Milgrom et al.

BMC Psychiatry 2011, 11:95

http://www.biomedcentral.com/1471-244X/11/95

RESEARCH ARTICLE Open Access

Treating postnatal depressive symptoms in

primary care: a randomised controlled trial of
GP management, with and without adjunctive
counselling
Jeannette Milgrom1,2*, Christopher J Holt2, Alan W Gemmill2, Jennifer Ericksen2, Bronwyn Leigh2, Anne Buist3,4 and
Charlene Schembri2

Abstract
Background: Postnatal depression (PND) is under-diagnosed and most women do not access effective help. We aimed
to evaluate comparative management of (PND) following screening with the Edinburgh Postnatal Depression Scale,
using three best-practice care pathways by comparing management by general practitioners (GPs) alone compared to
adjunctive counselling, based on cognitive behavioural therapy (CBT), delivered by postnatal nurses or psychologists.
Methods: This was a parallel, three-group randomised controlled trial conducted in a primary care setting (general
practices and maternal & child health centres) and a psychology clinic. A total of 3,531 postnatal women were
screened for symptoms of depression; 333 scored above cut-off on the screening tool and 169 were referred to
the study. Sixty-eight of these women were randomised between the three treatment groups.
Results: Mean scores on the Beck Depression Inventory (BDI-II) at entry were in the moderate-to-severe range.
There was significant variation in the post-study frequency of scores exceeding the threshold indicative of mild-to-
severe depressive symptoms, such that more women receiving only GP management remained above the cut-off
score after treatment (p = .028). However, all three treatment conditions were accompanied by significant
reductions in depressive symptoms and mean post-study BDI-II scores were similar between groups. Compliance
was high in all three groups. Women rated the treatments as highly effective. Rates of both referral to the study
(51%), and subsequent treatment uptake (40%) were low.
Conclusions: Data from this small study suggest that GP management of PND when augmented by a CBT-
counselling package may be successful in reducing depressive symptoms in more patients compared to GP
management alone. The relatively low rates of referral and treatment uptake, suggest that help-seeking remains an
issue for many women with PND, consistent with previous research.
Trial Registration: The study is registered at ClinicalTrials.gov, Trial Registration Number NCT01002027.

Background and infants [2]. Universal assessment of PND is becoming

Postnatal depression (PND), defined as an episode of best practice in many countries around the world [3-5].
major or minor depression occurring in the first 12 Whilst assessment methods recommended vary (e.g.,
months postpartum, has a point prevalence of 13% at 3 psychometric screening questionnaires, case-finding ques-
months postpartum [1] and early intervention is indicated tions), these developments in practice will see increasing
to prevent long-term impact on women, their partners numbers of cases of PND identified, making widespread
availability of effective PND care pathways a pressing
* Correspondence: [email protected] public health issue in many countries.
1
Department of Psychology, Psychological Sciences, University of Melbourne, General Practitioners (GPs) and postnatal nurses are
Victoria 3010, Australia key primary care professionals engaged with mothers
Full list of author information is available at the end of the article

© 2011 Milgrom et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Milgrom et al. BMC Psychiatry 2011, 11:95 Page 2 of 9
http://www.biomedcentral.com/1471-244X/11/95

