Академический Документы
Профессиональный Документы
Культура Документы
44
September 2010
10.3310/hta14440
Additionally the HTA website allows you to either print out your order or download a blank order form.
Payment methods
Paying by cheque
If you pay by cheque, the cheque must be in pounds sterling, made payable to University of Southampton
and drawn on a bank with a UK address.
Paying by credit card
You can order using your credit card by phone, fax or post.
Subscriptions
NHS libraries can subscribe free of charge. Public libraries can subscribe at a reduced cost of 100 for
each volume (normally comprising 4050 titles). The commercial subscription rate is 400 per volume
(addresses within the UK) and 600 per volume (addresses outside the UK). Please see our website for
details. Subscriptions can be purchased only for the current or forthcoming volume.
The website also provides information about the HTA programme and lists the membership of the v arious
committees.
Group cognitive behavioural therapy
for postnatal depression: a systematic
review of clinical effectiveness, cost-
effectiveness and value of information
analyses
*Corresponding author
Stevenson MD, Scope A, Sutcliffe PA, Booth A, Slade P, Parry G, et al. Group cognitive
behavioural therapy for postnatal depression: a systematic review of clinical effectiveness,
cost-effectiveness and value of information analyses. Health Technol Assess 2010;14(44).
Abstract
Group cognitive behavioural therapy for postnatal
depression: a systematic review of clinical
effectiveness, cost-effectiveness and value of
information analyses
MD Stevenson,* A Scope, PA Sutcliffe, A Booth, P Slade, G Parry,
D Saxon, E Kalthenthaler and the group cognitive behavioural therapy
for postnatal depression advisory group
School of Health and Related Research (ScHARR), The University of Sheffield, Sheffield, UK
*Corresponding author
Background: Postnatal depression (PND) describes standardised clinical assessment tool used to define
a wide range of distressing symptoms that can occur PND. All full papers were read by two reviewers (AS
in women following childbirth. There is substantial and DS) who made independent decisions regarding
evidence to support the use of cognitive behaviour inclusion or exclusion, and consensus, where possible,
therapy (CBT) in the treatment of depression, and was obtained by meeting to compare decisions. In the
psychological therapies are recommended by the event of disagreement, a third reviewer (EK) read the
National Institute for Health and Clinical Excellence paper and made the decision. All data from included
as a first-line treatment for PND. However, access is quantitative studies were extracted by one reviewer
limited owing to expense, waiting lists and availability (AS) using a standardised data extraction form. All
of therapists. Group CBT may, therefore, offer a data from included qualitative studies were extracted
solution to these problems by reducing therapist by two reviewers (AS and AB) using a standardised
time and increasing the number of available places for data extraction form with disagreements resolved by
treatment. discussion. Two different data extraction forms were
Objectives: To evaluate the clinical effectiveness used, one for the quantitative papers and a second for
and cost-effectiveness of group CBT compared with the qualitative papers.
currently used packages of care for women with PND. Results: Six studies met the inclusion criteria for
Data sources: Seventeen electronic bibliographic the quantitative review. Three were randomised
databases were searched (for example MEDLINE, controlled trials (RCTs) and three were non-
MEDLINE In-Process & Other Non-Indexed Citations, randomised trials. Two studies met the inclusion
EMBASE, PsycINFO, etc.), covering biomedical, health- criteria for the qualitative review. These were both
related, science, social science and grey literature treatment evaluations incorporating qualitative
(including current research). Databases were searched methods. Only one study was deemed appropriate
from 1950 to January 2008. In addition, the reference for the decision problem; therefore a meta-analysis
lists of relevant articles were checked and various was not performed. This study indicated that the
health services related resources were consulted via reduction in the EPDS score through group CBT
the internet. compared with routine primary care (RPC) was 3.48
Review methods: The study population included [95% confidence interval (CI) 0.23 to 6.73] at the end
women in the postpartum period (up to 1 year), of the treatment period. At 6-month follow-up the
meeting the criteria of a standardised PND diagnosis relative reduction in EPDS score was 4.48 (95% CI
using the Diagnostic and Statistical Manual of Mental 1.01 to 7.95). Three studies showed the treatment to
Disorders-Fourth Edition, or scoring above cut-off on be effective in reducing depression when compared
the Edinburgh Postnatal Depression Scale (EPDS). No to RPC, usual care or waiting list groups. There was
exclusion was made on the basis of the standardised no adequate evidence on which to assess group CBT
depression screening/case finding instrument of compared with other treatments for PND. Two
iii
studies of group CBT for PND were included in the evidence to assess the effectiveness of group CBT
qualitative review. Both studies demonstrated patient for PND. The evidence that was available was of low
acceptability of group CBT for PND, although negative quality in the main because of poor reporting of the
feelings towards group CBT were also identified. A results. Furthermore, little information was reported
de novo economic model was constructed to assess on concurrent treatment used in the studies, which
the cost-effectiveness of group CBT. The base-case was controlled for in only two of the studies.
results indicated a cost per quality-adjusted life-year Conclusions: Evidence from the clinical effectiveness
(QALY) of 46,462 for group CBT compared with review provided inconsistent and low quality
RPC. The 95% CI for this ratio ranged from 37,008 information on which to base any interpretations
to 60,728. There was considerable uncertainty in for service provision. Although three of the included
the cost per woman of running a CBT course, of studies provided some indication that group psycho-
the appropriateness of efficacy data to the decision education incorporating CBT is effective compared
problem, and the residual length of benefit associated with RPC, there is enough doubt in the quality of the
with group CBT. These were tested using univariate study, the level of CBT implemented in the group
sensitivity analyses. Supplementary analyses that fitted programmes, and the applicability to a PND population
distributions to the cost of treatment and the duration to limit any interpretations significantly. It is also
of comparative advantage reported a cost per QALY of considered that the place of group CBT in a stepped
36,062 (95% CI 20,464 to 59,262). care programme needs to be identified, as well as
Limitations: The cost per QALY ratio for group CBT there being a need for a clearer referral process for
in PND was uncertain because of gaps in the evidence group CBT.
base. There was little quantitative or qualitative RCT
iv
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Contents
Glossary and list of abbreviations ........... vii 7 Conclusions ............................................... 53
Implications for service provision .............. 53
Executive summary .................................. ix Suggested research priorities ..................... 53
Glossary
Postnatal depression (also known as Primipara A woman who is pregnant for the
postpartum depression) A non-psychotic first time, or has given birth to only one child.
depressive episode meeting standardised
diagnostic criteria for a minor or major The Beck Depression Inventory A 21-item
depressive disorder, beginning in or extending self-report scale used to determine depression
into the postnatal period. The term puerperal is severity. Items are scored on a 03 scale giving a
also used to describe the postnatal period. total range of 063. Total scores within the 19
range indicate minimal depression, 1018 mild
Cognitive behaviour therapy (CBT)The depression, 1929 moderate depression, and
pragmatic combination of concepts and 3063 severe depression.
techniques from cognitive and behaviour
therapies common in clinical practice. CBT The Edinburgh Postnatal Depression
aims to facilitate, through collaboration and Scale The most widely used self-report scale
guided discovery, recognition and re-evaluation designed to measure postnatal depression
of negative thinking patterns and practising new symptomology. The scale consists of 10-item
behaviours. Likert format relating to depression and anxiety
symptomology. Items are scored on a 03 scale
Interpersonal psychotherapy A time-limited, to give a total range of 030. Total scores within
structured and psycho-educational therapy the 1230 range suggest significant depression.
which links depression to role transitions,
interpersonal disputes, interpersonal sensitivity The Center for Epidemiological Studies
or losses. It facilitates understanding of recent Depression Scale A short self-report scale
events in these interpersonal terms and explores designed to measure depressive symptomology
alternative ways of handling interpersonal in the general population. The 20-item scale
situations. has a possible range of score from 0 to 60,
with higher scores indicating more symptoms,
Multipara A woman who has given birth two or weighted by frequency of occurrence during the
more times. past week.
vii
List of abbreviations
All abbreviations that have been used in this report are listed here unless the abbreviation is well
known (e.g. NHS), or it has been used only once, or it is a non-standard abbreviation used only in
figures/tables/appendices, in which case the abbreviation is defined in the figure legend or in the
notes at the end of the table.
viii
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Executive summary
Background Results
Postnatal depression (PND) describes a wide Number and quality of studies
range of distressing symptoms that can occur in Clinical effectiveness
women following childbirth. A clinical diagnosis Six studies met the inclusion criteria for the
of the disorder is often made using the Diagnostic quantitative review. Three were randomised
and Statistical Manual of Mental Disorders-Fourth controlled trials (RCTs) and three were non-
Edition which describes a range of diagnostic randomised trials. Two studies met the inclusion
categories indicative of a depressive disorder. criteria for the qualitative review. These were both
There is substantial evidence to support the use of treatment evaluations incorporating qualitative
cognitive behaviour therapy (CBT) in the treatment methods.
of depression, and psychological therapies are
recommended by the National Institute for Health Cost-effectiveness
and Clinical Excellence as a first-line treatment for No studies were identified that were deemed
PND. However, access is limited owing to expense, relevant to the decision problem.
