’). At the three- six- and 12-month assessments, in addition to the
previously mentioned assessments, the current tobacco use, gambling impacts, AUDIT-C, K-
10 and WHOQol-8 were re-administered. At the 12-month assessment point, participants
were also re-administered the DAST and PRIME-MD mood module as well as being asked to
reflect on their overall experience during the past 12 months in seeking and receiving help for
gambling and making changes in their lives. They were also asked about the cultural
appropriateness of the help they received (linked to source/type of help) and, if inappropriate,
27
what could be changed to make it more appropriate. They were invited to comment on other
changes that could make help more accessible, appropriate and effective.
3.5.3 Collateral assessments
After the three-month assessment, and again at 12 months, at least one collateral person per
participant (where details for collateral participants had been provided by the trial
participants) were contacted by telephone and asked about the participant’s involvement with
gambling over the last month. They were also asked how confident they were about the
accuracy of their reports (‘not at all’, ‘somewhat’, ‘fairly’, ‘extremely’).
3.6 Sample size
A power of 80% in finding a primary hypothesised effect is considered acceptable in most

well-designed clinical studies, and this level of power was taken as the minimum for the
current trial. Based on Hodgins and colleagues similar Canadian study (Hodgins et al., 2001;
2004) it was expected that 8.6%, 11.1% and 15.0% baseline participants would attrite from
the study at the three-month, six-month and 12-month assessment points, respectively.
Exploiting the longitudinal design (with baseline and three follow-up measurements), and
information for the MI+W treatment group reported by Hodgins and colleagues for the three
primary outcome variables, a significance level of 5%, power of 80%, and the sample size
calculations described by Twisk (2003), a total sample size of 110 per treatment group with
attrition occurring as described above has a time-averaged minimal detectable difference
between treatment groups as tabulated below. This implies that the study has 80% power to
significantly detect a one-day difference in mean days gambled between treatment groups
(after accounting for time changes), a $20/day difference in dollars gambled between
treatment groups and a quit or improved gambling rate difference of 0.13. These are
meaningful minimal detectable differences allowing the primary hypotheses to be
investigated with acceptable power.
Mean baseline Average Intra-participant Minimal

score follow-up S.D. correlation detectable
σ (σ2) ρ difference
Primary variable of interest
Days gambled 10.2 4.9 (24.0) 0.25 1 day
Dollars/gambling day 158 96 (9,216) 0.25 $20/day
Proportion in Minimal
control group detectable
difference
Gambling -Quit or improved* 0.74 0.13
* By definition there were no ‘quit or improved’ outcomes at baseline. Changes were only assessed
in the follow-up component of the study.
3.7 Randomisation
To allocate participants to intervention groups, a computer-generated block randomisation

procedure was used. The block size was 20, allocating participants to one of the four
treatment groups in a 1:1:1:1 ratio. Thus, five of every 20 callers were allocated to each of
28
the four groups. No stratification procedure was used due to the large number of criteria on
which to stratify. Once a minimum of 110 participants had been recruited to each of the four
groups, the randomisation procedure was terminated.
The computer programme for treatment allocation was written by an independent computer
specialist. Treatment allocation was concealed from counsellors until they activated the
programme, after the initial assessment and immediately prior to intervention delivery.
3.8 Blinding
Counsellors were aware of allocated interventions as they had to deliver the interventions to
the participants. AUT research assistants conducting the follow-up assessments were kept
blinded to the allocations; the blinding was only broken at the end of the trial for data
analyses.
3.9 Trial hypotheses and statistical methods
3.9.1 Study hypotheses
Study hypotheses are briefly described in this section. They are fully described together with
statistical notation in Appendix 2.
Efficacy hypotheses
Primary equivalence hypothesis
A. The Motivational Interview (MI) group will show similar improvement to Treatment
as Usual (TAU).
Primary superiority hypotheses

B. (Superiority is associated to lower values in the generic hypothesis statements below)
a. The Motivational Interview plus Workbook group (MI+W) will show greater
improvement than the TAU group.
b. The Motivational Interview plus Workbook group (MI+W) will show greater
improvement than the MI group.
c. The Motivational Interview plus Workbook plus Booster group (MI+W+B) will
show greater improvement than the TAU group.
d. The Motivational Interview plus Workbook plus Booster group (MI+W+B) will
show greater improvement than the MI group.
C. (Superiority is associated to lower values in the hypothesis statements below)

a. The MI+W+B group will show greater improvement than the TAU group at the
12-month follow-up.
b. The MI+W+B group will show greater improvement than the MI group at the 12-
month follow-up.
c. The MI+W+B group will show greater improvement than the MI+W group at the
12-month follow-up.
29
Secondary efficacy hypotheses
C*. (Superiority is associated to lower values in the hypothesis statements below)
d. The MI+W+B group will show greater improvement than the TAU group between
three and 12 months.
e. The MI+W+B group will show greater improvement than the MI group between
f. The MI+W+B group will show greater improvement than the MI+W group
between three and 12 months.
D. (Superiority is associated to lower values in the generic hypothesis statements below)

a. The TAU group will evince significant reduction in gambling.
b. The MI group will evince significant reduction in gambling.
c. The MI+W group will evince significant reduction in gambling.
d. The MI+W+B group will evince significant reduction in gambling.
E.
(Superiority is associated to lower values in the generic hypothesis statement below)
High levels of engagement within conditions will be associated with better gambling
outcomes (gambling participation, attainment of goal and sense of control over gambling).
Engagement secondary hypotheses

F.
(Superiority is associated to higher values in the generic hypothesis statements below)
a. The highest level of engagement will be in the ‘booster’ condition (MI+W+B),
followed by the non-‘booster’ experimental condition (MI+W).
b. The level of engagement will be higher in the non-‘booster’ experimental
condition (MI+W) then in the standard treatment group (TAU).
G.
Use of, and degree of, engagement in other treatment services will be significantly
lower in the two conditions involving motivational interviewing and workbooks
(MI+W; MI+W+B) than in the standard (TAU) and motivational interview (MI)
groups. This difference is expected to be greatest during the first three months.
3.9.2 Study endpoints
Efficacy endpoints
The primary efficacy endpoints are the primary efficacy outcomes (self-reported average
monthly numbers of days gambled, average monthly amount of money lost and Gambling-
quit or improved indicator), time-averaged over the three assessment points. In the case of
the number of days gambled in one month and amount of money spent gambling in one
month, these endpoints correspond to an annual average of monthly values over the
12 months post-randomisation. The Gambling-quit or improved indicator is defined as 1 if
either the average number of days in the previous three months of follow-back is zero (quit)
or the average amount of money spent in the previous three months of follow-back is less
than half the declared 12-month average at baseline. It has value zero otherwise. The time-
averaged Gambling-quit or improved is thus a value between 0 and 1 inclusively.
30
Secondary efficacy endpoints consist of the primary outcomes at the individual assessment
time points, as well as the endpoints listed in the Summary table of analyses (Appendix 3).
Engagement endpoints
The trial introduces engagement endpoints. The engagement endpoints are the time-averaged
versions of the workbook usage variables in the relevant treatment groups, and the endpoints
listed in the Summary table of analyses (Appendix 3).
Safety and tolerability endpoints

None.
3.9.3 Analysis sets
The Intention to Treat (ITT) and Per Protocol (PP) analysis sets are defined for analysis of
efficacy data. The Workbook Engagement (WE) set is defined for the analysis of some
engagement outcomes.
If a participant was randomised but not treated, or randomised but did not complete the
treatment schedule, then she or he was included in both the efficacy and workbook analysis
sets to the extent that she or he did provide data (as clarified below).
Intention to Treat
The primary analysis set of interest will be the Intention to Treat (ITT) set, which consists of
all randomised participants who have at least one baseline measurement. All randomised
participants were analysed in the group to which they were randomised, even if they did not
receive the allocated treatment, did not commence treatment, or were lost to follow-up.
Per Protocol
Participants were included in the Per Protocol (PP) analysis set if they fulfilled the criteria of
the ITT set, had completed at least one primary endpoint measurement and did not present
any major protocol violation. Participants in the PP set were assigned to the treatment group
corresponding to the intervention they actually received at the baseline telephone call.
The following describes the major protocol deviations that excluded participants from the PP
population:
 Eligibility violation
 Absence of any efficacy data.
All other protocol deviations were considered as minor and did not lead to excluding
participants from the PP population for analysis.
Workbook Engagement
Participants were included in the Workbook Engagement (WE) analysis set if they fulfilled
the criteria of the ITT set and received the workbook.
31
3.9.4 Statistical methods
The types of analyses of treatment effects were categorised as outcomes into primary and
secondary efficacy and engagement analyses, confirmatory analyses and exploratory analyses.
Primary and secondary analyses relate to the reported treatment effects on all primary and
secondary efficacy endpoints respectively. Confirmatory analyses were focused on ancillary
statistics and informed methodological choices in the primary and secondary analyses.
Exploratory analyses included all other analyses, for the essential planned and unplanned
variations on the primary and secondary analyses.
Descriptive statistics
All continuous measurements subject to descriptive statistics have been reported by treatment
group as number and proportion of non-missing observations, mean, standard deviation,
minimum, first quartile, median, third quartile and maximum.
All categorical measurements have been reported as number and proportion of non-missing
observations, and proportion in each valid category.
Covariates
We distinguish between (independent) predictors related to outcome and unrelated to the
allocation; potential confounders, related to outcome and imbalanced by chance across the
treatment groups; and potential effect modifiers, that may moderate the treatment group
effect.
Independent predictors
Generalities
Independent predictors are covariates that were sometimes included in the model for
interpretive purposes. The baseline value of the outcome being modelled was an independent
predictor. The time point, seen as an ordered factor or as a (continuous) time since baseline,
was another independent predictor. Independent predictors were sometimes included in the
model as a result of a specific hypothesis being tested. They are identified as such in
Appendix 3. The baseline outcome value was included in all models when available.
Baseline data collected post-randomisation

Due to feasibility considerations, some baseline data were collected seven days post-
randomisation, post-intervention. These concern the primary efficacy outcomes Days
Gambled and Money Lost, and the secondary efficacy outcomes PRIME-MD (Major
Depression, Dysthymia, Minor Depression, Bipolar Disorder) and New Zealand Deprivation
Index (NZDI). These values were used as baseline adjustments.
Potential confounders
Prior selection of potential confounders
Potential confounders were selected from amongst baseline outcomes and demographic and
personal history covariates as detailed below and identified in context in Appendix 3.
Comparisons of important and known potential confounders were conducted at baseline to
ensure that distributions were approximately equal between groups. If baseline separation
between groups in some covariates emerged, outcome differences were adjusted for these
covariates if they proved significant (but all potential confounders were assessed if the model
retained did not involve an identity or exponential link). Baseline separation between groups
in a covariate occurred for this purpose if any two mean estimates amongst the groups
differed by one or more pooled standard deviations. In the case of the age group, this
criterion was applied to the continuous age covariate instead.
32
Significance testing of potential confounders
All potential confounders were assessed for significance first as a single block, the
significance of which was assessed using an appropriate F or x 2 test. If the block was retained
as a result of this test, the individual covariates were tested using backward selection based on
the appropriate t, z, F or x2 tests. The significance threshold for retention of potential
confounders for adjustment purposes was 0.1 for all tests. The estimates associated with the
retained confounders were not reported but the retained confounders were reported by name:
 Gender
 Marital status, dichotimised
 Age group at randomisation
 Primary ethnicity
 Gross family income in last 12 months, dichotomised
 PGSI-12 (at baseline only)
 Electronic gaming machines as primary gambling type
 Current goal, dichotomised (quit vs. control gambling)
 Dichotomised level of belief in success within 12 months
Effect modifiers
Effect modifiers are covariates that may affect the treatment group effect (interactions
involving treatment). Some analyses involve the time point as a treatment effect modifier.
The subgroup analyses implicitly define variables that may modify the treatment effect.
There were no other considerations of effect modification.
Variance-covariance structure
In repeated measures analysis, the clustering between repeated measures were accounted for
by participant-specific and counsellor-specific random effects. No further correlation
structure was imposed.
Follow-back data resulting from the average of several measurements (as with the primary
outcomes) were associated with a weight corresponding to the number of valid observations
entering into the average. This overall weight was also applied in the time point-specific
analyses.
Inferential framework
Significance threshold
All tests of significance of hypotheses concerning treatment effect parameters were carried
out using a level of significance of 5% and two-sided alternatives. The significance threshold
of potential confounders was set at 10%, to promote unbiased and conservative inference. All
estimates were produced as point estimates and as 95% confidence intervals. Unless
otherwise noted, model selection, when required, was performed using backward selection
from the largest model dictated by the situation.
Family-wise error rate adjustment

Each composite hypothesis (Hypotheses B, C, D, F, G) was assessed by controlling False
Discovery Rate at the stated significance threshold, in accordance with the procedure outlined
in Benjamini and Hochberg (1995). Sub-hypotheses thus retained were deemed statistically
significant. The composite hypothesis was deemed statistically significant if all sub-
hypotheses were retained. No FWER adjustment was carried out across outcomes.
33
Analytical framework for continuous endpoint analysis
Normality assumption
The analysis described below assumes that normality of residuals is a reasonable assumption.
Contingencies for non-normality are described below.
Regression model
Repeated measures analyses fitted available endpoints as repeated measures over the three
assessment time points (excluding baseline) to an appropriate normal mixed effects model
using residual maximum likelihood (REML).
Baseline outcome value was included as an independent predictor in all models, when
available.
Specific covariates and interactions were included in specific analyses, such as subgroup
analyses.
As per potential confounders above, models may potentially have been adjusted for baseline
covariates, subject to achieving significance as per the significance threshold above.
Inclusion of treatment group; univariate and multivariate settings

When time point-specific (TPS) estimation was required, the assessment time (0, 3, 6 and
12 months) was entered as an ordinal factor in interaction with the treatment group. The
analysis-appropriate estimand (e.g. effect at three months) was retained for estimation and
reporting.
When time-averaged outcome (TAO)-based estimation was required, the 3, 6 and 12 month
levels were collapsed into a single level, yielding a baseline/post-baseline dichotomous factor.
In the cases when there were no repeated measures, this analysis reduced to a least-squares
regression.
Variance structure
A zero-mean, normally distributed random effect was assigned to participants based on their
counsellor’s identity, to account for heterogeneity between counsellors.
A nested, zero-mean, participant-specific normally distributed random effect was assigned to

observations from a single participant to account for within-participant correlation in a simple
compound-symmetry structure. This random effect was only used when there was more than
one measurement per participant (e.g. not in the case of PGSI-12).
When the outcome was an average of other observations, the number of valid observations
entered into the average was included as a weight in the regression.
Results
In most cases the estimated treatment contrasts represented differences in location,
themselves interpretable as differences in changes from baseline under the adjustment for
baseline value. In the case of treatment interactions with continuous covariates, the contrasts
were differences in slopes. Estimated treatment contrasts were produced as point estimates
and as 95% confidence intervals.
Trend models
34
The analyses described herein do not account for a time trend.
Absence of repeated measures

When outcome data were collected only at 12 months (e.g. PGSI-12), the above framework
reduced to a baseline-adjusted ANCOVA, with variance estimated in the full repeated
measures setting across the counsellors. For such analyses the individual random effects were
removed from the model, although the counsellor-specific random effects were retained.
Alternative analytical frameworks for continuous endpoints under non-normality I:

Alternative family and transformation
This section also applies to binomial outcomes with logit link and multinomial outcomes with
cumulative logit link.
If non-normality of residuals was demonstrated or a non-normal family and/or non-identity

link were called for, analyses equivalent to the analytical framework for continuous endpoint
analyses (above) using an alternative generalised linear model as a first choice, a data
transformation as a second choice, or both as a third choice, was investigated based on the
estimated variance function from the residuals.
If a generalised linear model was selected, potential confounders were automatically assessed
for significance in the model, without verification of baseline separation.
Any estimate produced under a non-identity link was converted to natural units with first-
degree bias correction, and their confidence intervals produced by applying the inverse link to
the confidence interval bounds of the linear predictor, rather than use of the delta method.
Alternative analytical frameworks for continuous endpoints under non-normality II:

Dichotomisation
If the provisions of non-normality I failed to apply satisfactorily, the outcomes were
dichotomised based on thresholds commonly held in the literature, or failing the existence of
such a threshold on the basis of the approximate median of the outcome in the TAU group,
without consideration of the time point. The analyses then proceeded according to non-
normality I using a binomial family and logit link, i.e. using mixed effects logistic regression.
In most cases the estimated treatment contrasts represented odds ratios with respect to a
reference category, usually TAU, adjusted for baseline odds. In the case of treatment
interactions with continuous covariates, the estimand was odds ratio per unit difference of the
continuous covariate. Estimated odds ratios were produced as point estimates and as 95%
confidence intervals.
Analytical framework for dichotomous (polytomous) endpoint analysis

The analyses proceeded according to non-normality I using a binomial (respectively,
multinomial) family and logit (respectively, cumulative logit) link, i.e. using mixed effects
logistic regression. Participant-level random effects were only used in the presence of
repeated measures.
In most cases the estimated treatment contrasts represented odds ratios with respect to a
reference category, usually TAU, adjusted for baseline odds. In the case of treatment
interactions with continuous covariates, the estimand was odds ratio per unit difference of the
continuous covariate. Estimated odds ratios were produced as point estimates and as
95% confidence intervals.
35
Software
Analyses were undertaken with R version 13.0 or higher, SAS version 9.2 or higher and SPSS
(PASW) version 16.0 or higher.
Detail of the efficacy and engagement analyses

Primary vs. secondary analyses
The primary analyses consist of analyses of primary outcomes and primary hypotheses in the
ITT analysis set.
The secondary analyses consist of the following:

 PGSI-12 and Hypothesis A in the ITT analysis set
 Primary outcomes and PGSI-12 within primary hypotheses in the PP analysis set
 Primary outcomes and PGSI-12 within secondary hypotheses in the ITT analysis set
 Primary and selected secondary outcomes (PGSI-12, attainment of goal and control)
and Hypothesis E
 Secondary outcomes and primary superiority hypotheses in the ITT analysis set
 Engagement outcomes and engagement hypotheses in the ITT analysis set.
Description of the main analyses

I: Time-averaged continuous endpoints
Time-averaged analysis, as per analytical framework for continuous endpoint analysis
(above), of a continuous primary outcome.
II.1: Time point-specific continuous endpoint, in the presence of repeated measures

Time point-specific analysis, as per analytical framework for continuous endpoint analysis
(above), of a continuous primary outcome in the ITT analysis set.
II.2: Time point-specific continuous endpoint, in the absence of repeated measures

Time point-specific analysis, as per analytical framework for continuous endpoint analysis
(above), of a continuous primary outcome in the ITT analysis set.
III: Time-averaged dichotomous endpoints

Time-averaged analysis, as per analytical framework for dichotomous (polytomous) endpoint
analysis (above), of a dichotomous primary outcome in the ITT analysis set.
IV.1: Time point-specific dichotomous endpoint, in the presence of repeated measures

Time point-specific analysis, as per analytical framework for dichotomous (polytomous)
endpoint analysis (above), of a dichotomous primary outcome in the ITT analysis set.
IV.2: Time point-specific dichotomous endpoint, in the presence of repeated measures

endpoint analysis (above), of a dichotomous primary outcome in the ITT analysis set.
V: Time-averaged multinomial endpoint

analysis (above), of a multinomial family random variable with cumulative logit link and
weight variable corresponding to the number of valid responses over which the response was
computed.
36
VI: Time point-specific multinomial endpoint
endpoint analysis (above), of a multinomial family random variable with cumulative logit link
and weight variable corresponding to the number of valid responses over which the response
was computed.
VII: Time-averaged binomial endpoint

analysis (above), of a binomial family random variable with logit link and weight variable
corresponding to the number of valid responses over which the response was computed.
VIII: Time point-specific binomial endpoint

endpoint analysis (above), of a binomial family random variable with logit link and weight
variable corresponding to the number of valid responses over which the response was
computed.
Subgroup analyses
The primary analyses were repeated by considering possible interaction of the treatment
group with the following subgroups defined at baseline:
1. Gender
2. Ethnicity: “Yes” responses only to each of:
a. European
b. Maori
c. Pacific
d. Asian or Other
Ethnicity subgroups were defined according to response, so that participants could
contribute data to more than one subgroup.
3. Gambling problem severity based on PGSI-12 score (past 12-month time frame):
Dichotomised with 17 (median score at baseline) as cut-off score. Note that the usual
cut-off of 8 to identify problem gambling is not material here, as problem gambling at
baseline was an inclusion criterion.
4. EGM anywhere or any other as primary gambling type
5. Mental health comorbidities based on Kessler-10 score: Dichotomised with
30 (median score at baseline) as cut-off score.
6. Alcohol abuse/dependence based on AUDIT-C score, cut point of 4 for males and
3 for females: Dichotomised
7. Current goal (quit or control gambling): Dichotomised
8. Belief level in treatment success: Dichotomised with 10 (median score at baseline) as
cut-off score
Analyses involving Hypothesis E

The assistance- and engagement-related variables in the analyses involving Hypothesis E
were collected at the post-randomisation time points. As such their status as covariates is
questionable.
Missing data
Assessment of the significance of potential confounders and effect modifiers was based on
complete-case analysis. If any confounder or effect modifier was retained based on the
complete-case analysis, the final model relied on multiple imputation to produce adjusted
37
treatment effect estimates. Confounders or effect modifiers with significance beyond the
stated threshold after the multiple imputation stage were removed from the model.
Missing outcome values were accommodated without further adjustment in mixed effects
models, under an assumption of missingness completely at random or missingness at random.
Modelling of missingness and outcomes were performed in confirmatory analyses.
Confirmatory Analyses
Normality assessment
Normality of continuous outcomes was assessed using q-q plots, Kolmogorov-Smirnov and
Shapiro-Wilks tests on the residuals of the mixed effects models involving treatment and time
point interaction, as well as baseline outcome value, when available. If the normal family
proved unsuitable, visual assessment of the estimated variance function was used to determine
whether a transformation of the data or a different generalised linear model was required. All
analyses (univariate at each time point and repeated measures) associated with an outcome
were effected using the same transformation and/or generalised linear model.
Influence and outlier analyses

All presented analyses had residual checks and influence diagnostics examined to ensure
model validity and robustness.
Influence and outlier analyses were carried out but in accordance with the ITT and PP
population definitions; no case was removed from the analyses if they proved overly
influential or to be outliers.
Collateral data
Correlations or polychoric correlations of collateral data were produced to inform discussions
of the reliability of the outcomes. Collateral data were not entered in the primary or
secondary analyses.
Missingness
Confirmatory analyses regarding missingness included survival analysis of attrition (drop-
outs) based on treatment group, baseline primary outcomes and demographic covariates. It
was not expected that pattern-mixture analysis would be used but the possibility of doing so
was retained. The purpose of these analyses was to identify or discount possible links
between treatment assignment and attrition.
The results of the confirmatory analyses served to inform the interpretation of the primary and
secondary analysis results, by corroborating or weakening the assumption of ignorable
missingness.
Baseline data collected post-randomisation

In the case of primary outcomes Days Gambled and Money Lost, partial pre-randomisation
baseline data were available to potentially identify bias in the baseline data collected post-
randomisation that was used for baseline adjustment. The correlation between pre- and post-
randomisation data was reported, as well as the estimate of their difference and their
respective variances, pooled and by treatment group. These results served to inform the
discussion.
Testing of random effects
38
Random effects associated with counsellors and participants were tested using likelihood ratio
tests against equivalent null models not involving the target random effect (but involving the
remaining random effect) in the main analyses, concerning the non-composite primary
outcomes Days Gambled and Money Lost under the time-averaged scheme.
The random effects were tested based on a likelihood ratio test, with models fitted using
maximum likelihood only (not REML). The resulting p-value was based on a null
distribution of (χ21 + χ22)/2 distribution.
Random effects that did not appear significant were removed from the model. If a random
effect was removed from both models it was removed from all analyses.
Exploratory analyses
Any other analyses were deemed exploratory. In particular, analyses of association
(correlation or otherwise) between endpoints were deemed exploratory.
39
5. RESULTS
This chapter details the results of data analyses from this Randomised Controlled Trial.
Section 4.1 shows the number of participants including the flow of participants through the
clinical trial and study key dates and timings.
In section 4.2 are the descriptive statistics covering participant numbers in the Intention To
Treat data set; socio-demographic characteristics; gambling characterisation; treatment
assistance, goal and prospects; co-existing issues at baseline; primary efficacy outcomes;
secondary efficacy outcomes; and treatment engagement.
Primary analyses for the Intention To Treat data set are detailed in section 4.3 relating to
Hypotheses A, B and C for the three primary variables (Days Gambled, Money Lost
gambling, and Gambling-quit or improved).
Subgroup analyses for the Intention To Treat data set are detailed in section 4.4 relating to
Hypotheses A, B and C for gender, ethnicity, gambling mode, baseline PGSI score, baseline
Kessler-10 score, baseline AUDIT-C score, baseline gambling goal, baseline belief in
treatment success, and goal achievement.
Analyses for the Per Protocol data set are detailed in section 4.5 relating to Hypotheses A, B
and C for the three primary variables (Days Gambled, Money Lost gambling, and Gambling-
quit or improved).
Secondary analyses for the Intention To Treat data set are detailed in section 4.6 relating to
Hypotheses C*, D and E for the three primary variables; PGSI-12 (PGSI, 12-month time
frame); motivation to overcome gambling problem; control over gambling; Kessler-10,
AUDIT-C, DAST, WHOQoL-8 and NZDI; PRIME-MD; tobacco use; treatment for co-
existing issues; gambling impacts; and legal problems. In addition, analyses for hypotheses F
and G are presented for workbook engagement and engagement in other formal treatment
services.
Section 4.7 details the correlation between collateral and participants’ reports of gambling,
whilst section 4.8 details treatment integrity, fidelity and inter-rater reliability results.
4.1 Participants
4.1.1. Participant flow and study sample
A total of 1,298 gambler callers to the gambling helpline was assessed for eligibility in the
trial; 836 were excluded as they did not meet inclusion criteria, met exclusion criteria or they
declined to participate. A total of 462 participants was randomised: 116 to the TAU group,
112 to the MI group, 118 to the MI+W group and 116 to the MI+W+B group. These were the
participants included in the Intention To Treat analyses. The number of participants receiving
the full intervention after randomisation was 115, 107, 116 and 113 for the TAU, MI, MI+W
and MI+W+B groups respectively; these participants comprise the Per Protocol analysis set.
Participant flow is detailed in Figure 1.
40
Figure 1: Participant flow
Note: Not contactable participants were not contactable at those assessment points but may have been
contacted at subsequent assessments.
4.1.2. Recruitment dates
Eligible participants were recruited from August 2009 to February 2011 and received their
randomly allocated intervention at recruitment. The median length of intervention delivery
was approximately half an hour (30 to 34 minutes) across all groups. Data are presented in
41
Appendix 7, Table 7.1 which also details mean, standard deviation, and minimum and
maximum length of intervention delivery time.
Participants received a follow-up assessment telephone call at three- (median 92 to 93 days),

six- (median 181 to 183 days) and 12-months (362 to 364 days) post recruitment. Data are
presented in Appendix 7, Table 7.2 which also details mean, standard deviation, and minimum
and maximum number of days in which participants were contacted at each assessment time
point.
Participants in the MI+W+B group received motivational booster sessions from a helpline
counsellor at one week after the initial interview (median 7 days) and at one- (median 31.5
days), three- (median 93 days) and six-months (median 183 days) after the initial intervention.
The number of participants who received a booster call at each time point was 79, 78, 73 and
62 respectively. Data are presented in Appendix 7, Table 7.3 which also details mean,
standard deviation, and minimum and maximum number of days in which participants were
contacted at each booster call time point.
Of the 116 participants randomised to the MI+W+B group, 34% received all four booster
calls, 22% received three booster calls, one-fifth (20%) received two booster calls, 10% only
received one booster call and 14% did not receive any booster calls (not contactable) ( Table
1).
Table 1: Number of booster calls received

No. booster No. of Percentage
calls received participants
0 16 13.8%
1 12 10.3%
2 23 19.8%
3 26 22.4%
4 39 33.6%
Total 116
42
4.2 Descriptive statistics
This section details the number of participants in the Intention To Treat (ITT) and Per
Protocol (PP) data sets for each of the four groups, and demographic characteristics of the
participants in the ITT groups.
4.2.1. Number of participants
Intention To Treat data set

In total, 462 participants were recruited into the trial with between 112 and 118 participants
allocated per group; this is the Intention to Treat data set as detailed in section 3.9.3. A total
of 373 participants (81%) remained in the trial at the three-month assessment, 340
participants (74%) at the six-month assessment and 295 participants (64%) at the 12-month
assessment. Participant retention varied between the groups at each assessment. Numbers are
detailed in Table 2.
The 36% attrition at the 12-month assessment is greater than the 15% attrition predicted prior
to study commencement and means that for outcomes at 12 months there was 70% power or a
30% Type II error rate compared with the expected 80% power, 20% Type II error rate.
Table 2: ITT data set at each time point

Time point
Group Baseline 3 months 6 months 12 months
(% retention) (% retention) (% retention)
TAU 116 100 (86%) 92 (79%) 78 (67%)

MI 112 88 (79%) 78 (70%) 66 (59%)
MI+W 118 98 (83%) 88 (75%) 78 (66%)
MI+W+B 116 87 (75%) 82 (71%) 73 (63%)
N 462 373 (81%) 340 (74%) 295 (64%)
Per Protocol data set

In total, 451 participants were included in the Per Protocol analyses (as detailed in
section 3.9.3) with between 107 and 116 participants per group. A total of 289 participants
remained in the Per Protocol data set at the 12-month assessment, indicating a 64% retention
rate. Individual group retention varied between 59% (MI group) and 67% (TAU group).
Numbers are detailed in Table 3.
Table 3: PP data set at each time point

