Cite this article as: BMJ, doi:10.1136/bmj.38940.355602.

80 (published 25 August 2006)

Research

BMJ: first published as 10.1136/bmj.38940.355602.80 on 25 August 2006. Downloaded from https://www.bmj.com/ on 21 November 2024 by guest. Protected by copyright.
BMJ

Telephone administered cognitive behaviour therapy for treatment

of obsessive compulsive disorder: randomised controlled
non-inferiority trial
Karina Lovell, Debbie Cox, Gillian Haddock, Christopher Jones, David Raines, Rachel Garvey, Chris Roberts, Sarah
Hadley

Abstract access to patients who cannot attend clinic appointments for

geographical, social, medical, or economic reasons. Telephone
Objectives To compare the effectiveness of cognitive behaviour delivery of cognitive behaviour therapy is growing.7–9 A pilot
therapy delivered by telephone with the same therapy given study of telephone delivery of such treatment in obsessive com-
face to face in the treatment of obsessive compulsive disorder. pulsive disorder showed potential with regard to effectiveness
Design Randomised controlled non-inferiority trial. and reduced therapist time, and a larger open study found a
Setting Two psychology outpatient departments in the United good outcome.10 11
Kingdom.
Participants 72 patients with obsessive compulsive disorder.
Intervention 10 weekly sessions of exposure therapy and Methods
response prevention delivered by telephone or face to face. Design, objectives, and randomisation
Main outcome measures Yale Brown obsessive compulsive We carried out a randomised controlled non-inferiority trial that
disorder scale, Beck depression inventory, and client satisfaction compared exposure therapy and response prevention delivered
questionnaire. either face to face during traditional 60 minute appointments or
Results Difference in the Yale Brown obsessive compulsive by telephone with reduced contact with the therapist. We
disorder checklist score between the two treatments at six hypothesised that exposure therapy and response prevention
months was − 0.55 (95% confidence interval − 4.26 to 3.15). delivered by either of these methods will have similar clinical
Patient satisfaction was high for both forms of treatment. outcomes in the treatment of obsessive cognitive disorder.
Conclusion The clinical outcome of cognitive behaviour
therapy delivered by telephone was equivalent to treatment Participants
delivered face to face and similar levels of satisfaction were We recruited patients during 2001 and 2002 from two psychol-
reported. ogy outpatient treatment units in greater Manchester. All
Trial registration Current Controlled Trials patients were assessed at screening clinics, and patients whose
ISRCTN500103984. main problem was obsessive compulsive disorder were invited to
take part. Inclusion criteria were diagnosis of obsessive compul-
sive disorder; obsessive compulsive disorder as the main
Introduction presenting problem; score of 16 or more on the Yale Brown
Obsessive compulsive disorder is a disabling mental health obsessive compulsive checklist; and age 16-65. We excluded
illness that tends to be chronic unless adequately treated.1 The patients who had obsessional slowness (a variant of obsessive
economic burden of this disorder is high—the estimated direct compulsive disorder), organic brain disease, a diagnosis of
and indirect costs are $8.4m (£4.5m, €6.6m) in the United States substance misuse, or severe depression with suicidal intent, and
each year.2 Cognitive behaviour therapy, particularly graded patients who had been on a stable dose of antidepressants or
exposure and response prevention, is effective in treating obses- anxiolytics for less than three months.
sive compulsive disorder.3 The current mode of delivery is a Outcomes
45-60 minute face to face session with the therapist each week, Primary outcome measure was the Yale Brown obsessive
during the hours of 9 am and 5 pm. Such a mode of delivery compulsive checklist (self report version).12 This is a 10 item
results in long waiting lists and precludes access to treatment. questionnaire, and each question is scored between 0 and 4 (0 no
Recent mental health policy in the United Kingdom demands symptoms, 4 severe symptoms). The total score range is 0-7 very
more accessible and effective treatments. Thus, alternative mod- mild, 8-15 mild, 16-23 moderate, 24-31 marked, and 32-40
els of delivery have been proposed that aim to reduce contact severe. A secondary outcome measure was the Beck depression
with therapists and make services more accessible.4 Innovations inventory.13 Satisfaction with treatment was measured using the
such as computerised cognitive behaviour therapy and facilitated client satisfaction questionnaire at the initial follow-up visit.14
self help still often require patients to attend scheduled clinic
appointments.5 6 Although useful, these systems increase Procedure
throughput and access only for patients who can attend the To establish baseline data we assessed patients twice, with four
clinic. Providing treatment over the telephone could increase weeks in between. We used permuted blocks with a block size of