during the postnatal period. It is therefore important to isolation from other service improvements, screening for
determine whether best-practice management of PND depression in primary care will generally be ineffective
in primary care can offer an effective pathway resulting in reducing morbidity or improving outcomes [16].
in alleviation of depression for the majority of women. The present study similarly sought to examine the
Further as many women are reluctant to take antide- effectiveness of counselling informed by the principles of
pressants during lactation, due to potential side effects CBT and delivered by primary care practitioners to
on the newborn [6] readily available non-pharmacological women with PND. In addition, this study sought to
treatments are essential. Systematic and meta-analytic address currently unanswered questions: Is the same
reviews support the efficacy of psychological therapy for treatment delivered by different professionals similarly
PND [7,8]; however, there have generally been too few effective (e.g. trained nurses versus psychologists)? In this
studies included to draw conclusions about the relative RCT we compare three model care pathways: manage-
effects of various types of psychological treatments. ment by trained GPs alone and management by trained
Nevertheless, cognitive-behavioural therapy (CBT) is GPs augmented with a counselling-CBT intervention
clearly one of the most effective treatments for depres- delivered either by a trained nurse or a psychologist.
sion at other life stages [9].
Whilst CBT is generally delivered by mental health spe- Methods
cialists such as psychologists, some evidence for the abil- Sample & Procedures
ity of nurses to deliver psychological interventions for The study (Trial Registration Number NCT01002027)
PND in primary care has been published. However, stu- took place in three municipalities in Melbourne, Austra-
dies conducted to date have not explicitly compared such lia with approval from Austin Health Human Ethics
interventions to management by GPs. To our knowledge, Research Committee. Postnatal women with infants < 12
in the postnatal period, five controlled trials have evalu- months of age were screened by nurses working in pri-
ated psychologically-informed interventions delivered by mary care at maternal child health centres during regular
primary care practitioners (generally nurses) [10-14]. routine visits. The Edinburgh Postnatal Depression Scale
Only one study [14] has compared non-specialists with (EPDS) [17], is a simple 10-item questionnaire designed
specialists (allocation to specialists versus non-specialists to screen for symptoms of PND. The EPDS has good
was not random).The interventions were CBT-based or acceptability [18] and is used worldwide. Women scoring
counselling-based (psychodynamic therapy was also eval- ≥13 on the EPDS were invited into the study. Once base-
uated in one study), and the nurses were trained in these line data were secured, a woman’s GP was contacted and
approaches. With the exception of one study [12], nurse offered training, prior to their first patient being allocated
delivered interventions were shown to be more effective to one of the three study groups. Inclusion criteria were:
in the short-term than routine care (which consisted in screening score above cut-off on the EPDS; infant aged
most cases of standard nursing practices in place for peri- 6 weeks to 4 months. Exclusion criteria were: insufficient
natal women). Morrell et al. [11] also found that benefits English; psychotic symptoms; need for immediate crisis
for women in the intervention group were maintained at management. Having been trained in diagnosis and
12 months postpartum. Interestingly, Cooper et al. [10] management of postpartum mood disorders (see next
found an expertise effect, such that women treated by section), GPs were asked to conduct a diagnostic assess-
non-specialists showed significantly greater reduction in ment on all women to confirm that their patients were
depressive symptoms compared with those treated by depressed and would require treatment. A coded, vari-
specialists (however treatment allocation was not able-length permuted blocks allocation schedule was pre-
randomised). generated by an independent person and administered
Effective and manualised psychological interventions centrally by administrative staff. Women were rando-
can be successfully translated to widespread delivery by mised with a 1:1:1 allocation ratio to the three groups. At
a range of primary care professionals and could be a entry, each participant agreed to randomization to either
valuable resource for health systems around the world. treatment by the GPs themselves, or with adjunctive ses-
For example, in Australia, the advent of the National sions with a nurse or a psychologist. Irrespective of
Perinatal Depression Initiative (NPDI [15]) will see the group allocation all women were asked to schedule at
implementation of universal screening for perinatal least 3, fortnightly check-up visits with their GP and all
mood disorders. As a large number of depressed women participants remained under the overall care of their own
will be identified following screening, it is important to GP.
establish which primary care pathways commonly pro- Training
vided in most countries can provide effective treatment Each participant’s GP received brief, focussed training,
of PND. Assessment without evidence-based treatment consisting of a face-to-face session with a psychologist
being readily available raises duty of care issues and, in in the GP’s practice (about 45-60 minutes), supported
Milgrom et al. BMC Psychiatry 2011, 11:95 Page 3 of 9
http://www.biomedcentral.com/1471-244X/11/95