Time point
Group Baseline 3 months 6 months 12 months % retention at
12 months
TAU 115 99 91 77 67%

MI 107 84 75 63 59%
MI+W 116 96 87 77 66%
MI+W+B 113 85 81 72 64%
N 451 364 334 289 64%
43
4.2.2. Socio-demographic characteristics
Socio-demographic characteristics of the participants are detailed in Appendix 4, Table 4.1.
There were slightly more females than males in all groups (53% to 59% female) apart from
the MI+W+B group (45% female). These profiles differed slightly from the overall gambler
caller profile to the gambling helpline from 2009 to 2011 (the period of trial recruitment)
where the gender split across the years was 47% to 49% female and 52% to 53% male
(Gambling Helpline, 2012).
Marital status was similar across the groups apart from the MI+W+B group. Generally just
over half of the participants were partnered (52% to 58%) apart from the MI+W+B
participants of whom 48% were partnered.
The median age of participants across the groups was similar and ranged from 36 years to
40 years with the youngest participant aged 18 years and the oldest aged 79 years. The
proportion of participants of the major ethnicities was also similar across groups with just
under half identifying primary ethnicity as European (42% to 47%), one-third to two-fifths
(36% to 41%) identifying primarily as Maori, eight percent to 16% identifying as Pacific and
three percent to six percent identifying as Asian/Other. The ethnicity profiles differed slightly
from the overall new gambler caller profile to the gambling helpline from 2009 to 2011
where: 28% to 35% (in each individual year) were European, 18% to 23% were Maori, seven
percent to nine percent were Pacific, six percent to nine percent were Asian, and 25% to 27%
were Other/multiple ethnicity (Gambling Helpline, 2012). Based on this, the trial recruited
more participants of European, Maori or Pacific ethnicity and less of Asian/Other ethnicity
than the general gambling helpline gambler caller profile at that time.
The highest educational qualification achieved was similar across groups with one-quarter or
slightly less of participants (18% to 26%) having no qualification, approximately one-third
(31% to 37%) being educated to secondary school level, one-quarter or slightly less of
participants (18% to 24%) having a trade or technical certificate, and the remainder having a
tertiary or professional qualification.
Employment status of the participants was similar across the groups with 42% to 49% in full-
time employment, 10% to 14% in part-time employment and 11% to 18% unemployed.
Gross family income in the past 12-months was also similar across groups with one-third to
two-fifths (33% to 43%) of participants’ family income being $30,000 or less.
Participants were recruited from around the country with a larger proportion residing in the
three major cities of Auckland, Christchurch and Wellington (Appendix 4, Table 4.2).
4.2.3. Gambling characterisation
Gambling characteristics of the participants at the baseline assessment are detailed in

Appendix 4, Table 4.3.
Participants’ primary gambling mode was similar across the groups with a majority citing
electronic gaming machines (85% to 89%) in pubs (69% to 74%), clubs (6% to 12%) and
casinos (4% to 12%). The remaining participants cited casino table games, track betting,
sports betting, card gambling, Lotto, keno or other forms of gambling.
44
The median duration of participants’ gambling problems was 60 months (range 0.3 to
588 months) across all groups apart from the MI+W+B group where it was lower at
48 months (range 0.5 to 360 months). The median number of days since the last gambling
session was similar across groups at 1.0 to 1.5 days (range 0 to 97 days).
Almost all participants responded affirmatively to both of the questions in the Lie-Bet screen.
Between 93% and 99% had ever felt the need to bet more and more money, and 86% to 93%
had ever had to lie to people important to them about how much they gambled.
4.2.4. Treatment assistance, goal and prospects
At the baseline assessment, between 16% and 20% of participants across the four groups were
currently receiving assistance for their gambling problems with one-third to two-fifths (31%
to 41%) having previously received assistance (Table 4).
Table 4: Current and past treatment

Group TAU MI MI+W MI+W+B
Currently receiving Yes 17.3% 19.8% 19.1% 15.7%
assistance N 110 111 115 115
N MISSING 6 1 3 1
Previously received Yes 34.4% 31.5% 41.2% 31.3%
assistance N 93 92 97 96
N MISSING 23 20 21 20
There was little variation between the four groups in regard to participants’ treatment goal, at
the baseline assessment, being to quit all/some modes of gambling or control their gambling.
Three-quarters to four-fifths (74% to 85%) of the participants reported a desire to quit
gambling (Table 5).
Table 5: Treatment goal

Current goal, Quit 79.1% 82.9% 74.4% 84.5%
dichotomised Control 20.9% 17.1% 25.6% 15.5%
N 115 111 117 116
N MISSING 1 1 1 0
At the baseline assessment, participants were asked to rate their level of belief in success in
achieving their treatment goal in six- and 12-months’ time, rated on a scale of 0 (‘none at all’)
to 10 (‘extremely’). The median for belief in success within six months was between 8 and
10 (range 0 to 10) and for belief within 12 months was 10 (range 0 to 10). The median score
for the level of difficulty expected in achieving the treatment goal was 8 (range 0 to 10 where
0 = ‘not very’ and 10 = ‘very’) (Appendix 4, Table 4.4).
4.2.5. Co-existing issues
At the baseline assessment, four-fifths or greater (79% to 91%) of participants across the
groups reported some psychological distress in the past four weeks, measured by the Kessler-
10 scale. Over half (61% to 65%) of the participants showed some level of alcohol abuse or
45
dependence (via the AUDIT-C scale) in the past 12-months, and approximately one-quarter to
two-fifths (23% to 37%) had thoughts of suicide in the past 12-months. Four percent to eight
percent had actually made a suicide plan and one percent to six percent had tried to harm
themselves (Table 6).
Table 6: Co-existing issues

Psychological Little or no disorder (K10<20) 15.5% 21.4% 9.3% 15.5%
distress Some disorder (K10≥20)
Kessler-10, past 84.5% 78.6% 90.7% 84.5%
4 weeks, N 116 112 118 116
dichotomised N MISSING 0 0 0 0
Alcohol abuse Little or no disorder 38.8% 34.8% 39.3% 36.2%
or dependence, Some disorder
past 12 months 61.2% 65.2% 60.7% 63.8%
dichotomised N 116 112 117 116
N MISSING 0 0 1 0
Suicidal No thoughts in last 12 months 67.0% 65.8% 52.1% 59.5%
thoughts in the Just thoughts
previous 12 28.7% 23.4% 36.8% 26.7%
months Plan 3.5% 5.4% 6.0% 7.8%
Tried to harm myself 0.9% 5.4% 5.1% 6.0%
N 115 111 117 116
N MISSING 1 1 1 0
4.2.6. Primary efficacy outcomes
Primary efficacy outcomes of the participants are detailed in Appendix 4, Table 4.5.
The self-reported number of days per month when gambling occurred (Days Gambled) at
each time point was similar across the groups with the median between 6.0 to 7.5 days at
baseline, decreasing to 1.7 to 2.2 days at the three month assessment and remaining fairly
static at this level at the six and 12 month assessments (Figure 2).
Figure 2: Median Days Gambled per month
The self-reported amount of money lost per day (Money Lost) at each time point was similar
across the groups with the median between $29 to $33 at baseline, decreasing to $2.50 to $3
46
per day at the three month assessment and remaining fairly static at this level at the six and 12
month assessments (Figure 3). The maximum amount of money lost per day by individuals
was variable with no trends apparent at the follow-up time points.
Figure 3: Median Money Lost per day
Four-fifths (82% to 83%) of participants in the TAU, MI and MI+W groups self-reported that
they had ceased gambling or improved control over their gambling (Gambling-quit or
improved) at the three month assessment. At the six-month assessment the percentage of
participants decreased slightly for the TAU and MI+W groups (72%), increasing again to
similar levels to the three-month assessment at the 12-month assessment (87%,
85% respectively). The percentage was slightly lower for the MI+W+B group at all follow-
up assessments at 76%, 73% and 75% respectively (Figure 4).
Figure 4: Percentage Gambling-quit or improved
47
4.2.7. Secondary efficacy outcomes
Problem Gambling Severity Index

Problem Gambling Severity Index data are presented in Appendix 4, Table 4.6.
At the baseline assessment, almost all participants across the four groups were categorised as
problem gamblers via the past 12-month PGSI (95% to 97%), with a median PGSI score of
17 (of a possible 27). At the 12-month assessment, improvement was noted for all groups
with just over half of the participants being categorised as problem gamblers (55% to 67%)
with a median score of 9 to 10.
When a past three-month PGSI was administered there was some evidence of a trend for
reduction in problem severity across time for all groups. However, a greater reduction was
noted for the MI+W and MI+W+B groups with a median PGSI score of 2.5 and
2.0 respectively at the 12-month assessment, in comparison with the TAU and MI groups
which showed median scores of 6.0 and 4.5 respectively (Figure 5).
Figure 5: Median PGSI score, past 3-month time frame
Control over gambling behaviour

Control over gambling behaviour data are presented in Appendix 4, Table 4.7.
Participants were asked to rate their control over their gambling on a scale of 0 to 10 (0 = ‘no
control’, 10 = ‘total control’). At the baseline assessment, the median was 2.0 to 3.0 across
the four groups. At each of the follow-up assessments for each group, the median was 7.0 to
8.5, though the range was from 0 to 10 (Figure 6).
48
Figure 6: Control over gambling behaviour
Co-existing issues
Various psychological distress and substance abuse/dependence screens were administered to
participants at the baseline and follow-up assessments. The data are presented in Appendix 4,
Table 4.8.
Psychological distress
Using the Kessler-10 screen, the median score for participants in each group at baseline
ranged from 28.5 to 32 (of a total score of 50). At the three-month assessment, the median
score had decreased to 14.5 to 17, and this appeared generally stable at the six-month
assessment. A further slight improvement in score was noted at the 12-month assessment
with median scores ranging from 11.5 to 14 (Figure 7).
Figure 7: Median Kessler-10 score
Similarly, the percentage of participants in each group showing major or minor depressive
disorder or dysthymia decreased at the 12-month assessment in relation to the baseline
assessment. However, there did not appear to be much difference in percentages of
participants across the groups with bipolar disorder or who were receiving treatment or
49
prescriptions for mental health (past 12-month time frame) at the 12-month assessment in
relation to the baseline assessment.
Substance abuse/dependence
Median scores for alcohol abuse/dependence using AUDIT-C were similar across groups
(3.0 to 5.0 of a total score of 12) at the baseline and 12-month assessments (Figure 8).
Figure 8: Median AUDIT-C score
Note: The TAU line is masked by the MI line as both are identical
Very few participants scored on the DAST screen for drug abuse/dependence. Just over half
(55% to 60%) of the participants smoked tobacco at the baseline assessment; this remained
fairly constant across time. Of those smokers, the majority (at least 85%) smoked at least
once a day.
Quality of life
Quality of life of the participants across the groups was similar at each assessment (measured
using WHOQoL-8). At the baseline assessment the median score was 24 to 26 (maximum
40), increasing slightly at the other assessment points to 30 to 33.
Gambling impacts
Participants were asked how their gambling had impacted on various life domains in the past
month with impacts rated from 0 to 10, where 0 represented ‘not at all’ and 10 represented
‘very severely’. The data are presented in Appendix 4, Table 4.9.
The measured impacts related to how the following were affected: work, social life, family
and home, and health. At the baseline assessment for each domain, the median impact score
was similar across the groups. It was relatively low for impacts on work (median range 2.0 to
3.5) and higher for the other domains (median range 5.0 to 8.5). The median values indicated
zero impacts at all follow-up assessments, although some individuals did report impacts. This
indicates sustained improvement over the year following treatment intervention.
Participants were also asked if they had experienced any legal problems in the past 12 months
(baseline assessment) or past three months (follow-up assessments). Less than one-fifth
(10% to 17%) of participants at the baseline assessment reported legal problems. The
percentage was slightly lower at the three-month assessment (6% to 11%), lower still at the
50
six-month assessment (1.4% to 5.2%) and increased slightly at the 12-month assessment (9%
to 10%).
Median levels of deprivation, measured with the New Zealand Deprivation Index, did not
show differences between the groups with baseline scores ranging from 1.19 to 1.69 (possible
total score of 8). Median scores at the 12-month follow-up assessment again showed no
major difference between the groups but were lower than at the baseline assessment (range
0.63 to 0.90).
Goal setting and motivation

At each follow-up assessment, participants were asked if they had met their goal (to quit some
or all gambling or to control their gambling). Data are presented in Appendix 4, Table 4.10.
There were no major differences noted over time for the TAU and MI+W+B groups with
between one-fifth and two-fifths (22% to 40%) each reporting that their goal had been met
partly, mostly or completely at each follow-up assessment. In these groups, the percentage of
participants reporting that their goal had not been met at all ranged from nine percent to
14% at the three- and six-month assessments, but increased slightly to 23% at the 12-month
assessment. A slightly different profile was noted for the MI group whereby a greater
percentage reported ‘not at all’ across the assessments (24% to 26%) and a lower percentage
reported ‘mostly’ (14% to 16%). Participants in the MI+W group also differed in that 18%/
19% reported ‘not at all’ at the three- and six-month assessments respectively, with the
percentage decreasing to 13% at the 12-month assessment (Figure 9).
Figure 9: Percentage goal met in past three months
Participants across all groups and assessment periods remained motivated to overcome their
gambling problems with a median score of 9.5 or 10 (where 0 = ‘not at all’ and 10 =
‘extremely’).
51
4.2.8. Treatment engagement
Workbook reception and use

Participants in the MI+W and MI+W+B groups were sent, by post, a self-help workbook
(‘Becoming a Winner: Defeating Problem Gambling’) which was discussed as part of the
intervention at the booster calls for participants in the MI+W+B group. Participants in the
TAU and MI groups were not sent any workbook.
At each of the follow-up assessments, all participants were asked if they had received the
‘Becoming a Winner’ workbook and were asked about workbook use. Data are presented in
Appendix 4, Table 4.11.
Almost all of the participants (89% to 94%) in the MI+W and MI+W+B groups recalled
receiving the workbook and this recall stayed constant over time. Interestingly, half (52% to
54%) of the TAU participants apparently ‘recalled’ receiving the workbook when asked at the
three- and six-month assessments; the percentage apparently ‘recalling’ receiving the
workbook increased to 65% at the 12-month assessment. Similarly, a proportion of
participants in the MI group apparently ‘recalled’ receiving the workbook; 21%, 28% and
41% at the three-, six-, and 12-month assessments respectively. No participants in the TAU
and MI groups were sent the workbook,
Participants were asked if they had read the workbook ‘not at all’, ‘some sections’ or
‘completely’ (scored as 1, 2 or 3). The median score was 1 or 2 for respondents in all groups
and at all assessment points.
When asked whether they had completed the exercises in the workbook, participants in the
MI+W and MI+W+B groups did not differ despite the workbook being discussed as part of
the booster calls for the latter group. At the three-month assessment approximately half
(51% MI+W, 45% MI+W+B) of the participants reported completing some of the exercises
with nine percent and 13% respectively, reporting completing all the exercises. The
percentages decreased over time. Again, whilst the majority of participants in the TAU and
MI groups (who did not receive the ‘Becoming a Winner’ workbook) stated that completing
the exercises was not applicable, 12% or less indicated completing some or all of the
exercises.
A similar profile was noted for participants reporting using some or all of the strategies in the
workbook as was noted for participants reporting completing the exercises (though the actual
percentages were slightly different).
Treatment service assistance

At each follow-up assessment, participants were asked if they had received any assistance
(formal or informal) (additional to their initial gambling helpline intervention) in the previous
three months for their gambling problems. Overall data are presented in Appendix 4,
Table 4.12. Data for individual formal services are presented in Appendix 4, Table 4.13.
Overall, approximately one-fifth to one-quarter (20% to 28%) of participants had received

some form of formal assistance (from a professional person) for their gambling problems over
the past three months, at the three-month assessment. For all groups apart from the MI group,
the percentage decreased slightly at the subsequent assessments (15% to 18%). For
participants in the MI group, the percentage receiving formal assistance at the six- and 12-
month assessments remained fairly constant at 23% and 26% respectively.
52
Overall, a slightly higher percentage of participants reported receiving some form of informal
assistance (e.g. from family, friends or other non-professional person) than those receiving
formal assistance at the three-month assessment (37% to 42%). The percentage remained at a
similar level for participants in all groups at the six- and 12-month assessments (30%
to 44%).
53
4.3 Primary analyses ITT data set
This section details the primary analyses of the trial for the Intention To Treat data set,
focusing on results relating to Hypotheses A, B and C as detailed in section 3.9 and re-iterated
below. The data are analysed by the three primary variables (Days Gambled, Money Lost
gambling, and Gambling-quit or improved).
Primary equivalence hypothesis A

A. The Motivational Interview (MI) group will show similar improvement to Treatment as
Usual (TAU).
Linear mixed effects

No statistically significant differences were noted between the MI and the TAU participants in
regard to time-averaged money lost gambling or time-averaged days spent gambling, i.e. the
MI group showed similar improvement to the TAU group (Table 7).
Table 7: MI vs. TAU Days Gambled, Money Lost

TEST 95% Confidence
Limits
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 -0.40 1.61
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 -2.38 8.15
Logistic mixed effects

Participants in the MI group showed similar improvement to the TAU group in regard to time-
averaged Gambling-quit or improved (Table 8).
Table 8: MI vs. TAU time-averaged Gambling-quit or improved

TEST Odds Odds Odds
ratio Ratio Ratio
CILB CIUB
Gambling-quit or hyp. A : TAU vs MI, δ=0.13 0.70 0.30 1.66
improved, time-
averaged
Conclude in inequivalence at 5% significance level if CIUB<0.88 or CILB>1.14
Primary superiority hypotheses B and C

B.
c. The Motivational Interview plus Workbook plus Booster group (MI+W+B) will show
greater improvement than the TAU group.
d. The Motivational Interview plus Workbook plus Booster group (MI+W+B) will show
greater improvement than the MI group.
C.
a. The MI+W+B group will show greater improvement than the TAU group at the 12-
month follow-up.
b. The MI+W+B group will show greater improvement than the MI group at the 12-month
follow-up.
54
c. The MI+W+B group will show greater improvement than the MI+W group at the 12-
month follow-up.

No statistically significant differences were noted for hypotheses B and C in regard to Days
Gambled, Money Lost gambling or in relation to PGSI-12 scores (PGSI, past 12-month time
frame) at the 12-month assessment (Table 9).
Table 9: Hypotheses B and C - Days Gambled, Money Lost, PGSI

TEST Estimated Standard P-value Alternative
change error (one- accepted
sided)
Days Gambled, hyp. B.a: TAU vs MI+W 0.32 0.49 0.74 No
time-averaged hyp. B.b: MI vs MI+W -0.29 0.51 0.29 No
hyp. B.c: TAU vs MI+W+B -0.02 0.51 0.49 No
hyp. B.d: MI vs MI+W+B -0.34 0.51 0.25 No
Days Gambled, hyp. C.a: TAU vs MI+W+B -0.13 0.59 0.41 No
at 12 months hyp. C.b: MI vs MI+W+B -0.83 0.62 0.09 No
hyp. C.c: MI+W vs MI+W+B -0.17 0.59 0.38 No
Money Lost, hyp. B.a: TAU vs MI+W -1.35 2.60 0.30 No
hyp. B.d: MI vs MI+W+B 1.30 2.66 0.69 No
Money Lost, at hyp. C.a: TAU vs MI+W+B -1.41 3.07 0.32 No
12 months hyp. C.b: MI vs MI+W+B -4.57 3.20 0.08 No
hyp. C.c: MI+W vs MI+W+B 1.76 3.07 0.72 No
PGSI-12, at 12 hyp. B.a: TAU vs MI+W -0.03 0.99 0.49 No
months hyp. B.b: MI vs MI+W -0.44 1.03 0.33 No
hyp. C.a: TAU vs MI+W+B 0.78 1.01 0.78 No
hyp. C.b: MI vs MI+W+B 0.37 1.04 0.64 No

No statistically significant differences were noted for hypotheses B and C in regard to time-
averaged self-reported Gambling-quit or improved and self-reported Gambling-quit or
improved at the 12-month assessment (Table 10).
55
Table 10: Hypotheses B and C - Gambling-quit or improved
TEST Odds Odds Odds P-value Alternative
ratio Ratio Ratio (one- accepted
CILB CIUB sided)
Gambling-quit hyp. B.a: TAU vs MI+W 1.25 0.53 2.96 0.31 No
or improved, hyp. B.b: MI vs MI+W 1.78 0.74 4.29 0.10 No
time-averaged
hyp. B.c: TAU vs 1.23 0.51 2.93 0.32 No
MI+W+B
hyp. B.d: MI vs MI+W+B 0.98 0.40 2.39 0.52 No
Gambling-quit hyp. C.a: TAU vs 1.52 0.57 4.03 0.20 No
or improved, at MI+W+B
12 months hyp. C.b: MI vs MI+W+B 2.67 0.99 7.24 0.03 No*
hyp. C.c: MI+W vs 0.36 0.13 1.04 0.97 No
MI+W+B
* False discovery rate control requires the p-value to be smaller than 0.017 for acceptance of the
alternative in this particular instance
56
4.4 Subgroup analyses ITT data set
This section details the subgroup analyses of the trial for the Intention To Treat data set,
focusing on results relating to hypotheses A, B and C as detailed in section 3.9 and re-iterated
below. The data are analysed by gender, ethnicity, gambling mode, baseline PGSI score,
baseline Kessler-10 score, baseline AUDIT-C score, baseline gambling goal, baseline belief in
treatment success, and goal achievement.
4.4.1. Gender subgroups
The number of participants by gender is detailed in Table 11.
Table 11: Number of participants by gender

Valid number of participants
Gender
Male 48 53 53 64
Female 68 59 64 52

Data are presented in Appendix 5, Tables 5.1 and 5.2.

When examined by gender, no statistically significant differences were noted between the MI
and the TAU participants in regard to time-averaged money lost gambling or time-averaged
days spent gambling, i.e. the MI group showed similar improvement to the TAU group.

When examined by gender, participants in the MI group showed similar improvement to the
TAU group in regard to time-averaged Gambling-quit or improved.

The MI+W group showed statistically significant (p=0.011) greater improvement than the MI
group for males in relation to time-averaged money lost gambling. No statistically significant
differences were noted for males for each of the other hypotheses B and C tested in regard to
days gambled, money lost gambling or in relation to PGSI-12 scores at the 12-month
assessment (Table 12). No statistically significant differences were noted for females for
hypotheses B and C (Appendix 5, Table 5.3).
Table 12: Hypotheses B and C - days gambled, money lost, PGSI - males
sided)
hyp. B.c: TAU vs MI+W+B 0.13 0.75 0.57 No
time-averaged
57
No statistically significant differences were noted when analysed by gender for hypotheses B
and C in regard to time-averaged self-reported Gambling-quit or improved and self-reported
Gambling-quit or improved at the 12-month assessment (Appendix 5, Table 5.4).
4.4.2. Ethnicity subgroups
The number of participants by ethnicity is detailed in Table 13.
Table 13: Number of participants by gender

Ethnicity
European 58 53 64 62
Maori 47 44 51 42
Pacific 13 18 10 12
Asian & Other 5 4 5 8


When examined by ethnicity, no statistically significant differences were noted between the
MI and the TAU participants in regard to time-averaged money lost gambling or time-
averaged days spent gambling, i.e. the MI group showed similar improvement to the TAU
group.

When examined by ethnicity, participants in the MI group showed similar improvement to the
TAU group in regard to time-averaged Gambling-quit or improved.

The MI+W+B group showed statistically significant (p=0.004) greater improvement than the
MI group for Maori in relation to money lost gambling at the 12-month assessment (Table
14). No statistically significant differences were noted when examined by ethnicity for each
of the other hypotheses B and C in regard to days gambled, money lost gambling or in
relation to PGSI-12 scores at the 12-month assessment (Appendix 5, Table 5.7).
58
Table 14: Hypothesis C - Money Lost gambling at 12-months by ethnicity
sided)
hyp. C.a: TAU vs MI+W+B -2.36 4.11 0.28 No
European
Maori
Pacific
Asian & Other
hyp. C.b: MI vs MI+W+B -0.71 4.36 0.44 No
European
hyp. C.b: MI vs MI+W+B -14.33 5.46 0.004 Yes
Money Lost, at Maori
Pacific
Asian & Other
European
Maori
Pacific
Asian & Other

No statistically significant differences were noted when analysed by ethnicity for hypotheses
B and C in regard to time-averaged self-reported Gambling-quit or improved and self-
reported Gambling-quit or improved at the 12-month assessment (Appendix 5, Table 5.8).
4.4.3. Gambling mode
The number of participants by dichotomised primary gambling mode causing the gambling
problem (EGM vs. non-EGM) is detailed in Table 15.
Table 15: Number of participants by gambling mode

Mode
EGM 108 102 108 107

Non-EGM 8 10 10 9


When examined by primary gambling mode, no statistically differences were noted between
the MI and the TAU participants in regard to time-averaged money lost gambling or time-
group.
59
When examined by primary gambling mode, participants in the MI group showed similar
improvement to the TAU group in regard to time-averaged gambling-quit or improved.

No statistically significant differences were noted when examined by gambling mode for
hypotheses B and C in regard to days gambled, money lost gambling or in relation to PGSI-12
scores at the 12-month assessment (Appendix 5, Table 5.11).

No statistically significant differences were noted when analysed by gambling mode for
hypotheses B and C in regard to time-averaged self-reported Gambling-quit or improved and
self-reported Gambling-quit or improved at the 12-month assessment (Appendix 5, Table
5.12).
4.4.4. Baseline PGSI score
As previously detailed, at the baseline assessment, almost all participants across the four
groups were categorised as problem gamblers via the past 12-month PGSI (95% to 97%),
with a median PGSI score of 17. Therefore, the equivalence and superiority hypotheses by
PGSI score were examined using baseline PGSI scores dichotomised to ≤ 17 or > 17
(i.e. either side of the median score).
The number of participants by dichotomised baseline PGSI score is detailed in Table 16.
Table 16: Number of participants by dichotomised PGSI baseline score

PGSI
Baseline ≤17 66 62 68 63
Baseline >17 50 50 50 53


When examined by dichotomised baseline PGSI score, no statistically significant differences
were noted between the MI and the TAU participants in regard to time-averaged money lost
gambling or time-averaged days spent gambling, i.e. the MI group showed similar
improvement to the TAU group.

When examined by dichotomised baseline PGSI score, participants in the MI group showed
similar improvement to the TAU group in regard to time-averaged Gambling-quit or
improved.
60
The MI+W+B group with a dichotomised PGSI baseline score of > 17 (i.e. those with higher
than the median PGSI score) showed statistically significant (p=0.005) greater improvement
than the MI group in relation to money lost gambling at the 12-month assessment (Table 17).
No statistically significant differences were noted when examined by dichotomised baseline
PGSI score of > 17 or of ≤ 17 for each of the other hypotheses B and C in regard to days
gambled, money lost gambling or in relation to PGSI scores at the 12-month assessment
(Appendix 5, Table 5.15).

The MI+W+B group with a dichotomised PGSI baseline score of > 17 showed statistically
significant greater improvement than both the TAU and MI groups (P=0.004 and p=0.001
respectively) in relation to self-reported Gambling-quit or improved at the 12-month
assessment (Table 18).
No statistically significant differences were noted for time-averaged Gambling-quit or

improved for hypotheses B and C in the PGSI score > 17 groups, or the score ≤ 17 groups for
time-averaged Gambling-quit or improved or Gambling-quit or improved at the 12-month
assessment (Appendix 5, Table 5.16).
Table 17: Hypotheses B and C - Days Gambled, Money Lost, PGSI by baseline PGSI >17
sided)
Days Gambled, hyp. B.a: TAU vs MI+W -0.36 0.75 0.32 No
Days Gambled, hyp. C.a: TAU vs MI+W+B 0.86 0.86 0.57 No

12 months hyp. C.b: MI vs MI+W+B -12.02 4.71 0.005 Yes
hyp. B.c & C.a: TAU vs -0.50 1.53 0.37 No
MI+W+B
hyp. B.d & C.b: MI vs -0.79 1.62 0.31 No
MI+W+B
61
Table 18: Hypotheses B and C - Gambling-quit or improved by baseline PGSI > 17
CILB CIUB sided)
time-averaged
hyp. B.c: TAU vs 4.43 1.19 16.54 0.0134 No
MI+W+B
Gambling-quit hyp. C.a: TAU vs 7.88 1.76 35.33 0.004 Yes
12 months hyp. C.b: MI vs MI+W+B 11.00 2.42 50.08 0.001 Yes
hyp. C.c: MI+W vs 0.88 0.18 4.44 0.56 No
MI+W+B
4.4.5. Baseline Kessler-10 score
As previously detailed, at the baseline assessment, four-fifths or greater (79% to 91%) of

participants across the groups had some psychological distress in the past four weeks, as
indicated by the Kessler-10 scale; the median score was 30. Therefore, the equivalence and
superiority hypotheses by Kessler-10 score were tested using baseline Kessler-10 scores
dichotomised to ≤ 30 or > 30 (i.e. either side of the median score).
The number of participants by dichotomised baseline Kessler-10 score is detailed in Table 19.
Table 19: Number of participants by dichotomised Kessler-10 baseline score

K-10
Baseline ≤ 30 60 67 55 67
Baseline > 30 56 45 63 59


When examined by dichotomised baseline Kessler-10 score, no statistically significant
differences were noted between the MI and the TAU participants in regard to time-averaged
money lost gambling or time-averaged days spent gambling, i.e. the MI group showed similar
improvement to the TAU group.