waiting lists and availability of therapists. Group
CBT may, therefore, offer a solution to these Evidence of effectiveness
problems by reducing therapist time and increasing
the number of available places for treatment. Clinical effectiveness
Six studies of group CBT for PND were included
in the quantitative review as part of a narrative
Objectives analysis. Only one study was deemed appropriate
for the decision problem; therefore a meta-analysis
The overall aims of the review were to evaluate the was not performed. This study indicated that the
clinical effectiveness and cost-effectiveness of group reduction in the Edinburgh Postnatal Depression
CBT compared with currently used packages of Scale (EPDS) score through group CBT compared
care for women with PND. with routine primary care (RPC) was 3.48 [95%
confidence interval (CI) 0.23 to 6.73] at the end
of the treatment period. At 6-month follow-up the
Methods relative reduction in EPDS score was 4.48 (95% CI
1.01 to 7.95). Three studies showed the treatment
Clinical effectiveness
to be effective in reducing depression when
A systematic review of the literature was performed compared to RPC, usual care or waiting list groups.
to identify all studies describing trials of group There was no adequate evidence on which to assess
CBT for PND. Databases were searched (for group CBT compared with other treatments for
example MEDLINE, MEDLINE In-Process PND. Two studies of group CBT for PND were
& Other Non-Indexed Citations, EMBASE, included in the qualitative review. Both studies
PsycINFO, etc.) from 1950 to January 2008 for demonstrated patient acceptability of group CBT
both quantitative and qualitative studies. for PND, although negative feelings towards group
CBT were also identified.
Cost-effectiveness
Cost-effectiveness
A systematic review of the literature was performed A de novo economic model was constructed to
to identify all cost-effectiveness studies of group assess the cost-effectiveness of group CBT.
CBT for PND. Databases were searched from 1950
to January 2008.
ix
SF-6D, to determine the effectiveness of group of the size of the participant group; the effect of
CBT compared with RPC and individual CBT the session duration; the effect of the setting; the
(preferably in terms of a utility measure to obviate qualifications and involvement of the facilitator;
the transformation from the EPDS) and to the effectiveness of group CBT on the different
determine the duration of comparative advantage subtypes of PND; whether effectiveness is
by following up the women 1 year, or longer, after dependent on patient background, comorbidity,
randomisation. the number of children, previous PND, pre-
pregnancy or antenatal depression; and the
If the sample size is large enough, data on the indirect effects of the treatment on the infant and
following aspects should be recorded: the effect other family members.
xi
Chapter 1
Background
identify PND; although this scale is the most widely longer periods tend to identify higher prevalence.
used self-report scale for the identification of It is noted, however, that the EPDS is not, in
PND administered by the health-care provider, it itself, a diagnostic test. It should be followed by a
should be noted that further research is required diagnostic interview or longer structured measure
to establish the measure as a tool of identification if a diagnosis of PND is required.
or diagnosis for PND. The scale consists of 10-item
Likert format relating to depression symptomology Mental illness associated with childbirth can
and has also been shown to measure anxiety occur in the form of new episodes but also as a
symptomology.6 Items are scored on a 03 scale, recurrence of pre-existing illnesses.7 The risk of
giving a total range of 030. Total scores within suffering from severe affective disorders, including
the range 1230 suggest significant depression. PND, is elevated in women who have recently given
The Beck Depression Inventory (BDI) is also used birth compared to the general population.7 Women
in the screening of PND. It is a 21-item self-report with a history of severe mental illness, whether
scale used to determine depression severity. Items associated with childbirth or not, have an increased
are scored on a 03 scale, giving a total range of risk of a recurrence of their condition of between
063. Total scores within the 19 range indicate 33% and 50% following the birth of a child. This
minimal depression, 1018 mild depression, 1929 risk is at its greatest during the first 30 days after
moderate depression, and 3063 severe depression. birth.17 PND is distinguished from both postnatal
blues and postnatal psychosis: PND is considered
A clinical definition in use in the UK is non- to be more severe and has a longer duration of
psychotic depression occurring during the first depressive symptoms in comparison to postnatal
3 months postpartum.7 Symptoms of PND may or maternity blues, as they are sometimes called.1
spontaneously resolve 36 months after onset,8 However, Beck18 suggests that it is the timing of the
although some symptoms of depression are depressive symptoms that differentiates PND and
common in sufferers up to a year after delivery.9 postnatal blues.
It should also be noted that there are strong links
between prenatal depression and anxiety, and PND Up to 80% of women experience emotional
and anxiety,10,11 and that the presentation of PND lability, known as postnatal blues, in the first
may be comorbid with other mental disorders. 2 weeks postpartum, making this experience
extremely common.1 For those with postnatal blues,
Morrell et al.12 provide UK data on EPDS levels symptoms occur in the first few days after delivery
at 6 weeks postpartum. Based on a sample of and can last for up to 10 days. Evaluation should
3449 postnatal women, 595 had an EPDS13 take place if symptoms continue beyond 10 days to
score of 12 or more at 6 weeks postpartum; an identify PND. However, the symptoms of postnatal
estimated proportion of 17.3% [95% confidence blues and PND can be difficult to distinguish at this
interval (CI) 16.0 to 18.5]. However, it should be early stage. Symptoms of postnatal blues include
noted that the EPDS does not yet have a proven crying, irritability, fatigue, anxiety and emotional
role in the identification, screening or diagnosis lability, and it is suggested that maternity blues
of PND. Therefore, prevalence rates based on may be a normal reaction following the physiologic
the EPDS should be treated with caution. This changes associated with childbirth.18
is comparable with previous reports that have
suggested PND affects approximately 14.5% of Postnatal depression is also distinguished
women in developed countries during the first from postpartum psychosis which has a much
3 months postpartum,14 and 13% of new mothers less frequent incidence and is more severe.1
in developing countries.15 At 6 months postpartum The prevalence of postpartum psychosis has
it is reported that the prevalence of PND in the been reported as one to two women per 1000
UK is 9.1% in new mothers compared to 8.2% in deliveries.19 Symptoms can include delusions,
women who had not given birth within the previous hallucinations, extreme agitation, confusion,
6 months.2 Milgrom et al.1 report that prevalence inability to eat or sleep, exhilaration and rapid
rates of PND are affected by the measurement tool mood swings, and women with postpartum
used such as self-report measures of depression psychosis are regarded as a danger to themselves
including the EPDS and BDI;16 sampling; timing and their infant.18
of the assessment; differing diagnostic criteria
used in clinical interviews, including the DSM-IV A multifactorial aetiology of PND has been
criteria and the ICD-103 criteria; and by the length suggested as no single causative factor has
of the postpartum period under evaluation, as emerged. There is little evidence for a biological
2
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
basis of PND;9,20 however, a number of psychosocial month, have you often been bothered by having
factors have emerged as risk factors. Prenatal little interest or pleasure in doing things?. If the
depression and anxiety, a history of previous woman answers yes to either question then a third
depression, and maternity blues have been shown question should be considered, Is this something
to be strongly related to PND.10,11,15,21 Further, that you feel you need or want help with?. Health-
psychosocial variables, such as self-esteem,10 care professionals may also consider the use of
stressful life events,11,15,21,22 childcare stress,10 marital self-report measures such as the EPDS, Hospital
conflict,10,15,21,22 a lack of social support,10,11,15,23 Anxiety and Depression Scale or Patient Health
low social status,10,15 infant temperament10 and Questionnaire-9 items.
unplanned or unwanted pregnancy,10 have emerged
as significant predictors of PND. In Sheffield, midwives visit postnatal women up
to 28 days after the birth, although they do not
Impact of health problem necessarily have to visit the women at home every
day, and often only visit until the 10th day. They
Significance for patients in terms of ill- do not usually use any formal tool for the detection
health (burden of disease) of PND, but are required to ask questions (as
Postnatal depression is a major health issue for the outlined in the previous paragraph) to assess how
affected individual but also represents a significant the woman is feeling. If the midwife feels there is
risk to the child of the sufferer. Impaired maternal a significant mental health problem he or she can
infant interactions24 can lead to attachment refer the woman to her GP for further assessment.
insecurity,25 and impaired cognitive26 and social- Women should not be discharged by the midwife
emotional development.27 Fewer positive mother until the health visitor has made contact, which
child interactions are reported in dyads where the usually occurs by 28 days after birth, although
mothers depression persists beyond 6 months practice is variable. Some health visitors use self-
postpartum than in those whose depressive report measures such as the EPDS typically at
symptoms end before 6 months.28 In addition 6 weeks postpartum if they feel PND may be an
to the impacts on mother and child, findings issue, although use of the EPDS is not a universal
have shown that there are links between womens practice. If they are concerned about the mental
depression and their partners mental health.29,30 health of the women and believe this is beyond
In men, partner depression has been found to be their scope they may consult the GP who could
associated with a higher probability of reporting refer the patient to the community mental health
depression,29 and PND in men has been reported team. Diagnosis is usually undertaken by the GP,
as associated with depression in their partners using a formal diagnostic framework, such as
during pregnancy and after delivery.30 DSM-IV criteria, for depression (source: Sheffield
Teaching Hospitals, Jessop Wing).