by detailed printed materials, to enhance their ability to This counselling-CBT was delivered by an experienced
manage PND. This involved systematically working psychologist at a hospital Psychology department. Again
through a 25-page training manual covering screening, this was delivered as an adjunct to GP management.
diagnosis with standard psychiatric criteria (DSM-IV), Outcome Measures
risk assessment and management, engagement, a biopsy- The main outcomes were levels of depressive symptoms
chosocial model of PND, medication during lactation, and the proportion of participants with symptoms below
common patient concerns, onward referral and princi- the cut-off score indicative of mild to severe depressive
ples of treatment (including supportive counselling stra- symptoms. Two validated measures of depressive symp-
tegies and cognitive-behavioural strategies). Telephone toms were used and were administered at baseline, again
consultation with a psychiatrist was available to provide after 3 weeks, and immediately post-study. The Beck
additional advice on medication for PND. A GP-specific, Depression Inventory II (BDI-II [22]) was the main mea-
one-page PND Management Guide (developed by sure. The BDI-II is a well-validated, 21-item self-report
beyondblue; available at http://www.beyondblue.org.au/ questionnaire that provides a clinical measure of depres-
index.aspx?link_id=7.102) was also provided. GPs were sive symptoms and threshold scores for classifying symp-
free to prescribe antidepressant medication in all three toms into minimal, mild, moderate and severe categories.
groups (as in other RCTs of psychological interventions The BDI-II has good internal consistency (a = 0.91) and
for PND in primary care settings [11]). A total of 46 good test-retest reliability (r = .96).The short form of the
GPs received the training (some had more than one of Depression Anxiety and Stress Scales (DASS 21 SF) [23]
their patients in the study). was used to monitor levels of stress and anxiety, which
Twenty two nurses completed a half-day training work- commonly occur co-morbidly with depression. The stress
shop in the counselling-CBT intervention [19]. The train- and anxiety scales have alpha values of 0.81 and 0.73
ing drew on an evaluated CBT program for PND [20,21] respectively [23].
adapted for routine application in primary care using a In addition, women completed questionnaires rating
counselling framework. The training was conducted by a the perceived effectiveness of treatment on binary (Yes/
senior psychologist, with several years experience in deli- No) and Likert-type (1 to 10) scales. Information on
vering CBT for PND, and covered three phases of the medication use was collected post-study. As all outcome
intervention: assessment, goal setting and treatment, measures were self-report, it was not possible to obtain
addressing the key skills and therapist pitfalls in each blinded measures of symptomatology.
stage. The sessions focussed on: psycho-education about
PND, goal setting and problem solving, behavioural inter- Power & Sample Size
ventions (e.g. encouraging pleasant activities, relaxation) Based on the average baseline BDI-II score in a previous
basic cognitive techniques (e.g. link between thoughts study (BDI-II = 23.8, SD = 8.4) a post-treatment
and feelings, challenging unhelpful beliefs and thoughts). improvement of 30% (7.1 points) would take average
Additional components included: the partner relation- scores to the midpoint of the “mild” range of depressive
ship, social support and the mother-baby relationship. symptoms (BDI-II = 14-19). Applying these numbers we
The Overcoming Postnatal Depression manual [19] pro- calculated: n = 2(0.84 + 1.96)2 (8.4/7.1)2 = 22.0, at 80%
vided detailed step-by-step, prompted, six-session con- power with p = 0.05. We therefore continued recruitment
tent. The psychologists delivered the same intervention until at least n = 22 had been achieved in all 3 groups.
package.
Treatment Groups Statistical Analysis
Group A: GP management Women allocated to this The BDI-II score classifications given by Beck et al [22]
group were managed as usual by their own GP (trained were used to categorise cases as either above (score ≥14 =
in PND management). mild, moderate or severe depressive symptoms) or below
Group B: Adjunctive counselling-CBT from a nurse threshold (score < 14 = zero or minimal depressive symp-
Women allocated to this group received six sessions (one toms). Between-group differences were tested by Analysis
per week over six weeks) of the manualised Overcoming of Covariance (ANCOVA) controlling for baseline scores.
Postnatal Depression Program. This counselling-CBT We asked if GP management differed from adjunctive
program was delivered by a trained nurse at maternal counselling-CBT per se, and also whether there was a dif-
and child health centres and was an adjunct to GP ference between counselling-CBT by psychologists com-
management. pared to nurses. This required two, a priori, orthogonal
Group C: Adjunctive counselling-CBT from a psychol- contrasts as follows: Contrast i) Group A vs [Group B +
ogist Women allocated to this group received six ses- Group C]/2. Contrast ii) Group B vs Group C.
sions (one per week over six weeks) of the same The primary analysis was by intention-to-treat [24]
Overcoming Postnatal Depression Program as group B. using maximum likelihood imputation of missing values
Milgrom et al. BMC Psychiatry 2011, 11:95 Page 4 of 9
http://www.biomedcentral.com/1471-244X/11/95