BMJ Online First bmj.com page 1 of 5

Research

BMJ: first published as 10.1136/bmj.38940.355602.80 on 25 August 2006. Downloaded from https://www.bmj.com/ on 21 November 2024 by guest. Protected by copyright.
four to randomise patients.15 Randomisation sheets were drawn
Contacted patients from referral letters (n=91)
up at the first baseline visit and kept by the principal investigator.
Therapists randomised patients to treatment groups four weeks
Refused (n=5)
later after telephoning the principal investigator to obtain the
treatment allocation. Researchers who were blinded to treatment Assessed for eligibility for the trial (n=86)
allocation assessed patients at both of the baseline visits, the ini-
tial visit after treatment, and at one, three, and six months of Excluded (n=9)
follow-up.
Met inclusion criteria.
Assessed and started 1 month baseline period (n=77)
Interventions
Face to face therapy consisted of 10 one hour sessions with the Lost between the two baseline visits (n=5)
therapist on an individual basis. In the first session the therapist
explained the rationale of graded exposure and response Completed baseline period and randomised (n=72)

prevention, which aims to reduce the patient’s anxious and fear-

ful reactions through gradual, repeated exposure to anxiety pro-
Allocated to face to Allocated to telephone
ducing situations. In collaboration with the patient, therapists face intervention (n=36) intervention (n=36)
used the assessment data to devise a hierarchy of fears. From this
hierarchy, patients and therapists set weekly targets to be Did not complete Withdrawn by
intervention (n=3) clinician (n=1)
completed between sessions. The therapist encouraged the
patient to progress though the hierarchy of fears by practising
Completed intervention (n=33) Completed intervention (n=35)
their targets for at least one hour a day and monitoring their
progress on a homework sheet. The therapist reviewed
Completed 6 month Completed 6 month
homework, helped devise weekly targets, encouraged the use of a follow-up (n=30) follow-up (n=35)
co-therapist (relative or friend), pre-empted difficulties, and
helped solve problems. Fig 1 Flow of participants through the trial
Telephone therapy consisted of one face to face session with
the therapist that covered the same material as the first session of
the face to face arm, followed by eight scheduled weekly Results
telephone calls of up to 30 minutes in length. Treatment was the Flow of participants, follow-up, and sample characteristics
same as in the face to face arm, but it was delivered in a shorter We invited 91 patients to be assessed and five declined (fig 1).
period of time and the therapist sent homework sheets to the Thus, we assessed 86 patients for eligibility and excluded nine
patient. The therapist’s role was the same as in the face to face (two with a Yale Brown checklist score > 16; two with suicidal
arm. The final session consisted of a one hour face to face treat- intent; one with substance misuse; and four with problems not
ment session. connected with obsessive compulsive disorder: one health anxi-
Treatment was delivered by two trained and experienced ety, two posttraumatic stress disorder, and one social phobia).
cognitive behaviour therapists (one therapist at each site Five of the 77 patients that we recruited did not attend for base-
delivered both forms of treatment). Consistency of treatment was line assessment. Four of the 72 participants allocated to
maintained by therapist manuals, fortnightly supervision of both treatment (36 for each arm) did not complete their treatment
therapists (where notes were scrutinised), and training days every (three did not attend appointments and one was withdrawn from
the telephone arm owing to increased depression and suicidal
four months during the first year of the study.
ideation deemed by the therapist to warrant a face to face
appointment) and three patients were lost to follow-up at six
Sample size and statistical methods
months. Table 1 shows the key baseline characteristics for each
We analysed the data on an intention to treat basis and assumed
group.
that missing data were missing at random. Because the Yale
Brown obsessive compulsive checklist score would be expected Clinical outcome
to change over time, we did not use the “last observation carried Figure 2 shows the mean scores for the obsessive compulsive
forward” method to imput data.16 17 To assess non-inferiority of disorder and depression scales in the two treatment groups. A
the two treatments, we computed the two sided 95% confidence mean Yale Brown checklist score of 25 before treatment
interval of the difference between treatments.18 Using this indicates obsessive compulsive disorder of marked severity. Table
method, the experimental treatment is not inferior to the control 2 gives the mean values for each treatment group. Differences
treatment at a 2.5% level if the upper boundary is below a between the two sets of baseline scores for the obsessive compul-
prespecified margin of non-inferiority,18 19 in this case 5 units on sive disorder checklist and depression inventory were not statisti-
the Yale Brown checklist. With 40 participants in each group cally significant (mean difference for Yale Brown checklist 1.4
(allowing for attrition of eight in each group) and a within group (95% confidence interval 0.32 to 3.13) and for Beck depression
standard deviation of 7.9,20 the study would have 80% power to inventory 1.3 (1.96 to 4.56)).
reject the null hypothesis that telephone therapy is inferior to Clinical outcome at all four time points was equivalent for
both treatment arms. At six months of follow-up the adjusted
face to face therapy.21 Adjustment for baseline values of the Yale
estimate of the effect of treatment was 0.70 ( − 2.71 to 4.11) for
Brown checklist and Beck depression inventory would be
the Yale Brown obsessive compulsive checklist and 1.51 ( − 2.23
expected to increase power by giving narrow confidence
to 5.25) for the Beck depression inventory—a slight reduction in
intervals. We used Stata 8 to analyse the data.
the mean value for telephone compared with face to face deliv-
ery. All confidence intervals for the Yale Brown checklist score