stated that self-help strategies [guided self-help, managed in a particular NHS trust and how this
computerised cognitive behaviour therapy (CCBT) may potentially contrast with the management of
or exercise], non-directive counselling delivered at PND in other areas of the UK. Rotherham Primary
home (listening visits), brief cognitive behaviour Care Mental Health Service provides a service
therapy (CBT) or interpersonal therapy (IPT) are based in GP practices for common mental health
recommended by NICE.5 problems, including PND. Women can be referred
to the service by any practitioner, obstetrician,
Antidepressant drugs are considered for women midwife, health visitor or other health professional
with mild depression during pregnancy or the during both the antenatal and postnatal periods,
postnatal period if they have a history of severe if it is felt necessary. Rotherham Primary Care
depression and they decline, or their symptoms do Mental Health Service provides a service based in
not respond to psychological treatments. However, GP practices for common mental health problems,
it is noted that, to minimise the risk of harm to including PND. Women can be referred to the
the fetus or child, drugs should be prescribed service by any practitioner, obstetrician, midwife,
cautiously.5 There is also evidence that women health visitor or other health professional during
prefer non-pharmacological modes of intervention both the antenatal and postnatal periods, if it is
at this time.33 felt necessary. The NICE clinical guidance for
antenatal and postnatal mental health is used by
For women with a moderate depressive episode practitioners where PND is suspected and they are
or a history of depression, or those with a severe aware of the primary care mental health service
depressive episode during pregnancy or in the and how to refer into it (although it should be
postnatal period, it is recommended by NICE that noted that this service is not specific to PND). The
structured psychological treatment specifically for EPDS is not used. Once referred to the service,
depression (CBT or IPT) should be considered. women may attend the GP practice or be visited at
If the woman has expressed a preference for it home for assessment; women may then be offered
antidepressant treatment will be considered as six to eight sessions of individual treatment in
an alternative, or combination treatment will which CBT approaches and counselling are utilised
be considered if there is no response, or there by the primary care mental health service staff (J
is a limited response to psychological or drug Hunter, Head of Service, Primary Care Mental
treatment alone. Health Service, Rotherham Community Health
Services, 2008, personal communication). This
Services are ideally provided in a timely fashion service may differ from other services provided
to ensure that adverse effects on the health in the UK in the following ways: health visitors in
of the woman and her baby can be avoided.34 other services may routinely administer the EPDS,
Specifically, it is recommended that women which was previously used in the Rotherham service
requiring psychological treatment for PND should and may be used again in the future; there may
be seen for treatment normally within 1 month of not be a dedicated GP-based service for common
initial assessment, and no longer than 3 months mental health disorders; and individual CBT may
afterwards.5 not be routinely administered. The applicability
of the Rotherham model to other areas is also
Variation in services and/or likely to be limited owing to the wide variation in
uncertainty about best practice health service provision amongst Primary Care
Trusts (PCTs) with a reported range of whole time
The NICE guidance states that the structure of equivalent health visitors per child under 5 years
services varies in different parts of the country old of 165 in County Durham PCT to 894 in
because of local factors including the organisation Lambeth PCT.35
of existing mental health services, the demographic
profile of the population and geographical issues. As the section on current service provision
Recommendations are made to ensure local indicates, psychological interventions to treat
needs are met and integrated care is delivered, by pregnant and breastfeeding women are preferable
developing managed clinical networks involving to the use of psychotropic medication because of
linked groups of services in primary, secondary and the risks of harm to the fetus or child. However,
tertiary care. in reality there is a significant mismatch between
provision of psychological therapies and the
As services vary widely across the UK it is recommendations for their provision.34 Although
appropriate to provide details of how PND is undocumented it is widely held that conventional
4
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
antidepressants are the usual first-line treatment with a psychiatric disorder (21%). Forty-two per
prescribed by GPs for women with PND. However, cent of trusts had no access to a specialist perinatal
women have been found to prefer psychological mental health service. Midwives provided most
intervention, rather than antidepressants, during antenatal and postnatal care but only 70% of trusts
the postnatal period.33,36 Furthermore, it is were able to refer women directly to mental health
common to prescribe antidepressants and provide specialists, this was not possible for the remaining
psychological therapies together, although a report 30%. Ninety-five per cent of trusts had access to a
suggests that there is no advantage in receiving mother and baby unit, it is assumed that the other
both, and that cognitive behavioural counselling 5% do not have any access to a mother and baby
and a separate antidepressant are equally unit, although this is not detailed in the report.
effective.33
The Healthcare Commission report concluded that
Previous attempts to improve services have there are inadequate provisions for mental health
had only limited success. A Royal College of needs in many trusts maternity services, including
Psychiatrists report suggests that, despite efforts to booking, speciality training, streamlining referral
improve the recognition of and screening for PND pathways and access to specialist services. If services
in primary care, little has changed.37 In a more are lacking for those with severe postnatal illnesses,
recent report by the Healthcare Commission31 the likelihood is that this will be the case with those
(now known as the Care Quality Commission) treated only in primary care for mild to moderate
it is stated that the recording of mental health depression associated with pregnancy and the
needs by maternity staff in trusts is inconsistent, postpartum period, although this is not explicitly
making it problematic to assess the prevalence covered in the Healthcare Commission report.
of mental health problems associated with child
birth. It reported the number of women receiving There is also uncertainty around the number of
a postnatal check-up of their own health and well- women with PND who may be undiagnosed or
being at 6 weeks postpartum as ranging between unidentified. It is reported that women are often
71% and 97%. Half of the trusts reported a rate of reluctant to pursue health care for PND for a
89% or below, showing that many women may not variety of reasons. These include a lack knowledge
be receiving postnatal checks with the GP. about the condition meaning they are not aware
they have it, thinking they could or were expected
The Healthcare Commission report in relation to cope with it without help, stigma and a fear of
to mental health focuses on input from perinatal failure a fear of losing their baby if they admit
psychiatry and puerperal psychosis and suggests to having PND, the fear of giving the family
that PND can be treated with support from a bad name, and the fear of being labelled as
mainstream services and does not usually require mentally ill.13,36 Cultural reasons have also been
specialist services. As women with a previous reported, these include the fact that the family may
history of mental health problems and those with discourage women from obtaining help as it is seen
depression during pregnancy are reported as at as unacceptable to discuss such issues with people
higher risk of developing postnatal illnesses,15,21 the external to the family. Furthermore, it is reported
data reported by the Healthcare Commission may that health professionals may limit the number of
have some relevance to PND. women who come forward for treatment for PND
by making inappropriate assessments and having
Data for the Healthcare Commission report31 were insufficient knowledge of PND to provide adequate
provided from 40 trusts, and of these the median care. It is also reported that women with PND feel
trust reported that 8% of women were identified health professionals have a tendency to normalise
at booking as having personal or family history depressive symptoms making women less likely to
of mental illness (range 230% across trusts). pursue treatments. They also feel that they have
Twenty-nine trusts provided data on referrals limited time with health professionals and are not
to mental health teams following booking; the taken seriously.36 These reports suggest that there
median number of women referred by these may be a significant number of women with PND
trusts to a mental health team was 1.6% (range who remain undiagnosed and that a clearer referral
07%). It was also reported that about a third of process may help address this.
trusts had joint clinics with mental health teams
for previous puerperal psychosis, and some had It is beneficial to improve the commissioning of
specialist midwives for women with previous effective antenatal and postnatal mental health
puerperal psychosis (19%) or to support women services for a number of reasons outlined in the
5
commissioning guide. These include improving the Trust, 2008, personal communication). The cost of
motherchild relationship, reducing inequalities a CBT session has been estimated as 6238 (based
and improving timely access to services in primary on a 55-minute session), therefore we estimate
care, mental health and maternity services; the cost per hour to be 68. Assuming 25 hours
reducing the risk of relapse; reducing the risk of of treatment and clinical supervision, the cost per
women stopping medication in an unplanned way; patient would be 1700. An alternative method
reducing the number of inappropriate referrals and based on health visitor hourly rate provides a
readmissions and the length of inpatient stays, and larger cost; the cost per hour of health visitor time
offering alternatives to admission; reducing the was estimated at 89 (based on information from
risk of self-harm and suicide; preventing avoidable Morrell et al.13 amended using inflation indices
separation of mother and baby; and improving to represent current prices), which equates to an
performance and person-centred clinical care.34 estimated cost of 2225 assuming a health visitor
was required for 25 hours per patient, although it
Current service cost is unclear whether these resources would be used in
reality and may be an overestimation.