(expectation maximisation: EM). All computations were between-group differences in post-study scores for the
carried out in SPSS 16. three DASS 21 SF scales of depression, anxiety and stress
(p > 0.05).
Results Depressive symptoms above threshold
Participants at Baseline An observed-case frequency analysis of remittance rates
Figure 1 shows the flow of participants through the based on categorising BDI-II scores as above or below
study. Of 3,531 women screened, 333 scored ≥ 13 on threshold (Table 3), found that the frequency of above-
the EPDS. One hundred and sixty four of these women threshold cases did vary significantly post-study, such that
were not referred to the study. Reasons for non-referral those women in GP management (Group A) appeared
by nurses were not recorded systematically. However, more likely to exhibit symptoms of depression (Table 3,
the reasons for non-participation among those referred c2, df = 2, p = .028). The same information is re-expressed
to the study are detailed in Figure 1. Ultimately, sixty in terms of Relative Risk at the bottom of Table 3. Man-
eight women were randomised. The mean baseline agement in Group B (adjunctive counselling-CBT from a
EPDS of these 68 women (16.98, SD 4.49) was not sig- nurse) lowered the risk of an above-threshold outcome
nificantly different from the 101 referred women not relative to GP management, but as numbers are small
randomised (16.36, SD 3.56). Twenty-three women were these findings should be interpreted cautiously.
allocated to Group A (GP management), 22 to Group B Services accessed and Medication Use
(adjunctive counselling-CBT with nurse) and 23 to There was a poor return rate from women regarding other
Group C (adjunctive counselling-CBT with psycholo- services accessed and medication use with only one third
gist). Table 1 shows baseline characteristics of each of the sample returning these questionnaires. Based on the
group. As is appropriate in a RCT, no between-group available data there was no difference in post-study out-
significance tests were conducted on baseline values come between women known to be taking antidepressants
[24,25]. Mean baseline scores on the BDI-II were in the (mean BDI-II score = 10.3, 95% CIs 6.4 - 14.1) and
moderate to severe range for all groups indicating the all other women (mean BDI-II score = 9.0, 95% CIs 5.6 -
presence of clinically significant depressive symptoms. 12.3).
For the 66 women in total the average BDI-II score at Participant Ratings
baseline was 29.14 (SD 10.12) with scores ranging from Forty six women responded to the questions on treat-
12 to 51 points. Group averages are given in Table 2. ment efficacy. A majority in all groups indicated that
Compliance treatment was sufficient (9/14, 16/18 and 12/14 in
Seventy one percent of GP appointments were kept groups A, B and C respectively). On a scale of 1 to 10,
(67%, 87% and 67% in groups A, B and C respectively). respondents rated perceived effectiveness of their treat-
Similarly, attendance at the 6 counselling-CBT sessions ment highly in all groups (6.9, 8.6 and 7.4 respectively
averaged 4.6 and 4 sessions for groups B and C respec- in groups A, B and C), and significantly more highly in
tively. Of the 68 participants, 50 returned post-study group B (Kruksall Wallis test, p = 0.04).
questionnaires. This attrition was demonstrably random
with respect to group (c2 = 1.59, df = 2, p = .45). Discussion
Symptoms of depression, anxiety and stress This study compared three pathways of care for managing
Graphical inspection of Figure 2a shows that BDI-II scores PND, all treatments requiring training the key primary
across all treatments dropped on a similar trajectory. This care health professionals involved. An important question
constituted a significant drop between baseline and post- in the management of perinatal mood disorders is whether
study (mean reduction in BDI-II scores for all treatment different “real world” care pathways actually result in ame-
groups combined = 17.3 points, 95% CI 14.2-20.5). Table 2 lioration of depressive symptoms, and whether they differ
gives the mean baseline and post-study BDI-II scores for consistently in efficacy [26]. On average, women who were
each treatment group. The results of the intention-to-treat offered GP management in the present study had similar
contrasts of post-study BDI-II scores controlling for base- improvements in symptoms of depression and anxiety to
line scores showed that variation between treatments was those receiving adjunctive counselling-CBT per se. Possi-
non-significant (F= 1.051, df = 2,45, p = .358). Neither bly, the GP training component made any additional effect
planned Contrast i) GP management versus counselling- of adjunctive counselling-CBT more difficult to detect.
CBT, nor planned Contrast ii) Adjunctive counselling-CBT Nonetheless, we also found that women in GP manage-
from nurse versus psychologist, were significant (p = 0.347 ment continued to exhibit a higher frequency of above-
and p = .247 respectively). threshold depressive symptoms post-study. These data
Figure 2b shows the significant (p < 0.05) overall drop may suggest that adjunctive counselling-CBT involving
in anxiety over the course of the study. Similarly to the either psychologists or nurses could be a promising model
results for BDI-II scores, there were no significant of collaborative PND management in primary care.
Milgrom et al. BMC Psychiatry 2011, 11:95 Page 5 of 9
http://www.biomedcentral.com/1471-244X/11/95

N = 3,531 WOMEN SCREENED

EPDS •13, n = 333

Not referred to study, n = 164
Improved mood, n = 33
Referred to study, n = 169
Not responding to contact, n = 33
Already in treatment, n = 14
Declined participation, n = 8
RANDOMISATION
Hospitalised, n = 2
n = 68
Other, n = 11

Group A Group B Group C

Routine GP Adjunctive Adjunctive
Management Counselling-CBT Counselling-CBT
with nurse with Psychologist
n = 23 n = 22 n = 23

• 3 fortnightly GP appointments in all groups

• Mid-point data collection at 3-weeks post-
randomisation
• End-point data collection at 8- weeks post-
randomisation

Group A Group B Group C

Observed case Observed case Observed case
analysis n = 15. analysis n = 17. analysis n = 17.