When examined by dichotomised baseline Kessler-10 score, participants in the MI group
showed similar improvement to the TAU group in regard to time-averaged Gambling-quit or
improved.
62
The MI+W+B group with a dichotomised Kessler-10 baseline score of > 30 (i.e. those with
higher than the median Kessler-10 score) showed statistically significant (p=0.0053) greater
improvement than the MI group in relation to money lost at the 12-month assessment (Table
20). No statistically significant differences were noted when examined by dichotomised
baseline Kessler-10 score of > 30 or of ≤ 30 for the other hypotheses B and C in regard to
days gambled, money lost gambling or in relation to PGSI-12 scores at the 12-month
assessment (Appendix 5, Table 5.19).
Table 20: Hypotheses B and C - Days gambled, Money lost, PGSI by baseline K-10 score > 30
sided)
time-averaged -1.25 0.74 0.046 No
hyp. B.b: MI vs MI+W
Days Gambled, hyp. C.a: TAU vs MI+W+B 0.13 0.85 0.56 No

PGSI-12, at 12 hyp. B.a: TAU vs MI+W 0.10 1.41 0.53 No
hyp. B.c & C.a: TAU vs 1.34 1.41 0.83 No
MI+W+B
hyp. B.d & C.b: MI vs -1.79 1.61 0.13 No
MI+W+B

The MI+W+B group with a dichotomised Kessler-10 baseline score of > 30 showed
statistically significant greater improvement than the MI group (p=0.00005) in relation to
self-reported Gambling-quit or improved at the 12-month assessment (Table 21). No
statistically significant differences were noted for time-averaged Gambling-quit or improved
in the score > 30 groups, or the score ≤ 30 groups for hypotheses B and C for time-averaged
Gambling-quit or improved or Gambling-quit or improved at the 12-month assessment
63
Table 21: Hypotheses B and C - Gambling-quit or improved by Kessler-10 score > 30
CILB CIUB sided)
time-averaged
hyp. B.c: TAU vs 1.45 0.41 5.22 0.28 No
MI+W+B
hyp. C.c: MI+W vs 1.33 0.27 6.61 0.36 No
MI+W+B
4.4.6. Baseline AUDIT-C score
As previously detailed, at the baseline assessment, over half (61% to 65%) of the participants
showed some level of alcohol abuse or dependence using the AUDIT-C. The equivalence and
superiority hypotheses by AUDIT-C results were tested using baseline AUDIT-C scores
dichotomised to low risk or high risk (score 3 or more for females, 4 or more for males).
The number of participants by dichotomised baseline AUDIT-C score is detailed in Table 22.
Table 22: Number of participants by dichotomised AUDIT-C baseline score

AUDIT-C
Low risk 45 39 46 42
High risk 71 73 71 74


When examined by dichotomised baseline AUDIT-C score, no differences were noted
between the MI and the TAU participants in regard to time-averaged money lost gambling or
time-averaged days spent gambling, i.e. the MI group showed similar improvement to the
TAU group.

When examined by dichotomised baseline AUDIT-C score, participants in the MI group
showed similar improvement to the TAU group in regard to time-averaged Gambling-quit or
improved.

AUDIT-C score for hypotheses B and C in regard to days gambled, money lost gambling or in
relation to PGSI scores at the 12-month assessment (Appendix 5, Table 5.23).
64
The MI+W+B group with a low risk AUDIT-C baseline score showed statistically significant
greater improvement than the MI group (P=0.013) in relation to self-reported Gambling-quit
or improved at the 12-month assessment (Table 23). No statistically significant differences
were noted for time-averaged Gambling-quit or improved in the low risk score groups ( Table
23), or the high risk score groups for time-averaged Gambling-quit or improved or Gambling-
quit or improved at the 12-month assessment (Appendix 5, Table 5.24).
Table 23: Hypotheses B and C - gambling-quit or improved by low risk AUDIT-C score
CILB CIUB sided)
time-averaged
hyp. B.c: TAU vs 0.96 0.21 4.43 0.52 No
MI+W+B
hyp. C.c: MI+W vs 0.65 0.11 3.84 0.68 No
MI+W+B
4.4.7. Baseline gambling goal
As previously detailed, at the baseline assessment, three-quarters to four-fifths (74% to 85%)

of the participants reported a desire to quit all/some modes of gambling versus controlling
their gambling. The equivalence and superiority hypotheses by gambling goal results were
tested using data dichotomised to quit (all/some modes) or control gambling.
The number of participants by dichotomised baseline gambling goal is detailed in Table 24.
Table 24: Number of participants by dichotomised baseline gambling goal

Goal
Quit 91 92 87 98
Control 24 19 30 18

When examined by dichotomised baseline gambling goal, no differences were noted between
the MI and the TAU participants in regard to time-averaged money lost gambling or time-
group (Table 25).
Table 25: TAU vs. MI days gambled, money lost by dichotomised baseline gambling goal
TEST 95% Confidence
Limits
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 Quit -0.76 1.43
65
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 Quit -4.87 6.77
TEST 95% Confidence

Limits
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 Control -0.57 4.16
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 Control -0.45 24.69

When examined by dichotomised baseline gambling goal, participants in the MI group whose
goal was to quit some or all modes of gambling showed similar improvement to the TAU
group in regard to time-averaged Gambling-quit or improved. However, participants in the
MI group whose goal was to control their gambling, gambled significantly more (Odds
Ratio 0.06) than participants in the TAU group in regard to time-averaged Gambling-quit or
improved (Table 26).
Table 26: TAU vs. MI Gambling-quit or improved by dichotomised baseline gambling goal
TEST Odds Odds Odds
ratio Ratio Ratio
CILB CIUB
Gambling-quit or hyp. A : TAU vs MI, δ=0.13 Quit 1.39 0.53 3.65
improved, time-
averaged
Gambling-quit or hyp. A : TAU vs MI, δ=0.13 Control 0.06 0.01 0.46
improved, time-
averaged

The MI+W+B group by dichotomised baseline gambling goal to control gambling showed
statistically significant (p=0.009) greater improvement than the MI group in relation to money
lost at the 12-month assessment. The MI+W+B group by dichotomised baseline gambling
goal to control gambling also showed statistically significant greater improvement than both
the MI and the MI+W groups (p=0.006 and p=0.004 respectively) in relation to number of
days gambled at the 12-month assessment. No statistically significant differences were noted
when examined by dichotomised baseline gambling goal to control gambling for the other
hypotheses B and C in regard to days gambled, money lost gambling or in relation to PGSI
scores at the 12-month assessment (Table 27).

gambling goal to quit some or all modes of gambling for hypotheses B and C in regard to
days gambled, money lost gambling or in relation to PGSI scores at the 12-month assessment
Table 27: Hypotheses B and C - Days gambled, Money lost, PGSI by baseline control gambling
goal
sided)
66
Money Lost, hyp. B.a: TAU vs MI+W 4.04 5.56 0.77 No
at 12 months hyp. C.b: MI vs MI+W+B -3.72 1.49 0.006 Yes
hyp. C.c: MI+W vs MI+W+B -3.39 1.29 0.004 Yes
PGSI-12, at 12 hyp. B.a: TAU vs MI+W 2.22 2.04 0.86 No
hyp. B.c & C.a: TAU vs -0.51 2.28 0.41 No
MI+W+B
hyp. B.d & C.b: MI vs -4.71 2.54 0.03 No
MI+W+B

The MI+W and MI+W+B groups by dichotomised baseline gambling goal to control
gambling showed statistically significant greater improvement than the MI group (p=0.015
and p=0.0008 respectively) in relation to time-averaged Gambling-quit or improved (Table
28). No statistically significant differences were noted for self-reported Gambling-quit or
improved at the 12-month assessment, or for groups by dichotomised baseline gambling goal
to quit some/all modes of gambling for time-averaged Gambling-quit or improved or
Gambling-quit or improved at the 12-month assessment (Appendix 5, Table 5.26).
Table 28: Hypotheses B and C - gambling-quit or improved by control gambling goal

CILB CIUB sided)
or improved, hyp. B.b: MI vs MI+W 7.18 1.22 42.20 0.015 Yes
time-averaged
hyp. B.c: TAU vs 1.50 0.16 13.74 0.36 No
MI+W+B
hyp. B.d: MI vs MI+W+B 11.85 2.57 54.55 0.0008 Yes
12 months hyp. C.b: MI vs MI+W+B Unreliable results due to numerical instability
hyp. C.c: MI+W vs 3.59 0.31 41.33 0.15 No
MI+W+B
4.4.8. Baseline belief in treatment success
As previously detailed, at the baseline assessment, participants were asked to rate their level
of belief in success in achieving their treatment goal in 12-months; the overall median value
was 10. The equivalence and superiority hypotheses by gambling goal results were tested
67
using data dichotomised to low belief level (lower than median) or high belief level (median
value).
The number of participants by dichotomised belief in treatment success is detailed in Table

29.
Table 29: Number of participants by dichotomised belief in treatment success

Belief
Low level 55 61 72 65
High level 61 51 46 51

When examined by dichotomised belief in treatment success, no differences were noted
between the MI and the TAU participants in regard to time-averaged money lost gambling or
time-averaged days spent gambling, i.e. the MI group showed similar improvement to the
TAU group (Table 30).
Table 30: TAU vs. MI days gambled, money lost by belief in treatment success
TEST 95% Confidence
Limits
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 Low -0.82 2.17
belief
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 Low -4.54 11.58
belief
TEST 95% Confidence

Limits
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 High -1.04 1.57
belief
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 High -5.15 8.94
belief

When examined by dichotomised belief in treatment success, participants in the MI group
whose baseline belief in treatment success at 12-months was high showed similar
improvement to the TAU group in regard to time-averaged Gambling-quit or improved.
However, participants in the MI and TAU groups whose baseline belief in treatment success
was low showed inequivalence in regard to time-averaged Gambling-quit or improved,
i.e. participants in the MI group whose baseline belief in treatment success was low gambled
(days and money) significantly more (Odds Ratio 0.20) than participants in the TAU group in
regard to time-averaged Gambling-quit or improved (Table 31).
68
Table 31: TAU vs. MI Gambling-quit or improved by dichotomised belief in treatment
success
TEST Odds Odds Odds
ratio Ratio Ratio
CILB CIUB
Gambling-quit or hyp. A : TAU vs MI, δ=0.13 Low 0.20 0.05 0.73
improved, time- belief
averaged
Gambling-quit or hyp. A : TAU vs MI, δ=0.13 High 2.64 0.78 8.93
improved, time- belief
averaged

No statistically significant differences were noted when examined by dichotomised belief in
treatment success for hypotheses B and C in regard to days gambled, money lost gambling or
in relation to PGSI scores at the 12-month assessment (Appendix 5, Table 5.27).

The MI+W+B group whose belief in treatment success was low showed statistically
significant greater improvement than the MI group (p=0.0007) in relation to self-reported
Gambling-quit or improved at the 12-month assessment (Table 32). No statistically
significant differences were noted for self-reported time-averaged Gambling-quit or improved
in the low belief groups, or the high belief groups for time-averaged Gambling-quit or
improved or Gambling-quit or improved at the 12-month assessment (Appendix 5, Table
5.28).
Table 32: Hypotheses B and C - Gambling-quit or improved by low belief treatment success
CILB CIUB sided)
time-averaged
hyp. B.c: TAU vs 0.64 0.16 2.52 0.74 No
MI+W+B
hyp. B.d: MI vs 3.19 0.89 11.35 0.26 No
MI+W+B
12 months hyp. C.b: MI vs 10.87 2.53 46.74 0.0007 Yes
MI+W+B
hyp. C.c: MI+W vs 0.41 0.09 1.89 0.87 No
MI+W+B
4.4.9. Goal achievement
Multinomial mixed regression

No statistically significant differences were noted when examined by whether the
participants’ goal was met in the past three-months (time-averaged) for hypotheses B or C, or
whether the goal was met in the past three-months at the 12-month assessment (Appendix 5,
Table 5.29).
69
70
4.5 Primary analyses PP data set
This section details analyses of the trial for the Per Protocol data set, focusing on results
relating to hypotheses A, B and C as detailed in section 3.9 and re-iterated below. The data
are analysed by the three primary variables (Days Gambled, Money Lost gambling, and
Gambling-quit or improved).

B. The Motivational Interview (MI) group will show similar improvement to Treatment as
Usual (TAU).

No statistically significant differences were noted between the MI and the TAU participants in
regard to time-averaged money lost gambling or time-averaged days spent gambling, i.e. the
MI group showed similar improvement to the TAU group (Table 33).
Table 33: TAU vs. MI Days Gambled, Money Lost

TEST 95% Confidence
Limits
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 -0.59 1.46
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 -3.90 4.70

Participants in the MI group showed similar improvement to the TAU group in regard to time-
averaged Gambling-quit or improved (Table 34).
Table 34: TAU vs. MI time-averaged Gambling-quit or improved

TEST Odds Odds Odds
ratio Ratio Ratio
CILB CIUB
Gambling-quit or hyp. A : TAU vs MI, δ=0.13 0.85 0.35 2.06
improved, time-
averaged

B.
b. The Motivational Interview plus Workbook (MI+W) group will show greater
c. The Motivational Interview plus Workbook plus Booster (MI+W+B) group will show
d. The Motivational Interview plus Workbook plus Booster (MI+W+B) group will show
C.
month follow-up.
follow-up.
71
month follow-up.

No statistically significant differences were noted for hypotheses B and C in regard to days
gambled, money lost gambling or in relation to PGSI-12 scores (PGSI, past 12-month time
frame) at the 12-month assessment (Table 35).
Table 35: Hypotheses B and C - Days Gambled, Money Lost, PGSI

sided)
time-averaged hyp. B.b: MI vs MI+W 0.02 0.52 0.51 No
Money Lost, hyp. B.a: TAU vs MI+W 0.56 2.11 0.61 No
time-averaged hyp. B.b: MI vs MI+W 0.96 2.20 0.67 No

No statistically significant differences were noted for hypotheses B and C in regard to time-
averaged self-reported Gambling-quit or improved and self-reported Gambling-quit or
improved at the 12-month assessment (Table 36).
72
Table 36: Hypotheses B and C - Gambling-quit or improved
CILB CIUB sided)
time-averaged
hyp. B.c: TAU vs 0.83 0.34 1.99 0.67 No
MI+W+B
12 months hyp. C.b: MI vs MI+W+B 2.19 0.79 6.08 0.07 No
hyp. C.c: MI+W vs 0.36 0.13 1.05 0.97 No
MI+W+B
73
4.6 Secondary analyses ITT data set
This section details the secondary efficacy analyses of the ITT data set, focusing on results
from analyses relating to hypotheses C*, D, and E for the three primary variables (Days
Gambled, Money Lost gambling, and Gambling-quit or improved); hypotheses A, B and C for
PGSI which underwent the same analyses as the primary outcomes; hypotheses B and C for
other secondary outcomes (such as motivation to overcome gambling problem, co-existing
issues, and gambling impacts); and secondary engagement hypotheses F and G as discussed
in section 3.9 and re-iterated below.
4.6.1 Primary variables

C*.
e. The MI+W+B group will show greater improvement than the MI group between three
and 12 months.
f. The MI+W+B group will show greater improvement than the MI+W group between
D.
E.
Mixed linear regression

No statistically significant differences were noted for hypothesis C* in regard to days
gambled and money lost gambling between the three and 12-month assessments. However,
all treatment groups showed a statistically significant reduction (p≤0.0001) for time-averaged
days gambling and money lost gambling as well as for time-averaged PGSI (past 12-month
time frame) (hypothesis D) (Table 37).
In regard to hypothesis E, high levels of workbook engagement were associated with less
time-averaged money lost gambling (p=0.03) and high levels of receiving informal assistance
for the gambling problem were associated with higher time-averaged control over gambling
(p=0.01). No statistically significant differences were noted between levels of engagement
and PGSI (past 12-month time frame) at the 12-month assessment (Table 38).
74
Table 37: Hypotheses C* and D: Days Gambled, Money Lost and PGSI
sided)
Days Gambled, hyp. C*.d: TAU vs MI+W+B -0.13 0.64 0.42 No
between 3 and hyp. C*.e: MI vs MI+W+B -0.62 0.67 0.18 No
12 months
hyp. C*.f: MI+W vs 0.10 0.65 0.56 No
MI+W+B
Money Lost, hyp. C*.d: TAU vs MI+W+B -0.56 3.18 0.43 No
between 3 and hyp. C*.e: MI vs MI+W+B -4.05 3.31 0.11 No
12 months
hyp. C*.f: MI+W vs 1.66 3.20 0.70 No
MI+W+B
Days Gambled, hyp. D.a: TAU -6.08 0.66 <0.0001 Yes
time-averaged hyp. D.b: MI -4.67 0.71 <0.0001 Yes
hyp. D.c: MI+W -5.79 0.66 <0.0001 Yes
hyp. D.d: MI+W+B -5.30 0.69 <0.0001 Yes
Money Lost, hyp. D.a: TAU -35.38 5.58 <0.0001 Yes
time-averaged hyp. D.b: MI -37.64 5.99 <0.0001 Yes
hyp. D.c: MI+W -42.41 5.61 <0.0001 Yes
hyp. D.d: MI+W+B -37.16 5.91 <0.0001 Yes
PGSI-12, time- hyp. D.a: TAU -7.32 0.76 <0.0001 Yes
averaged hyp. D.b: MI -7.01 0.82 <0.0001 Yes
hyp. D.c: MI+W -7.48 0.75 <0.0001 Yes
hyp. D.d: MI+W+B -6.53 0.78 <0.0001 Yes
75
Table 38: Hypothesis E - Days Gambled, Money Lost, control over gambling and PGSI
sided)
Days gambled, time- hyp. E: Workbook -0.33 0.24 0.09 No
averaged engagement (per unit
change)
Days gambled, time- hyp. E: Formal -0.02 0.33 0.47 No
averaged assistance
Days gambled, time- hyp. E: Informal -0.31 0.27 0.12 No
averaged assistance
Money lost, time- hyp. E: Workbook -1.68 0.90 0.03 Yes
change)
Money lost, time- hyp. E: Formal 0.93 1.56 0.72 No
averaged assistance
Money lost, time- hyp. E: Informal -1.98 1.29 0.06 No
averaged assistance
Control over hyp. E: Workbook 0.20 0.14 0.08 No
gambling, time- engagement (per unit
averaged change)
Control over hyp. E: Formal 0.09 0.21 0.34 No
gambling, time- assistance
averaged
Control over hyp. E: Informal 0.44 0.17 0.01 Yes
gambling, time- assistance
averaged
PGSI-12, at 12 hyp. E: Workbook 0.72 0.65 0.87 No
months engagement (per unit
change)
PGSI-12, at 12 hyp. E: Formal 1.95 0.95 0.98 No
months assistance
PGSI-12, at 12 hyp. E: Informal -0.22 0.78 0.39 No
months assistance
Mixed logistic regression

The MI+W+B group showed statistically significant greater improvement than the MI+W
group (p=0.0001) in relation to self-reported Gambling-quit or improved at the 12-month
assessment (Hypothesis C*). All treatment groups showed a statistically significant reduction
in gambling (p≤0.0001) when time-averaged (Hypothesis D) (Table 39).
In relation to Hypothesis E, high levels of workbook engagement were associated with less
time-averaged gambling (p=0.03) and high levels of receiving informal assistance for the
gambling problem were associated with higher time-averaged goal being met (p≤0.01) (Table
39).
76
Table 39: Hypotheses C*, D and E - Gambling-quit or improved and goal met
CILB CIUB sided)
Gambling-quit hyp. C*.d: TAU vs 0.71 0.31 1.63 0.79 No
12 months hyp. C*.e: MI vs 0.49 0.21 1.14 0.95 No
MI+W+B
hyp. C*.f: MI+W vs 5.76 2.31 14.38 0.0001 Yes
MI+W+B
Gambling-quit hyp. D.a: TAU 11.32 6.23 20.57 <0.0001 Yes
or improved, hyp. D.b: MI 7.94 4.26 14.79 <0.0001 Yes
time-averaged
hyp. D.c: MI+W 14.15 7.61 26.31 <0.0001 Yes
hyp. D.d: MI+W+B 13.87 7.33 26.24 <0.0001 Yes
Gambling-quit hyp. E: Workbook 1.42 0.97 2.08 0.03 Yes
or improved, engagement (per unit
time-averaged change)
Gambling-quit hyp. E: Formal 0.95 0.59 1.51 0.41 No
or improved, assistance
time-averaged
Gambling-quit hyp. E: Informal 1.05 0.71 1.54 0.41 No
or improved, assistance
time-averaged
Goal met, time- hyp. E: Workbook 1.03 0.73 1.47 0.43 No
change)
Goal met, time- hyp. E: Formal 1.11 0.67 1.85 0.34 No
averaged assistance
Goal met, time- hyp. E: Informal 1.76 1.15 2.69 0.00 Yes
averaged assistance
4.6.2 PGSI
Equivalence hypothesis A
A. The Motivational Interview (MI) group will show similar improvement to Treatment as
Usual (TAU).

No differences were noted between the MI and the TAU participants in regard to PGSI (past
12-month time frame) (Table 40).
Table 40: TAU vs. MI PGSI

TEST 95% Confidence
Limits
PGSI-12 hyp. A : TAU vs MI, δ=1 -2.45 1.62

B.
77
b. The Motivational Interview plus Workbook (MI+W) group will show greater
c. The Motivational Interview plus Workbook plus Booster (MI+W+B) group will show
d. The Motivational Interview plus Workbook plus Booster (MI+W+B) group will show
C.
month follow-up.
follow-up.
month follow-up.

No statistically significant differences were noted for hypotheses B and C in regard to PGSI
(past three-month time frame) time-averaged or at the 12-month assessment (Table 41).
Table 41: Hypotheses B and C - PGSI

sided)
PGSI-3, time- hyp. B.a: TAU vs MI+W -0.57 0.81 0.24 No
averaged hyp. B.b: MI vs MI+W -1.37 0.84 0.05 No
PGSI-3, at 12 hyp. C.a: TAU vs MI+W+B -0.45 1.02 0.33 No
months hyp. C.b: MI vs MI+W+B -1.46 1.05 0.08 No

No statistically significant differences were noted for hypotheses B and C in regard to
dichotomised PGSI scores (≤ 17 or > 17) either for PGSI in a past 12-month time frame at the
12-month assessment, or for PGSI in a past three-month time frame time-averaged or at the
12-month assessment (Table 42).
78
Table 42: Hypotheses B and C - dichotomised PGSI
TEST Odds 95% CI 95% CI P-value Alternative
ratio Lower Upper (two- accepted
sided)
PGSI-12 hyp. B.a: TAU vs MI+W 1.13 0.58 2.20 0.72 No
dichotomised, at hyp. B.b: MI vs MI+W 1.41 0.71 2.79 0.32 No
12 months
hyp. B.c: TAU vs 1.29 0.65 2.56 0.47 No
MI+W+B
hyp. B.d: MI vs 1.61 0.80 3.25 0.18 No
MI+W+B
hyp. C.a: TAU vs 1.29 0.65 2.56 0.47 No
MI+W+B
hyp. C.b: MI vs 1.61 0.80 3.25 0.18 No
MI+W+B
hyp. C.c: MI+W vs 1.14 0.58 2.26 0.70 No
MI+W+B
PGSI-3 hyp. B.a: TAU vs MI+W 0.85 0.46 1.60 0.62 No
dichotomised, hyp. B.b: MI vs MI+W 0.77 0.40 1.48 0.44 No
time-averaged
hyp. B.c: TAU vs 0.89 0.47 1.68 0.71 No
MI+W+B
hyp. B.d: MI vs 0.80 0.42 1.55 0.51 No
MI+W+B
PGSI-3 hyp. C.a: TAU vs 0.85 0.27 2.68 0.78 No
dichotomised, at MI+W+B
12 months hyp. C.b: MI vs 0.90 0.27 2.97 0.87 No
MI+W+B
hyp. C.c: MI+W vs 1.13 0.35 3.62 0.84 No
MI+W+B
4.6.3 Motivation to overcome gambling problem
At each assessment, participants were asked how motivated they were to overcome their
gambling problem. Responses were reported on a scale of 0 to 10 where 0 = ‘not at all’ and
10 = ‘extremely’.
Superiority hypotheses B and C

motivation to overcome gambling problem, time-averaged or at the 12-month assessment
(Appendix 6, Table 6.1)
4.6.4 Control over gambling

The MI+W and MI+W+B groups showed statistically significant (p=0.016 and p=0.009
respectively) greater improvement than the MI group in relation to time-averaged control over
gambling. No statistically significant differences were noted for hypothesis C when
examined by control over gambling at the 12-month assessment (Table 43).
79
Table 43: Hypotheses B and C - Control over gambling
sided)
Control over hyp. B.a: TAU vs MI+W 0.42 0.33 0.10 No
gambling, hyp. B.b: MI vs MI+W 0.74 0.34 0.016 Yes
time-averaged
hyp. B.d: MI vs MI+W+B 0.83 0.35 0.009 Yes
Control over hyp. C.a: TAU vs MI+W+B 0.25 0.43 0.28 No
gambling, at 12 hyp. C.b: MI vs MI+W+B 0.62 0.45 0.08 No
months
4.6.5 Psychological distress, alcohol abuse/dependence, drug abuse, quality of life and
deprivation index

psychological distress (measured by Kessler-10), alcohol abuse/dependence (AUDIT-C) or
quality of life (WHOQoL) scores time-averaged or at the 12-month assessment. Similarly, no
statistically significant differences were noted for hypotheses B and C in regard to drug abuse
(DAST) and deprivation index (NZDI) scores at the 12-month assessment (Appendix 6,
Table 6.2).
4.6.6 Mental disorders mood module

No statistically significant differences were noted for hypotheses B and C in regard to any of
the PRIME-MD mood modules (major depressive disorder, dysthymia, minor depressive
disorder, bipolar disorder) at the 12-month assessment (Appendix 6, Table 6.3).
4.6.7 Tobacco use

currently not smoking tobacco or decreasing smoking frequency, both time-averaged and at
the 12-month assessment (Appendix 6, Table 6.4).
4.6.8 Treatment for co-existing issues
At the 12-month assessment participants were asked if, in the previous 12-months, they had
received any treatment for a mental health issue (other than gambling), if they had been
prescribed medication for an emotional, nervous or mental health issue or if they had received
treatment for an alcohol or drug problem.
80
treatment for co-existing issues at the 12-month assessment (Appendix 6, Table 6.5).
4.6.9 Gambling impacts
At each assessment, participants were asked if their gambling, in the past month, had affected
their work, social life, family life/home responsibilities or physical health. Responses were
reported on a scale of 0 to 10 where 0 = ‘not at all’ and 10 = ‘very severely’.

gambling impacts time-averaged or at the 12-month assessment (Appendix 6, Table 6.6).
4.6.10 Legal problems
At each assessment, participants were asked if they had experienced any legal problems (in
the past three or 12-months) as a result of their gambling.