It is assumed that usual care (UC) for PND includes
visits by midwives and health visitors, visits to
the GP, prescriptions for medication, and other Description of technology
health contacts, such as community mental health under assessment
contacts, clinical mental health contacts and social
services contacts. Based on these contacts, Morrell Summary of intervention
et al.13 report that costs at 6 months postpartum Cognitive behavioural therapy is a psychotherapy
for women scoring 12 or above on the EPDS are commonly practised in the NHS. CBT refers to
374 per patient. Health visitor costs per hour of a combination of concepts and techniques from
client time were reported as 77 for UC, and 79 cognitive and behaviour therapies. Cognitive
for those trained in using a cognitive behavioural therapy is derived from cognitive theories and
or person-centred approach. Overall costs at seeks to challenge negative automatic thoughts
6 months were 339 for those receiving CBT or with an aim of changing maladaptive thoughts
person-centred therapy. These prices were based and beliefs.39 Behavioural therapy refers to
on 20034 unit costs: prices using 20078 inflation a therapy derived from learning theory and
indices38 would equate to health visitor costs of works on symptoms by changing behaviour and
86 for UC and 89 for those trained to deliver an environmental factors that control behaviour. The
intervention, and overall costs as 419 for UC and patient works collaboratively with a therapist to
380 for intervention care. The findings of Morrell identify the types and effects of thoughts, beliefs
et al.13 provide some evidence that a psychological and interpretations on current symptoms, feelings
intervention delivered by health visitors is cost- states and/or problem areas. They develop skills to:
effective compared to UC. The costs related to UC identify, monitor and then counteract problematic
did not include any formal CBT treatment. thoughts, beliefs and interpretations related to the
target symptoms/problems; learn a repertoire of
The current NICE guidance recommends coping skills appropriate to the target thoughts,
psychological intervention such as CBT or beliefs and/or problem areas; and test out new
IPT for women with PND. On occasions where behavioural patterns.40
formal CBT is provided it is assumed in current
practice to be on an individual basis. If a course Cognitive behavioural therapy has an important
of individual CBT were offered this would most role to play in helping people with mental
likely be delivered by a CBT therapist, and would health problems. There is evidence to support
consist of around 12 sessions, 90 minutes in the use of CBT in the treatment of several
duration. One or two follow-up sessions may be mental health problems (e.g. depression, panic/
included and the therapist would be required to agoraphobia, social phobia, generalised anxiety
undertake clinical supervision for approximately disorder, obsessive compulsive disorder, bulimia,
1030 minutes per session; however, it should be etc.).39 However, it has also been reported that
noted that the current service provision of CBT psychological therapy is effective in the treatment
may vary widely (P Slade, Professor of Clinical of mild to moderate, non-childbirth related
Psychology, University of Sheffield and J Curran, depression.41 There is no evidence that CBT is
Consultant Cognitive Behavioural Psychotherapist, more effective than other psychological therapies
Sheffield Health and Social Care NHS Foundation in the treatment of the same condition. Specific
6
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
to PND, a systematic review has indicated that include a flip chart, audio-visual equipment, and
psychosocial and psychological interventions are equipment to display powerpoint presentations (J
effective treatments.42 Furthermore, a recent trial Curran, personal communication). It is likely that
has demonstrated that psychologically informed services of this kind are very limited.
treatments delivered by trained health visitors are
clinically effective at 6 and 12 months for women The resources required using the delivery methods
with PND compared with UC.12 deemed by the authors to be most likely were group
CBT to become widely available would include
Cognitive behavioural therapy can be practised two group facilitators, a recently qualified clinical
in an individual or group setting; the potential psychologist and a health visitor.
benefits of providing CBT in a group setting
include increasing the availability of therapists, The criteria used for entry to the treatment
reducing waiting times and reducing costs. Group would normally include a diagnosis of DSM-IV
CBT differs from individual CBT only in the depression, or an elevated score on a self-report
respect that participants are treated in small measure such as the EPDS. However, those with
groups of around eight people, rather than in a subthreshold symptoms of PND or those with
one-to-one situation with their therapist. Group a history of depression may also be referred at
CBT treatment usually runs for 12 weeks, and the discretion of the GP (J Curran, personal
is often preceded by one individual session of communication).
2-hour duration with the purpose of assessing the
patient and briefing the patient regarding group Identification of important
treatment, and one or two sessions follow-up the subgroups
treatment. Thus, approximately 13 sessions are
required for the group treatment, each typically From a clinical perspective, PND includes four
of 2-hour duration. The group facilitators are subgroups of women whose management may
likely to require 12 hours for preparation and differ: (1) those who develop depression only after
supervision. Follow-ups may take place at 6 months childbirth; (2) those who have developed antenatal
and sometimes at 12 months, but may vary to depression which continues into the postnatal
a large extent. Group psycho-educational CBT period; (3) those with pre-existing chronic or
is lower impact than normal group CBT and is relapsing depression; and (4) subthreshold groups.
usually delivered in a smaller number of sessions, It was not possible to assess the efficacy of group
four to six opposed to 1012 (J Curran, personal CBT for these subgroups separately because of a
communication). lack of available data.
health visitor per session, and a further hour per per session per health visitor would be required
session per health visitor would be required for for travelling to and from the sessions (G Parry,
travelling to and from the sessions. Based on these University of Sheffield, P Slade, University
parameters the health visitor time required would of Sheffield, J Hamilton, St Johns Hospital,
be 74 hours, cost per hour of health visitor time West Lothian, Clinical experts, 2008, personal
was estimated at 89 (based on information from communication). Facilitator time required would
Morrell et al.13 amended using inflation indices to be 112 hours, cost per hour of facilitator time
represent 20078 prices). This equates to a total was estimated at 89 (based on information from
health visitor cost of 6586 and a total cost per Morrell et al.13 amended using inflation indices
participant of 1317. to represent current prices). This equates to a
total facilitator cost of 9968 and a total cost per
The authors estimated that two group facilitators participant of 1246.
would be required, a recently qualified clinical
psychologist and a health visitor. The programme We assume the group facilitators would be
would consist of 12 sessions occurring once per undertaking their normal duties relating to UC
week for a duration of 2 hours. These would during the rest of the week. The costs presented
be preceded by a 2-hour individual session for may be slightly underestimated as they do not
the initial assessment. The average number of include any set up costs or additional running
participants for the treatment was estimated as costs, such as room hire and crche facilities, which
eight. Preparation time was estimated as 1 hour may be incurred (J Hamilton, Psychiatrist, personal
per health visitor per session, and a further hour communication).
8
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Chapter 2
Definition of the decision problem
Chapter 3
Assessment of clinical effectiveness
Methods for reviewing the same strategy was used with minor alterations
effectiveness necessary for specific databases. The searches were
Identification of studies undertaken in January 2008. The databases were
searched from 1950 to 2008, the actual date range
Search strategies for each of the databases searched depended on
The search aimed to identify all references relating the coverage of the individual database.
to the clinical effectiveness of group CBT for PND.
The original intention was to synthesise evidence Search restrictions
within the framework of a mixed-treatment The searches were intended to be as broad as
comparison;44 however, during the early stages possible, and whilst they were restricted to human
of the research it became clear that the clinical studies where possible, they were not restricted by
evidence regarding group CBT was relatively language, date, publication type or study design.
poor. As such, confidence in building a coherent Non-English papers were excluded at the sifting
network that contained comparable study designs stage rather than setting this as an inclusion
and homogeneous participants was low. The use criterion.
of substantial resources to construct a comparison
with potential low internal validity was not deemed Inclusion and exclusion criteria
appropriate.
Population
Sources searched Included: Women in the postpartum period (up
Seventeen electronic bibliographic databases were to 1 year), meeting the criteria of a standardised
searched, covering biomedical, health-related, PND diagnosis using DSM-IV, or scoring above
science, social science and grey literature (including cut-off on the EPDS. No exclusion was made on
current research). A list of the databases searched is the basis of the standardised depression screening/
provided in Appendix 1. case finding instrument of standardised clinical
assessment tool used to define PND.
In addition, the reference lists of relevant articles
were checked and various health service-related Excluded: Prenatal women, women with other
resources were consulted via the internet. comorbid psychiatric disorders or major medical
These included health technology assessment problems, and women who have been involved in a
organisations, guideline producing bodies, generic previous psychological programme.
research and trials registers, and specialist mental
health sites. A list of these additional resources is Intervention
given in Appendix 1. Included: All interventions that included elements
designated as deriving from cognitive behavioural
Search terms principles including those that are purely psycho-
A combination of free-text and thesaurus terms education (i.e. any psycho-educational activity
were used. Key papers identified through initial which is informed by cognitive behavioural theory
scoping searches were used to develop keyword or techniques) in a group setting.
strategies. Population search terms (e.g.
depression, postpartum, postnatal depression and Setting
post pregnancy depression) were used to identify Included: All settings.
any references related to this population. The
searches were not restricted by intervention because Comparator
of the complexity of defining the intervention and Included: All comparators were considered. These
to prevent omission of relevant references. Copies included RPC, waiting list, individual CBT, group-
of the search strategies used in the major databases based counselling, medication, group behaviour
are included in Appendix 1, for the other databases therapy and group IPT.
11
12
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Tables relating to those studies excluded at full that the groups followed a structured format;
paper sift with reasons for exclusion are presented however, the use of a manual was not reported.
in Appendix 4 (see Tables 4750). The use of antidepressants also formed part of
the intervention and medication use proved to be
Study characteristics much higher in the intervention group than in the
Study characteristics for the six studies are control group. The Clark et al.61 study examined
described in Appendix 2, and a summary of this a group that provided therapeutic intervention
information is provided in Table 2. RCTs are and peer support. Exercises and strategies were
presented followed by non-randomised trials in drawn from CBTs, although it was reported that
date order. the intervention was not proscribed by a manual.