Intention-to-treat Intention-to-treat Intention-to-treat

analysis n = 23. analysis n = 22. analysis n = 23.

Figure 1 Participant Flowchart.

A number of other positive outcomes were found. symptoms of depression. Interestingly, there is some
Firstly, anxiety, (which is often a co-morbid problem suggestion that adjunctive counselling-CBT was most
with PND) was also effectively reduced by treatment. effective when delivered by nurses. This is consistent
Secondly, compliance rates were good and women in all with some previous findings on the effectiveness of PND
groups showed significant reductions in post-study treatment programs delivered by both specialist and
Milgrom et al. BMC Psychiatry 2011, 11:95 Page 6 of 9
http://www.biomedcentral.com/1471-244X/11/95

Table 1 Baseline Characteristics of Participants

Treatment Condition A (GP) Treatment Condition B (GP+ nurse) Treatment Condition C
(n = 23) (n = 22) (GP+ psychologist) (n = 23)
Mean Screening EPDS 17.1 (4.5) 16.8 (4.8) 17.0 (4.5)
(SD)
Mean Mothers’ Age (SD) 30 (3.3) 33.1 (4.4) 31.4 (5.6)
Mean Infant age in 17.03 (9.22) 14.84 (11.44) 20.68 (9.15)
weeks (SD)
*Marital Status, n (%)
Married/De Facto 18 (85.8%) 20 (90.9%) 21 (91.3%)
No partner 3 (14.3%) 2 (9%) 2 (8.6%)
*Born In Australia, n (%) 17 (81%) 19 (90.5%) 21 (91.3%)
*English Speaking, n (%) 21 (100%) 19 (90.5%) 21 (91.3%)
*Education, n (%)
High School only 5 (23.8%) 3 (13.6%) 7 (30.4%)
Degree or Higher 15 (71.4%) 14 (63.6%) 14 (60.9%)
*Income, n (%)
< $40,000 7 (33.4) 4 (18.1%) 2 (8.6%)
$40,000-80,000 11 (52.3%) 1 (45.4%) 12 (52.1%)
> $80,001 2 (9.5%) 5 (22.7%) 6 (26.1%)
Wouldn’t divulge 1(4.8%) 3 (13.6%) 3 (13%)
*Number of Children,
n (%)
1 13 (61.9%) 8 (36.4%) 9 (39.1%)
2 5 (23.8%) 10 (45.5%) 11 (47.8%)
>2 3 (14.3%) 4 (18.2%) 3 (13%)
*indicates missing data on these variables.

trained non-specialist practitioners [11,12,27,14]. In the improvements in mood (a drop of 17.3 BDI-II points on
present study, psychologists worked from treatment average) are of a magnitude at least as large as post-treat-
rooms in a public hospital whilst nurses conducted the ment effect sizes observed in studies involving psychologi-
first counselling-CBT session at home and subsequent cal interventions versus routine care for PND [8].
sessions in a health centre. Conceivably, this difference Furthermore, in our previous RCT of psychological treat-
may have contributed to the possible advantage of coun- ments for PND [21] we found that, following routine care,
selling-CBT delivered by nurses. Baseline BDI-II scores symptoms of depression and anxiety were essentially
may also have influenced these results, as they were unchanged after 12 weeks. Thus, spontaneous improve-
somewhat higher in group C (counselling-CBT with ment seems an insufficient explanation for the large drop
psychologists). in symptomatology following treatment observed in the
The study has a number of limitations. First, the sample present study. Third, GP report of depressive symptoms
size was relatively small, and attrition reduced this further rather than a standardized diagnostic interview was used
at follow-up, limiting our ability to generalise from the for inclusion. However, all GPs were trained in diagnosis
results. Second, the “control” group itself involved an according to standard criteria and baseline BDI-II scores
enhancement of current care, by training GPs. For ethical in all three groups reflected moderate to severe levels of
reasons it was inappropriate to include a wait-listed symptomatology. Furthermore, a single psychologist deliv-
control group in this study. However the observed ered the intervention, again limiting the generalisability of