No statistically significant differences were noted for hypotheses B and C in regard to legal
problems time-averaged or at the 12-month assessment (Appendix 6, Table 6.7).
4.6.11 Workbook and other formal treatment engagement
As detailed previously in section 4.2.8, as part of their intervention, participants in the MI+W
and MI+W+B groups were sent, by post, a self-help workbook (‘Becoming a Winner:
Defeating Problem Gambling’) which was the workbook referred to in the follow-up
assessment interviews. Participants in the TAU and MI groups were not sent any workbook.
The number of participants who reported receiving the workbook (‘Becoming a Winner’) is
detailed in Table 44.
Table 44: Number of participants reporting receiving workbook

Time point
3 months 46 17 85 78
6 months 45 20 75 72
12 months 49 26 70 67
81
F.
followed by the non-booster experimental condition (MI+W).
b. The level of engagement will be higher in the non-booster experimental condition
(MI+W) then in the standard treatment group (TAU).
G.
a. Use of, and degree of, engagement in other treatment services will be significantly
lower (and higher for engagement in workbook) in the two conditions involving
motivational interviewing and workbooks (MI+W and MI+W+B) than in the standard
(TAU) and motivational interview (MI) groups.
b. This difference is expected to be greatest during the first three months.
There were no statistically significant differences for hypotheses F and G in relation to

reading the workbook, although a level of statistical significance was only just missed for
participants in the MI+W and MI+W+B groups in comparison with the TAU and MI groups.
There were also no statistically significant differences noted for using the strategies in the
workbook. However, in relation to completing some or all of the exercises in the workbook
(time-averaged), a level of statistical significance was attained for the MI+W group in relation
to the TAU group (p=0.0002), and for the MI+W and MI+W+B groups in comparison with
the TAU and MI groups (p=0.008) at the three-month assessment (Table 45).
There were no statistically significant differences for hypotheses F and G in relation to

participant engagement in formal (professional) treatment services (other than the gambling
helpline) for gambling problems (Appendix 6, Table 6.8).
Table 45: Hypotheses F and G - Workbook engagement, time-averaged and at 3-months
ratio Lower Upper (two-sided) accepted
Read workbook, hyp. F.a: MI+W vs 0.94 0.63 1.41 0.78 No
time-averaged MI+W+B
hyp. F.b: TAU vs MI+W 1.51 0.95 2.40 0.08 No
hyp. G.a: TAU/MI vs 2.13 0.99 4.60 0.053 No
MI+W/MI+W+B
Read workbook, hyp. G.b: TAU/MI vs 2.49 0.57 10.84 0.22 No
at 3 months MI+W/MI+W+B
Completed hyp. F.a: MI+W vs 0.85 0.39 1.87 0.68 No
workbook MI+W+B
exercises, time- hyp. F.b: TAU vs MI+W 7.90 2.74 22.82 0.0002 Yes
averaged hyp. G.a: TAU/MI vs 25.85 4.55 146.80 0.0003 Yes
MI+W/MI+W+B
Completed hyp. G.b: TAU/MI vs 26.20 2.37 289.20 0.008 Yes
workbook MI+W/MI+W+B
exercises, at 3
months
Used workbook hyp. F.a: MI+W vs 0.62 0.21 1.77 0.37 No
strategies, time- MI+W+B
averaged hyp. F.b: TAU vs MI+W 1.35 0.40 4.61 0.63 No
hyp. G.a: TAU/MI vs 1.04 0.13 8.27 0.97 No
MI+W/MI+W+B
Used workbook hyp. G.b: TAU/MI vs 0.81 0.04 15.38 0.890 No
strategies, at 3 MI+W/MI+W+B
months
82
83
4.7 Collateral assessments
At the three- and 12-month assessments, collateral participants were asked about the
respective gambler’s gambling (days gambled and dollars gambled) over the previous two
months. There was moderate correlation between gambler participants’ self-reports of
gambling and collateral reports (Table 46).
Table 46: Gambler and collateral reports of gambling

Gambler Collateral
N Mean SD N Mean SD N Spearman p-value
Correlation
At 3 Days Most recent 109 2.79 4.24 95 3.83 11.43 94 0.33 0.001
months gambled month
Month prior 109 2.74 4.62 92 3.23 6.02 91 0.41 <0.0001
Dollars Most recent 109 220.04 490.81 94 215.36 589.22 93 0.35 0.001
gambled month
Month prior 108 177.41 416.28 92 317.40 708.95 91 0.53 <0.0001
At 12 Days Most recent 97 2.82 3.87 86 4.53 4.53 84 0.43 <0.0001
Month prior 97 2.68 4.28 86 4.56 4.56 84 0.39 0.0002
Dollars Most recent 97 275.21 605.79 87 262.16 262.16 85 0.47 <0.0001
gambled month
Month prior 97 253.59 558.54 87 306.46 306.46 85 0.39 0.0003
Collateral participants were asked how confident they were in their responses. Those who
reported being ‘fairly’ or ‘extremely’ confident also showed moderate correlation with
gamblers’ self-reports of gambling (Table 47). However, there was substantially less
correlation between the gambler and collateral reports when the collateral participant was ‘not
at all’ or only ‘somewhat’ confident in their responses (Table 48).
Table 47: Gambler and collateral reports of gambling by collateral confidence ‘fairly’ or ‘extremely’
Gambler Collateral
Correlation
Month prior 77 2.69 4.67 73 2.40 4.72 72 0.43 0.0002
gambled month
Month prior 77 188.04 454.09 75 337.03 764.55 74 0.53 <0.0001
At 12 Days Most recent 69 2.97 4.27 70 2.46 4.88 69 0.45 <0.0001
Month prior 69 2.87 4.85 69 2.61 4.86 68 0.41 0.0005
Dollars Most recent 69 255.29 568.30 70 129.59 253.76 69 0.53 <0.0001
gambled month
Month prior 69 231.62 544.60 69 146.16 263.75 68 0.41 0.0006
84
Table 48: Gambler and collateral reports of gambling by collateral confidence ‘not at all’ or ‘somewhat’
Gambler Collateral
Correlation
Month prior 25 2.24 2.73 18 6.33 9.22 18 0.46 0.05
gambled month
Month prior 25 100.40 185.86 17 230.82 381.25 17 0.51 0.04
Month prior 15 2.73 2.19 15 3.60 3.18 14 0.08 0.78
Dollars Most recent 15 293.20 650.30 16 226.25 297.25 15 -0.08 0.78
gambled month
Month prior 15 302.87 637.53 16 339.38 431.87 15 0.12 0.67
85
4.8 Treatment integrity and fidelity
Seventy-four digital recordings of intervention delivery were listened to for treatment

integrity and fidelity purposes. All counsellors and all treatment conditions were represented
in the recordings which were made throughout the recruitment and intervention delivery
period. Additionally, nine booster calls were recorded and listened to.
Adherence to treatment intervention was very good with no difference between shared
elements across the treatments groups and with minimal motivational interviewing elements
in the TAU group (expected), and minimal TAU only elements in the three treatment groups
(expected) (Figure 10).
Figure 10: Treatment adherence
N=74
Inter-rater reliability was very good across the 30 recordings assessed throughout data
collection (Figure 11 and Table 49). Most counsellors and treatment conditions were
represented in the recordings.
Figure 11: Mean elements
86
Table 49: Reliability
Element Pearson r ICC
Shared 0.962 0.959
MI 0.980 0.979
TAU 0.982 0.982
87
6. DISCUSSION
5.1. Limitations
5.1.1. Differential attrition and imbalance
No difference between the proportions in each group having received partial interventions
was found to be significant (smallest unadjusted p-value 0.11). Likewise there was no
significant differential loss to follow-up between the groups overall (p=0.16) nor over time
(p=0.08). There is, thus, no evidence for the presence of a bias in the outcome summary
statistics due to differential attrition.
There was no imbalance found amongst baseline covariates across the groups in the sense
expressed in section 3.9. There is, therefore, no reason to believe that chance confounding
has occurred.
5.1.2. Multiplicity
The method chosen to adjust for multiplicity was to control false discovery rate for families of
hypotheses tested using the same model on the same outcome. This is not an especially
conservative approach, but given that: a) the p-values used in the application of false
discovery rate control were all derived from likelihood ratio-based statistics and were thus
monotonic for the likelihood (and therefore evidentially interpretable to a degree); b) that a
number of hypotheses were excluded from significance, as compared with per comparison
error rate control; and c) that only a modest number of hypotheses were found significant, the
approach displayed face usefulness.
The analysis plan was completed before unblinding of the data and with only the knowledge
supplied by an allocation-blinded analysis of the data collected at three months post-
randomisation. The analyses selected, therefore, cannot introduce bias in the conclusions
reached.
5.1.3. Ascertainment bias
Counsellors were responsible both for delivering the interventions and carrying out the initial
assessment. As the initial assessment was conducted prior to intervention allocation the
counsellor had no knowledge of intervention group at the time of initial assessment data
collection and thus ascertainment bias is unlikely. In the course of the analyses, we found that
the random effects corresponding to counsellors in the models for the primary outcomes and
PGSI at 12 months, did not reach significance. This lack of significance is consistent with
homogeneity of bias across the counsellors and intervention groups. In particular it is
consistent with, though it does not entail, a bias of 0. Participants and research assistants
conducting the follow-up assessments were blind to treatment intervention and again
ascertainment bias is unlikely.
88
5.1.4. Selection bias
As counsellors did not activate the intervention random allocation computer programme until
after eligible clients had agreed to participate in the trial and had responded to the initial
assessment questions, selection bias is unlikely to have occurred.
5.1.5. Post-intervention baseline assessment
As detailed in section 3.5, a few baseline assessment questions were asked of participants
within seven days of receiving their intervention. These questions related to a detailed time
line follow-back of gambling/problem gambling history over the previous two months,
comorbidity and substance abuse (measured by PRIME-MD), and the New Zealand Index of
Socio-economic Deprivation for Individuals. This may have biased the affected ANCOVA-
derived estimates toward the null by causing nominal baseline values to align more closely
with their values at three-, six- and 12 months than a pre-intervention baseline value might
have.
5.2. Discussion and interpretation
5.2.1 Scene setting
The major purpose of the present study is to examine the effectiveness of three brief
telephone interventions relative to standard helpline treatment (TAU) and to compare their
performance relative to each other. One of the interventions (motivational interview and
workbook; MI+W) had been shown to produce clinically significant outcomes in two
previous efficacy studies (Hodgins et al., 2001; 2004; 2009). Both studies were conducted by
teams that included one of the developers of these interventions (Hodgins) and involved
volunteers recruited through the mass media and other avenues rather than people seeking
help from problem gambling or other clinical services. The present study differed from these
earlier trials in that it involved embedding this and related brief interventions within the
everyday operations of a national gambling helpline. Hodgins et al. (2009) expressed the
view that these interventions would fit “very well with the existing helplines that provide
information and support to pathological gamblers in most Canadian provinces, U.S. states,
and elsewhere”. The study was developed to examine the feasibility of integrating these brief
therapies into a helpline service and to evaluate them in this context. To our knowledge this is
the first time that a manualised psychological intervention for problem gambling, previously
assessed in two efficacy trials, has been formally and independently evaluated in a real-life
service setting. The study is also distinctive in that relative to previous gambling trials it
included a large number of participants, enabling potential subgroup differences in treatment
response to be assessed.
As mentioned in the literature review, in the initial efficacy trial (Hodgins et al., 2001) a
cognitive behavioural self-help workbook (Hodgins & Makarchuk, 1997) provided with or
without a motivational enhancement interview was compared with a wait-list control group.
At one month follow-up, participants who received the interview and the workbook (MI+W)
substantially reduced their frequency of gambling participation and expenditure. These
reductions were statistically and clinically significant. Less substantial reductions were also
found both in the group that received the workbook only and the wait-list control group. For
ethical reasons the wait-list group was terminated at one month and participants were given
the option of receiving an intervention. While at three and six month follow-ups MI+W
participants continued to show less frequent gambling and lower expenditure than those in the
89
workbook-only group, at 12 months there were no significant differences between the two
groups. However, when participants were followed up at 24 months, group differences were
again evident (Hodgins et al., 2004). Although participants in both groups generally
maintained therapeutic gains (overall 77% improved, 37% reported six months abstinence and
55% were below the SOGS-R cut-off for pathological gambling) MI+W participants gambled
significantly less often, lost less money, had lower SOGS-R scores and were more likely to be
categorised as improved.
In the second trial (Hodgins et al., 2009), in addition to the MI+W intervention, participants
were also allocated to a group that received this intervention plus six follow-up motivational
booster sessions spread over a nine-month period. As in the earlier study, there were also
workbook only and wait-list (six week) control groups. Again, at the time the wait-list group
was terminated, MI+W participants (in both the original and booster conditions) reduced their
gambling frequency and losses more than those in the control groups. They were also
significantly more likely to be abstinent or improved at six weeks than were wait-list and
workbook only participants. As hypothesised, participants in the MI+W groups gambled
significantly less often than the workbook only participants during the first nine months of the
trial. However, contrary to expectation, workbook only participants were as likely as other
participants to have significantly reduced their gambling losses over the year of the follow-up,
to be abstinent, and to not meet the criteria for pathological gambling. The investigators had
hypothesised that the addition of six booster phone calls would help motivate or maintain
changes in gambling behaviour. Although participants who received booster calls reported
slightly higher self-efficacy ratings and generally reported calls as helpful and wanted more,
there were no significant differences in gambling outcomes between booster and non-booster
participants.
In addition to examining the effectiveness of the interventions used in the previous efficacy
trials (MI+W with and without booster sessions) the present study included a single
motivational interview without the workbook (MI). This was added to see whether it was the
motivational interview per se rather than the combination of interview and workbook that was
responsible for the treatment effect. As indicated previously in this report, in the wider
addictions field, as well as from a small number of gambling studies, it is apparent that ‘more’
treatment is not necessarily ‘better’. While some studies indicate that many people benefit
equally well from brief or longer interventions, research suggests that other people benefit
more from longer, more intensive interventions. The latter may include those with more
serious problems and/or comorbid disorders. In the present trial, further to seeking to
determine how well each of the brief interventions performed relative to each other and to
helpline standard care, there was an interest in identifying groups of people that do better with
different types and levels of intervention. This is of particular importance to the development
of evidence-based stepped-care treatment models that match clients to treatments and engage,
in a cost-effective manner, the wide spectrum of problem gamblers, including the majority
that do not currently access specialist problem gambling services. Little is known about this
topic with respect to gambling. Given that problem gambling is highly comorbid with a range
of other addictions and mental health disorders, there was particular interest in seeing whether
the presence of substance misuse and mental health disorders compromised response to
treatment and whether or not reduced gambling problems are associated with improved
mental health status. A further object of the present study is to determine how well the
standard helpline interventions, which have not been evaluated relative to natural or self-
recovery or to other interventions, perform relative to the brief interventions included in the
trial that has been compared with wait-list and placebo control groups.
90
5.2.2 Study interpretation
While not without initial challenges, the large majority of helpline counsellors were
successfully trained to deliver motivational interviews and conduct follow-up booster sessions
as well as deliver standard (TAU) helpline interviews in a consistent manner. Prior to the trial
most had limited or no experience of motivational interviewing or its application to encourage
the use of workbook cognitive behavioural interventions. Following training, the new
counselling approach and other trial procedures became integrated into the operations of the
helpline. Over a third of potentially eligible clients were recruited into the trial and
counsellors delivered the interventions with a very high level of integrity, i.e. the four
interventions were for the most part delivered as intended. Treatment adherence was high, the
counsellors were proficient in delivering both standard care and the new interventions and the
standard and new interventions were differentiated in the ways intended. The great majority
of the 462 clients recruited into the trial and randomised to the four groups received the
applicable interview intervention and, where appropriate, the workbook or helpline manual.
Delivery of the four booster sessions to MI+W+B participants, however, was partial with only
a third receiving all four and 14% receiving none. The involvement of Hodgins in the
counsellor training and integrity assessment helped ensure that the MI, MI+W and MI+W+B
interventions corresponded to those that had been included in the previous efficacy trials.
Some of the major study hypotheses were corroborated, others not. It was predicted that the
MI and TAU participants would show similar improvements on the three primary measures
and that there would be a variety of outcome differences between these participants and
participants in the other, more intensive, groups (MI+W; MI+W+B). While there were no
significant differences between MI and TAU participants on the three primary measures, the
most notable study finding was that participants in all four intervention groups evidenced
statistically and clinically significant and sustained improvement on the three primary
variables (days gambled, money lost gambling and having quit gambling or improved control
over gambling) and, that contrary to expectation, no group was superior to any of the others.
This applied both when performance on these measures was time-averaged across the
duration of the trial and when considered at 12 months. This was also the case for problem
gambling severity as measured by PGSI and a number of other outcome measures including
participant self-ratings of control over gambling, gambling impacts on work, social life,
family and home and health, psychological distress (Kessler-10) and quality of life
(WHOQoL). There was also a substantial reduction in participants in each group meeting the
criteria for major and minor depressive disorder and dysthymia. These disorders were
assessed at intake into the trial and at 12 months. In contrast to these significant
improvements little or no change was evident in any of the groups with respect to alcohol
misuse (AUDIT-C and PRIME-MD) and tobacco use. Overall more than half of participants
reported that their treatment goals had been completely or mostly met at three months and
slightly more reported similarly at 12 months. At intake almost all participants (95% to 97%
across the groups) met PGSI criteria for problem gambling. This reduced to 65% to 67%
across the groups at the 12-month assessment.
Thus it can be concluded, as proposed by Hodgins et al. (2009), that these brief interventions
can be integrated into the routine operations of an existing helpline service. Furthermore, it is
evident that these interventions are effective, producing clinically significant outcomes in a
help-seeking population. With regard to the primary outcome measures and problem
gambling severity (PGSI) the effects appear to be comparable to those achieved in the earlier
efficacy studies that involved volunteers who responded to advertisements to participate.
While not involving a population seeking formal help from an existing problem gambling
service, these volunteers did have substantial gambling and other mental health problems.
While it appears that they were similar to client groups in this regard, the study authors
91
indicated that they differed in that most were not interested in formal treatment and wanted to
overcome their problems in other ways.
It was hypothesised that the more intensive MI+W+B intervention would have a greater
impact than the other three interventions during the latter part of the trial. As predicted,
between the three- and 12-month assessment points, the MI+W+B group showed greater
improvement than the MI+W group in relation to reporting having quit gambling or improved
control over gambling. Participants in this group did not, however, evidence more
improvement on this measure than those in the MI or TAU groups. During this time period,
the MI+W and MI+W+B participants reported experiencing greater time-averaged control
over gambling than MI participants. They did not, however, do better in this regard than TAU
participants. Outcome differences between groups were not found for days gambled, money
lost or problem gambling severity (time-averaged PGSI). Increased workbook involvement
and receiving informal assistance for gambling problems were found to be associated with
better outcomes (the former in regard to time-averaged money lost and time-averaged quit
gambling or improved control over gambling; the latter in regard to time-averaged control
over gambling and time-averaged goal being met).
Although there were no significant primary outcome differences between participants in each
of the four treatment groups overall, differences were found for a number of subgroups.
Males in the MI+W group (but not MI+W+B group) showed significantly more improvement
than males in the MI group with respect to time-averaged money lost gambling. Maori
participants in the MI+W+B group performed significantly better than Maori in the MI group
with respect to money reported lost gambling at the 12 month assessment. On this measure,
participants in MI+W+B group with more serious gambling problems also had better
outcomes than their counterparts in the MI group. Significantly more participants with more
serious problems in this group than in the MI and TAU group also reported that they had
stopped gambling or improved at 12 months. People in the MI+W+B group with higher
levels of psychological disorder (Kessler-10) had better outcomes with respect to both money
lost and reporting having quit gambling or improved control over gambling than their MI
counterparts. There was, however, no difference between the MI+W+B and TAU groups in
this regard. In the case of alcohol misuse (AUDIT-C) MI+W+B participants with lower
problem levels had better Gambling-quit or improved outcomes at 12 months than their
counterparts in the MI group.
Subgroup differences were also found in relation to treatment goal and belief in treatment
success. While the majority of people who entered the trial reported that they wanted to quit
all or some modes of gambling, a significant minority indicated that they sought to decrease
or control their gambling. With respect to time-averaged self-assessment of having quit or
improved control over their gambling, participants who sought to control their gambling did
significantly better in the MI group than they did in TAU. The MI+W+B participants who
sought to control their gambling also did better than their MI counterparts with respect to
money losses at 12 months, number of days gambled and time-averaged assessment of having
quit gambling or improved control over gambling at 12 months. The MI+W participants in
this category also did significantly better than those with this treatment goal in the MI
treatment group. Participants with lower levels of belief in their success in achieving their
treatment goal in 12 months were also found to do significantly worse in the MI group than in
the TAU group with respect to time-averaged self-assessment of having quit gambling or
improved control over gambling and those with lower levels of belief in the MI+W+B group
showed significantly more improvement than their counterparts in the MI group on this
measure.
92
While noting these subgroup differences in response to the trial interventions on some
measures, as indicated, the most important finding is that overall, on the primary outcome
measures and problem gambling severity, all groups evidenced statistically and sustained (to
12 months) clinically significant improvement. Given that TAU performed similarly to two
of the interventions that had previously been shown in efficacy studies to produce
significantly better outcomes than waiting for treatment, it is likely that it would too. All
study participants had sought help because they wanted to stop or reduce gambling and the
large majority had long-standing, serious gambling problems as measured by the PGSI and
the self-reported impacts of gambling on work, social life, family and heath, as well as
significant comorbidity. Most had problems primarily with electronic gaming machines
(EGMs). Approximately a third had previously received assistance for a gambling problem,
17% were receiving assistance at the time they contacted the helpline and 20% had received
treatment for a mental health problem in the past 12 months. At intake over 80% reported
significant psychological distress, over 60% met diagnostic criteria for major or minor
depressive disorder and the same percentage was classified as likely to abuse or be dependent
on alcohol. Over a half smoked tobacco. All four trial interventions can be regarded as brief
or minimal in that they involved a telephone interview, and in some cases receipt of a
workbook with or without follow-up booster phone calls, rather than face-to-face therapy
sessions with a clinician. Participants not only evidenced substantial improvement with
regard to their gambling problems, they improved considerably in a number of other areas
including psychological distress and depression. These findings are consistent with those
from the earlier efficacy studies and problem gambling treatment literature generally. Less
change was evident in the case of tobacco and alcohol use/misuse. Both are highly comorbid
with problem gambling and significant health issues in their own right. Given that they do
not respond, or respond minimally, to the interventions offered, consideration needs to be
given to whether or not they should be addressed via gambling helplines in conjunction with
gambling interventions or responded to in some other way. This could be a focus for future
research.
A growing literature supports the view that motivational interviewing makes a significant
contribution to behaviour change, both on its own and as part of other interventions (Hodgins
et al., 2004; Wulfert et al., 2006). Diskin and Hodgins (2009) found that a motivational
interview had a larger impact on gambling than a non-motivational interview. However, in
the present study it was hypothesised that TAU and MI alone would be equally effective.
This was because both were expected to be of similar length and that while MI on its own was
considered likely to have more impact, TAU was more intensive in that participants also
received an information pack. There was also emphasis on seeking face-to-face and other
forms of support. In the present study, the addition of workbook and workbook and booster
sessions to motivational interviewing was not found to produce superior outcomes for
participants overall, although they did for some categories of participant. This suggests that
for many callers the motivational interview, or other factors common to all four interventions,
were the important ingredients. Hodgins et al. (2009) found that both MI and MI+W+B
participants reported less gambling at six weeks than both those in a wait-list control and
those who received the workbook alone. However, while superiority was demonstrated at six
weeks, as mentioned earlier, 12 months post- treatment the workbook only participants were
just as likely as the MI+W+B group to have reduced their past year gambling losses and no
longer meet the criteria for pathological gambling. All of these findings support the value of
offering brief interventions and suggest that even ‘less’ within already minimal interventions
can produce comparable outcomes to ‘more’. However, it remains unclear what the most
important factors are in producing therapeutic gain.
Why did the four interventions produce similar overall outcomes on a number of measures?
From prospective general population studies it is evident that many people with gambling
93
problems, especially if less severe, overcome them without professional or specialist
intervention (Abbott & Clark, 2007). However, fluctuation over time including relapse is
common, especially when problems are more severe and comorbid with alcohol misuse
disorders (Abbott, Williams & Volberg, 2004). Thus it is likely that a significant proportion
of participants would have improved during the 12 month trial period, whether they
participated or not. In addition, all participants recognised that they had gambling problems
and were seeking help. In this situation it is likely that many people already had, or would
subsequently, take measures, informally and or formally, to stop or reduce their gambling and
make other changes in their lives. The present study did not include a wait-list control group.
The helpline does not have a wait-list and to include one in the trial would have been
unacceptable and unethical. However, as mentioned, the two previous efficacy studies did.
In these studies between 14% and 18% of wait-list controls were abstinent and between 26%
and 45% improved. In the treatment groups approximately an additional ten percent were
abstinent at four or six weeks.
Westphal and Abbott (2006) have also pointed out that previous trials have found high rates of
non-specific or ‘placebo’ response, even higher than those generally in trials involving other
mental health disorders. In other words, many participants seeking to change do well
irrespective of the particular form of intervention offered. Therapist characteristics and the
perceived credibility of the intervention are also important. The challenge is to identify
interventions that add additional value by enhancing outcomes further, either overall, or for
one or more groups of client. As mentioned, the four interventions were differentiated and
delivered as intended, other than there being a short-fall in the number of follow-up booster
sessions in the MI+M+B group. It is possible that increasing the percentage of participants
who received all or most of the four booster sessions would have made a difference.
However, this was not the case in the Hodgins et al. (2009) trial which involved six booster
calls over a longer time period. Hodgins et al. (2009) were of the view that the assessment
process per se may have an impact, for example by increasing caller awareness of the
negative effects of gambling and by serving a “motivational-supportive purpose”. They
noted that more consistent differences were evident over the two years of their earlier trial
that involved much briefer and less frequent assessment. The present study more resembled
that 2009 trial in this regard so it might be that the researcher contact, which was fairly
considerable and consistent across the four interventions, played a role in diminishing
outcome differences. The present design does not allow the impact of these different factors
to be assessed.
The helpline has a role both in providing direct service to callers, by way of receiving initial
calls and providing an interview along the lines of TAU in the present study, sending out a
self-help manual and indicating other options including face-to-face counselling from other
service providers. Callers may also re-contact the helpline for further counselling or support.
As mentioned above, in the present study it was expected that TAU participants would make
greater use of other treatment services than participants who received the workbook and
workbook and booster sessions. It was also considered likely MI participants would seek
other forms of help more frequently because they would not receive the workbook or booster
sessions. It was expected that these differences would be greatest during the first three
months. As it transpired, relatively few participants in any of these conditions re-contacted
the helpline during the trial (1.3% to 7.1% in any three month period). While approximately
20% received some other type of formal assistance for gambling problems, contrary to
prediction there were no differences between the four intervention groups with regard to
engagement in non-helpline formal gambling treatment services. There were also no
differences between groups with respect to participants who reported receiving treatment for
mental health problems, including medication, and treatment for alcohol or drug problems.
Approximately a third of trial participants also indicated that they had received informal
94
support or help for their gambling problems within the past three month period. These
findings underline the reality that people who seek help for gambling problems from
helplines, or other specialist services, concurrently and subsequently seek formal and informal
assistance for their gambling problems as well as other related and non-related mental health
and other problems. This makes it difficult to determine the impact that any particular
intervention makes.
The identification of the major ingredients for effective gambling treatment remains an
important focus for further investigation. The earlier trials also indicated that treatment
effects and differences between groups change over time. Given that problem gambling is
often relapsing in nature, the value of different interventions is best judged, both clinically
and financially, over a longer time frame than most studies employ. In this regard it is
anticipated that participants in the present study will be re-assessed three years after they
entered the study. It may be found that some interventions included in the study have more
enduring effects, at least for some client subgroups.
As discussed earlier the identification of subgroups of problem gamblers who do better with
different forms and intensities of intervention is also an important focus for future research.
The present study sheds some light on this topic, an area that has been little investigated in the
gambling field to date, largely because of the small sample size of previous trials and various
methodological shortcomings indicted in the literature review. Where there were significant
outcome differences between different groups in the various interventions it mostly applied to
differences between the MI (the least intensive) and MI+W+B (most intensive) interventions.
Thus, if MI was the primary intervention being used in a helpline, or perhaps other service
setting, there may be added value in including the option of the workbook and booster
sessions for some clients including Maori, people with more serious gambling problems and
those with higher levels of psychological disturbance. People with more serious gambling
problems also had better outcomes in the MI+W+B condition than in TAU, as did those who
sought to control/reduce their gambling rather than quit. This group also did better in
MI+W+B than in MI. For clients with initially low levels of belief in their success in
achieving their treatment goal it was foundthat MI alone produced worse outcomes than TAU.
While a number of the foregoing differences were only found with respect to one or a small
number of outcome measures, they provide an indication of groups of clients who may do
significantly better with particular types and intensities of intervention. This requires more
focused investigation in future studies that include cost-benefit analysis.
95
7. OTHER INFORMATION
6.1 Registration
The trial is registered with the Australian New Zealand Clinical Trials Registry (ANZCTR),
study registration number: ACTRN12609000560291.
6.2 Protocol
Full details of the trial protocol are maintained by the Gambling and Addictions Research
Centre, National Institute for Public Health and Mental Health Research, School of Public
Health and Psychosocial Studies, Faculty of Health and Environmental Sciences, Auckland
University of Technology, Private Bag 92006, Auckland 1142, New Zealand.
6.3 Funding
The trial was funded by the New Zealand Ministry of Health. The funder had no role in study
design, data collection and analysis, or reporting, although they approved each of those stages
and had the right to suggest changes. Final decision on content was exclusively retained by
the trial investigators.
96
8. REFERENCES
Abbott, M.W. (2001a). What do we know about gambling and problem gambling in New
Zealand? Wellington: Department of Internal Affairs.
Abbott, M.W. (2001b). Problem and non-problem gamblers in New Zealand: A report on
Phase Two of the 1999 National Prevalence Study. Wellington: Department of Internal
Affairs.
Abbott, M.W. (2006). Do EGMs and problem gambling go together like a horse and carriage?
Gambling Research, 18, 7-38.
Abbott, M.W. (2007). Situational factors that affect gambling behaviour. In G. Smith, D.C.
Hodgins, and R.J. Williams (Eds). Research and measurement issues in gambling studies (pp
251-278). Burlington, MA: Academic Press.
Abbott, M.W., & Clarke, D. (2007). Prospective gambling research: Contribution and
potential. International Gambling Studies, 7, 123-144.
Abbott, M.W., & McKenna, B. (2005). Gambling and problem gambling among recently
sentenced women prisoners in New Zealand. Journal of Gambling Studies, 21(4), 537-557.
Abbott, M.W., McKenna, B., & Giles, L. (2005). Gambling and problem gambling among
recently sentenced males in four New Zealand prisons. Journal of Gambling Studies, 21(4),
559-581.
Abbott, M.W. & Volberg, R.A. (1991). Gambling and problem gambling in New Zealand.
Research Series No. 12. Wellington: Department of Internal Affairs.
Abbott, M.W., & Volberg, R. (1992). Frequent and problem gambling in New Zealand.
Research Series No. 14. Wellington: Department of Internal Affairs.
Abbott, M.W., & Volberg, R. A. (1996). The New Zealand National Survey of Problem and
Pathological Gambling. Journal of Gambling Studies, 12, 143-160.
Abbott, M.W., & Volberg, R. (2000). Taking the pulse on gambling and problem gambling in
the community: Phase One of the 1999 National Prevalence Study. Wellington: Department of
Internal Affairs.
Abbott, M.W., & Volberg, R.A. (2006). The measurement of adult problem and pathological
gambling. International Gambling Studies, 6, 175-200.
Abbott, M., Volberg, R., Bellringer, M. & Reith, G. (2004). A review of research on aspects of
problem gambling: Final report. London: Responsibility in Gambling Trust.
Abbott, M.W., Williams, M., & Volberg, R.A. (2004). A prospective study of problem and
regular non-problem gamblers living in the community. Substance Use and Misuse, 39(6),
855-884.
Apodaca, T.R. (2007). A pilot study of bibliotherapy to reduce alcohol problems among
patients in a hospital trauma centre. Journal of Addictions Nursing,18(4), 167-173.
97
Arkowitz, H., Miller, W.R., Westra, H.A., & Rollnick, S. (2007). Motivational interviewing in
the treatment of psychological problems. Conclusions and future directions. In A. Arkowitz,
W.R. Miller, H.A. Westra, and S. Rollnick (Eds.), Motivational interviewing in the treatment
of psychological problems (pp. 324-342). New York: Guildford Press.
Babor, T.F. (1994). Avoiding the horrid and beastly sin of drunkenness: Does discussion make
a difference? Journal of Consulting and Clinical Psychology, 62, 1127-1140.
Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and
powerful approach to multiple testing. Journal of the Royal Statistical Society, Series B, 57
(1): 289–300.
Bennett, D. (2001). How can I deal with missing data in my study? Australian and New
Zealand Journal of Public Health, 25, 464-469.
Bertholet, N., Daeppen, J.B., Wietlisbach, V., Fleming, M, & Burnand, B. (2005). Reduction
of alcohol consumption by brief intervention in primary care: Systematic review and meta-
analysis. Archives of Internal Medicine, 165, 986-995.
Carlbring, P., Jonsson, J., Josephson, H., & Forsberg, L. (2010). Motivational interviewing
versus cognitive behavioral group therapy in the treatment of problem and pathological
gambling: A randomized controlled trial. Cognitive Behavioral Therapy, 39(2), 92-103.
Chambless, D.L., & Ollendick, T.H. (2001). Empirically supported psychological