In addition to the 1-hour womens group, there
Description of group CBT was an additional motherinfant dyadic group
Included studies were those whose interventions which lasted 30 minutes; therefore the findings
incorporated any psycho-educational activity which of the study may be confounded by this co-
is informed by cognitive behavioural theory or therapy. Information on the content of the group
technique, in a group setting. The included studies interventions extracted from the studies is provided
therefore were required to specifically refer the use in Appendix 2 (Tables 18 and 19).
of CBT when describing their intervention. Varying
degrees of detail regarding the description of the In summary, three studies43,59,60 specifically referred
group programmes were provided and in the main to at least a CBT component which appeared to
these descriptions were brief. Therefore, when we be a core, predefined aspect of the treatment. It
refer to group CBT we are referring to a group should be noted that this could not be claimed
programme that states that it incorporates some with any certainty for the Highet and Drummond60
level of CBT theory or technique. study because of poor reporting. The definitions
used in the other three studies58,61,62 were somewhat
It was deemed important to assess the degree ill-specified and it was unclear whether CBT was a
to which the interventions used in each study core aspect of the group treatment.
actually reflected and incorporated CBT theory or
technique. The CBT components of the studies are Study quality
described here and studies are presented in order The Downs and Black checklist55 was used to assess
of relevance to group CBT. The Milgrom et al.59 both the randomised and non-randomised studies.
study was judged to most accurately reflect group Key components of the quality assessment are listed
CBT for a number of reasons. The intervention in Table 3 and in Appendix 2 (see Tables 18 and 19).
was termed group-based CBT rather than a group The components of the checklist used to assess the
incorporating CBT theory or techniques, and studies included (1) the standard of reporting, (2)
it was reported to be clinic-based and delivered the external validity of the study, (3) the internal
according to detailed manuals. The Highet validity of the study, and (4) power to detect
and Drummond60 study specifically reported changes in depression.
the use of group CBT; however, no further
details were reported. The Honey43 study used 1. To assess the standard of reporting the
the term brief psycho-educational group and following issues were examined: whether there
specifically referred to use of cognitive behavioural were clearly described objectives, outcomes,
techniques as one of the three aspects of the patient characteristics, interventions and
group intervention; it also stated that although findings; whether estimates of random
the intervention was not proscribed by a manual, variability for main outcomes were assessed;
a predefined programme was employed. Meager and whether adverse events had been reported.
and Milgrom62 referred to their intervention as a 2. For external validity, the representativeness
cognitive behavioural treatment programme, and of the sample and representativeness of the
the cognitive behavioural component was reported intervention and its setting were assessed.
as one of eight components of the programme, 3. The following issues were considered to assess
they did not refer to the use of a manual. The internal validity (bias): blinding; whether
Rojas et al.58 study was less specific in describing data dredging had been used; whether follow-
the group intervention. The group was referred to up time was equivalent for controls and
as a psycho-educational group (PEG) and among experimental groups; whether appropriate
other aspects included behavioural activation statistical analyses had been applied; the
and cognitive techniques. The authors stated compliance with interventions; and the
14
TABLE 2 Summary of study characteristics for the six included studies
DOI: 10.3310/hta14440
continued
Health Technology Assessment 2010; Vol. 14: No. 44
15
16
TABLE 2 Summary of study characteristics for the six included studies (continued)
Drummond (2004),60 treatment treatment (which differed groups to wait at least 3 weeks to
non-RCT, Australia groups; 10 WLG in duration); 6 months receive group intervention
after end of treatment
Clark et al. (2003),61 13 MITG; 15 Baseline prior to MITG one session per MITG, IPT and infant development group WLG those waiting to
non-RCT, USA IPT; 11 WLG treatment; at the end of week for 12 weeks, 90 occurred simultaneously, followed by mother receive MITG
treatment 12 weeks minutes in duration (60 infant dyadic group. Based on interpersonal,
minutes for mothers psychodynamic, family systems, and cognitive
group, 30 minutes for behavioural approaches
motherinfant dyadic IPT group individual therapy, relating to partners,
activities); number of children and others
attendees not reported
Meager and Milgrom 10 group Baseline prior to Group treatment one Group treatment programme consisting WLG had the opportunity
(1996),62 non-RCT, treatment; 10 allocation and beginning session per week for of targets which take into consideration the to participate in the
Australia WLG of treatment; at the end 10 weeks, 90 minutes in risk factors for postpartum depression. An treatment programme
of treatment 10 weeks duration; 10 attendees environment of social and emotional support, an once the participants in
educational component, a cognitive behavioural the treatment group had
component, encouragement of networking, completed the programme
examination of patterns of communication,
normalising of feelings, involvement of spouse in
the group, practical homework
MCI, multicomponent intervention; MITG, motherinfant therapy group; NR, not reported; PEG, psycho-educational group; RPC, routine primary care; WLG, waiting list group.
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Author (date),
study type, setting Quality
Rojas et al. (2007), 58
Reporting: Objectives, outcomes, patient characteristics and interventions clearly described, results
RCT, Chile difficult to interpret. Estimates of random variability given for main outcomes. Adverse events were
not reported
External validity: Baseline characteristics of participants were not compared across groups using a
statistical test although they appeared to be well matched. The intervention was not representative
of UC for this population
Internal validity: Participants could not be blinded; recruiters and assessors were blind to treatment
allocation. Data dredging was not used. Follow-up times were equivalent for each group.
Appropriate statistical analyses were employed. It is unclear whether compliance with interventions
was reliable, as the experimental intervention was multicomponent making the assessment of the
effects of group treatment difficult, and the control group were not receiving identical treatment,
as is the case in UC. Outcome measures were reliable and valid. Participants were in different
intervention groups. Randomisation was individually based with use of computer-generated random
numbers. Numbers lost to follow-up were reported, but reasons for loss to follow-up not reported
Power: Calculation reported
Milgrom et al. Reporting: Objectives, outcomes, patient characteristics and interventions clearly described, results
(2005), 59 RCT, difficult to interpret as combined scores used. Estimates of random variability given for main
Australia outcomes although only for combined scores. Adverse events were not reported
External validity: Baseline characteristics of participants were not compared across groups, reported
for all participants together. The interventions were not representative of UC for this population
Internal validity: Assessors blinded. Participants blinded until treatment started. Data dredging was
not used. Follow-up times were equivalent for each group. Appropriate statistical analyses were
employed, although combined analyses were performed making interpretation regarding individual
interventions difficult. It is unclear whether compliance with interventions was reliable, as it is not
clear whether participants in the experimental conditions were receiving other treatment. The
control group may not have been receiving identical treatment, as is the case in routine primary
care. Outcome measures were reliable and valid. Participants were in different intervention groups.
Randomisation was performed by cycling allocation and by drawing lots (one coded slip of paper
drawn from a bag containing multiple slips coded in equal number for each of the four treatment
conditions). Numbers lost to follow-up were reported, but reasons for loss to follow-up not
reported
Power: Calculation reported
Honey (2002),43 RCT, Reporting: Objectives, outcomes, patient characteristics, interventions and results clearly described.
UK Estimates of random variability given for main outcomes. Adverse events were not reported
External validity: Baseline characteristics of participants were compared across groups using a
statistical test. The intervention was not representative of UC for this population
Internal validity: Details of blinding were not reported. Data dredging was not used. Follow-up times
were equivalent for each group. Appropriate statistical analyses were employed. It is not clear
whether compliance with interventions was reliable; antidepressant use was included as a covariate
in the analyses. However, the control group may not have been receiving identical treatment,
as is the case in routine primary care. Outcome measures were reliable and valid. Participants
were in different intervention groups. Randomisation was performed using a block randomisation
procedure. Numbers lost to follow-up were reported, but reasons for loss to follow-up not
reported
Power: Calculation not reported
Highet and Reporting: Objectives and outcomes clearly described, limited patient characteristics reported and
Drummond (2004),60 not clearly described, and interventions were not clearly described. The results were difficult to
non-RCT, Australia interpret because of participants being included in more than one intervention group. Estimates of
random variability given for main outcomes. Adverse events were not reported
External validity: Baseline characteristics of participants were not compared across groups using
a statistical test and it was difficult to ascertain whether they were well matched because of
the limited detail reported. The interventions were representative of the array of UC for this
population, due to the retrospective nature of the trial
continued
17
TABLE 3 Assessment of study quality for the six included studies (continued)
Author (date),
study type, setting Quality
Internal validity: No blinding was employed as the study was retrospective. Data dredging was
used. Follow-up times differed depending on the treatment given. Appropriate statistical analyses
were employed, although these were combined analyses making interpretations regarding specific
interventions difficult. Compliance with interventions was not reliable as the intervention groups
were overlapping, although data for some intervention groups were presented separately. The
control group was very small and participants were not receiving any treatment. Outcome
measures were reliable and valid. Participants were not in different intervention groups in all cases.