Table 2 Baseline and post-study depressive symptoms

Treatment Condition A Treatment Condition B Treatment Condition C
(Routine management) (CBT-counselling with nurse) (CBT-counselling with psychologist)
Mean Baseline BDI-II (SD; 95% CI) 27.9 (10.8; 23.3-32.6) 25.5 (8.3; 21.7-29.3) 30.9 (10.7; 26.2-35.6)
Mean Post-study BDI-II (SD; 95% CI) 11.8 (9.8; 6.4-17.2) 6.1 (4.8; 3.7-8.6) 10.9 (11.0; 5.2- 16.5)
*Mean Adjusted Post-study BDI-II (SD; 95% CI) 11.0 (8.0; 7.6-14.5) 6.7 (4.3; 4.8-8.6) 10.4 (9.5; 6.3-14.5)
*Estimate from intention-to-treat ANCOVA controlling for baseline symptoms with maximum likelihood imputation of missing values.
Milgrom et al. BMC Psychiatry 2011, 11:95 Page 7 of 9
http://www.biomedcentral.com/1471-244X/11/95

possible so that long-term maintenance of gains cannot be

a) assessed.
35 Early intervention for PND is essential due to the nega-
tive consequences for women and for their close family
Treatment A
30 members in terms of mental health and child socio-emo-
Treatment B tional development [28,29]. The results presented here
25 Treatment C add to a growing body of evidence that following a positive
screening result for PND many (indeed most) women do
20 not pursue further options for assessment and treatment.
BDI-II

Less than 50% of women affected by PND have been

15 reported by others to access treatment [18,30-32]. In this
study, only 20% of those screening positive did so and this
10
may have been partly due to nurse’s and women’s reluc-
tance to participate in a randomised research study.
5
Even among those who agreed to referral to this study,
most did not ultimately take up treatment, although
0
baseline week 3 week 8 (post-study) some cited improved mood or had already accessed other
Time treatment options. Low referral rates to, and participation
rates in a particular research study such as this may also
b) reflect the reluctance of women to take part in research.
However, given the current evidence, it seems clear that
15 specific research on how to increase women’s engagement
with treatment would be valuable. Whilst systematic
DASS 21 SF Anxiety Scale

screening for PND offers one possibility for increasing

detection (the first step to accessing treatment) data on the
10 ultimate usefulness of screening programs for PND in
terms of increased treatment uptake are still relatively
scarce. As has been pointed out elsewhere, the introduction
of screening in isolation will have little impact [16,33]. In
the only published RCT of screening effectiveness [34] a
5
significant reduction in morbidity was found due to the
implementation of screening. The key to effectiveness in
terms of improving women’s outcomes was to systemati-
cally follow up all positive screening results with further
0 clinical assessment for depression and access to effective
baseline w eek 3 w eek 8 (post-study) management. Recent meta-analyses of the effectiveness of
Time depression screening (not just for PND) suggest that it can
Figure 2 Changes in Symptoms of Depression and Anxiety. have its biggest impact on morbidity when deployed as part
a) Beck Depression Inventory II; b) DASS 21 SF Anxiety Sub-scale. of a well-coordinated health system effort towards identifi-
For each measure the means of the three groups are plotted across cation and treatment. A clear policy of acting on all positive
time. Only those cases with complete data are shown:- Group A (GP screening results plus a well-resourced treatment compo-
management), n = 12; Group B (Counselling-CBT with nurse), n =
nent appear to maximise the usefulness of screening for
12; Group C (Counselling-CBT with psychologist), n = 12. Error bars
are ± 1 SE. depressive disorders in general [16] and the effectiveness
and cost-effectiveness of PND screening in particular
[34-36].
results. The study is also limited in that no diagnostic pro-
cedure was carried out post-treatment, so that the num- Conclusions
bers of women meeting diagnostic criteria for a depressive In summary, for the majority of those who received treat-
disorder following treatment is not known. Referral to the ment, all three possible models of care appeared effective.
study was relatively low, and of those referred most either It therefore appears that for the management of moder-
could not be contacted (n = 33) or had experienced ate-to-severe PND, best practice primary care manage-
improved mood (n = 33). Only 8 women still experiencing ment routes are effective for the majority of women. GP
low mood and not accessing treatment refused involve- management coupled with adjunctive counselling-CBT
ment with the study. Lastly, no longer-term follow-up was yielded promising results. In practice these models of
Milgrom et al. BMC Psychiatry 2011, 11:95 Page 8 of 9
http://www.biomedcentral.com/1471-244X/11/95

Table 3 Baseline and Post-study frequencies of depressive symptoms

Treatment Condition Treatment Condition B Treatment Condition C
A (CBT-counselling with (CBT-counselling
(Routine nurse) psychologist)
management)
Baseline
*below: above threshold 1:22 2:19 0:21
Post-study
*below: above threshold 8:7 16:1 13:4
Absolute Risk (probability of being above threshold post- 0.47 0.06 0.24
treatment)
Relative Risk compared to Routine management (95% CIs) 1 (reference value) 0.13 (0.02 -0.91) 0.5 (0.18-1.4)
*For frequency analysis, BDI-II scores ≥14 were classed as ‘above-threshold’.