interventions: controversies and evidence. Annual Review of Psychology, 52, 685-716.
Cocco, K.M., & Carey, K.B. (1998). Psychometric properties of the Drug Abuse Screening
Test in psychiatric outpatients. Psychological Assessment, 10, 408-414.
Cowlishaw, S., Merkouris, S., Dowling, N., Anderson, C., Jackson, A., & Thomas, S. (2012).
Psychological therapies for pathological and problem gambling (Review). The Cochrane
Library, 11. John Wiley & Sons, Ltd.: The Cochrane Collaboration.
Diskin, K.M., & Hodgins, D.C. (2009). A randomized controlled trial of a single session
motivational intervention for concerned gamblers. Behaviour Research and Therapy, 47(5),
382-388.
Dupont, W.D. (2002). Statistical modelling for biomedical researchers. A simple introduction
to the analysis of complex data. Cambridge: Cambridge University Press.
Echeburua, E., Baez, C., & Fernandez-Montalvo, J. (1996). Comparative effectiveness of

three therapeutic modalities in the psychological treatment of pathological gambling.
Behavioural and Cognitive Psychotherapy, 24, 51-72.
Ferris, J., & Wynne, H. (2001). The Canadian Problem Gambling Index: Final report.
Ottawa: Canadian Centre on Substance Abuse.
Gambling Helpline. (2012). Report for national statistics to 31 December 2011. Auckland:
Gambling Helpline New Zealand. Retrieved 5 September 2012 from
http://www.health.govt.nz/our-work/preventative-health-wellness/problem-gambling/service-
user-data/gambling-helpline-client-data.
98
Gerstein, D., Volberg, R.A., Murphy, S., Toce, M. et al. (1999). Gambling impact and
behavior study. Report to the National Gambling Impact Study Commission. Chicago:
National Opinion Research Center at the University of Chicago.
Gooding, P., & Tarrier, N. (2009). A systematic review and meta-analysis of cognitive-
behavioural interventions to reduce problem gambling: Hedging our bets? Behaviour
Research and Therapy, 47(7), 592-607.
Gould, R., & Clum, G. (1993). A meta-analysis of self-help treatment approaches. Clinical
Psychology Review, 13, 169-178.
Grinols, E.L. (2007). Social and economic impacts of gambling. In G. Smith, D.C. Hodgins,
and R.J. Williams (Eds). Research and measurement issues in gambling studies (pp 515-539),
Burlington, MA: Academic Press.
Harel, O., & Zhou, X. (2007). Multiple imputation: review of theory, implementation and
software. Statistics in Medicine, 26(16), 3057-3077.
Hodgins, D., Currie, S., Currie, G., & Fick, G. (2009). Randomized trial of brief motivational
treatments for pathological gamblers: More is not necessarily better. Journal of Consulting
and Clinical Psychology, 77(5), 950-960.
Hodgins, D.C., Currie, S.R., & el-Guebaly. N. (2001). Motivational enhancement and self-
help treatments for problem gambling. Journal of Consulting and Clinical Psychology,
69(1), 50-57.
Hodgins, D.C., Currie, S., el-Guebaly, N., & Peden, N. (2004). Brief motivational treatment
for problem gambling: a 24-month follow-up. Psychology of Addictive Behaviors, 18(3),
293-296.
Hodgins, D.C., & Holub, A. (2007). Treatment of problem gambling. In G. Smith, D.C.
Hodgins, and R.J. Williams (Eds). Research and measurement issues in gambling studies (pp
237-297). Burlington, MA: Academic Press.
Hodgins, D.C., & Makarchuk, K. (1997). Becoming a winner: Defeating problem gambling.
Alberta: University of Calgary Press.
Hodgins, D.C., Peden, N., & Makarchuk, K. (2004) Self-efficacy in pathological gambling
treatment outcome: Development of a gambling abstinence self-efficacy scale (GASS).
International Gambling Studies, 4, 99-108.
Johnson, E.E., Hamer, R., & Nora, R.M. (1997). The lie-bet questionnaire for screening
pathological gamblers. Psychological Reports, 80, 83-88.
Kessler, R., & Mroczek, D. (1994). Final versions of our Non-Specific Psychological Distress
Scale. Written communication-memo dated 10/3/94. Ann Arbour (MI), Surrey Research
Center of the Institute for Social Research, University of Michigan.
Kobak, K.A., Taylor, L.H., Dottl, S.L., Greist, J.H., Jefferson, J.W., Burroughs, D., Mantle,
J.M., Katzelnick, D.J., Norton, R., Henk, H.J., & Serlin, R.C. (1997). A computer-
administered telephone interview to identify mental disorders. Journal of the American
Medical Association, 278, 905-910.
99
Kristman, V., Manno, M., & Cote, P. (2005). Methods to account for attrition in longitudinal
data: do they work? A simulation study. European Journal of Epidemiology, 20(8), 657-662.
LaBrie, R.A., Peller, A.J., LaPlante, D.A., Bernhard, B., Harper, A., Schrier, T., & Shaffer, H.
J. (2012). A brief self-help toolkit intervention for gambling problems: A randomized
multisite trial. American Journal of Orthopsychiatry, 82(2), 278-289.
Ladouceur, R., Sylvain, C., Boutin, C., Lachance, S., Doucet, C., Leblond, J., & Jacques, C.
(2001). Cognitive treatment of pathological gambling. Journal of Nervous and Mental
Disease, 189, 74-80.
Larimer, M.E., Neighbors, C., Lostutter, T.W., Whiteside, U., Cronce, J.M., Kaysen, D., &
Walker, D.D. (2012). Brief motivational feedback and cognitive behavioral interventions for
prevention of disordered gambling: A randomized clinical trial. Addiction, 107(6), 1148-1158.
Lee, J-H., Herzog, T.A., Meade, C.D., Webb, M.S., & Brandon T.H. (2007). The use of GEE
for analyzing longitudinal binomial data: A primer using data from a tobacco intervention.
Addictive Behaviors, 32, 187-193.
Ministry of Health. (2004). Problem gambling factsheet. Wellington: Ministry of Health.
Ministry of Health. (2005). Preventing and minimising gambling harm: Strategic plan 2004-
2010. Wellington: Ministry of Health.
Ministry of Health. (2006). Problem gambling in New Zealand: Analysis of the 2002/03 New
Zealand Health Survey. Wellington: Ministry of Health.
Ministry of Health. (2007a). Preventing and minimising gambling harm: Three-year service
plan 2007-2010. Wellington: Ministry of Health.
Ministry of Health. (2007b). Problem gambling intervention services in New Zealand: 2006
Service-user statistics. Wellington: Ministry of Health.
Ministry of Health. (2008). A portrait of health. Key results of the 2006/07 New Zealand
Health Survey. Wellington: Ministry of Health.
Ministry of Health. (2010). Preventing and Minimising Gambling Harm: Three-year service
plan 2010/11-2012/13. Wellington: Ministry of Health.
Ministry of Health. (2012). Intervention Client Data. Retrieved 26 June, 2012, from
http://www.health.govt.nz/our-work/preventative-health-wellness/problem-gambling/service-
user-data/intervention-client-data
Moyers, T.B., Martin, T., Manuel, J.K., Hendrickson, S.M.L., & Miller, W.R. (2004).
Assessing competence in the use of motivational interviewing. Journal of Substance Abuse
Treatment, 28(1), 19-26.
Pallesen, S., Mitsem, M., Kvale, G., Johnsen, B.H., & Molde, H. (2005). Outcome of
psychological treatments of pathological gambling: A review and meta-analysis. Addiction,
100, 1412-1422.
Petry, N. (2005). Pathological gambling: Etiology, comorbidity, and treatments. Washington,

DC: American Psychological Association.
100
Petry, N.M., Ammerman, Y., Bohl, J., Doersch, A., Gay, H., Kadden, R., Molina, C., &
Steinberg, K. (2006). Cognitive-behavioural therapy for pathological gamblers. Journal of
Consulting and Clinical Psychology, 74, 555-567.
Petry, N.M., & Weinstock, J. (2007). Comorbidity and mental illness. In G. Smith,
D.C.Hodgins, and R.J. Williams (Eds). Research and measurement issues in gambling Studies
(pp 305-322). Burlington, MA: Academic Press.
Petry, N.M., Weinstock, J., Ledgerwood, D.M., & Morasco, B. (2008). A randomized trial of
brief interventions for problem and pathological gamblers. Journal of Consulting and
Clinical Psychology, 76, 318-328.
Petry, N., Weinstock, J., Morasco, B., & Ledgerwood, D. (2009). Brief motivational
interventions for college student problem gamblers. Addiction, 104(9), 1569-1578.
Petry, N.M., & Weinstock, J. (2007). Comorbidity and mental illness. In D.C.H.G. Smith, and
R.J. Williams (Ed.), Research and measurement issues in gambling studies (pp. 305-322):
Burlington, MA: Academic Press.
Problem Gambling Research and Treatment Centre. (2011). Guideline for screening,
assessment and treatment in problem gambling. Clayton: Monash University.
Productivity Commission. (1999). Australia’s Gambling Industries, Report No. 10. Canberra:
AusInfo.
Productivity Commission. (2010). Gambling, Report No. 50. Canberra.
Reinert, D.F., & Allen, J.P. (2002). The Alcohol Use Disorders Identification Test (AUDIT): A
review of recent research. Alcoholosim: Clinical and Experimental Research, 26(2), 272-279.
Salmond, G. (2005). Cited in Ministry of Health (2005). The New Zealand Health Survey
2002/03. Wellington: Author.
Saunders, J.B., Aasland, O.G., Babor, T.F., De La Fuente, J.R., & Grant, M. (1993).
Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO collaborative
project on early detection of persons with harmful alcohol consumption–II. Addiction 88,791–
804.
Schmidt, S., Muhlan, H., & Power, M. (2005). The EUROHIS-QOL 8-item index:
Psychometric results of a cross-cultural field study. European Journal of Public Health,
16(4), 420-428.
Skinner, H.A. (1982). The Drug Abuse Screening Test. Addictive Behaviors, 7, 363-371.
Sobell, L.C., & Sobell, M.B. (1992). Time-line follow-back: A technique for assessing self-
reported alcohol consumption. In R.Z. Litten and J.P. Allen (Eds.), Measuring alcohol
consumption: Psychological and bio-chemical methods (pp. 41-72). Totowa, NJ: Humana
Press.
Spitzer, R.L., Williams, J.B.W., Gibbon, M., & First, M.B. (1992). The Structured Clinical
Interview for DSM-III-R (SCID), I: History, rationale, and description. Archives of General
Psychiatry, 49, 624-629.
101
Spitzer, R.L., Williams, J.B., Kroenke, K., Linzer, M., deGruy, F.V., 3rd, Hahn, S.R., Brody,
D., & Johnson, J.G. (1994). Utility of a new procedure for diagnosing mental disorders in
primary care. The PRIME-MD 1000 study, Journal of American Medical Association,
272(22), 1749-56.
Toneatto, T., & Ladouceur, R. (2003). Treatment of pathological gambling: A critical review
of the literature. Psychology of Addictive Behaviors, 42, 92-99.
Toneatto, T., & Millar, G. (2004). Assessing and treating problem gambling: Empirical status
and promising trends. Canadian Journal of Psychiatry, 49(8), 517-525.
Twisk, J. (2003). Applied longitudinal data analysis for epidemiology. A practical guide.
Cambridge: Cambridge University Press.
Volberg, R.A., & Abbott, M.W. (1997). Gambling and problem gambling among indigenous
peoples. Journal of Substance Use and Misuse, 32, 1525-1538.
Walker, M., Toneatto, T., Potenza, M.N., Petry, N., Ladouceur, R., Hodgins, D.C., el-Guebaly,
N., Echeburua, E., & Blaszczynski, A. (2006). A framework for reporting outcomes in
problem gambling treatment research: The Banff, Alberta Consensus. Addiction, 101, 504-
511.
Westphal, J.R., & Abbott, M.W. (2006). Models for multi-site problem gambling clinical
trials. International Gambling Studies, 6(2), 129-145.
Wulfert, E., Blanchard, E.B., Freidenberg, B., & Martell, R. (2006). Retaining pathological
gamblers in cognitive-behavioral therapy through motivational enhancement. Behavior
Modification, 30, 315-340.
102
APPENDIX 1
Ethical approval
103
104
APPENDIX 2
Trial hypotheses notation
The outcome as a function of time t from baseline assessment is denoted Y(t). The subscript
g =0,1,2,3 denotes the TAU, MI, MI+W and MI+W+B groups; the subscript j=0,1,2,3 denotes
the baseline, three-month, six-month and 12-month time points 7; the subscript k=1,…,Kg
identifies the participant in group g, that has size Kg. The quantities tkgj are the actual
assessment times elapsed for participant k in arm g since baseline assessment. We define Y kgj
= Y(tkgj), with common expectation μgj. A single subscript present refers to the treatment arm
and indicates that a common mean for the outcome involved is posited across the three non-
baseline time points.
Averaging over post-randomisation time points, or time-averaging, is represented by “,●”.

(Time-averaging was implemented indirectly through use of suitably weighted repeated
measures models rather than the computation of time-averaged outcomes).
Averaging over groups is represented by “●,”.
Averaging of a parameter indicates that it is assumed in the hypothesis concerned to be equal

over the distributions being averaged. It does not necessarily indicate that a composite
outcome is being computed, since the averaging can be effected in a repeated measures
setting with appropriate re-parameterisation.
Parameters identified by subscripted letters are understood to represent the effect of the level
in the category identified by the subscript. Thus μg,● is the true mean time-averaged outcome
associated with the gth treatment group.
Covariates entering hypotheses are expressed as continuous covariates for simplicity,

although they may in fact be categorical.
Interactions are indicated by a colon (:).
Note that baseline true means are assumed to be equal (μ g,0 = μ●.0) and so are not included in
several of the hypotheses.
The hypotheses are not fully detailed below. They detail, often implicitly, what underlying
distribution and other adjustments may enter into defining the hypotheses. Whatever these
additional aspects are, they appear in the null and alternative hypotheses simultaneously.
Efficacy hypotheses
With Yg,j,k an efficacy outcome, the basic model is either E(Y g,j,k)= μg,●, or E(Yg,j,k)= μg,j, where
E(.) represents expectation. Adjustments may be added to these models; adjustments for
baseline, when available, are systematic in continuous outcomes analysis: the interpretation of
the treatment effects are, therefore, as an average change from baseline in these cases. In
some cases (e.g. gambling-quit or improved), a link function may have been used and/or the
inequalities presented may be reversed to correctly reflect superiority.
7
Note that Ykg0 is treated as an independent covariate, systematically included in all continuous variable
models when available. Thus the subscript j=0 does not enter hypotheses below.
105
Primary equivalence hypothesis
B. The Motivational Interview (MI) group will show similar improvement to Treatment
as Usual (TAU).
HA0: |μ1,● –μ0,●|≤δ vs. HA1: |μ1,●-μ0,●|>δ
where δ is a present equivalence threshold for each outcome.
Primary superiority hypotheses

B. (Superiority is associated to lower values in the generic hypothesis statements below)
a. The Motivational Interview/Workbook group (MI+W) will show greater
HBa0: μ2,● ≥ μ0,● vs. HBa1: μ2,● < μ0,●
HBb0: μ2,● ≥ μ1,● vs. HBb1: μ2,● < μ1,●
c. The Motivational Interview plus Workbook plus Booster group (MI+W+B) will
show greater improvement than the TAU group.
HBc0: μ3,● ≥ μ0,● vs. HBc1: μ3,● < μ0,●
d. The Motivational Interview plus Workbook plus Booster group (MI+W+B)will
show greater improvement than the MI group.
HBd0: μ3,● ≥ μ1,● vs. HBd1: μ3,● < μ1,●
C. (Superiority is associated to lower values in the hypothesis statements below)

a. The MI+W+B group will show greater improvement than the TAU group at the
12-month follow-up.
HCa0: μ3,3 ≥ μ0,3 vs. HCa1: μ3,3 < μ0,3
b. The MI+W+B group will show greater improvement than the MI group at the 12-
month follow-up.
HCb0: μ3,3 ≥ μ1,3 vs. HCb1: μ3,3 < μ1,3
c. The MI+W+B group will show greater improvement than the MI+W group at the
12-month follow-up.
HCc0: μ3,3 ≥ μ2,3 vs. HCc1: μ3,3 < μ2,3

C*. (Superiority is associated to lower values in the hypothesis statements below)
3 and 12 months.
HCd0: μ3,3 – μ3,1 ≥ μ0,3 - μ0,1 vs. HCd1: μ3,3 – μ3,1 < μ0,3 - μ0,1
e. The MI+W+B group will show greater improvement than the MI group between 3
and 12 months.
HCe0: μ3,3 – μ3,1 ≥ μ1,3 – μ1,1 vs. HCe1: μ3,3 – μ3,1 < μ1,3 – μ1,1
f. The MI+W+B group will show greater improvement than the MI+W group
between 3 and 12 months.
HCf0: μ3,3 – μ3,1 ≥ μ1,3 – μ1,1 vs. HCf1: μ3,3 – μ3,1 < μ1,3 – μ1,1
D. (Superiority is associated to lower values in the generic hypothesis statements below)

HDa0: μ0,● ≥ μ0,0 vs. HDa1: μ0,● < μ0,0
HDb0: μ1,● ≥ μ1,0 vs. HDb1: μ1,● < μ1,0
106
HDc0: μ2,● ≥ μ2,0 vs. HDc1: μ2,● < μ2,0
HDd0: μ3,● ≥ μ3,0 vs. HDd1: μ3,● < μ3,0
E.
(Superiority is associated to lower values in the generic hypothesis statement below)
In the model, E(Yg,j,k)= μg,j +α Ag,j,k, where Ag,k represents the level of engagement of
participant k in group g=0,2,3, to which other adjustments may be added:
HE0: α ≥ 0 vs. HE1: α < 0.

F.
With Vg,j,k the level of engagement in other treatment services, the basic model is E(V g,j,k)= νg,j,
to which other adjustments may be brought.
followed by the non-‘booster’ experimental condition (MI+W).
HFa0: ν3,● u≤ ν2,● vs. HFa1: ν3,● > ν2,●
b. The level of engagement will be higher in the non-‘booster’ experimental
condition (MI+W) then in the standard treatment group (TAU).
HFb0: ν2,● ≤ ν0,● vs. HFb1: ν2,● > ν0,●
(Note that the third alternative, ν 3,● > ν0,● is not considered here, as the group
sizes are comparable and transitiveness is almost guaranteed, not warranting
family-wise error rate adjustment.)
G.
Use of, and degree of, engagement in other treatment services will be significantly lower in
the two conditions involving motivational interviewing and workbooks (MI+W; MI+W+B)
than in the standard (TAU) and motivational interview (MI) groups. This difference is
expected to be greatest during the first three months.
With Rg,j,k the level of engagement in other treatment services, the basic model is E(R g,j,k)= μg,j,
to which adjustments may be brought.
a. HGa0: μ3,● + μ2,● ≤ μ1,●+ μ0,● vs. HGa1: μ3,● + μ2,● > μ1,●+ μ0,●
b. HGa0: μ3,1 + μ2,1 ≤ μ1,1+ μ0,1 vs. HGa1: μ3,1 + μ2,1 > μ1,1+ μ0,1
Safety and tolerability hypotheses

None.
107
APPENDIX 3
Summary table of analyses
Code Endpoint Analysis Set Focus Statistical model Hypotheses Comment

ID
Primary analyses
01 Days Gambled, time- I ITT Tx Linear mixed effects, weighted A, δ=1 Report A with 95% CI
averaged Baseline-adjusted Ba,b,c,d FWER adjustment for B
No time point covariate
02 Days Gambled at 12 months II.1 ITT Tx: (Timepoint=12) Linear mixed effects, weighted Ca,b,c FWER adjustment
Baseline-adjusted Null is alternative less only the Tx:
Time point-adjusted (Timepoint=12) interaction
Time point-treatment interaction
03 Money Lost, time-averaged I ITT Tx Linear mixed effects, weighted A, δ=20 FWER adjustment for B
Baseline-adjusted Ba,b,c,d
04 Money Lost at 12 months II.1 ITT Tx: (Timepoint=12) Linear mixed effects, weighted Ca,b,c FWER adjustment
05 Gambling-quit or improved, III ITT Tx Logistic mixed effects, weighted A, δ=0_13 FWER adjustment for B
time-averaged No time point covariate Ba,b,c,d
06 Gambling-quit or improved, IV.1 ITT Tx: (Timepoint=12) Logistic mixed effects, weighted Ca,b,c FWER adjustment
at 12 months Time point-adjusted Null is alternative less only the Tx:
Time point-treatment interaction (Timepoint=12) interaction
Secondary analyses
Equivalence hypothesis for PGSI-12
07 PGSI-12, at 12 months I ITT Tx Linear regression, unweighted A, δ=1 Report A with 95% CI
Baseline-adjusted
Primary outcomes and PGSI-12 and primary hypotheses in the PP analysis set
08 Days Gambled, time- I PP Tx Linear mixed effects, weighted A, δ=1 FWER adjustment for B
averaged Baseline-adjusted Ba,b,c,d
09 Days Gambled at 12 months II.1 PP Tx: (Timepoint=12) Linear mixed effects, weighted Ca,b,c FWER adjustment
108
10 Money Lost, time-averaged I PP Tx Linear mixed effects, weighted A, δ=20 FWER adjustment for B
Baseline-adjusted Ba,b,c,d
11 Money Lost at 12 months II.1 PP Tx:(Timepoint=12) Linear mixed effects, weighted Ca,b,c FWER adjustment
12 Gambling-quit or improved, III PP Tx Logistic mixed effects, weighted A, δ=0_13 FWER adjustment for B
time-averaged No time point covariate Ba,b,c,d
13 Gambling-quit or improved, IV.1 PP Tx: (Timepoint=12) Logistic mixed effects, weighted Ca,b,c FWER adjustment
at 12 months Time point-adjusted Null is alternative less only the Tx:
14 PGSI-12, at 12 months II.2 PP Tx (no repeated Linear mixed effects, counsellor- Ca,b,c FWER adjustment
measures) specific random effects only, if
indicated, unweighted
Baseline-adjusted
Primary outcomes and PGSI-12 within secondary hypotheses in the ITT analysis set
15 Days Gambled, contrast at 3 II.1 ITT Tx:(Timepoint=12) – Linear mixed effects, weighted C*d,e,f FWER adjustment
and 12 months Tx:(Timepoint=3) Baseline-adjusted Null is alternative less the interaction
Time point covariate
16 Money Lost, time-averaged II.1 ITT Tx:(Timepoint=12) – Time point covariate C*d,e,f FWER adjustment
Tx:(Timepoint=3) Time point-treatment interaction Null is alternative less the interaction
17 Gambling-quit or improved, III ITT Tx:(Timepoint=12) – Logistic mixed effects, weighted C*d,e,f FWER adjustment
time-averaged Tx:(Timepoint=3) Time point covariate Null is alternative less the interaction
18 Days Gambled, time- I ITT Tx Linear mixed effects, weighted Da,b,c,d FWER adjustment
averaged minus baseline Baseline-adjusted Baseline subtraction not strictly necessary
No time point covariate but indicated for ease of interpretation
19 Money Lost, time-averaged I ITT Tx Linear mixed effects, weighted Da,b,c,d FWER adjustment
minus baseline Baseline-adjusted Baseline subtraction not strictly necessary
No time point covariate but indicated for ease of interpretation
20 Gambling-quit or improved, III ITT Tx Logistic mixed effects, weighted Da,b,c,d FWER adjustment
time-averaged No time point covariate HDx0: pg,● ≥ 0.05 vs.
HDx1: pg,● < 0.05
21 S1_1_1_PGSI-12, at 12 I ITT Tx Linear regression, unweighted Da,b,c,d FWER adjustment
months minus baseline Baseline-adjusted Baseline subtraction not strictly necessary
but indicated for ease of interpretation
109
Primary and selected secondary outcomes and Hypothesis E
22 Days Gambled, time- II.1 WE Workbook Linear mixed effects, weighted E Effect of workbook engagement; null is
averaged Engagement Baseline-adjusted alternative less engagement term
Time point-adjusted
23 Money Lost, time-averaged II.1 WE Workbook Linear mixed effects, weighted E Effect of workbook engagement; null is
Engagement Baseline-adjusted alternative less engagement term
Time point-adjusted
24 Gambling-quit or improved, IV.1 WE Workbook Logistic mixed effects, weighted E Effect of workbook engagement; null is
time-averaged Engagement Time point-adjusted alternative less engagement term
25 PGSI-12, at 12 months II.2 WE Workbook Linear mixed effects, counsellor- E Effect of workbook engagement; null is
Engagement specific random effects only, if alternative less engagement term
indicated, unweighted
Baseline-adjusted
26 Goal met in past 3 months, V WE Workbook Multinomial mixed effects, weighted E Effect of workbook engagement; null is
time-averaged Engagement Time point-adjusted alternative less engagement term
27 Control over gambling, time- II.1 WE Workbook Linear mixed effects, weighted E Effect of workbook engagement; null is
averaged Engagement Baseline-adjusted alternative less engagement term
Time point-adjusted
28 Days Gambled, time- II.1 ITT Assistance - formal, Linear mixed effects, weighted E Effect of workbook engagement; null is
averaged Assistance - informal Baseline-adjusted alternative less engagement term
(simultaneous Time point-adjusted
inclusion of terms) Time point-treatment interaction
29 Money Lost, time-averaged II.1 ITT Assistance - formal, Linear mixed effects, weighted E Effect of workbook engagement; null is
Assistance - informal Baseline-adjusted alternative less engagement term
30 Gambling-quit or improved, IV.1 ITT Assistance - formal, Logistic mixed effects, weighted E Effect of workbook engagement; null is
time-averaged Assistance - informal Time point-adjusted alternative less engagement term
(simultaneous Time point-treatment interaction
inclusion of terms)
110
31 PGSI-12, at 12 months II.2 ITT Assistance - formal, Linear mixed effects, counsellor- E Effect of workbook engagement; null is
Assistance - informal specific random effects only, if alternative less engagement term
(simultaneous indicated, unweighted
inclusion of terms) Baseline-adjusted
32 Goal met in past 3 months, V ITT Assistance - formal, Multinomial mixed effects, weighted E Effect of workbook engagement; null is
time-averaged Assistance - informal Time point-adjusted alternative less engagement term
(simultaneous Time point-treatment interaction
inclusion of terms)
33 Control over gambling, time- II.1 ITT Assistance - formal, Linear mixed effects, weighted E Effect of workbook engagement; null is
averaged Assistance - informal Baseline-adjusted alternative less engagement term
Secondary outcomes and primary superiority hypotheses in the ITT analysis set
34 PGSI-12, at 12 months II.2 ITT Tx Linear mixed effects, counsellor- Ba,b FWER adjustment across all 5 sub-
(no data collected at 3 specific random effects only, if Ca,b,c hypotheses
and 6 months) indicated, unweighted
Baseline-adjusted
35 PGSI-12-Dichotomised, at IV.2 ITT Tx Logistic mixed effects, counsellor- Ba,b FWER adjustment across all 5 sub-
12 months (no data collected at 3 specific random effects only, if Ca,b,c hypotheses
36 PGSI-3, time-averaged I ITT Tx Linear mixed effects, weighted Ba,b,c,d FWER adjustment
Baseline-adjusted
37 PGSI-3-Dichotomised, time- III ITT Tx Logistic mixed effects, weighted Ba,b,c,d FWER adjustment
averaged
38 PGSI-3, at 12 months II.1 ITT Tx: (Timepoint=12) Linear mixed effects, weighted Ca,b,c FWER adjustment
39 PGSI-3-Dichotomised, at 12 IV.1 ITT Tx: (Timepoint=12) Logistic mixed effects, weighted Ca,b,c FWER adjustment
months Time point-adjusted Null is alternative less only the Tx:
40 Control over gambling, time- I ITT Tx Linear mixed effects, weighted Ba,b,c,d FWER adjustment
averaged Baseline-adjusted
41 Control over gambling, at 12 II.1 ITT Tx: (Timepoint=12) Linear mixed effects, weighted Ca,b,c FWER adjustment
42 Kessler-10, time-averaged I ITT Tx Linear mixed effects, weighted Ba,b,c,d FWER adjustment
Baseline-adjusted
111
43 Kessler-10, at 12 months II.1 ITT Tx: (Timepoint=12) Linear mixed effects, weighted Ca,b,c FWER adjustment
44 AUDIT-C, time-averaged I ITT Tx Linear mixed effects, weighted Ba,b,c,d FWER adjustment
Baseline-adjusted
45 AUDIT-C, at 12 months II.1 ITT Tx: (Timepoint=12) Linear mixed effects, weighted Ca,b,c FWER adjustment
46 DAST, at 12 months II.2 ITT Tx Linear mixed effects, counsellor- Ba,b FWER adjustment across all 5 sub-
Baseline-adjusted
47 PRIME-MD Major IV.2 ITT Tx Logistic mixed effects, counsellor- Ba,b FWER adjustment across all 5 sub-
Depression, at 12 months (no data collected at 3 specific random effects only, if Ca,b,c hypotheses
48 PRIME-MD Dysthymia, at IV.2 ITT Tx Logistic mixed effects, counsellor- Ba,b FWER adjustment across all 5 sub-
12 months (no data collected at 3 specific random effects only, if Ca,b,c hypotheses
49 PRIME-MD Minor IV.2 ITT Tx Logistic mixed effects, counsellor- Ba,b FWER adjustment across all 5 sub-
Depression, at 12 months (no data collected at 3 specific random effects only, if Ca,b,c hypotheses
50 PRIME-MD Bipolar IV.2 ITT Tx Logistic mixed effects, counsellor- Ba,b FWER adjustment across all 5 sub-
Disorder, at 12 months (no data collected at 3 specific random effects only, if Ca,b,c hypotheses
51 Tobacco use current, time- III ITT Tx Logistic mixed effects, weighted Ba,b,c,d FWER adjustment
averaged
52 Tobacco use current, at 12 IV.1 ITT Tx: (Timepoint=12) Logistic mixed effects, weighted Ca,b,c FWER adjustment
53 Tobacco use frequency, time- V ITT Tx Multinomial mixed effects, weighted Ba,b,c,d FWER adjustment
averaged 2-sided
alternatives
54 Tobacco use frequency, at 12 V ITT Tx: (Timepoint=12) Multinomial mixed effects, weighted Ca,b,c FWER adjustment
months Time point-adjusted 2-sided Null is alternative less only the Tx:
Time point-treatment interaction alternatives (Timepoint=12) interaction
112
55 Mental health treatment in IV.2 ITT Tx Logistic mixed effects, counsellor- Ba,b FWER adjustment across all 5 sub-
past 12 months, at 12 months (no data collected at 3 specific random effects only, if Ca,b,c hypotheses
56 Prescribed medication for IV.2 ITT Tx Logistic mixed effects, counsellor- Ba,b FWER adjustment across all 5 sub-
mental health in past 12 (no data collected at 3 specific random effects only, if Ca,b,c hypotheses
months, at 12 months and 6 months) indicated, unweighted
57 Comorbity treatment in past IV.2 ITT Tx Logistic mixed effects, counsellor- Ba,b FWER adjustment across all 5 sub-
12 months, at 12 months (no data collected at 3 specific random effects only, if Ca,b,c hypotheses
58 WHOQoL-8, time-averaged I ITT Tx Linear mixed effects, weighted Ba,b,c,d FWER adjustment
Baseline-adjusted
59 WHOQoL-8, at 12 months II.1 ITT Tx: Linear mixed effects, weighted Ca,b,c FWER adjustment
(C1_2_Timepoint=12) Baseline-adjusted Null is alternative less only the Tx:
60 Affect on Work, time- I ITT Tx Linear mixed effects, weighted Ba,b,c,d FWER adjustment
61 Affect on Work, at 12 months II.1 ITT Tx: Linear mixed effects, weighted Ca,b,c FWER adjustment
(C1_2_Timepoint=12) Baseline-adjusted Null is alternative less only the Tx:
62 Affect on Social Life, time- I ITT Tx Linear mixed effects, weighted Ba,b,c,d FWER adjustment
63 Affect on Social Life, at 12 II.1 ITT Tx: Linear mixed effects, weighted Ca,b,c FWER adjustment
months (C1_2_Timepoint=12) Baseline-adjusted Null is alternative less only the Tx:
64 Affect on Family/Home, I ITT Tx Linear mixed effects, weighted Ba,b,c,d FWER adjustment
time-averaged Baseline-adjusted
65 Affect on Family/Home, at II.1 ITT Tx: Linear mixed effects, weighted Ca,b,c FWER adjustment
12 months (C1_2_Timepoint=12) Baseline-adjusted Null is alternative less only the Tx:
66 Affect on Health, time- I ITT Tx Linear mixed effects, weighted Ba,b,c,d FWER adjustment
113
67 Affect on Health, at 12 II.1 ITT Tx: (Timepoint=12) Linear mixed effects, weighted Ca,b,c FWER adjustment
months Baseline-adjusted Null is alternative less only the Tx:
68 Legal Problems, time- III ITT Tx Logistic mixed effects, weighted Ba,b,c,d FWER adjustment
averaged
69 Legal Problems, at 12 IV.1 ITT Tx: (Timepoint=12) Logistic mixed effects, weighted Ca,b,c FWER adjustment
70 NZDI, at 12 months II.2 ITT Tx Linear mixed effects, counsellor- Ba,b FWER adjustment across all 5 sub-
Baseline-adjusted
71 Goal met in past 3 months, V ITT Tx Multinomial mixed effects, weighted Ba,b,c,d FWER adjustment
time-averaged
72 Goal met in past 3 months, at VI ITT Tx: (Timepoint=12) Multinomial mixed effects, weighted Ca,b,c FWER adjustment
12 months Time point-adjusted Null is alternative less only the Tx:
73 Motivation level, time- I ITT Tx Linear mixed effects, weighted Ba,b,c,d FWER adjustment
74 Motivation level, at 12 II.1 ITT Tx: (Timepoint=12) Linear mixed effects, weighted Ca,b,c FWER adjustment
months Baseline-adjusted Null is alternative less only the Tx:
Engagement outcomes and engagement hypotheses in the ITT analysis set
75 E2_1_Wkbk_Read, time- V WE Tx Multinomial mixed effects, weighted Fa,b FWER of F adjustment based on whole
averaged by number of valid responses With 2- contrasts between treatments
sided Null is model of common mean.
alternative
Ga
76 E2_1_Wkbk_Read, at 3 VI WE Tx: (Timepoint=3) Multinomial mixed effects, weighted Gb Null is alternative less only the Tx:
months by number of valid responses (Timepoint=3) interaction
Time point-adjusted
77 E2_2_Wkbk_Exercise, time- V WE Tx Multinomial mixed effects, weighted Fa,b FWER adjustment of F
averaged by number of valid responses With 2- Null is model of common mean.
sided
alternative
Ga
114
78 Workbook Exercise, at 3 VI WE Tx: (Timepoint=3) Multinomial mixed effects, weighted Gb Null is alternative less only the Tx:
Time point-adjusted
79 Workbook Strategies, time- V WE Tx Multinomial mixed effects, weighted Fa,b FWER adjustment of F
averaged by number of valid responses With 2- Null is model of common mean.
sided
alternative
Ga
80 Workbook Strategies, at 3 VI WE Tx: (Timepoint=3) Multinomial mixed effects, weighted Gb Null is alternative less only the Tx:
Time point-adjusted
81 Assistandce Any, time- VII ITT Tx Binomial mixed effects, weighted Fa,b FWER adjustment of F
averaged Ga Null is model of common mean.
82 Assistance Any, at 3 months VIII ITT Tx: (Timepoint=3) Binomial mixed effects, weighted by Gb Null is alternative less only the Tx:
number of valid responses (Timepoint=3) interaction
Time point-adjusted
Notes: 1) All alternative hypotheses bar A are one-sided.