No randomisation took place because of the retrospective nature of the study. Numbers lost to
follow-up were reported and not included in the study, reasons for loss to follow-up were reported
Power: No power calculation was reported
Clark et al. (2003),61 Reporting: Objectives, outcomes, patient characteristics, interventions and results clearly described.
non-RCT, USA Estimates of random variability given for main outcomes. Adverse events were not reported
External validity: Baseline demographic characteristics of participants were compared across groups
using a statistical test, pretreatment depression scores were included as a covariate in the analyses.
The interventions were not representative of UC for this population
Internal validity: No blinding was reported. Data dredging was not used. Follow-up times were
equivalent for each group. Appropriate statistical analyses were employed. It was not clear
whether compliance with interventions was reliable; other treatments may have been prescribed
simultaneously. The motherinfant dyadic activities may have confounded the group intervention.
It was not clear whether the control group were receiving any treatment during the waiting period.
Outcome measures were reliable and valid. Participants were in different intervention groups. No
randomisation was performed; participants were matched and sequentially assigned to groups.
Numbers lost to follow-up were reported, but reasons for loss to follow-up not reported
Power: No power calculation was reported
Meager and Milgrom Reporting: Objectives, outcomes, patient characteristics and interventions clearly described, results
(1996),62 non-RCT, difficult to interpret because of statistical tests used. Estimates of random variability not given for
Australia main outcomes. Adverse events were not reported
External validity: Baseline characteristics of participants were compared across groups using a
statistical test. The intervention was not representative of UC for this population
Internal validity: No blinding was reported. Data dredging was not used. Follow-up times were
equivalent for each group. Appropriate statistical analyses were not employed or not reported.
Compliance with interventions appeared reliable. Medication use was reported and post hoc
examination revealed no significant differences between the groups on medication usage. Outcome
measures were reliable and valid. Participants were in different intervention groups. The study was
reported to be randomised but method was not reported. Numbers and reasons for loss to follow-
up provided
Power: No power calculation reported
reliability and validity of outcome measures. To all three RCTs, blinding of participants is not
assess internal validity confounding (selection possible for psychological interventions owing
bias), the following were considered: whether to their nature; however, two studies reported
participants were in different intervention blinded assessment,58,59 and two58,59 reported power
groups, whether randomisation had been calculations. All three RCTs reported numbers lost
used, whether adjustment for confounding in to follow-up, but none reported reasons for loss to
the analyses were employed [were intention- follow-up.
to-treat (ITT) analyses employed], and the
reporting of loss to follow-up. Non-randomised controlled
4. Power was also considered by assessing whether trials
the study had employed a power calculation.
The three non-randomised studies were Meager
Randomised controlled trials and Milgrom,62 Clark et al.61 and Highet and
Drummond.60 Participants included in the Meager
Of the six included studies three43,58,59 were RCTs. and Milgrom62 study were volunteers and were
The method of randomisation was reported in reported to be randomly assigned to either the
18
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
group treatment or a waiting list group (WLG); CBT to IPT (and RPC). One RCT59 compared
however, the randomisation method was not group CBT to group counselling and individual
detailed. The Highet and Drummond study60 was a counselling (and RPC). One non-RCT60 compared
retrospective study which examined patient records; a number of different conditions (with overlapping
therefore, no randomisation had taken place. populations); non-overlapping conditions were
In the Clark et al.61 study, suitable participants group CBT only, individual CBT only and
were referred for the treatment by a health-care medication only (but not to RPC as noted above).
provider. Sequential assignment to group treatment
or to the waiting list was performed on the basis Sample size and drop-out rates
of matching for sociodemographic variables. A Sample sizes are shown in Table 2. The sample sizes
third individual treatment group was added later. for the included studies were relatively large for two
Owing to the retrospective nature of the Highet of the RCTs58,59 and relatively small for Honey,43
and Drummond60 study, no blinded assessment Clark et al.61 and Meager and Milgrom.62 The
was performed. Meager and Milgrom62 and Clark Highet and Drummond60 study had a relatively
et al.61 did not report that the assessment had large sample size due to its retrospective nature
been blinded. None of the non-RCTs presented a and the large number of conditions analysed, but
power calculation. Meager and Milgrom62 detailed as noted above participants who dropped out of
numbers and reasons for loss to follow-up, these treatments were not included in the analyses. Of
included physical illness, need to support de facto the RCTs, Rojas et al.58 had a large sample size
husband who was on a methadone programme, (>200) with relatively low drop-out rates (21 at
difficulty in organising attendance and distance to 3 months, 22 at 6 months), Milgrom et al.59 had
travel. Clark et al.61 gave numbers but not reasons a large sample size (>192) but had a relatively
for loss to follow-up. Highet and Drummond60 was large number of dropouts prior to the start of
a retrospective study therefore participants who the interventions (52). Honey43 had a moderate
had been lost to follow-up were not included in sample size (<50) and relatively low drop-out rates
the study at all. Reasons were provided for loss to before intervention (four) but these participants
follow-up, these included not being contactable were followed-up, although three participants in
post treatment, not considered to have PND by each condition who did participate could not be
their health-care provider, refusal to take part followed-up (six). Of the non-RCTs, Meager and
in the study and stopping treatment prior to Milgrom62 had a small sample size (20) with only
completion. one dropout prior to intervention; Clark et al.61 had
a relatively small sample size (40) with a relatively
Co-therapy or medication low drop-out rate before intervention for the group
Concurrent use of antidepressants was reported in treatment (four). The Highet and Drummond60
Rojas et al.,58 Honey,43 and Meager and Milgrom,62 study had a relatively large sample size overall;
although not controlled for in Rojas et al.,58 however, the relevant treatment condition sample
making interpretations regarding the effects of was of moderate size (<60).
group treatment problematic. Both cotherapy
and medication use was reported in Highet Therapy details
and Drummond60 and was controlled for in the
analyses. No medication was detailed in Milgrom Table 3 and Tables 22 and 23 in Appendix 2 describe
et al.59 and Clark et al.,61 although the intervention the details of therapy for the three RCTs and three
group participants in the Clark et al.61 study were non-RCTs.
also receiving motherinfant dyadic therapy.
Recruitment
Comparators For the RCTs, participants were recruited from
Comparators are shown in Table 3 and in Appendix a community screening programme of newly
2 (see Tables 18 and 19). All six included studies delivered mothers at 618 weeks postnatal, and
had a comparison arm.43,5862 Five of the studies, the were invited to take part if they scored 12 or
three RCTs43,58,59 and two non-RCTs61,62 compared above59 on the EPDS (the cut-off used for the Rojas
group CBT to RPC or a WLG [although it should et al.58 study was 10 or above), from three clinics
be noted that definitions of RPC and waiting list at any stage during the first postnatal year,58 or
vary across the studies, details are provided in referred by health visitors if they were attending
Table 3 and in Appendix 2 (see Tables 20 and 21)]. mother and baby clinics.43 For the non-RCTs,
Only one non-RCT60 compared group CBT to recruitment was through health-care provider
individual CBT. One non-RCT61 compared group referrals and newspaper advertisements,61 and
19
20
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Inclusion and exclusion criteria six had a semi-professional occupation, six were
Five43,58,59,61,62 studies included in the review had in sales or business management, two worked in
clearly stated inclusion criteria; however, one skilled occupations and four were housewives.62
study61 did not report exclusion criteria. The sixth The final study reported the mean educational
study60 did not report selection of participants level for each group; this was 14.9 years for the
using a standardised diagnostic measure as part of motherinfant therapy group (MITG), 15.5 years
the inclusion criteria. for the IPT group and 16 years for the WLG.61 Two
studies did not report this information.43,60
Patient characteristics
Some details of patient history were reported in
Tables 26 and 27 in Appendix 2 describe the details four studies. These included mean number of
of patient characteristics for the three RCTs and children as 1.859 and two,58 and the percentage
three non-RCTs. of primiparas in each group, which was 50% for
the PEG and 59% for the RPC groups.43 A further
Diagnosis of disorder study also reported the mean number of children
For the RCTs, PND was indicated using the in each group, which was two in the treatment
EPDS in one study,43 and a DSM-IV diagnosis of group and 1.6 in the control group, the average
major or minor depression was given in the other age of the infant was 10.6 months, and the marital
two studies.58,59 For the non-RCTs diagnosis was status of the women was 15 married, four in de
performed using the DSM-IV criteria for major facto relationships and one separated.62 No details
depression in the Clark et al.61 study. Diagnosis of patient history were reported in the remaining
information was not supplied in the Highet and two studies.60,61
Drummond60 or Meager and Milgrom62 studies
other than that the participants had been referred Time postpartum was reported only in four43,5861
for treatment of PND, although all participants of the studies. In Honey,43 Rojas et al.58 and
included in the trials had an EPDS score of 12 or Milgrom et al.,59 all participants were <12 months
above. postpartum, and the time postpartum in the Clark
et al.61 study ranged from 1 to 24 months. Details of
Age, gender, ethnicity, background and time postpartum were not reported in Meager and
patient history Milgrom62 and Highet and Drummond.60
As PND follows childbirth, those diagnosed with
the disorder are exclusively female. The mean Baseline comparability
age of the women taking part in the treatment Four studies reported baseline comparability.43,6062
was around 30 years across five of the studies, the In one study it was reported that groups did
sixth60 did not provide details of participants ages. not differ significantly on sociodemographic
Three studies provided standard deviations around and baseline self-report measures,43 a second
the mean age.43,58,59 reported that groups had been matched on
sociodemographic characteristics and baseline
Information regarding ethnicity was reported in depression scores had been used as a covariate in
only two of the studies, indicating in one study the analyses,61 the third reported that groups were
that 80% of the participants were Australian born similar in terms of clinical status and social support
with the remaining born in Ireland or the UK62 across all scales,60 and the fourth reported no
and in the other that one participant was African significant differences between the groups on mean
American and the remaining participants were age of infant, medication usage, pretest BDI scores
Caucasian.61 or occupational background.62 Two studies did not
report any details of baseline comparability.58,59
Four studies provided information on either
the education or socioeconomic background of Outcomes and results
participants. It was reported in one study59 that
62.7% had received 12 or more years of education, Tables 28 and 29 in Appendix 2 describe the details
with 30.5% receiving higher education. The of the outcomes and results for the three RCTs and
majority of the participants in the Rojas et al.58 three non-RCTs.