page 2 of 5 BMJ Online First bmj.com

 Research

BMJ: first published as 10.1136/bmj.38940.355602.80 on 25 August 2006. Downloaded from https://www.bmj.com/ on 21 November 2024 by guest. Protected by copyright.
Brown checklist score at the assessment immediately after treat-
Table 1 Baseline data of patients with obsessive compulsive disorder
ment (mean 21, SD 3.5, n = 3) than those who were followed to
treated with cognitive behaviour therapy. Values are number (%) unless
stated otherwise
six months (mean 13.4, SD 7.2, n = 65) (P = 0.07). This suggests
that the mean Yale Brown checklist score at six months may be
Treatment delivered by Treatment delivered face
Characteristic telephone (n=36) to face (n=36)
slightly underestimated for the face to face treatment group.
Mean age (SD) 33.4 (9) 30.4 (10)
Treatment was deemed clinically relevant if the mean
Mean duration of obsessive 15.3 (11) 14.9 (11) pretreatment score was reduced by two standard deviations or
compulsive disorder in years (SD) more after treatment.22 Treatment was clinically relevant in 49
Marital status: patients (72%)—27 (77%) patients in the telephone treatment
Married 15 (42) 12 (33) arm and 22 (67%) in the face to face treatment arm.
Single, widowed, or divorced 16 (44) 15 (42)
Cohabiting 5 (14) 9 (36) Satisfaction
Sex: Total scores on the client satisfaction questionnaire ranged from
Female 20 (56%) 23 (61) 0 to 32 (higher score indicates greater satisfaction). Patients were
Male 16 (44) 13 (25) very satisfied with treatment, and the results were similar for both
Employment: treatments (table 2).
Employed 25 (69) 22 (66)
Unemployed 9 (25) 10 (28) Blindness
Other 2 (6) 4 (11) We assessed the level of blindness of the independent assessor by
Treatment site: asking them to guess the patients’ treatment status at one month
Stockport 20 (56) 19 (53) of follow-up. Nine of the 72 patients directly or indirectly
Macclesfield 16 (44) 17 (47) revealed their treatment status to the assessor. The assessors
Treatment history:
guessed 35 (56%) of the remaining 63 correctly and 28 (44%)
Past psychological treatment for 14 (39) 11 (31)
obsessive compulsive disorder
incorrectly.
Past psychological treatment for 14 (39) 14 (39)
other mental health disorder
Past drug treatment for obsessive 19 (53) 17 (47)
Discussion
compulsive disorder
The clinical outcome of cognitive behaviour therapy delivered by
Currently taking antidepressants 22 (61) 15 (42)
telephone was equivalent to treatment delivered face to face at all
four follow-up time points and patients reported similarly high
are within 5 units; this suggests that the treatments are levels of satisfaction. The effect size of treatment was 2.5, which is
equivalent. Between the start of the treatment and the six month as large or larger than other studies of face to face (individual or
follow-up visit, mean scores on the Yale Brown checklist dropped group) cognitive behaviour therapy in obsessive compulsive dis-
by about twice the prespecified margin of non-inferiority in both order.3
treatment groups.
Comparison with other studies
The data for the Yale Brown obsessive compulsive checklist
The characteristics of our patients (age, sex, current use of anti-
were less complete at the six month follow-up in the face to face
depressants, and duration of obsessive compulsive disorder) are
treatment group (30 of 36) than in the telephone group (35 of
similar to other studies on this disorder.23–25 Yale Brown checklist
36). Patients not followed up at six months had a worse Yale
scores before and after treatment are also similar to other studies
that have used exposure on its own or as part of a cognitive
30 behavioural intervention.20 23 24 Sample size in our study is equal
Yale Brown obsessive
compulsive disorder scale