2) Time points entered as covariates are entered as categorical covariates unless otherwise indicated.
115
APPENDIX 4
Tables - Descriptive statistics
Table 4.1: Socio-demographics

TAU MI MI+W MI+W+B
Male 41.4% 47.3% 45.3% 55.2%
Female 58.6% 52.7% 54.7% 44.8%
Gender
N 116 112 117 116
N MISSING 0 0 1 0
Never married 25.2% 30.3% 34.2% 32.8%
Married 23.5% 21.1% 22.8% 25.0%
De facto 24.3% 21.1% 25.4% 26.7%
Separated 13.9% 14.7% 10.5% 10.3%
Marital status
Divorced 8.7% 12.8% 4.4% 3.4%
Widowed 4.3% 0.0% 2.6% 1.7%
N 115 109 114 116
N MISSING 1 3 4 0
Partnered 52.2% 57.8% 51.8% 48.3%
Marital status, Not partnered 47.8% 42.2% 48.2% 51.7%
N MISSING 1 3 4 0
18-24 years 15.2% 11.6% 8.5% 19.0%
25-34 years 22.3% 33.0% 27.1% 29.3%
35-44 years 23.2% 19.6% 30.5% 17.2%
Age group 45-54 years 24.1% 22.3% 22.0% 25.0%
55+ years 15.2% 13.4% 11.9% 9.5%
N 112 112 118 116
N MISSING 4 0 0 0
MEAN YEARS 40.3 39.1 39.9 37.5
STD 13.6 13.1 11.7 13.1
MIN YEARS 18.0 19.0 19.0 19.0
Q1 YEARS 28.0 27.0 31.0 26.0
Age MEDIAN YEARS 40.0 37.0 39.0 36.0
Q3 YEARS 49.5 50.0 46.0 46.0
MAX YEARS 79.0 71.0 76.0 72.0
N 112 112 118 116
N MISSING 4 0 0 0
Maori 40.5% 39.3% 43.2% 36.2%
Pacific 11.2% 16.1% 7.6% 10.3%
European 47.0% 42.0% 44.9% 47.4%
Primary ethnicity
Asian & Other 3.0% 2.7% 4.2% 6.0%
N 116 112 118 116
N MISSING 0 0 0 0
European ethnicity Yes 50.0% 47.3% 54.2% 53.4%
Maori ethnicity Yes 40.5% 39.3% 43.2% 36.2%
Pacific ethnicity Yes 11.2% 16.1% 8.5% 10.3%
Asian ethnicity Yes 4.3% 1.8% 2.5% 4.3%
Other ethnicity Yes 0.9% 1.8% 1.7% 3.4%
Asian and Other ethnicity Yes 4.3% 3.6% 4.2% 6.9%
N 116 112 118 116
N MISSING 0 0 0 0
116
Table 4.1: Socio-demographics - continued
TAU MI MI+W MI+W+B
Full time 44.3% 44.1% 41.5% 48.7%
Part time 13.0% 11.7% 14.4% 9.6%
Homemaker 7.8% 8.1% 5.1% 9.6%
Student 5.2% 3.6% 5.1% 3.5%
Retired 2.6% 1.8% 2.5% 3.5%
Employment status Unemployed 11.3% 18.0% 11.9% 11.3%
Illness/sick leave 6.1% 3.6% 5.9% 3.5%
Maternity Leave 0.9% 0.0% 0.0% 0.0%
Other 8.7% 9.0% 13.6% 10.4%
N 115 111 118 115
N MISSING 1 1 0 1
None 25.9% 19.6% 21.4% 18.3%
Secondary school qualification 33.6% 31.3% 36.8% 36.5%
Trade or technical certificate 18.1% 24.1% 21.4% 22.6%
Professional qualification 3.4% 7.1% 4.3% 4.3%
Highest educational Undergrad. Dip. or Cert. 7.8% 6.3% 6.0% 8.7%
qualification achieved Undergrad. Degree 6.0% 3.6% 6.0% 7.0%
Postgrad. Dip. or Cert. 1.7% 1.8% 1.7% 0.0%
Postgrad. Degree 3.4% 6.3% 2.6% 2.6%
N 116 112 117 115
N MISSING 0 0 1 1
≤$20,000 26.9% 17.6% 23.0% 19.3%
$20,001-$30,000 16.3% 22.2% 14.2% 13.8%
$30,001-$50,000 23.1% 19.4% 32.7% 33.9%
$50,001-$100,000 24.0% 31.5% 23.0% 23.9%
Gross family income in last
$100,001-$200,000 7.7% 7.4% 7.1% 8.3%
12 months
$200,001-$500,000 1.9% 1.9% 0.0% 0.9%
$500,001+ 0.0% 0.0% 0.0% 0.0%
N 104 108 113 109
N MISSING 12 4 5 7
≤$30,000 43.3% 39.8% 37.2% 33.0%
Gross family income in last >$30,000 56.7% 60.2% 62.8% 67.0%
12 months, dichotomised N 104 108 113 109
N MISSING 12 4 5 7
117
Table 4.2: Area of residence
TAU MI MI+W MI+W+B
Ashburton 1 2 2
Auckland 29 27 27 32
Bay of Plenty 1 2
Blenheim 1 1 2 1
Cambridge 1
Canterbury 1
Christchurch 12 9 14 13
Clutha 1
Coromandel 1
Dannevirke 2
Dunedin 2 5 4 1
Fielding 1
Foxton 2
Franklin 1
Geraldine 1
Gisborne 2 2 1 3
Hamilton 3 9 6 4
Hastings 2 1 3 1
Havelock 1
Havelock North 1
Hawera 1
Hawkes Bay 1
Helensville 1
Hokitika 1
Huntly 1
Inglewood 1
Invercargill 1 1 1
Kaikohe 1
Kaitaia 1 1
Kapiti 1
Kapiti Coast 1 1
Kaukapakapa 1
Kawerau 1
Levin 1 1
Lower Hutt 3 3 1
Manukau 5 5 3 3
Morrinsville 1 2
Motueka 1
Table 4.2: Area of residence - continued
TAU MI MI+W MI+W+B
118
Mt Maunganui 1 1
Napier 1 4 3
Nelson 1 2 1
New Plymouth 1 2 4
North Auckland 1
North Canterbury 2
North Shore 1 1 1
Northland 1 1 1 1
Oamaru 1
Otaki 1
Otane 1
Paeroa 1 1
Palmerston 1
Palmerston North 4 6 3
Papakura 1
Papamoa 1 2
Paraparaumu 1
Pukekohe 1
Putaruru 1
Rotorua 10 2 1 3
South Auckland 1
Southland 1
Stratford 1
Taihape 1
Taranaki 1 1 1
Taupo 2 2
Tauranga 2 4 6 1
Te Awamutu 2
Te Kuiti 1 1
Te Puke 1
Thames 1 1 1
Timaru 2 1
Upper Hutt 2 2
Waikato 1 1
Waimate 1 1
Waitakere 1 1
Waitara 1
Wanganui 1 2 4 2
Table 4.2: Area of residence - continued

TAU MI MI+W MI+W+B
Warkworth 1 1
Wellington 11 11 7 6
119
Westport 1
Whakatane 1 1 1 1
Whangamata 1
Whanganui 1
Whangaparaoa 1 1
Whangarei 1
Whitby 1
Whitianga 1
N 116 111 118 116
N MISSING 0 1 0 0
120
Table 4.3: Gambling characterisation
TAU MI MI+W MI+W+B
Yes 0.9% 4.5% 2.5% 0.9%
Gambling type:
N 116 112 118 116
Cards
N MISSING 0 0 0 0
Gambling type: Yes 19.0% 25.9% 15.3% 17.2%
Casino gaming N 116 112 118 116
machines N MISSING 0 0 0 0
Yes 6.9% 4.5% 2.5% 7.8%
Gambling type:
N 116 112 118 116
Casino tables
N MISSING 0 0 0 0
Yes 28.4% 22.3% 23.7% 25.9%
Gambling type:
N 116 112 118 116
Club machines
N MISSING 0 0 0 0
Yes 81.9% 83.0% 83.9% 82.8%
Gambling type:
N 116 112 118 116
Pub machines
N MISSING 0 0 0 0
Yes 0.9% 0.9% 0.0% 0.0%
Gambling type:
N 116 112 118 116
Housie
N MISSING 0 0 0 0
Yes 1.7% 0.9% 0.8% 0.9%
Gambling type:
N 116 112 118 116
Keno
N MISSING 0 0 0 0
Yes 6.9% 5.4% 5.1% 6.0%
Gambling type:
N 116 112 118 116
Lotto
N MISSING 0 0 0 0
Yes 2.6% 6.3% 3.4% 7.8%
Gambling type:
N 116 112 118 116
Sports betting
N MISSING 0 0 0 0
Yes 5.2% 15.2% 16.9% 15.5%
Gambling type:
N 116 112 118 116
Track
N MISSING 0 0 0 0
Yes 5.2% 3.6% 4.2% 2.6%
Gambling type:
N 116 112 118 116
Other
N MISSING 0 0 0 0
Cards 0.9% 0.0% 0.9% 0.0%
Casino Gaming Machines 3.5% 11.7% 6.0% 5.4%
Casino Tables 4.4% 1.0% 1.7% 3.6%
Club Gaming Machines 11.5% 5.8% 9.5% 9.9%
Pub Gaming Machines 73.5% 71.8% 69.0% 72.1%
Primary Gambling Keno 0.9% 0.0% 0.0% 0.0%
Type Lotto 0.0% 1.0% 0.0% 0.9%
Sports Betting 0.9% 2.9% 0.9% 2.7%
Track 3.5% 4.9% 7.8% 3.6%
Other 0.9% 1.0% 4.3% 1.8%
N 113 103 116 111
N MISSING 3 9 2 5
Electronic gaming Yes 93.1% 91.1% 91.5% 92.2%
machines as N 116 112 118 116
gambling type N MISSING 0 0 0 0
121
Table 4.3: Gambling characterisation - continued
TAU MI MI+W MI+W+B
MEAN 81.5 84.0 86.4 73.9
STD 85.3 98.9 99.9 70.1
MIN 1.0 1.0 0.3 0.5
How long has Q1 18.0 18.0 18.0 18.0
gambling been a MEDIAN 60.0 60.0 60.0 48.0
problem (months) Q3 120.0 120.0 120.0 108.0
MAX 480.0 588.0 564.0 360.0
N 113 112 115 115
N MISSING 3 0 3 1
MEAN 4.8 4.0 3.0 3.2
STD 11.5 9.2 4.5 5.8
MIN 0.0 0.0 0.0 0.0
How long since the Q1 1.0 1.0 1.0 1.0
last time you MEDIAN 1.0 1.5 1.0 1.0
gambled (days) Q3 4.0 3.5 3.0 3.0
MAX 97.0 67.0 30.0 35.0
N 116 112 118 116
N MISSING 0 0 0 0
Yes 93.1% 95.5% 99.2% 94.8%
Ever felt the need
N 116 112 118 116
to bet more
N MISSING 0 0 0 0
Yes 86.1% 88.2% 93.2% 86.2%
Ever felt the need
N 115 110 118 116
to lie?
N MISSING 1 2 0 0
122
Table 4.4: Treatment prospects
TAU MI MI+W MI+W+B
Level of belief in MEAN 8.6 8.2 7.9 8.3
success within six
STD 2.0 2.2 2.2 1.9
months
(10-point Likert MIN 0.0 0.0 0.0 3.0
scale)
Q1 8.0 7.0 6.0 7.0
MEDIAN 10.0 9.0 8.0 9.0
Q3 10.0 10.0 10.0 10.0
MAX 10.0 10.0 10.0 10.0
N 111 106 107 111
N MISSING 5 6 11 5
Level of belief in MEAN 9.1 8.9 8.6 9.0
success within 12
STD 1.7 1.7 2.0 1.7
months
(10-point Likert MIN 0.0 0.0 0.0 1.0
scale)
Q1 9.0 8.0 7.0 8.0
MEDIAN 10.0 10.0 10.0 10.0
Q3 10.0 10.0 10.0 10.0
MAX 10.0 10.0 10.0 10.0
N 112 106 109 109
N MISSING 4 6 9 7
Level of belief in Lower level 47.4% 54.5% 61.0% 56.0%
success within 12
Higher level 52.6% 45.5% 39.0% 44.0%
months,
N MISSING 0 0 0 0
Level of difficulty MEAN 7.5 8.0 8.1 7.8
expected in next 12
STD 2.4 2.3 2.2 2.1
months
(10-point Likert MIN 0.0 0.0 0.0 0.0
scale)
Q1 6.0 7.0 7.0 6.0
MEDIAN 8.0 8.0 8.0 8.0
Q3 10.0 10.0 10.0 10.0
MAX 10.0 10.0 10.0 10.0
N 115 111 114 116
N MISSING 1 1 4 0
123
Table 4.5: Primary efficacy - gambling, money lost and gambling cessation/improvement
TAU MI MI+W MI+W+B
Baseline 3 6 12 Baseline 3 6 12 Baseline 3 6 12 Baseline 3 6 12
months months months months months months months months months months months months
Days Gambled MEAN 9.0 3.2 2.6 3.0 8.2 3.3 3.4 3.5 8.8 3.5 3.4 3.2 8.3 3.1 2.6 2.7
per month
STD 7.2 4.3 3.5 4.0 6.0 4.4 4.8 4.8 6.6 5.2 5.7 4.4 6.2 3.8 3.1 3.2
MIN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.5 0.0 0.0 0.0 0.5 0.0 0.0 0.0
Q1 3.5 0.3 0.0 0.5 4.0 0.8 0.7 0.3 4.5 0.3 0.0 0.5 4.5 0.3 0.0 0.2
MEDIAN 7.5 1.7 1.0 1.5 7.0 2.2 2.0 2.0 6.0 1.7 1.0 1.8 6.5 1.7 1.5 1.5
Q3 13.0 4.7 4.0 4.3 11.5 4.3 4.3 4.3 12.0 4.7 4.5 4.3 10.0 5.0 4.3 4.5
MAX 30.0 25.3 16.7 25.3 30.0 28.7 30.0 29.7 31.5 30.0 30.0 24.0 30.0 18.0 16.3 13.2
N 111 100 92 78 95 88 78 66 109 98 88 78 109 87 82 73
N MISSING 5 0 0 0 17 0 0 0 9 0 0 0 7 0 0 0
Money Lost per MEAN 42.9 9.5 7.2 9.7 53.2 9.8 14.3 13.4 48.6 9.2 9.0 7.5 49.2 9.0 10.6 8.4
day
STD 20.8 13.2 18.3 17.4 42.2 35.4 18.3 20.9 12.0 14.6 25.0 16.1
45.7 58.8 69.1 59.5
MIN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 3.9 0.0 0.0 0.0 0.3 0.0 0.0 0.0
Q1 14.1 0.6 0.0 0.4 16.4 0.4 0.7 0.3 18.1 0.2 0.0 0.3 16.5 0.0 0.0 0.1
MEDIAN 29.3 2.6 1.6 2.7 33.1 2.9 2.9 2.5 32.1 3.2 1.7 1.8 31.8 2.5 2.1 2.2
Q3 60.6 9.3 6.3 7.3 70.7 11.5 8.6 9.3 52.6 9.9 7.4 6.6 59.5 9.8 10.7 10.1
MAX 263.6 166.7 52.6 85.4 327.2 99.1 320.7 260.1 646.4 131.6 156.8 55.8 388.2 66.0 168.9 110.7
N 111 100 92 78 95 88 78 66 109 98 88 78 110 87 82 73
N MISSING 5 0 0 0 17 0 0 0 9 0 0 0 6 0 0 0
Gambling-quit or YES . 82.0% 71.7% 87.2% . 83.0% 87.2% 84.8% . 82.7% 71.6% 84.6% . 75.9% 73.2% 75.3%
improved
N . 100 92 78 . 88 78 66 . 98 88 78 . 87 82 73
N MISSING . 0 0 0 . 0 0 0 . 0 0 0 . 0 0 0
124
Table 4.6: Secondary efficacy outcomes - PGSI
TAU MI MI+W MI+W+B
Problem MEAN 16.8 . . 9.2 17.2 . . 9.7 17.3 . . 9.3 16.6 . . 10.0
Gambling
STD 4.6 . . 6.3 4.7 . . 7.3 4.8 . . 6.1 4.6 . . 6.4
Severity Index
- 12 month MIN 3.0 . . 0.0 7.0 . . 0.0 0.0 . . 0.0 5.0 . . 0.0
time frame
Q1 14.0 . . 3.0 14.0 . . 4.0 14.0 . . 4.0 13.0 . . 5.5
MEDIAN 17.0 . . 9.0 17.0 . . 9.0 17.0 . . 10.0 17.0 . . 10.0
Q3 20.0 . . 13.0 21.0 . . 14.0 20.0 . . 13.0 20.0 . . 15.0
MAX 25.0 . . 23.0 26.0 . . 25.0 27.0 . . 25.0 25.0 . . 24.0
N 111 . . 74 104 . . 65 106 . . 77 110 . . 72
N MISSING 5 . . 4 8 . . 1 12 . . 1 6 . . 1
Problem MEAN 17.3 7.8 6.9 6.4 17.5 8.5 7.5 7.1 18.2 7.6 6.7 5.3 17.2 7.6 5.6 5.8
Gambling
STD 5.3 7.0 6.7 6.2 5.3 6.9 6.8 7.5 4.7 6.3 7.0 6.1 5.2 5.8 6.2 6.7
Severity Index,
3 month time MIN 0.0 0.0 0.0 0.0 3.0 0.0 0.0 0.0 6.0 0.0 0.0 0.0 5.0 0.0 0.0 0.0
frame
Q1 14.0 1.0 1.0 0.0 14.0 3.0 1.0 0.0 15.0 2.0 0.0 0.0 14.0 2.0 1.0 0.0
MEDIAN 18.0 7.0 5.0 6.0 17.0 7.0 6.0 4.5 18.0 6.0 4.0 2.5 18.0 7.0 3.0 2.0
Q3 21.0 13.0 11.0 10.0 22.0 13.0 12.0 13.0 22.0 13.0 12.0 10.5 21.0 12.0 9.0 11.0
MAX 27.0 26.0 24.0 23.0 27.0 27.0 26.0 22.0 27.0 23.0 25.0 21.0 27.0 19.0 21.0 27.0
N 110 92 85 76 106 82 71 66 110 89 83 76 111 83 77 73
N MISSING 6 8 7 2 6 6 7 0 8 9 5 2 5 4 5 0
125
Table 4.6: Secondary efficacy outcomes - PGSI - continued
TAU MI MI+W MI+W+B
PGSI, 12 ≥8 97.2% . . 60.8% 97.7% . . 55.4% 96.9% . . 63.6% 95.2% . . 66.7%
month time
frame, N 111 . . 74 104 . . 65 106 . . 77 110 . . 72
dichotomised
(≥8 vs. <8) N MISSING 5 . . 4 8 . . 1 12 . . 1 6 . . 1
PGSI, 3 ≥8 96.2% 43.5% 40.0% 40.8% 94.4% 48.8% 43.7% 39.4% 99.0% 44.9% 36.1% 34.2% 96.2% 48.2% 35.1% 37.0%
month time
frame, N 110 92 85 76 106 82 71 66 110 89 83 76 111 83 77 73
dichotomised
(≥8 vs. <8) N MISSING 6 8 7 2 6 6 7 0 8 9 5 2 5 4 5 0
Table 4.7: Secondary efficacy outcomes - Control over gambling