study had received 812 years of education [73%
in the multicomponent intervention (MCI) group, Improvement in psychological symptoms
75% in the UC group]. A further study reported The outcomes to be reported in the quantitative
that two women had a professional background, part of the review were clinical effectiveness in
21
terms of improvement in psychological symptoms to these requests but were not able to provide
of PND. these additional data.
One paper58 included a confounding factor,
Instruments antidepressant treatment in the intervention
All six studies reported outcome measures relating group, making it insufficiently similar to the
to depression. The main outcomes related to other studies.
improvement in depression symptoms in five of The delivery settings varied widely for each of
the six studies. The main outcome in the Clark the studies.
et al. study61 was infant development, although There was a mix of RCT and non-RCT data.
depression was measured. Studies reported various The level of CBT use was undetermined in a
other outcome measures such as social support, number of the studies.
self-esteem, mood, parenting stress and infant
development; however, these were not consistently Only one study was deemed appropriate for data
reported across all included studies and therefore extraction relevant to our decision problem. This
the data were not available to review these outcome was due to the lack of appropriate data in two
measures. The full range of outcome measures cases,59,62 and the lack of sufficient similarity in
reported is listed in Tables 30 and 31 of Appendix study type and comparator in one,60 a retrospective
2. Measures of depression were reported in each of study which did not provide appropriate data
the six included studies using either the EPDS or for the WLG. Two studies were suitable for the
the BDI. These are well-recognised and frequently meta-analysis of change in depression between
used scales to measure depression. baseline and follow-up;43,58 the third study61 did
not report follow-up data. A final study58 included
Results for psychological a confounding factor in the intervention arm.
symptoms Therefore, it was concluded that there was not
enough commonality of intervention, service
Meta-analysis setting, population and antidepressant use to
Meta-analyses using the six quantitative studies perform a meta-analysis.
were considered. Data were available to assess
group CBT against RPC. However, data were not For the study that was deemed appropriate, the
available to assess group CBT against individual reduction in the EPDS score through group CBT
CBT, or any other intervention. The suitability of compared with RPC was 3.48 (95% CI 0.23 to 6.73)
these data for meta-analysis was assessed and the at the end of the treatment period. At 6-month
following issues were encountered: follow-up the relative reduction in EPDS score was
4.48 (95% CI 1.01 to 7.95).
The outcome measure, depression, was
measured using two different scales, the EPDS Narrative analysis
and the BDI. Although this could be overcome As meta-analyses could not be performed the
using a standardised mean difference statistic, results are presented in narrative format.
assumptions regarding the standard deviation
of each scale would be required. Group psycho-education incorporating
Depression is measured at baseline, after CBT versus RPC, UC or WLG
treatment or at 3 months, and at follow-up Honey43 reported that depression symptomatology
(usually 6 months); however, this was not the as measured by the EPDS was significantly reduced
case for all six papers. Some of the studies only in the intervention group compared to the RPC
measured depression twice.59,61,62 See Tables 28 group. Depression scores improved over time for
and 29 in Appendix 2 for details. those in the intervention group, but not for those
One paper62 did not provide any measures in the RPC group. Six months after the end of
of variability around the mean (standard the intervention, significantly more women scored
deviations or CIs) and attempts were made below cut-off than in the RPC group, although
to gain this information from the author. there were no differences immediately post
A further paper59 did not provide separate intervention. The benefit in terms of improved
means and measures of variability around the depression score was not related to antidepressant
mean for each treatment group, again further use and was maintained 6 months after the
information was requested from the author. group had ended. However, some women in the
The authors of these papers have responded intervention group continued to show evidence
of depressive symptomatology at this 6-month
22
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
follow-up. Improvements in EPDS score were not When further examining these findings they do
accompanied by changes in coping, perceptions give an indication that group CBT may be just
of social support or the marital relationship. as effective as individual treatment. Clark et al.61
Rojas et al.58 reported that the MCI group had reported that the two treatment groups did not
significantly improved scores on EPDS and Short differ from each other on any of the measures
Form questionnaire-36 items measures compared administered, although the small sample size may
to the UC group overall, and when examining have contributed to the null findings.
simple effects this was true at 3 months post
randomisation, significant simple effects were not Other comparisons
reported at the 6-month follow-up. Meager and Highet and Drummond60 reported that receiving
Milgrom62 reported significant improvements treatment, compared with being part of the WLG,
on the EPDS and BDI in the intervention group achieved significant decreases in depression
compared with the control group. Clark et al.61 between pre and post treatment; a decrease in
reported no significant differences between the psychological anxiety also approached statistical
groups on the BDI, although BDI was lower post significance (p=0.081). There were no differences
treatment in both intervention groups than in between the groups for physiological anxiety
the control group. Also, both the women in the following treatment. Highet and Drummond60
MITG and those in the IPT condition reported also reported that for CBT versus behaviour
significantly fewer symptoms on the Center for therapy, following treatment, there were significant
Epidemiological Studies Depression Scale (CES-D) reductions in depression, psychological anxiety
measure post treatment than did those in the WLG, and physiological anxiety. Treatment gains were
there were no differences between the intervention maintained 6 months later for both treatment
groups. Specific analyses comparing group CBT conditions. The study suggested that whilst CBT
with the WLG or RPC group were not reported by was no more effective than behavioural-based
Highet and Drummond60 or Milgrom et al.59 supportive counselling, confounding effects of
greater medication use and greater treatment
Group versus individual treatment duration for those in the latter group may have
Highet and Drummond60 reported comparisons resulted in underestimation of the efficacy
of participants receiving individual treatment with and efficiency of CBT. It was also reported
those receiving treatment in groups. Depression that medication was no more effective than
decreased significantly following treatment for both CBT. Participants treated with CBT (alone or
groups and these treatment gains were maintained in combination with medication) had greater
at follow-up. Comparison of participants treated in decreases in depression and psychological anxiety
groups (alone and in conjunction with individual following treatment than those who received
treatment) versus those treated individually medication alone. Milgrom et al.59 reported that
revealed that depression was significantly lower changes in depression and anxiety immediately
at post treatment in subjects treated individually post intervention significantly differed between
than in those who received group or combined psychological interventions (combined data)
intervention. At follow-up there was also a compared with RPC, although it was not possible to
significant decrease in depression particularly in assess effects of the group CBT intervention alone
those treated in both group and individual settings. as data for the interventions were not presented
Depression continued to decrease for those separately. Milgrom et al.59 also reported data
who had been treated in the combined settings, showing that interventions based on a counselling
whilst there was no change for those treated in approach may be more effective when delivered on
groups only. Psychological anxiety declined at an individual rather than a group basis.
post treatment and during the 6 months follow-
up, particularly in those who received individual Patient preference, satisfaction and
treatment only. Specifically, anxiety decreased acceptability
more for those treated only on an individual basis Only one of the six studies provided data on
than for subjects treated in groups. The Milgrom et patient satisfaction.60 Similar ratings of satisfaction
al.59 study reported significantly better BDI scores were reported when comparing CBT with
for those undertaking individual treatment than medication, with neither being preferred over the
for those receiving group treatment, although other. Individual treatment was preferred to group
these were combined group treatment scores treatment, and similar ratings of satisfaction were
and they did not provide direct comparisons of reported when comparing group CBT with group
specific group CBT versus individual treatment. behaviour therapy, with neither being preferred
23
over the other. These data are summarised in Tables Study characteristics
32 and 33 in Appendix 2. Study characteristics for the two support group
with a specific CBT basis are described in Table 36
Qualitative papers of Appendix 3, and a summary of this information
Quantity and quality of papers available is provided in Table 4. Study characteristics for
support groups without any theoretical CBT basis
Figure 2 shows the QUOROM flowchart for the are provided in Table 37 of Appendix 3.