Face to face delivery to or greater than most other studies of cognitive behaviour
Telephone delivery therapy in this disorder.23–25 The attrition rate in our study was
25
95% CI low compared with other studies; this contrasts with reports that
patients with obsessive compulsive disorder often refuse
20 exposure treatment.25 Only one patient was lost from the
telephone arm of the trial. Reasons for the low attrition rate in
15 both treatment arms are unclear. Both clinics had long waiting
lists (12 months or more), so perhaps the participants were all
highly motivated. It is also possible that the experienced
10
therapists were particularly good at engaging patients.
20
Beck depression inventory

Implications
Telephone sessions were 30 minutes (50%) shorter than face to
15
face sessions; this equates to more than a 40% saving in the
therapist’s time. This has important economic implications. Our
10 findings support the National Institute for Health and Clinical
Excellence guidelines for obsessive compulsive disorder,3 which
5 encourage cognitive behaviour therapy delivered by telephone.
1 2
ely

th

s
th

th
on
iat

on

on

Limitations
m
ed

m
1
m

6
Im

We did not include a control (no treatment) group. However, we

Baseline After treatment found no differences between the two sets of baseline scores so
Fig 2 Scores for Yale Brown obsessive compulsive disorder checklist and Beck few improvements were made in the absence of treatment. This
depression inventory from first baseline visit to six months of follow-up finding is consistent with other studies.26 27 We did not compare

BMJ Online First bmj.com page 3 of 5

Research

BMJ: first published as 10.1136/bmj.38940.355602.80 on 25 August 2006. Downloaded from https://www.bmj.com/ on 21 November 2024 by guest. Protected by copyright.
Table 2 Main outcome measures and effect of treatment in patients with obsessive compulsive disorder
Treatment delivered by telephone Treatment delivered face to face Adjusted* mean difference between treatment groups
Measure
Mean (SD) No Mean (SD) No (95% CI)
Yale Brown obsessive compulsive disorder score
Before randomisation:
1st baseline visit 25.9 (4.9) 36 25.5 (5.5) 36
2nd baseline visit 24.9 (4.7) 36 23.7 (5.8) 36
After randomisation:
Immediately after treatment 14 (6.9) 35 13.4 (7.7) 33 −0.59 (−3.51 to 2.34)
1 month follow-up visit 14 (7.3) 33 13.7 (8.5) 32 −0.92 (−4.31 to 2.47)
3 month follow-up visit 12.6 (7.5) 34 12.9 (7.7) 29 −1.11 (−4.60 to 2.37)
6 month follow-up visit 14.2 (7.8) 35 13.3 (8.6) 30 −0.55 (−4.26 to 3.15)
Beck depression inventory score
Before randomisation:
1st baseline visit 20.2 (10.4) 36 15.7 (8.5) 36
2nd baseline visit 19.1 (10.6) 36 14.1 (9.1) 36
After randomisation:
Immediately after treatment 11.2 (8.0) 35 9.3 (8.5) 33 −0.52 (−3.66 to 2.63)
1 month follow-up visit 12.7 (10.1) 33 10.3 (8.4) 32 0.13 (−4.01 to 4.27)
3 month follow-up visit 10.1 (8.4) 34 10.6 (8.4) 29 −1.79 (−5.65 to 2.08)
6 month follow-up visit 11.5 (9.5) 35 11.1 (9.1) 29 −2.46 (−6.38 to 1.47)
Score on client satisfaction questionnaire
Immediately after treatment 28.74 (3.6) 34 29.84 (2.9) 32 −0.81 (−2.46 to 0.84)
*Analysis of covariance: adjusted for baseline Beck depression inventory score, baseline Yale Brown obsessive compulsive disorder score, and site.