PND management are deliverable by existing primary 2. Murray L, Cooper PJ: Postpartum depression and child development.
Psychol Med 1997, 27:253-260.
care professionals. However, rates of both referral to 3. Guidelines Expert Advisory Committee: Draft Clinical practice guidelines for
treatment (51%), and subsequent treatment uptake (40%) depression and related disorders - anxiety, bipolar disorder and puerperal
were low, suggesting help-seeking remains an issue in psychosis - in the perinatal period Melbourne: beyondblue: the national
depression initiative; 2010.
clinical practice that needs to be addressed by compre- 4. NICE: Antenatal and postnatal mental health: Clinical management and
hensive research on methods to overcome this obstacle. service guidance The British Psychological Society & The Royal College of
Training key primary care professionals and strengthen- Psychiatrists; 2007 [http://www.nice.org.uk/CG45].
5. SIGN: Postnatal Depression and Puerperal Psychosis: A national clinical
ing their collaboration is likely to remain centrally impor- guidance: Edinburgh Royal College of Physicians; 2002.
tant for improving current treatment pathways for PND 6. Pearlstein T: Perinatal depression: treatment options and dilemmas. J
following screening, under Australia’s National Perinatal Psychiat & Neurosci 2008, 33:302-318.
7. Cuijpers P, Brannmark J, Gvan Straten A: Psychological treatment of
Depression Initiative, and for similar universal programs postpartum depression: a meta-analysis. J Clinical Psychol 2008, 64:103-118.
in other countries. 8. Dennis CL, Hodnett E: Psychosocial and psychological interventions for
treating postpartum depression. Cochrane Database of Systematic Reviews
2007, 4:CD006116.
Acknowledgements 9. Tolin D: Is cognitive-behavioral therapy more effective than other
Our thanks to the beyondblue Victorian Centre of Excellence in Depression therapies? A meta-analytic review. Clin Psychol Rev 2010, 710-720.
and Related Disorders and to the Royal Australian and New Zealand College 10. Cooper PJ, Murray L, Wilson A, Romaniuk H: Controlled trial of the short-
of Psychiatrists for funding this project and to Yolanda Romeo for delivering and long-term effect of psychological treatment of post-partum
training to nurses. Our late colleague Rachel McCarthy contributed much to depression. 1. Impact on maternal mood. Brit J Psychiat 2003,
the treatment manual used in this study. 182:412-419.
11. Morrell CJ, Slade P, Warner R, Paley G, Dixon S, Walters SJ, Brugha T,
Author details Barkham M, Parry GJ, Nicholl J: Clinical effectiveness of health visitor
1 training in psychologically informed approaches for depression in
Department of Psychology, Psychological Sciences, University of Melbourne,
Victoria 3010, Australia. 2Parent-Infant Research Institute, Department of postnatal women: pragmatic cluster randomised trial in primary care.
Clinical & Health Psychology, Heidelberg Repatriation Hospital, Austin Health, BMJ 2009, 338:a3045.
300 Waterdale Road, Heidelberg West, Victoria 3081, Australia. 3Northpark 12. Prendergast P, Austin MP: Early childhood nurse-delivered cognitive
Hospital, Victoria, Australia. 4Department of Medicine, University of behavioural counselling for post-natal depression. Aust Psychiat 2001,
Melbourne, Victoria, Australia. 9:255-259.
13. Wickberg B, Hwang CP: Counselling for postnatal depression: A
Authors’ contributions controlled study on a population based Swedish sample. J Affect Dis
JM, JE, AG and AB conceived the study. JM, JE, and BL contributed to the 1996, 39:209-216.
design of the GP training. CS and BL delivered the training. CH and BL 14. Holden JM, Sagovsky R, Cox JL: Counselling in a general practice setting:
oversaw data collection and monitored the adherence to study protocols. controlled study of health visitor intervention in the treatment of
AG and CH designed and executed data analyses. AG (50%) BL (25%) and postnatal depression. BMJ 1989, 298:223-226.
CH (25%) wrote a first draft of the manuscript. JM, JE, AG, CS, AB, CH and BL 15. Department of Health and Ageing: National Perinatal Depression Framework
all edited subsequent drafts for important intellectual content and all 2009 [http://www.health.gov.au/internet/main/publishing.nsf/content/
authors agreed on the submitted version. mental-perinat].
16. Gilbody S, House A, Sheldon T: Screening and case finding instruments
Competing interests for depression. Cochrane Database of Systematic Reviews 2009, 4:CD002792.
The authors declare that they have no competing interests. 17. Cox J, Holden J: Perinatal Mental Health A Guide to the Edinburgh Postnatal
Depression Scale (EPDS). London: Gaskell; 2003.
Received: 12 November 2010 Accepted: 27 May 2011 18. Gemmill AW, Leigh B, Ericksen J, Milgrom J: A survey of the clinical
Published: 27 May 2011 acceptability of screening for postnatal depression in depressed and
non-depressed women. BMC Public Health 2006, 6:211.
19. PIRI: Overcoming Postnatal Depression Melbourne: Parent-Infant Research
References
Institute; 2003.
1. Gavin NI, Gaynes BN, Lohr KN, Meltzer-Brody S, Gartlehner G, Swinson T:
20. Milgrom J, Martin PR, Negri LM: Treating Postnatal Depression. A
Perinatal depression: a systematic review of prevalence and incidence.
Psychological Approach for Health Care Practitioners Chichester: Wiley; 1999.
Obstet & Gynecol 2005, 106:1071-1083.
Milgrom et al. BMC Psychiatry 2011, 11:95 Page 9 of 9
http://www.biomedcentral.com/1471-244X/11/95