TAU MI MI+W MI+W+B
Control over MEAN 3.4 6.9 7.2 7.1 2.8 6.7 6.6 6.7 2.4 6.9 7.4 7.6 2.5 7.3 7.7 7.2
gambling
STD 2.9 2.6 2.8 2.8 2.3 2.8 3.0 2.9 2.3 2.8 3.0 2.4 2.4 2.5 2.3 3.0
behaviour
(10-point MIN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
scale)
Q1 1.0 5.0 5.0 6.0 0.0 5.0 5.0 5.0 0.0 5.0 5.5 7.0 0.0 5.0 7.0 5.0
MEDIAN 3.0 7.0 8.0 8.0 2.5 7.0 7.0 7.5 2.0 7.5 8.5 8.0 2.0 8.0 8.0 8.0
Q3 5.0 9.0 10.0 10.0 5.0 9.0 9.0 9.0 4.0 9.0 10.0 10.0 4.0 10.0 10.0 10.0
MAX 10.0 10.0 10.0 10.0 8.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0
N 114 93 87 78 109 84 74 66 118 94 84 78 113 84 77 73
N MISSING 2 7 5 0 3 4 4 0 0 4 4 0 3 3 5 0
126
Table 4.8: Secondary efficacy outcomes - Co-existing issues
TAU MI MI+W MI+W+B
Psychological MEAN 30.2 18.3 18.2 16.2 29.5 19.1 17.5 17.1 32.1 19.6 18.7 15.9 30.3 16.7 16.6 15.1
distress
STD 8.1 8.4 7.7 8.6 9.3 8.0 8.4 7.7 8.9 9.6 8.8 7.4 9.2 7.0 7.5 6.7
Kessler-10,
4 week time frame MIN 14.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 12.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0
Q1 24.0 12.0 12.0 10.0 24.0 13.0 11.0 11.0 26.0 11.0 12.0 10.0 23.0 11.0 10.0 10.0
MEDIAN 31.0 16.0 16.0 13.0 28.5 17.0 15.0 14.0 32.0 16.0 16.0 13.0 31.0 14.5 14.0 11.5
Q3 36.0 22.0 24.0 19.0 36.0 25.0 21.0 24.0 39.0 26.0 23.0 20.0 37.0 20.0 21.0 19.0
MAX 50.0 48.0 39.0 50.0 50.0 38.0 45.0 36.0 50.0 43.0 47.0 41.0 49.0 42.0 39.0 34.0
N 111 92 87 78 104 83 73 65 117 93 84 78 114 84 76 72
N MISSING 5 8 5 0 8 5 5 1 1 5 4 0 2 3 6 1
AUDIT-C MEAN 4.5 3.5 3.3 3.7 5.1 3.8 3.5 3.6 4.9 3.5 3.6 3.6 5.2 4.4 4.0 4.5
(12-point score)
STD 3.5 3.2 3.2 3.1 3.5 3.3 3.4 3.4 3.6 3.2 3.6 3.4 3.7 2.8 3.1 3.4
MIN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Q1 0.0 0.0 0.0 0.0 2.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0 2.0 2.0 2.0 2.0
MEDIAN 5.0 4.0 3.0 3.0 5.0 4.0 3.0 3.0 5.0 3.0 3.0 3.0 5.0 4.0 4.0 5.0
Q3 7.0 6.0 6.0 6.0 8.0 7.0 6.0 6.0 8.0 6.0 7.0 6.0 8.0 7.0 6.0 7.0
MAX 12.0 12.0 10.0 10.0 12.0 12.0 12.0 12.0 12.0 11.0 11.0 12.0 12.0 10.0 10.0 12.0
N 109 93 87 78 104 83 73 66 109 93 84 78 110 83 77 70
N MISSING 7 7 5 0 8 5 5 0 9 5 4 0 6 4 5 3
127
Table 4.8: Secondary efficacy outcomes - Co-existing issues - continued
TAU MI MI+W MI+W+B
Drug Abuse MEAN 0.76 . . 0.23 0.92 . . 0.38 0.82 . . 0.44 0.78 . . 0.46
Screening Test
STD 1.83 . . 1.21 2.20 . . 1.08 1.95 . . 1.39 1.94 . . 1.27
(DAST)
(10-point score) MIN 0.00 . . 0.00 0.00 . . 0.00 0.00 . . 0.00 0.00 . . 0.00
Q1 0.00 . . 0.00 0.00 . . 0.00 0.00 . . 0.00 0.00 . . 0.00
MEDIAN 0.00 . . 0.00 0.00 . . 0.00 0.00 . . 0.00 0.00 . . 0.00
Q3 0.00 . . 0.00 0.00 . . 0.00 0.00 . . 0.00 0.00 . . 0.00
MAX 8.00 . . 10.00 10.00 . . 6.00 8.00 . . 8.00 9.00 . . 6.00
N 106 . . 77 105 . . 66 108 . . 78 110 . . 71
N MISSING 10 . . 1 7 . . 0 10 . . 0 6 . . 2
PRIME-MD Yes 57.4% . . 17.9% 49.0% . . 22.7% 62.9% . . 16.7% 50.9% . . 20.5%
Major depressive
N 108 . . 78 96 . . 66 105 . . 78 110 . . 73
disorder
N MISSING 8 . . 0 16 . . 0 13 . . 0 6 . . 0
PRIME-MD Yes 43.5% . . 32.1% 44.2% . . 30.3% 44.2% . . 39.7% 39.1% . . 29.2%
Dysthymia
N 108 . . 78 95 . . 66 104 . . 78 110 . . 72
N MISSING 8 . . 0 17 . . 0 14 . . 0 6 . . 1
PRIME-MD Yes 13.0% . . 3.8% 15.8% . . 4.5% 15.4% . . 2.6% 16.4% . . 1.4%
Minor depressive
N 108 . . 78 95 . . 66 104 . . 78 110 . . 72
disorder
N MISSING 8 . . 0 17 . . 0 14 . . 0 6 . . 1
PRIME-MD Yes 23.3% . . 17.3% 19.1% . . 10.6% 28.3% . . 16.2% 20.2% . . 22.9%
Alcohol abuse
N 103 75 89 66 99 74 109 70
N MISSING 13 . . 3 23 . . 0 19 . . 4 7 . . 3

128
TAU MI MI+W MI+W+B
PRIME-MD Yes 2.9% . . 4.1% 3.2% . . 3.1% 3.9% . . 6.8% 4.5% . . 7.2%
Bipolar disorder
N 105 . . 74 94 . . 65 102 . . 73 110 . . 69
N MISSING 11 . . 4 18 . . 1 16 . . 5 6 . . 4
Current smoking Yes 59.6% 57.0% 55.2% 52.6% 54.6% 48.2% 45.9% 40.9% 56.0% 47.9% 51.2% 51.3% 60.0% 58.3% 58.4% 55.6%
status
N 99 93 87 78 108 83 74 66 116 94 84 78 90 84 77 72
N MISSING 17 7 5 0 4 5 4 0 2 4 4 0 26 3 5 1
Frequency of 1 95.5% 98.1% 95.8% 92.7% 88.3% 90.0% 85.3% 96.3% 93.9% 93.3% 93.0% 92.5% 93.0% 91.8% 90.9% 95.0%
smoking
2 3.0% 1.9% 2.1% 4.9% 6.7% 7.5% 8.8% 0.0% 4.5% 6.7% 4.7% 7.5% 2.8% 6.1% 6.8% 2.5%
1=At least once a
day 3 1.5% 0.0% 2.1% 0.0% 3.3% 2.5% 0.0% 3.7% 0.0% 0.0% 0.0% 0.0% 1.4% 0.0% 0.0% 2.5%
2=At least once a
4 0.0% 0.0% 0.0% 2.4% 1.7% 0.0% 5.9% 0.0% 1.5% 0.0% 2.3% 0.0% 2.8% 2.0% 2.3% 0.0%
week
3=At least once a N 66 53 48 41 60 40 34 27 66 45 43 40 71 49 44 40
month
4=Less than once a N MISSING 50 47 44 37 52 48 44 39 52 53 45 38 45 38 38 33
month
Treatment received Yes 20.9% . . 17.9% 24.3% . . 19.7% 18.6% . . 19.2% 21.6% . . 16.4%
for mental health in N 115 . . 78 111 . . 66 118 . . 78 116 . . 73
previous 12 months N MISSING 1 . . 0 1 . . 0 0 . . 0 0 . . 0
Prescription Yes 22.8% . . 18.2% 27.7% . . 28.8% 27.6% . . 26.9% 24.8% . . 19.4%
received for mental N 101 . . 77 101 . . 66 105 . . 78 105 . . 72
health in previous N MISSING 15 . . 1 11 . . 0 13 . . 0 11 . . 1
12 months
Treatment received Yes 6.4% . . 3.8% 8.0% . . 9.1% 5.2% . . 6.4% 8.6% . . 7.0%
for drugs or alcohol N 110 . . 78 112 . . 66 116 . . 78 116 . . 71
in previous 12 N MISSING 6 . . 0 0 . . 0 2 . . 0 0 . . 2
months
129
TAU MI MI+W MI+W+B
WHO Quality of MEAN 25.0 29.5 31.2 31.1 24.8 30.1 31.3 30.8 23.9 30.1 30.1 31.8 24.4 31.0 31.5 31.8
Life,
STD 5.6 5.4 5.8 6.2 6.3 6.3 5.3 5.5 6.1 5.7 6.0 5.5 6.6 5.6 5.7 5.8
8-item questionnaire
MIN 10.0 17.0 16.0 11.0 9.0 16.0 19.0 20.0 9.0 14.0 14.0 18.0 9.0 16.0 11.0 15.0
Q1 21.0 26.0 27.0 27.0 21.0 26.0 28.0 27.0 20.0 27.0 26.0 28.0 19.0 27.0 29.0 28.0
MEDIAN 26.0 30.0 32.0 32.0 25.0 31.0 31.0 31.0 24.0 31.0 30.0 32.0 25.0 32.0 32.0 33.0
Q3 29.0 33.0 36.0 35.0 29.0 36.0 35.0 35.0 28.0 35.0 34.5 36.0 31.0 35.0 36.0 36.0
MAX 39.0 39.0 40.0 40.0 38.0 40.0 40.0 40.0 38.0 40.0 40.0 40.0 40.0 40.0 40.0 40.0
N 109 93 87 78 106 83 73 65 107 92 84 77 111 84 77 72
N 7 7 5 0 6 5 5 1 11 6 4 1 5 3 5 1
MISSING
130
Table 4.9: Secondary efficacy outcomes - Gambling impacts
TAU MI MI+W MI+W+B
How was work MEAN 3.2 1.4 1.1 0.7 3.4 1.7 1.1 1.1 3.8 1.7 1.1 1.0 3.6 0.8 1.1 0.6
affected in past one
STD 3.5 2.8 2.5 1.7 3.6 3.1 2.4 2.4 3.6 3.2 2.4 2.3 3.8 2.2 2.6 1.7
month
(10-point Likert MIN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
scale) Q1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
MEDIAN 2.0 0.0 0.0 0.0 3.0 0.0 0.0 0.0 3.5 0.0 0.0 0.0 2.0 0.0 0.0 0.0
Q3 6.0 1.0 0.0 0.0 6.0 1.5 0.0 0.0 7.0 1.0 0.0 0.0 7.0 0.0 0.0 0.0
MAX 10.0 10.0 10.0 7.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0
N 102 88 83 71 102 76 71 60 110 84 74 70 106 79 73 70
N MISSING 14 12 9 7 10 12 7 6 8 14 14 8 10 8 9 3
How was social life MEAN 5.1 1.6 1.5 1.2 5.0 2.6 1.9 1.8 5.8 1.9 1.9 1.7 5.7 1.2 1.5 1.7
STD 3.4 3.0 3.0 2.4 3.7 3.5 2.8 3.0 3.5 3.3 3.1 2.9 3.5 2.4 2.9 3.0
month?
(10-point Likert MIN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
scale)
Q1 2.0 0.0 0.0 0.0 0.5 0.0 0.0 0.0 3.0 0.0 0.0 0.0 3.0 0.0 0.0 0.0
MEDIAN 5.5 0.0 0.0 0.0 5.0 0.0 0.0 0.0 7.0 0.0 0.0 0.0 7.0 0.0 0.0 0.0
Q3 8.0 2.0 1.0 0.0 8.0 5.0 3.0 3.0 9.0 3.0 4.0 2.0 8.0 1.0 1.0 3.0
MAX 10.0 10.0 10.0 9.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0
N 114 93 87 78 112 84 74 66 117 93 84 78 113 84 77 72
N MISSING 2 7 5 0 0 4 4 0 1 5 4 0 3 3 5 1
Table 4.9: Secondary efficacy outcomes - Gambling impacts - continued

131
TAU MI MI+W MI+W+B
How were family & MEAN 6.4 2.1 2.1 1.3 6.5 2.4 1.8 2.3 7.5 2.3 2.0 2.2 6.7 1.9 1.9 1.7
home affected in past
STD 3.3 3.2 3.3 2.6 3.3 3.3 3.0 3.3 2.8 3.4 3.2 3.3 3.4 3.1 3.1 3.1
one month?
(10-point Likert scale MIN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Q1 5.0 0.0 0.0 0.0 4.0 0.0 0.0 0.0 6.0 0.0 0.0 0.0 5.0 0.0 0.0 0.0
MEDIAN 7.0 0.0 0.0 0.0 7.5 0.0 0.0 0.0 8.5 0.0 0.0 0.0 8.0 0.0 0.0 0.0
Q3 9.0 4.0 4.0 2.0 9.0 5.0 3.0 5.0 10.0 5.0 3.5 5.0 10.0 3.0 3.0 3.0
MAX 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0
N 116 93 87 78 112 84 74 66 118 94 84 78 116 84 77 73
N MISSING 0 7 5 0 0 4 4 0 0 4 4 0 0 3 5 0
How was health MEAN 5.4 1.6 1.8 1.4 5.2 1.7 1.5 1.8 5.7 1.8 1.4 1.6 5.0 1.1 1.4 1.3
STD 3.0 2.8 2.9 2.5 3.0 2.7 2.7 2.8 3.3 3.0 2.6 2.6 3.6 2.2 2.7 2.6
month?
(10-point Likert MIN 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
scale)
Q1 3.0 0.0 0.0 0.0 3.0 0.0 0.0 0.0 3.0 0.0 0.0 0.0 1.0 0.0 0.0 0.0
MEDIAN 6.0 0.0 0.0 0.0 5.3 0.0 0.0 0.0 6.0 0.0 0.0 0.0 5.0 0.0 0.0 0.0
Q3 8.0 3.0 3.0 2.0 8.0 2.5 2.0 4.0 8.0 3.0 2.0 3.0 8.0 0.0 1.0 2.0
MAX 10.0 10.0 10.0 9.0 10.0 10.0 9.0 9.0 10.0 10.0 10.0 10.0 10.0 9.0 10.0 10.0
N 114 93 87 78 110 84 74 66 114 93 84 78 116 84 77 73
N MISSING 2 7 5 0 2 4 4 0 4 5 4 0 0 3 5 0
Table 4.9: Secondary efficacy outcomes - Gambling impacts - continued
132
TAU MI MI+W MI+W+B
Legal problems Yes 12.6% 9.7% 3.5% 9.0% 17.1% 10.7% 1.4% 9.1% 17.2% 8.6% 4.8% 9.1% 9.6% 6.0% 5.2% 9.6%
experienced in past
12 months (baseline)/
133
N 111 93 86 78 111 84 74 66 116 93 84 77 115 84 77 73
N MISSING 5 7 6 0 1 4 4 0 2 5 4 1 1 3 5 0
New Zealand MEAN 1.44 . . 1.09 1.48 . . 1.19 1.65 . . 1.21 1.34 . . 1.11
Deprivation Index
STD 1.18 . . 1.30 1.26 . . 1.13 1.30 . . 1.33 1.21 . . 1.25
MIN 0.00 . . 0.00 0.00 . . 0.00 0.00 . . 0.00 0.00 . . 0.00
Q1 0.58 . . 0.00 0.00 . . 0.00 0.58 . . 0.00 0.00 . . 0.00
MEDIAN 1.23 . . 0.63 1.24 . . 0.90 1.69 . . 0.66 1.19 . . 0.63
Q3 2.36 . . 1.57 2.42 . . 1.98 2.56 . . 1.98 2.35 . . 1.89
MAX 4.25 . . 4.77 4.25 . . 4.19 4.77 . . 4.77 4.25 . . 4.25
N 100 . . 76 93 . . 66 103 . . 77 108 . . 72
N MISSING 16 . . 2 19 . . 0 15 . . 1 8 . . 1
134
Table 4.10: Secondary efficacy outcomes - Goal setting and motivation
TAU MI MI+W MI+W+B
Baseline 3 6 12 Baseline 3 6 12 Baseline 3 6 12 Baseline 3 6 12 months
months months months months months months months months months months months
Goal met in Not at all . 10.8% 13.8% 23.1% . 25.0% 24.3% 26.2% . 18.1% 19.0% 12.8% . 14.3% 9.1% 23.3%
the last 3
Partly . 37.6% 26.4% 21.8% . 31.0% 28.4% 23.1% . 29.8% 20.2% 20.5% . 35.7% 23.4% 27.4%
months
Mostly . 23.7% 25.3% 26.9% . 14.3% 16.2% 15.4% . 18.1% 21.4% 30.8% . 23.8% 27.3% 21.9%
Completely . 28.0% 34.5% 28.2% . 29.8% 31.1% 35.4% . 34.0% 39.3% 35.9% . 26.2% 40.3% 27.4%
N . 93 87 78 . 84 74 65 . 94 84 78 . 84 77 73
N MISSING . 7 5 0 . 16 18 13 . 6 8 0 . 16 15 5
How MEAN 8.9 8.8 8.9 8.8 9.2 8.6 8.5 8.5 9.0 8.9 9.0 8.8 9.0 8.5 9.1 8.7
motivated
STD 1.6 1.7 1.9 1.8 1.4 2.1 2.2 2.2 1.9 1.8 1.7 2.0 1.8 2.0 1.4 2.0
are you to
overcome MIN 4.0 4.0 0.0 1.0 4.0 1.0 0.0 0.0 0.0 3.0 2.0 0.0 2.0 0.0 5.0 0.0
your Q1 8.0 8.0 8.0 8.0 9.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 9.0 8.0
gambling
problem? MEDIAN 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 9.5 10.0 10.0
(10-point Q3 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0
Likert scale)
MAX 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0
N 116 92 87 78 112 84 74 66 117 94 84 78 115 84 76 73
N MISSING 0 8 5 0 0 4 4 0 1 4 4 0 1 3 6 0
135
Table 4.11: Secondary efficacy outcomes - Workbook reception and use
TAU MI MI+W MI+W+B
Did you Yes . 51.7% 54.2% 65.3% . 21.0% 28.2% 40.6% . 93.4% 89.3% 93.3% . 94.0% 93.5% 93.1%
receive the
N . 89 83 75 . 81 71 64 . 91 84 75 . 83 77 72
workbook?
N MISSING . 11 9 3 . 7 7 2 . 7 4 3 . 4 5 1
Have you MEAN . 2.02 1.40 1.47 . 2.00 4.76 1.35 . 2.21 1.63 1.44 . 2.14 1.64 1.42
read the
STD . 0.77 0.62 0.65 . 0.87 0.77 0.56 . 0.67 0.73 0.69 . 0.66 0.70 0.61
workbook?
MIN . 1.00 1.00 1.00 . 1.00 1.00 1.00 . 1.00 1.00 1.00 . 1.00 1.00 1.00
Q1 . 1.00 1.00 1.00 . 1.00 1.00 1.00 . 2.00 1.00 1.00 . 2.00 1.00 1.00
MEDIAN . 2.00 1.00 1.00 . 2.00 2.00 1.00 . 2.00 1.00 1.00 . 2.00 2.00 1.00
Q3 . 3.00 2.00 2.00 . 3.00 2.00 2.00 . 3.00 2.00 2.00 . 3.00 2.00 2.00
MAX . 3.00 3.00 3.00 . 3.00 3.00 3.00 . 3.00 3.00 3.00 . 3.00 3.00 3.00
N . 47 45 47 . 17 21 26 . 85 75 71 . 78 72 67
N MISSING . 53 47 31 . 71 57 40 . 13 13 7 . 9 10 6
Did you Not . 61.8% 80.2% 79.7% . 86.4% 84.7% 87.9% . 19.6% 55.4% 70.1% . 18.1% 50.6% 67.6%
complete Applicable
some of the None . 21.3% 15.1% 12.2% . 7.4% 5.6% 7.6% . 20.7% 12.0% 9.1% . 24.1% 16.9% 9.9%
exercises in
Some . 12.4% 3.5% 8.1% . 4.9% 8.3% 4.5% . 51.1% 25.3% 15.6% . 44.6% 28.6% 21.1%
the
workbook? All . 4.5% 1.2% 0.0% . 1.2% 1.4% 0.0% . 8.7% 7.2% 5.2% . 13.3% 3.9% 1.4%
N . 89 86 74 . 81 72 66 . 92 83 77 . 83 77 71
N MISSING . 11 6 4 . 7 6 0 . 6 5 1 . 4 5 2
136
Table 4.11: Secondary efficacy outcomes - Workbook reception and use - continued
TAU MI MI+W MI+W+B
Have you Not . 63.2% 80.5% 78.7% . 87.7% 85.9% 89.2% . 21.8% 56.1% 70.1% . 18.5% 50.6% 67.6%
used any of Applicable
the strategies None . 8.0% 10.3% 9.3% . 6.2% 4.2% 1.5% . 25.3% 14.6% 6.5% . 27.2% 19.5% 11.3%
recommended
in the Some . 19.5% 6.9% 8.0% . 3.7% 5.6% 6.2% . 37.9% 11.0% 16.9% . 39.5% 20.8% 19.7%
workbook? All . 9.2% 2.3% 4.0% . 2.5% 4.2% 3.1% . 14.9% 18.3% 6.5% . 14.8% 9.1% 1.4%
N . 87 87 75 . 81 71 65 . 87 82 77 . 81 77 71
N MISSING . 13 5 3 . 7 7 1 . 11 6 1 . 6 5 2
Workbook MEAN . 1.53 0.79 0.82 . 1.35 1.22 0.76 . 1.78 1.20 0.90 . 1.79 1.17 0.86
engagement
STD . 0.90 0.66 0.73 . 0.91 0.93 0.72 . 0.74 0.97 0.86 . 0.74 0.87 0.75
(numerical 1-
3 averaged MIN . 0.33 0.33 0.33 . 0.33 0.33 0.33 . 0.33 0.33 0.33 . 0.33 0.33 0.33
over 3
Q1 . 0.33 0.33 0.33 . 0.33 0.33 0.33 . 1.67 0.33 0.33 . 1.33 0.33 0.33
outcomes)
MEDIAN . 1.67 0.33 0.33 . 1.33 1.17 0.33 . 2.00 0.33 0.33 . 2.00 1.33 0.33
Q3 . 2.33 1.33 1.33 . 2.17 2.00 1.33 . 2.33 2.00 1.67 . 2.33 2.00 1.67
MAX . 3.00 2.67 2.67 . 2.67 3.00 2.33 . 3.00 3.00 3.00 . 3.00 3.00 2.33
N . 43 44 44 . 16 20 25 . 80 73 71 . 76 72 66
N MISSING . 57 48 34 . 72 58 41 . 18 15 7 . 11 10 7
137
Table 4.12: Secondary efficacy outcomes - Treatment service assistance
TAU MI MI+W MI+W+B
Received any Yes - 39.8% 31.0% 29.5% - 36.9% 31.5% 34.8% - 41.5% 32.1% 34.6% - 40.5% 44.2% 35.6%
assistance
N - 93 87 78 - 84 73 66 - 94 84 78 - 84 77 73
(formal and
informal) in past N MISSING - 7 5 0 - 4 5 0 - 4 4 0 - 3 5 0
3 months
Received formal Yes - 28.0% 18.4% 15.4% - 25.0% 23.0% 25.8% - 20.2% 15.5% 16.7% - 23.8% 18.2% 19.2%
assistance from
any treatment N - 93 87 78 - 84 74 66 - 94 84 78 - 84 77 73
service in the
past 3 months N MISSING - 7 5 0 - 4 4 0 - 4 4 0 - 3 5 0
Received Yes - 39.8% 31.0% 29.5% - 36.9% 31.5% 34.8% - 41.5% 32.1% 34.6% - 41.7% 44.2% 35.6%
informal
N - 93 87 78 - 84 73 66 - 94 84 78 - 84 77 73
assistance from
any person in N MISSING - 7 5 0 - 4 5 0 - 4 4 0 - 3 5 0
past 3 months
138
Table 4.13: Secondary efficacy outcomes - Treatment service assistance
Received formal assistance from: TAU MI MI+W MI+W+B
Base- 3 6 12 Base- 3 6 12 Base- 3 6 12 Base- 3 6 12
line months months months line months months months line months months months line months months months
Gambling Yes (N) . 4 2 1 . 3 3 2 . 6 2 3 . 6 2 2
Helpline No Sessions:
3 3 9 4 1 2 1 3.5 1.5 2.5 2 6
Median . . . .
(2-6) (2-4) (9-9) (1-10) (1-1) (2-2) (1-4) (1-6) (1-2) (1-4) (1-3) (2-10)
(Min-Max)
Problem Yes (N) . 9 3 3 . 7 4 3 . 7 4 3 . 4 3 3
Gambling No Sessions:
5 6 12 6 4.5 6 6 4 2 7 12 12
Foundation Median . . . .
(1-13) (2-7) (12-24) (1-10) (1-6) (3-48) (1-12) (1-8) (1-3) (3-12) (12-12) (2-12)
(Min-Max)
Salvation Army Yes (N) . 5 7 2 . 3 1 2 . 1 2 2 . 1 4 2
Oasis Centres No Sessions:
6 7 13.5 1 10 2.5 3 5 4 12 5.5 24
Median . . . .
(2-12) (1-12) (3-24) (1-5) (10-10) (2-3) (3-3) (4-6) (2-6) (12-12) (1-12) (24-24)
(Min-Max)
Gamblers Yes (N) . 4 2 2 . 5 4 5 . 5 3 4 . 1 3 1
Anonymous No Sessions:
7 12 22 12 11 24 12 6 10 1 2 2
Median . . . .
(1.5-12) (12-12) (20-24) (2-12) (10-12) (3-504#) (6-12) (2-12) (4-24) (1-1) (1-3) (2-2)
(Min-Max)
Other problem Yes (N) . 5 3 5 . 5 6 7 . 1 2 3 . 6 4 7
gambling support No Sessions:
6 6 4 2 5 5 12 7.5 6 1.5 4 12
service(s) Median . . . .
(3-12) (2-12) (1-12) (1-12) (1-270) (2-24) (12-12) (3-12) (2-24) (1-12) (1-48) (2-21)
(Min-Max)
Online/internet- Yes (N) . 1 0 0 . 0 0 0 . 0 1 0 . 2 0 0
based service No Sessions:
2
Median . - - . - - - . - NR - . NR - -
(2-2)
(Min-Max)
Total Formal Yes (N)## . 26 16 12 . 21 17 17 . 19 13 13 . 20 14 14
Assistance Yes (%) . 28.0% 18.4% 15.4% . 25.0% 23.0% 25.8% . 20.2% 15.5% 16.7% . 23.8% 18.2% 19.2%
N . 93 87 78 . 84 74 66 . 94 84 78 . 84 77 73
N MISSING . 7 5 0 . 4 4 0 . 4 4 0 . 3 5 0
#
Involved multiple text messages per day
##
Note some participants received assistance from multiple agencies, therefore not mutually exclusive and total number of participants does not equal the sum of individual agency assistance
NR Not reported
139
APPENDIX 5
Tables - Subgroup analyses
Table 5.1: TAU vs. MI Days Gambled, Money Lost by gender

TEST 95% Confidence
Limits
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 Male -0.14 2.88
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 Male -0.13 15.70
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 Female -1.34 1.36
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 Female -8.44 5.67
Table 5.2: TAU vs. MI Gambling-quit or improved by gender

TEST Odds Odds Odds
ratio Ratio Ratio
CILB CIUB
Gambling-quit or hyp. A : TAU vs MI Male 0.36 0.09 1.42
improved, time-averaged
Gambling-quit or hyp. A : TAU vs MI Female 1.04 0.22 3.76
Table 5.3: Hypotheses B and C - Days Gambled, Money Lost , PGSI - females
change error (one-sided) accepted
hyp. B.a: TAU vs MI+W 0.18 0.66 0.61 No
hyp. B.b: MI vs MI+W 0.17 0.71 0.59 No
Days Gambled,
time-averaged hyp. B.c: TAU vs MI+W+B 0.00 0.70 0.50 No
Money Lost, time- hyp. B.a: TAU vs MI+W -1.24 3.46 0.36 No
averaged
hyp. B.b: MI vs MI+W 0.14 3.69 0.52 No

Days Gambled, at hyp. C.b: MI vs MI+W+B -0.12 0.87 0.44 No
12 months
Money Lost, at 12 hyp. C.b: MI vs MI+W+B -2.32 4.49 0.30 No
months
hyp. B.b: MI vs MI+W 0.40 1.40 0.61 No
PGSI-12, at 12 hyp. C.a: TAU vs MI+W+B 2.63 1.36 0.97 No
months
140
Table 5.4: Hypotheses B and C - Gambling-quit or improved by gender
Males
ratio Ratio Ratio (one-sided) accepted
CILB CIUB
hyp. B.a: TAU vs MI+W 0.59 0.15 2.31 0.77 No
Gambling-quit or hyp. B.b: MI vs MI+W 1.63 0.45 5.97 0.23 No
improved, time-
hyp. B.c: TAU vs MI+W+B 0.63 0.16 2.39 0.75 No
averaged
hyp. C.a: TAU vs MI+W+B 0.64 0.14 2.95 0.72 No
Gambling-quit or
hyp. C.b: MI vs MI+W+B 1.80 0.42 7.76 0.21 No
improved, at 12
months hyp. C.c: MI+W vs 0.08 0.02 0.46 1.00 No
MI+W+B
Females
CILB CIUB
improved, time-
averaged
Gambling-quit or
improved, at 12
MI+W+B
Table 5.5: TAU vs. MI Days Gambled, Money Lost by ethnicity

TEST 95% Confidence
Limits
hyp. A : TAU vs MI, δ=1 European -0.78 1.97
hyp. A : TAU vs MI, δ=1 Maori -0.98 2.39
Days Gambled, time-averaged
hyp. A : TAU vs MI, δ=1 Pacific -3.89 2.00
hyp. A : TAU vs MI, δ=1 Asian & Other -4.33 6.53
hyp. A : TAU vs MI, δ=20 European -6.26 8.25
hyp. A : TAU vs MI, δ=20 Maori -1.53 16.09
Money Lost, time-averaged
hyp. A : TAU vs MI, δ=20 Pacific -18.35 12.82
hyp. A : TAU vs MI, δ=20 Asian & Other -20.39 37.03
141
Table 5.6: TAU vs. MI Gambling-quit or improved by ethnicity
TEST Odds Odds Odds
ratio Ratio Ratio
CILB CIUB
hyp. A : TAU vs MI European 0.49 0.15 1.65
hyp. A : TAU vs MI Maori 0.84 0.20 3.44
Gambling-quit or
improved, time-averaged hyp. A : TAU vs MI Pacific 0.36 0.02 5.72
hyp. A : TAU vs MI Asian & Other Number of observations too
small
142
Table 5.7: Hypotheses B and C - Days Gambled, Money Lost , PGSI - by ethnicity
hyp. B.a: TAU vs MI+W European -0.18 0.67 0.40 No

hyp. B.a: TAU vs MI+W Maori 0.08 0.78 0.54 No
hyp. B.a: TAU vs MI+W Pacific 0.76 1.66 0.68 No
hyp. B.a: TAU vs MI+W Asian & 2.38 2.57 0.82 No
Other
hyp. B.b: MI vs MI+W European -0.77 0.70 0.14 No
hyp. B.b: MI vs MI+W Maori -0.62 0.86 0.23 No
hyp. B.b: MI vs MI+W Pacific 1.71 1.52 0.87 No
hyp. B.b: MI vs MI+W Asian & Other 1.28 2.57 0.69 No
Days Gambled,
time-averaged hyp. B.c: TAU vs MI+W+B European -0.38 0.69 0.29 No
hyp. B.c: TAU vs MI+W+B Maori -0.34 0.82 0.34 No
hyp. B.c: TAU vs MI+W+B Pacific 0.57 1.63 0.64 No
hyp. B.c: TAU vs MI+W+B Asian & 0.38 2.52 0.56 No
Other
hyp. B.d: MI vs MI+W+B European -0.98 0.72 0.09 No
hyp. B.d: MI vs MI+W+B Maori -1.05 0.89 0.12 No
hyp. B.d: MI vs MI+W+B Pacific 1.52 1.50 0.84 No
hyp. B.d: MI vs MI+W+B Asian & -0.72 2.52 0.39 No
Other
hyp. B.a: TAU vs MI+W Maori -1.37 4.08 0.37 No
Other
hyp. B.b: MI vs MI+W Asian & Other -6.14 13.48 0.32 No
Money Lost,
time-averaged hyp. B.c: TAU vs MI+W+B European -0.68 3.62 0.43 No
hyp. B.c: TAU vs MI+W+B Maori 0.23 4.30 0.52 No
hyp. B.c: TAU vs MI+W+B Pacific 0.27 8.66 0.51 No
hyp. B.c: TAU vs MI+W+B Asian & 3.01 13.28 0.59 No
Other
hyp. B.d: MI vs MI+W+B European -5.35 3.97 0.09 No
hyp. B.d: MI vs MI+W+B Maori -7.05 4.68 0.07 No
hyp. B.d: MI vs MI+W+B Pacific 3.03 7.87 0.65 No
hyp. B.d: MI vs MI+W+B Asian & -5.32 13.24 0.34 No
Other
143
Table 5.7: Hypotheses B and C - Days Gambled, Money Lost, PGSI - by ethnicity -
continued
hyp. C.a: TAU vs MI+W+B European -0.55 0.79 0.24 No

hyp. C.a: TAU vs MI+W+B Maori -0.24 0.96 0.40 No
hyp. C.a: TAU vs MI+W+B Pacific 0.71 2.08 0.63 No
hyp. C.a: TAU vs MI+W+B Asian & 0.61 2.86 0.58 No