treatment with another psychological intervention. However, 4 Lovell K, Richards DA. Multiple access points and levels of entry (MAPLE): ensuring
choice, acceptability and equity for CBT services. Behav Cogn Psychother 2000;28:379-
other studies that have used interventions not based on cognitive 91.
behaviour therapy, such as relaxation and anxiety management, 5 Kaltenthaler E, Parry G, Beverley C. Computerized cognitive behaviour therapy: a sys-
tematic review. Behav Cogn Psychother 2004;32:31-55.
have shown poor results.23 28 Patients in our study were followed 6 Den Boer PCAM, Wiersma D, Van Den Bosch RJ. Why is self-help neglected in the
up for six months only, which precludes conclusions on the long treatment of emotional disorders? A meta-analysis. Psychol Med 2004;34:959-71.
term efficacy of telephone treatment. Finally, our findings may 7 Mohr DC, Lokosky W, Bertagnoli A, Goodkin DE, Wende J, Dwyer P, et al. Telephone-
administered cognitive behavioural therapy for the treatment of depressive symptoms
only be relevant to settings in which patients are treated by expe- in multiple sclerosis disorder. J Clin Consult Psychol 2000;68:356-61.
rienced therapists in departments that specialise in cognitive 8 Palmer RL, Birchall H, McGrain L, Sullivan V. Self-help for bulimic disorders: a
randomised controlled trial comparing minimal guidance with face-to-face or
behaviour therapy. Further investigations are needed into the telephone guidance. Br J Psychiatry 2002;181:230-5.
acceptability by clinicians of delivering treatment by telephone. 9 McNamee G, O’Sullivan G, Lelliott P, Marks I. Telephone-guided treatment for house-
bound agoraphobics with panic disorder: exposure vs relaxation. Behav Ther
Contributors: KL conceived and designed the study, obtained funding, 1989;20:491-7.
10 Lovell K, Fullalove L, Garvey R, Brooker C. Telephone treatment of obsessive-
drafted the manuscript, and is guarantor. DC, DR, and RG helped conduct
compulsive disorder. Behav Cogn Psychother 2000;28:87-91.
the study. CR helped design the study and analyse the data. GH helped 11 Taylor S, Thordarson DS, Spring T, Yeh AH, Corcoran KM, Eugster K, et al. Telephone-
design the study. SH and CJ helped interpret and analyse the data. All administered cognitive behaviour therapy for obsessive-compulsive disorder. Cogn
authors contributed to the final manuscript. Behav Ther 2003;32:13-25.
Funding: NHS Executive North West (Research and Development Fund). 12 Goodman WK, Ramussen SA, Price LH, Mazure C, Henninger GR, Charney DS. The
Yale-Brown obsessive-compulsive scale: development, use, and reliability. Arch Gen Psy-
Competing interests: None declared. chiatry 1989;46:1006-16.
Ethical approval: South Cheshire local research committee M125/01; 13 Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring
Stockport ethics committee RJC/SPR/1809. depression. Arch Gen Psychiatry 1961;4:561-71.
14 Attkinson C, Zwick R. The client satisfaction questionnaire: psychometric properties
and correlations with service utilisation and psychotherapy outcome. Eval Program
1 Marks IM. Fears, phobias and rituals: panic anxiety, and their disorders. Oxford: Oxford Uni- Planning 2003;5:233-7.
versity Press, 1987. 15 Pocock S. Clinical trials. Chichester, UK: John Wiley & Sons, 1990.
2 Dupont RL, Rice DP, Shiraki S, Rowland CR. Economic costs of obsessive compulsive 16 Lavori PW. Clinical trials in psychiatry: should protocol deviation censor patient data?
disorder. Med Interface 1995;8:102-9. Quality of life assessment in clinical trials: methodolocigal issues. Neuropsychopharmacol-
3 National Institute for Health and Clinical Excellence. Obsessive compulsive disorder. Clini- ogy 1992;6:29-48.
cal guideline 31. www.nice.org.uk/CG031 (last accessed 18 Aug 2006). 17 Shrout PE. Clinical trials in psychiatry: a comment. Quality of life assessment in clinical
trials: methodological issue. Neuropsychopharmacology 1992;6:49-50.
18 Jones B, Jarvis P, Lewis A, Ebbutt AF. Trials to assess equivalence: the importance of rig-
orous methods. BMJ 1996;313:36-9.
What is already known on this topic 19 Piaggio G, Elbourne DR, Altman DG, Pocock SJ, Wang SJW. Reporting of noninferior-
ity and equivalence of randomized trials. JAMA 2006;295:1152-60.
Cognitive behaviour therapy delivered by telephone may 20 Van Balkom AJ, De Haan E, Van Oppen P, Spinhoven P, Hoogduin KAL, Van Dyck R.
Cognitive and behavioural therapies alone versus in combination with fluvoxamine in
offer help to people with obsessive compulsive disorder the treatment of obsessive compulsive disorder. J Nervous Ment Dis 1998;186:492-9.
21 Elashoff JD. nQuery advisor version 5.0 user’s guide. Los Angeles, CA, 2002.
What this study adds 22 Jacobsen NS, Traux P. Clinical significance: a statistical approach to defining meaning-
ful change in psychotherapy research. J Consult Clin Psychol 1991;59:12-9.
Telephone delivery of cognitive behaviour therapy for 23 Lindsay M, Crino R, Andrews G. Controlled trial of exposure and response prevention
in obsessive-compulsive disorder. Br J Psychiatry 1997;171:135-9.
obsessive compulsive disorder results in equivalent clinical 24 Cordioli AV, Heldt E, Bochi DB, Margis R, Basso de Sousa M, Tonello JF, et al.
outcome and similar levels of patient satisfaction as Cognitive-behavior group therapy in obsessive-compulsive disorder: a randomized
clinical trial. Psychother Psychosomat 2003;72:211-6.
traditionally delivered face to face approaches 25 Vogel PA, Stiles TC, Gotestam KG. Adding cognitive therapy elements to exposure
therapy for obsessive compulsive disorder: a controlled study. Behav Cogn Psychother
This form of treatment delivery is an accessible intervention 2004;32:275-90.
26 O’Connor K, Todorov C, Robillard S, Borgeat F, Brault M. Cognitive-behaviour therapy
that can reduce costs and therapist time and medication in the treatment of obsessive-compulsive disorder; a controlled study.
Can J Psychiatry 1999;44:64-71.