21. Milgrom J, Negri LM, Gemmill AW, McNeil M, Martin PR: A randomized
controlled trial of psychological interventions for postnatal depression.
Brit J Clin Psychol 2005, 44:529-542.
22. Beck AT, Steer RA, Brown GK: BDI-II manual. San Antonio: The Psychological
Corporation; 1996.
23. Lovibond SL, Lovibond PF: Manual for the Depression Anxiety Stress Scales
Sydney: Psychological Foundation; 1995.
24. Moher D, Hopewell S, Schulz KF, Montori V, Gotzsche PC, Devereaux PJ,
Elbourne D, Egger M, Altman DG: CONSORT 2010 explanation and
elaboration: updated guidelines for reporting parallel group randomised
trials. BMJ 2010, 340:c869.
25. Senn R: Base logic: tests of baseline balance in randomized controlled
trials. Clin Res Regulatory Affairs 1995, 12:171-182.
26. Highet N, Drummond P: A comparative evaluation of community
treatments for post-partum depression: implications for treatment and
management practices. Aust & NZ J Psychiat 2004, 38:212-218.
27. Cooper PJ, Murray L, Wilson A, Romaniuk H: Controlled trial of the short-
and long-term effect of psychological treatment of post-partum
depression. Brit J Psychiat 2003, 182:412-419.
28. Goodman JH: Postpartum depression beyond the early postpartum
period. JOGNN 2004, 33:410-420.
29. Murray L, Cooper PJ: The impact of postpartum depression on child
development. Int Rev Psychiat 1996, 8:55-63.
30. Carter FA, Carter JD, Luty SE, Wilson DA, Frampton CM, Joyce PR: Screening
and treatment for depression during pregnancy: a cautionary note. Aust
& NZ J Psychiat 2005, 39:255-261.
31. Buist A, Bilszta J, Barnett B, Milgrom J, Ericksen J, Condon J, Hayes B.
Brooks J: Recognition and management of perinatal depression in
general practice–a survey of GPs and postnatal women. Australian Family
Physician 2005, 34:787-790.
32. MacLellan A, Wilson D, Taylor A: The self-reported prevalence of postnatal
depression. Aust & NZ J Obstet Gynaecol 1996, 36:313.
33. Kroenke K: Depression screening is not enough. Annals of Internal
Medicine 2001, 134:418-420.
34. Leung S, Leung C, Lam T, Hung T, Chan R, Yeung T, et al: Outcome of a
postnatal depression screening programme using the Edinburgh
Postnatal Depression Scale: a randomized controlled trial. Journal of
Public Health 2010.
35. Milgrom J, Mendelsohn J, Gemmill AW: Does postnatal depression
screening work? Throwing out the bathwater, keeping the baby. Journal
of Affective Disorders 2010.
36. Paulden M, Palmer S, Hewitt C, Gilbody S: Screening for postnatal
depression in primary care: cost effectiveness analysis. BMJ 2009, 339:
b5203.

Pre-publication history
The pre-publication history for this paper can be accessed here:
http://www.biomedcentral.com/1471-244X/11/95/prepub

doi:10.1186/1471-244X-11-95
Cite this article as: Milgrom et al.: Treating postnatal depressive
symptoms in primary care: a randomised controlled trial of
GP management, with and without adjunctive counselling. BMC
Psychiatry 2011 11:95.

Submit your next manuscript to BioMed Central

and take full advantage of:

• Convenient online submission

• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution

Submit your manuscript at

www.biomedcentral.com/submit