Other
hyp. C.b: MI vs MI+W+B European -0.53 0.84 0.27 No
hyp. C.b: MI vs MI+W+B Maori -2.08 1.06 0.02 No
Days Gambled,
at 12 months hyp. C.b: MI vs MI+W+B Pacific 0.05 1.84 0.51 No
hyp. C.b: MI vs MI+W+B Asian & -0.51 2.86 0.43 No
Other
European
hyp. C.c: MI+W vs MI+W+B Maori -0.57 0.94 0.27 No
hyp. C.c: MI+W vs MI+W+B Pacific -2.65 2.00 0.09 No
hyp. C.c: MI+W vs MI+W+B Asian & 1.87 2.77 0.75 No
Other
hyp. B.a: TAU vs MI+W Maori 0.10 1.58 0.48 No
Other
hyp. B.b: MI vs MI+W Asian & Other -8.07 5.00 0.05 No
hyp. B.c & C.a: TAU vs MI+W+B -1.21 1.33 0.18 No
European
hyp. B.c & C.a: TAU vs MI+W+B 2.06 1.65 0.89 No
Maori
PGSI-12, at 12 hyp. B.c & C.a: TAU vs MI+W+B 6.34 3.57 0.96 No
months Pacific
hyp. B.c & C.a: TAU vs MI+W+B 2.87 4.67 0.73 No
Asian & Other
hyp. B.d & C.b: MI vs MI+W+B -0.95 1.40 0.25 No
European
hyp. B.d & C.b: MI vs MI+W+B Maori 1.22 1.82 0.75 No
hyp. B.d & C.b: MI vs MI+W+B 3.93 3.15 0.89 No
Pacific
hyp. B.d & C.b: MI vs MI+W+B Asian -5.79 4.67 0.11 No
& Other
European
hyp. C.c: MI+W vs MI+W+B Maori 1.97 1.59 0.89 No
hyp. C.c: MI+W vs MI+W+B Pacific -1.27 3.38 0.35 No
hyp. C.c: MI+W vs MI+W+B Asian & 2.28 4.66 0.69 No
Other
144
Table 5.8: Hypotheses B and C - Gambling-quit or improved by ethnicity
CILB CIUB
European
hyp. B.a: TAU vs MI+W Maori 1.71 0.44 6.54 0.22 No
hyp. B.a: TAU vs MI+W Pacific 0.13 0.01 2.44 0.91 No
hyp. B.a: TAU vs MI+W Asian &
Number of observations too small
Other
hyp. B.b: MI vs MI+W European 2.40 0.71 8.15 0.08 No
hyp. B.b: MI vs MI+W Maori 2.03 0.48 8.61 0.17 No
hyp. B.b: MI vs MI+W Pacific Unreliable results due to numerical instability
hyp. B.b: MI vs MI+W Asian &
Gambling-quit or Other
improved, hyp. B.c: TAU vs MI+W+B 1.10 0.33 3.71 0.44 No
time-averaged European
Maori
hyp. B.c: TAU vs MI+W+B
Unreliable results due to numerical instability
Pacific
hyp. B.c: TAU vs MI+W+B Asian
& Other
European
hyp. B.d: MI vs MI+W+B Maori 2.99 0.77 11.60 1.00 No
hyp. B.d: MI vs MI+W+B Pacific 1.14 0.20 6.63 1.00 No
hyp. B.d: MI vs MI+W+B Asian
& Other
European
Maori
hyp. C.a: TAU vs MI+W+B
Pacific
hyp. C.a: TAU vs MI+W+B Asian
& Other
European
Gambling-quit or hyp. C.b: MI vs MI+W+B Maori 3.15 0.61 16.24 0.09 No
improved, at 12
hyp. C.b: MI vs MI+W+B Pacific Unreliable results due to numerical instability
months
hyp. C.b: MI vs MI+W+B Asian
& Other
hyp. C.c: MI+W vs MI+W+B 0.29 0.07 1.27 0.95 No
European
hyp. C.c: MI+W vs MI+W+B 0.27 0.05 1.37 0.94 No
Maori
hyp. C.c: MI+W vs MI+W+B
Pacific
hyp. C.c: MI+W vs MI+W+B
Asian & Other
145
Table 5.9: TAU vs. MI Days Gambled, Money Lost by gambling mode
TEST 95% Confidence
Limits
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 EGM -0.57 1.55
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 EGM -2.26 8.90
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 Non-EGM -1.50 4.80
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 Non-EGM -17.90 15.15
Table 5.10: TAU vs. MI Gambling-quit or improved by gambling mode

TEST Odds Odds Odds
ratio Ratio Ratio
CILB CIUB
Gambling-quit or hyp. A : TAU vs MI EGM 0.87 0.35 2.18
Gambling-quit or hyp. A : TAU vs MI Non-EGM 0.06 0.00 1.29
Table 5.11: Hypotheses B and C - Days Gambled, Money Lost, PGSI by gambling mode
EGM
hyp. B.b: MI vs MI+W -0.13 0.55 0.41 No
Days Gambled,
time-averaged hyp. B.c: TAU vs MI+W+B -0.03 0.54 0.48 No
averaged
hyp. B.b: MI vs MI+W -4.37 2.89 0.07 No

12 months
months
hyp. B.b: MI vs MI+W 0.04 1.11 0.49 No
PGSI-12, at 12 hyp. B.c & C.a: TAU vs 1.14 1.09 0.15 No

months MI+W+B
hyp. B.d & C.b: MI vs 0.75 1.13 0.25 No
MI+W+B
Non-EGM
146
hyp. B.b: MI vs MI+W -1.29 1.49 0.19 No
Days Gambled,
averaged
hyp. B.b: MI vs MI+W -1.05 7.84 0.45 No
12 months
months
hyp. B.a: TAU vs MI+W -1.82 2.74 0.25 No
hyp. B.b: MI vs MI+W -2.56 2.75 0.18 No
PGSI-12, at 12 hyp. B.c & C.a: TAU vs -0.60 2.79 0.41 No

months MI+W+B
hyp. B.d & C.b: MI vs -1.34 2.79 0.32 No
MI+W+B
Table 5.12: Hypotheses B and C - Gambling-quit or improved by gambling mode
EGM
CILB CIUB
improved, time-
averaged
Gambling-quit or
improved, at 12
MI+W+B
147
Non-EGM
CILB CIUB
improved, time-
averaged
hyp. C.a: TAU vs MI+W+B
Gambling-quit or
hyp. C.b: MI vs MI+W+B
improved, at 12 Number of valid observations too small
months hyp. C.c: MI+W vs
MI+W+B
Table 5.13: TAU vs. MI Days Gambled, Money Lost by dichotomised baseline PGSI score
TEST 95% Confidence
Limits
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 Baseline ≤ 17.0 -1.11 1.61
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 Baseline ≤ 17.0 -4.31 10.06
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 Baseline > 17.0 -0.60 2.39
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 Baseline > 17.0 -5.95 9.77
Table 5.14: TAU vs. MI Gambling-quit or improved by dichotomised baseline PGSI

score
TEST Odds Odds Odds
ratio Ratio Ratio
CILB CIUB
Gambling-quit or hyp. A : TAU vs MI Baseline ≤ 17.0 0.43 0.13 1.44
Gambling-quit or hyp. A : TAU vs MI Baseline > 17.0 1.26 0.368 4.35
148
Table 5.15: Hypotheses B and C - Days Gambled, Money Lost , PGSI by baseline PGSI ≤
17
hyp. B.b: MI vs MI+W 0.58 0.71 0.79 No
Days Gambled,
Money Lost, time- hyp. B.a: TAU vs MI+W 1.52 3.43 0.67 No
averaged
hyp. B.b: MI vs MI+W -1.35 3.72 0.36 No

12 months
Money Lost, at 12 hyp. C.b: MI vs MI+W+B 1.48 4.36 0.63 No
months
hyp. B.b: MI vs MI+W 0.33 1.43 0.59 No

months MI+W+B
hyp. B.d & C.b: MI vs 1.42 1.48 0.83 No
MI+W+B
Table 5.16: Hypotheses B and C - Gambling-quit or improved by baseline PGSI ≤ 17

CILB CIUB
improved, time-
averaged
Gambling-quit or
improved, at 12
MI+W+B
149
Table 5.17: TAU vs. MI Days Gambled, Money Lost by baseline Kessler-10 score
TEST 95% Confidence
Limits
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 Baseline ≤ 30 -1.11 1.67
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 Baseline ≤ 30 -6.31 8.45
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 Baseline > 30 -0.48 2.43
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 Baseline > 30 -2.54 12.87
Table 5.18: TAU vs. MI Gambling-quit or improved by baseline Kessler-10 score

TEST Odds Odds Odds
ratio Ratio Ratio
CILB CIUB
Gambling-quit or hyp. A : TAU vs MI Baseline ≤ 30 1.18 0.35 4.01
Gambling-quit or hyp. A : TAU vs MI Baseline > 30 0.40 0.12 1.36
Table 5.19: Hypotheses B and C - Days Gambled, Money Lost, PGSI by baseline K-10 ≤
30
hyp. B.b: MI vs MI+W 0.71 0.72 0.84 No
Days Gambled,
averaged
hyp. B.b: MI vs MI+W -2.18 3.82 0.28 No

12 months
months
hyp. B.b: MI vs MI+W 0.47 1.44 0.63 No

months MI+W+B
hyp. B.d & C.b: MI vs 1.96 1.43 0.92 No
MI+W+B
150
Table 5.20: Hypotheses B and C - Gambling-quit or improved by baseline K-10 ≤ 30
CILB CIUB
improved, time-
averaged
Gambling-quit or
improved, at 12
MI+W+B
Table 5.21: TAU vs. MI Days Gambled, Money Lost by baseline AUDIT-C score
TEST 95% Confidence
Limits
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 Low risk -1.78 1.79
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 Low risk -8.43 10.32
Days Gambled, time-averaged hyp. A : TAU vs MI, δ=1 High risk -0.34 2.11
Money Lost, time-averaged hyp. A : TAU vs MI, δ=20 High risk -2.92 9.96
Table 5.22: TAU vs. MI Gambling-quit or improved by baseline AUDIT-C score

TEST Odds Odds Odds
ratio Ratio Ratio
CILB CIUB
Gambling-quit or hyp. A : TAU vs MI Low risk 0.27 0.06 1.26
Gambling-quit or hyp. A : TAU vs MI High risk 1.07 0.37 3.09
151
Table 5.23: Hypotheses B and C - Days Gambled, Money Lost, PGSI by baseline AUDIT-C
Low risk
hyp. B.b: MI vs MI+W 1.45 0.91 0.94 No
Days Gambled,
averaged
hyp. B.b: MI vs MI+W 1.79 4.77 0.65 No

Days Gambled, at hyp. C.b: MI vs MI+W+B 0.50 1.11 0.67 No
12 months
months
hyp. B.b: MI vs MI+W 0.11 1.86 0.52 No

months MI+W+B
hyp. B.d & C.b: MI vs 1.91 1.90 0.84 No
MI+W+B
152
High risk
hyp. B.b: MI vs MI+W -1.21 0.65 0.03 No
Days Gambled,
averaged
hyp. B.b: MI vs MI+W -6.97 3.35 0.019 No

12 months
months
hyp. B.b: MI vs MI+W -0.43 1.33 0.37 No

months MI+W+B
hyp. B.d & C.b: MI vs -0.42 1.32 0.38 No
MI+W+B
Table 5.24: Hypotheses B and C - Gambling-quit or improved by high risk AUDIT-C

score
CILB CIUB
improved, time-
averaged
Gambling-quit or
improved, at 12
MI+W+B
153
Table 5.25: Hypotheses B and C - Days Gambled, Money Lost , PGSI by baseline quit
gambling goal
hyp. B.b: MI vs MI+W -0.50 0.58 0.19 No
Days Gambled,
averaged
hyp. B.b: MI vs MI+W -4.04 3.06 0.09 No

12 months
months
hyp. B.b: MI vs MI+W -0.69 1.15 0.27 No

months MI+W+B
hyp. B.d & C.b: MI vs 1.33 1.13 0.88 No
MI+W+B
Table 5.26: Hypotheses B and C - Gambling-quit or improved by quit gambling goal

CILB CIUB
improved, time-
averaged
Gambling-quit or
improved, at 12
MI+W+B
154
Table 5.27: Hypotheses B and C - Days Gambled, Money Lost, PGSI by belief in treatment
success
Low belief
hyp. B.b: MI vs MI+W 0.18 0.71 0.60 No
Days Gambled,
averaged
hyp. B.b: MI vs MI+W -3.10 3.84 0.21 No

12 months
months
hyp. B.b: MI vs MI+W -1.03 1.52 0.25 No

months MI+W+B
hyp. B.d & C.b: MI vs -0.18 1.56 0.45 No
MI+W+B
155
High belief
hyp. B.b: MI vs MI+W -0.74 0.70 0.15 No
Days Gambled,
averaged
hyp. B.b: MI vs MI+W -5.37 3.77 0.08 No

Days Gambled, at hyp. C.b: MI vs MI+W+B 0.24 0.83 0.61 No
12 months
months
hyp. B.b: MI vs MI+W 0.10 1.42 0.53 No

months MI+W+B
hyp. B.d & C.b: MI vs 0.88 1.43 0.73 No
MI+W+B
Table 5.28: Hypotheses B and C - Gambling-quit or improved by high belief treatment

success
CILB CIUB
improved, time-
averaged
Gambling-quit or
improved, at 12
MI+W+B
156
Table 5.29: Hypotheses B and C - Goal met in past 3-months
CILB CIUB
Goal met in the hyp. B.b: MI vs MI+W 1.54 0.87 2.75 0.07 No
past 3-months,
time-averaged

ratio Ratio Ratio (two- accepted
CILB CIUB sided)
Goal met in the hyp. C.a: TAU vs MI+W+B 0.89 0.33 2.42 0.83 No
past 3-months, at
12 months
hyp. C.c: MI+W vs 0.57 0.21 1.52 0.26 No
MI+W+B
157
APPENDIX 6
Tables - Secondary analyses
Table 6.1: Hypotheses B and C - Motivation to overcome gambling

Motivation, time- hyp. B.a: TAU vs MI+W 0.06 0.20 0.39 No
averaged
hyp. B.b: MI vs MI+W 0.47 0.21 0.014 No*
Motivation, at 12 hyp. C.a: TAU vs MI+W+B -0.09 0.30 0.61 No
months
* FDR Threshold is 0.0125
158
Table 6.2: Hypotheses B and C - Kessler-10, AUDIT-C, DAST, WHOQoL and NZDI
Kessler-10, time- hyp. B.a: TAU vs MI+W 0.13 0.93 0.56 No
averaged
hyp. B.b: MI vs MI+W -0.48 0.97 0.31 No
Kessler-10, at 12 hyp. C.a: TAU vs MI+W+B -0.63 1.22 0.26 No
months
AUDIT-C, time- hyp. B.a: TAU vs MI+W -0.16 0.32 0.31 No
averaged
hyp. B.b: MI vs MI+W 0.19 0.34 0.71 No
AUDIT-C, at 12 hyp. C.a: TAU vs MI+W+B 0.32 0.38 0.80 No
months
DAST, at 12 hyp. B.a: TAU vs MI+W -0.11 0.18 0.26 No
months
hyp. B.b: MI vs MI+W -0.05 0.18 0.40 No
hyp. B.c & C.a: TAU vs -0.21 0.18 0.13 No
MI+W+B
hyp. B.d & C.b: MI vs MI+W+B -0.14 0.19 0.23 No
WHO-QOL-8, hyp. B.a: TAU vs MI+W 0.21 0.65 0.37 No
time-averaged
hyp. B.b: MI vs MI+W 0.50 0.67 0.23 No
WHOQoL-8, at hyp. C.a: TAU vs MI+W+B 0.74 0.82 0.18 No
12 months
NZDI, at 12 hyp. B.a: TAU vs MI+W 0.08 0.16 0.70 No
months
hyp. B.b: MI vs MI+W 0.11 0.16 0.74 No
hyp. B.c & C.a: TAU vs 0.02 0.16 0.56 No
MI+W+B
hyp. B.d & C.b: MI vs MI+W+B 0.05 0.17 0.62 No
159
Table 6.3: Hypotheses B and C - PRIME-MD
CILB CIUB
PRIME-MD major hyp. B.a: TAU vs MI+W 1.09 0.48 2.52 0.58 No
depressive
hyp. B.b: MI vs MI+W 1.47 0.64 3.38 0.82 No
disorder, at 12
months hyp. B.c & C.a: TAU vs 0.85 0.37 1.91 0.34 No
MI+W+B
hyp. B.d & C.b: MI vs 1.14 0.50 2.56 0.62 No
MI+W+B
hyp. C.c: MI+W vs 0.77 0.34 1.77 0.27 No
MI+W+B
PRIME-MD hyp. B.a: TAU vs MI+W 0.72 0.37 1.38 0.16 No
dysthymia, at 12
hyp. B.b: MI vs MI+W 0.66 0.33 1.32 0.12 No
months
hyp. B.c & C.a: TAU vs 1.15 0.57 2.30 0.65 No
MI+W+B
hyp. B.d & C.b: MI vs 1.06 0.51 2.20 0.56 No
MI+W+B
hyp. C.c: MI+W vs 1.60 0.81 3.17 0.91 No
MI+W+B
PRIME-MD minor hyp. B.a: TAU vs MI+W 1.52 0.25 9.43 0.67 No
depressive
hyp. B.b: MI vs MI+W 1.81 0.29 11.25 0.74 No
disorder, at 12
months hyp. B.c & C.a: TAU vs 2.84 0.29 28.21 0.81 No
MI+W+B
hyp. B.d & C.b: MI vs 3.38 0.34 33.66 0.85 No
MI+W+B
hyp. C.c: MI+W vs 1.87 0.16 21.27 0.69 No
MI+W+B
PRIME-MD hyp. B.a: TAU vs MI+W 0.57 0.13 2.51 0.23 No
bipolar disorder, at
hyp. B.b: MI vs MI+W 0.43 0.08 2.32 0.16 No
12 months
hyp. B.c & C.a: TAU vs 0.54 0.12 2.37 0.21 No
MI+W+B
hyp. B.d & C.b: MI vs 0.41 0.08 2.19 0.15 No
MI+W+B
hyp. C.c: MI+W vs 0.94 0.26 3.42 0.46 No
MI+W+B
160
Table 6.4: Hypotheses B and C - Tobacco use
Odds ratios of currently not smoking

CILB CIUB
Tobacco current, hyp. B.a: TAU vs MI+W 1.18 0.64 2.16 0.30 No
time-averaged
hyp. B.b: MI vs MI+W 0.79 0.42 1.49 0.76 No
Tobacco current, hyp. C.a: TAU vs MI+W+B 0.89 0.29 2.71 0.58 No
at 12 months
hyp. C.c: MI+W vs 0.84 0.28 2.57 0.62 No
MI+W+B
Odds ratios of decreasing smoking frequency

CILB CIUB sided)
Tobacco hyp. B.a: TAU vs MI+W 1.54 0.25 9.38 0.64 No
frequency, time-
hyp. B.b: MI vs MI+W 0.62 0.12 3.14 0.56 No
averaged
Tobacco hyp. C.a: TAU vs MI+W+B 1.50 0.23 9.63 0.67 No
frequency, at 12
months
hyp. C.c: MI+W vs 1.49 0.23 9.58 0.67 No
MI+W+B
161
Table 6.5: Hypotheses B and C - Treatment for co-existing issues
CILB CIUB sided)
Mental health hyp. B.a: TAU vs MI+W 0.92 0.41 2.07 0.42 No
treatment at 12
hyp. B.b: MI vs MI+W 1.03 0.45 2.37 0.53 No
months
hyp. B.c & C.a: TAU vs 1.11 0.48 2.60 0.60 No
MI+W+B
hyp. B.d & C.b: MI vs 1.25 0.52 2.98 0.69 No
MI+W+B
hyp. C.c: MI+W vs 1.21 0.52 2.80 0.67 No
MI+W+B
Prescribed hyp. B.a: TAU vs MI+W 0.60 0.28 1.30 0.10 No
medication at 12
hyp. B.b: MI vs MI+W 1.10 0.53 2.29 0.60 No
months
hyp. B.c & C.a: TAU vs 0.92 0.40 2.10 0.42 No
MI+W+B
hyp. B.d & C.b: MI vs 1.67 0.76 3.70 0.90 No
MI+W+B
hyp. C.c: MI+W vs 1.53 0.71 3.30 0.86 No
MI+W+B
Alcohol or drug hyp. B.a: TAU vs MI+W 0.58 0.13 2.55 0.24 No
treatment at 12
hyp. B.b: MI vs MI+W 1.46 0.42 5.05 0.73 No
months
hyp. B.c & C.a: TAU vs 0.53 0.12 2.31 0.20 No
MI+W+B
hyp. B.d & C.b: MI vs 1.32 0.38 4.57 0.67 No
MI+W+B
hyp. C.c: MI+W vs 0.90 0.25 3.28 0.44 No
MI+W+B
162
Table 6.6: Hypotheses B and C - Gambling impacts
Work impact, hyp. B.a: TAU vs MI+W 0.08 0.29 0.61 No
time-averaged
hyp. B.b: MI vs MI+W -0.10 0.30 0.37 No
Work impact, at hyp. C.a: TAU vs MI+W+B -0.31 0.40 0.22 No
12 months
Social impact, hyp. B.a: TAU vs MI+W 0.33 0.32 0.85 No
time-averaged
hyp. B.b: MI vs MI+W -0.40 0.33 0.12 No
Social impact, at hyp. C.a: TAU vs MI+W+B 0.50 0.46 0.86 No
12 months
Family/home hyp. B.a: TAU vs MI+W 0.04 0.36 0.54 No
impact, time-
averaged hyp. B.b: MI vs MI+W -0.24 0.38 0.27 No
Family/home hyp. C.a: TAU vs MI+W+B 0.47 0.50 0.83 No
impact, at 12
months
Health impact, hyp. B.a: TAU vs MI+W -0.12 0.31 0.35 No
time-averaged
hyp. B.b: MI vs MI+W -0.26 0.32 0.21 No
Health impact, at hyp. C.a: TAU vs MI+W+B 0.04 0.42 0.54 No
12 months
163
Table 6.7: Hypotheses B and C - Legal problems
CILB CIUB sided)
Legal problems, hyp. B.a: TAU vs MI+W 1.09 0.32 3.70 0.55 No
time-averaged
hyp. B.b: MI vs MI+W 1.26 0.34 4.74 0.63 No
Legal problems, at hyp. C.a: TAU vs MI+W+B 1.08 0.16 7.26 0.53 No
12 months
hyp. C.c: MI+W vs 1.06 0.16 7.16 0.52 No
MI+W+B
Table 6.8: Hypotheses B and C - Other formal service engagement

ratio Lower Upper (two- accepted
sided)
Other service hyp. F.a: MI+W vs 0.80 0.33 1.94 0.62 No
engagement, MI+W+B
time-averaged hyp. F.b: TAU vs MI+W 1.24 0.52 2.95 0.62 No
hyp. G.a: TAU/MI vs 1.79 0.52 6.19 0.36 No
MI+W/MI+W+B
Other service hyp. G.b: TAU/MI vs 1.98 0.40 9.76 0.40 No
engagement, at 3 MI+W/MI+W+B
months
164
APPENDIX 7
Tables - Call timings
Table 7.1: Intervention delivery timing (minutes)

Intervention group
TAU MI MI+W MI+W+B
MEAN 33.5 34.6 37.7 32.5
STD 16.1 17.3 19.4 14.8
MIN 10 5 10 6
Q1 20 21 20 20
MEDIAN 30 34 32.5 30
Q3 45 45 50 45
MAX 80 80 105 80
N 102 104 110 108
N MISSING 14 8 8 8
165
Table 7.2: Follow-up assessment timing (days)
TAU MI MI+W MI+W+B
3 6 12 3 6 12 3 6 12 3 6 12
Follow-up MEAN 96.3 185.7 369.2 94.8 183.2 367.8 94.4 183.7 369.0 97.1 185.6 368.7
assessment
STD 10.9 9.9 15.8 6.5 7.3 16.3 7.9 8.7 16.9 13.2 10.3 17.1
MIN 60 170 339 87 170 350 77 168 351 59 171 351
Q1 91 180 361 90 180 361 90 180 361 90.5 180 360
MEDIAN 93 183 364 92 181 362 92 181 364 92 182 363
Q3 99 186 369 98 186 368 96 185 370 101 187 370
MAX 137 223 431 118 217 456 123 224 466 157 237 476
N 93 87 78 84 74 66 94 84 78 84 77 73
N MISSING 7 5 0 4 4 0 4 4 0 3 5 0
166
Table 7.3: Booster call timing (days)
Booster calls
1 week 1 month 3 months 6 months
MEAN 8.9 34.0 96.0 186.1
STD 4.7 7.7 10.2 8.2
MIN 6 21 84 168
Q1 7 28 91 182
MEDIAN 7 31.5 93 183
Q3 8 40 98 187
MAX 36 64 139 223
N 69 70 66 58
N date not recorded 10 8 7 4
N successful calls 79 78 73 62
N unsuccessful calls 15 20 29 34
N not contacted 22 18 14 20
167

Effectiveness Problem Gambling Brief Telephone Interventions

Uploaded by

Copyright:

Available Formats

Effectiveness Problem Gambling Brief Telephone Interventions

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Effectiveness Problem Gambling Brief Telephone Interventions

Uploaded by

Copyright:

Available Formats

EFFECTIVENESS OF PROBLEM GAMBLING BRIEF

TELEPHONE INTERVENTIONS: A RANDOMISED

Provider Number: 467589

Contract Number: 326673/00 and 326673/01

Table 1: Number of booster calls received..............................................................................41

Figure 1: Participant flow.......................................................................................................40

The primary hypotheses are:

Problem gambling is a significant public health issue, contributing to a broad spectrum of

The four groups are:

The primary hypotheses were:

Some interventions, however, correspond to the “possibly efficacious” category (Chambless

Barriers to the development of evidence-based treatments identified by Westphal and Abbott

3.1 Ethics approval

3.2 Trial design

3.3.1 Eligibility criteria

The inclusion criteria were:

Callers were excluded from the trial if:

3.3.2 Setting and location

3.4.2 Group 2: Single brief motivational interview (MI)

3.4.3 Group 3: MI plus self-help workbook (MI+W)

3.4.4 Group 4: MI+W plus four follow-up motivational booster

3.4.5 Treatment integrity and fidelity

Assessment of the recordings was based on three categories: Motivational interviewing

The motivational interviewing elements included:

TAU only elements included:

3.5 Outcome measures

Primary outcome measures were self-reports of:

3.5.1 Initial assessment

Gambling/problem gambling history, impacts and past help-seeking

Comorbidity and substance use

Motivation and perceived control over gambling

Gambling/problem gambling history

Comorbidity and substance use

3.5.2 Follow-up assessments

Participants were contacted by an AUT researcher to complete a follow-up assessment at

3.5.3 Collateral assessments

3.6 Sample size

A power of 80% in finding a primary hypothesised effect is considered acceptable in most

Mean baseline Average Intra-participant Minimal

To allocate participants to intervention groups, a computer-generated block randomisation

3.9 Trial hypotheses and statistical methods

3.9.1 Study hypotheses

Primary superiority hypotheses

C. (Superiority is associated to lower values in the hypothesis statements below)

D. (Superiority is associated to lower values in the generic hypothesis statements below)

Engagement secondary hypotheses

3.9.2 Study endpoints

Safety and tolerability endpoints

3.9.3 Analysis sets

Baseline data collected post-randomisation

There were no other considerations of effect modification.

Family-wise error rate adjustment

Inclusion of treatment group; univariate and multivariate settings

A nested, zero-mean, participant-specific normally distributed random effect was assigned to

Absence of repeated measures

Alternative analytical frameworks for continuous endpoints under non-normality I:

If non-normality of residuals was demonstrated or a non-normal family and/or non-identity

Alternative analytical frameworks for continuous endpoints under non-normality II:

Analytical framework for dichotomous (polytomous) endpoint analysis