page 4 of 5 BMJ Online First bmj.com

 Research

BMJ: first published as 10.1136/bmj.38940.355602.80 on 25 August 2006. Downloaded from https://www.bmj.com/ on 21 November 2024 by guest. Protected by copyright.
27 Jones MK, Menzies RG. Danger ideation reduction (DIRT) for obsessive compulsive Sarah Hadley research assistant
washers: a controlled trial. Behav Res Ther 1998;36:959-70.
28 Fals-Stewart W, Marks A, Schafer J. A comparison of behavioural group therapy and CBT Department, Cheshire and Wirral NHS Partnership NHS Trust, Macclesfield
individual behaviour therapy in treating obsessive compulsive disorder. J Nerv Ment Dis District General Hospital, Macclesfield SK10 3BL
1993;181:189-93. Debbie Cox cognitive behaviour psychotherapist
(Accepted 21 July 2006)
British Psychological Society, Centre for Clinical, Outcomes, Research, and
doi 10.1136/bmj.38940.355602.80 Effectiveness, Sub-department of Clinical Health Psychology, University College
London, London WC1E 6BT
Christopher Jones health economist
Department of Nursing, Midwifery, and Social Work, University of Manchester,
Manchester M13 9PL Department of Clinical Psychology, Pennine Care NHS Trust, Stepping Hill
Karina Lovell professor of mental health Hospital, Stockport SK2 7JE

School of Psychological Sciences, University of Manchester, Manchester M15 6SZ

David Raines cognitive behaviour therapist
Gillian Haddock professor of clinical psychology Leicestershire Partnership NHS Trust, Department of Cognitive and Behavioural
Psychotherapy, Leicester LE3 1AR
Biostatistics Group, Division of Epidemiology and Health Sciences, University of
Manchester, Manchester M13 9PT Rachel Garvey cognitive behavioural psychotherapist
Chris Roberts senior lecturer in medical statistics Correspondence to: K Lovell [email protected]

BMJ Online First bmj.com page 5 of 5