included qualitative studies. A table of excluded
studies with reasons for exclusion is presented Description of the treatment
in Appendix 5 (see Tables 52 and 53), with the As in the quantitative review, included studies
reference for these studies provided in Appendix 6. were those where interventions incorporated any
psycho-educational activity informed by cognitive
Included studies behavioural theory or technique, in a group
Details of those studies included in the qualitative setting. Therefore, when we refer to group CBT
review are provided in Tables 34 and 35 of we are referring to a group programme which
Appendix 3. The support groups without any states that it incorporates some level of CBT theory
theoretical CBT basis were used as a collective or technique. Only two52,53 papers examined a
comparator against CBT groups, and are detailed group treatment that was informed using CBT
at the end of this section. The following sections techniques. Included studies reported varying
relate to the two CBT studies. degrees of thickness64 regarding the description
of the support groups, although in the main
Unavailable papers
n=5
Unavailable through interlibrary loans
Total full papers included from or via the authors. From the details
qualitative sift available to the authors, it is unlikely
n=2 that these papers would contain data
relevant to this project
these descriptions were brief; this information is study53 is not clear. The qualitative methodology
provided in Appendix 3. Most included detail of was therefore appropriate for evaluation purposes
the number of sessions, their frequency, duration, in both cases. Both studies provided explanation
the number of participants per group and details and justification of the recruitment strategy and
of the group facilitator. The degrees to which the setting for data collection. Data collection methods
interventions used in each study actually reflected were clearly explained in one study;52 however, such
and incorporated CBT theory or technique are detail was lacking in the second.53 Reflexivity and
detailed here in order of relevance to group CBT. ethical issues were addressed in only one of the
studies.52 Data analysis was presented in a rigorous
The Davies and Jasper52 study termed the way in only one of the studies,52 although there was
intervention as a therapeutic group, known as The no reference to particular qualitative methods of
Lifeskills Group. The group programme had aims analysis in either study. Findings were clearly stated
to encourage cognitive restructuring and self-help. in both studies.52,53
Although the use of a manual was not stated, the
group had a predefined programme based on the Co-therapy or medication
CBT model. The Morgan et al.53 study examined Only one of the studies provided information on
a group programme for postnatally distressed cotherapy or medication.53
women and their partners. Psychotherapeutic and
cognitive behavioural strategies were employed Comparators
as part of the group programme, although a Neither of the studies utilised a comparator group.
particular manualised or predefined structure was
not reported. Little information was provided on Numbers of participants
the level of CBT used in the programme. The Morgan et al.53 study had a reasonable number
of participants (34), whilst the Davies and Jasper52
Study quality study had numbers of participants that reflected
The CASP checklist for qualitative studies56 was the size of the group taking part in the treatment
used to assess the quality of the studies. The key (8).
components of quality assessment are listed in
Appendix 3. The studies were assessed on the Therapy details
following criteria: justification of the research
design; whether the recruitment strategy and Tables 38 and 39 in Appendix 3 describe the details
setting for data collection were explained and of therapy for the six studies.
justified; whether the data collection methods
were explained and justified; whether reflexivity Recruitment
and ethical issues were addressed; whether data In one study the women were referred to the group
analysis was sufficiently rigorous; and whether treatment by health visitors,52 in the other study53
findings were clearly stated. Neither study could the women were referred from another treatment.
be defined as a qualitative research study, although
both incorporated a qualitative approach and Number and length of sessions
were included on this basis. Both studies were One study ran eight weekly 2-hour sessions,53 whilst
evaluations of a group treatment and the research the other ran 12 weekly sessions of 90 minutes in
design was justified in one of the studies,52 duration.52
although justification of the design in the second
25
TABLE 5 Themes and findings directly relating to the PND, motherhood and the context of having a baby
feelings specifically relating to PND.52 The women The leaders are careful to praise even meagre
reported: attempts at self-care. Clearly as the weeks pass
they are rethinking their roles as wife and
you feel that you are the only one and mother. As well as emotional outpouring and
that the feelings and thoughts you have are frequent tears, sound cognitive work begins to
dreadful, yes, to people who have not had this take place.
they are, but to people who had PND these (p. 915)
feelings are normal.
(p. 431) The authors of this study also reported:
Being around others with PND enabled them to the women were set the weekly task of
feel more normal by applying social comparisons caring for themselves in some practical way.
and prototyping. The quotation emphasises how Initially some members found this difficult.
women specifically needed to compare themselves They became irritated when their own or their
to others with PND to achieve this effect, thus partners behaviour was not perfect. They had
illustrating the value of a group treatment difficulty too when their babies did not behave
specifically for PND. perfectly.
(p. 915)
Practicalities and knowledge
Author comments from one study53 appeared to This appeared to illustrate that the participants
show that issues with PND were being addressed initially found practical tasks difficult, but the
with the implication that CBT processes were group was instrumental in overcoming this.
responsible for the changes.
The environment demonstrated that it was not always easy for group
The authors of one study reported:52 members to participate fully in the treatment:52
The leadership qualities most appreciated were she found it difficult to participate in the
the caring and supportive attitudes and the group commenting I have been depressed
provision of a safe environment for the sharing for over 2 years I found it hard to talk openly
of feelings. within a large group after a long period of
(p. 431) depression.
(p. 431)
They further reported:52
The same study further demonstrated that
All eight members found small group work these difficulties were specifically related to the
helpful for sharing experiences and getting application of the CBT techniques learned in the
to know each other. Other peoples honesty. sessions:52
Sharing their darkest thoughts, not being alone
any more. Additionally she found it difficult to apply the
(p. 430) CBT techniques at home, commenting I find it
hard to put anything into practice with others
These comments appeared to illustrate the around.
utility of the group environment for overcoming (p. 431)
depression.
Other considerations
Community Data were also reported on the partner sessions
Author comments from one of the studies which occurred only as part of one study.53 These
illustrated the value of the development of findings appeared to demonstrate the usefulness of
community:52 partners attending the group at some point during
the programme. Author comments illustrated that
These were thereduction in isolation not the relationship with partners may be an important
being alone anymore. aspect of depression/treatment:
(p. 430)
The sessions on relationships are often
A case study also demonstrated this value: emotionally arduous often sad, angry tones
accompany their attempt to understand
Additionally, the experiential process of being the meaning of their current emotional
a valued group member improved her sense of experiences. Relationships with partners are
self-worth, and increased her self confidence. often perceived by the women as strained.
(p. 914)
Practicalities and knowledge
The practical aspects of the group were also Author comments also confirmed the impression of
acknowledged as being important in the treatment the usefulness of the couples session:
process. A second study confirmed that the group
sessions were of practical use. Author comments, Some women report their partner is now more
supported by study data, revealed:52 supportive; some men now look after the infant
for specified times, releasing the women to
Every session in the programme was have time to herself. One father has expressed
acknowledged as being helpful by at least one the desire to have counselling for himself.
group member. Even sessions that I didnt Another couple said that they were now more
expect to be helpful were helpful in ways that I appreciative of each others efforts and said so
didnt expect so I was glad to come. to each other.
(p. 429)
The men also report that their session was
Adverse effects and limitations useful, both from understanding more about
Adverse effects and limitations of the group their partners mood state, and from hearing
treatment were also reported. Author comments how other men experienced similar difficulties.
supported by study data from one study
28
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
The study found that the group members Comparison of the positive and negative
who practised the CBT techniques positively aspects of the group treatment
appraised the approach and appeared to Group members also provided information
gain more benefit from the overall group on positive and negative aspects of the group
programmeopinions can change after treatment. These are reported in Table 7 for each
reflecting and applying the techniques of the studies. Comments generally related to the
discussed and practised during the sessions. practical aspects of the group, such as the format
(p. 432) of the sessions, although the supportive aspects of
the group were raised as important. The comments
There were, however, a number of common were in the large part positive; however, some of
factors which may be attributed to any type of the comments raised were negative and related to
psychological intervention rather than specifically both the format of the sessions and personal issues
to CBT. Participants reported feeling that every associated with sharing concerns in a group.
session had been helpful in ways they had not
TABLE 7 Summary of the positive and negative aspects of the CBT group treatment
TABLE 9 A comparison of components specific to CBT-based support groups and non-theoretically based support groups
Components specific to CBT-based support groups Component common to CBT and non-theoretically-
referenced by CBT group based support groups referenced by support group
Adverse effects; difficulties in application (of CBT Community; solace, trust and safety (Beck66 and Pitts67 );
approaches) (Davies and Jasper)52 reduction of isolation (Duskin65)
Finding CBT techniques helpful (Davies and Jasper)52 Social comparison and prototyping; normalising (Duskin,65
Beck66 and Pitts67 )
Practicalities and knowledge; knowledge (Eastwood68)
Environment; trust and safety (Eastwood68); support
(Eastwood68)
will be used or reported. Finally, it is not typical to The design of the Milgrom et al.59 paper was shown
report adverse events in studies of psychological to be of high quality in the main, although a few
interventions. problems were noted. The reporting of results was
unclear as scores were combined across a number
There were further difficulties specific to the of interventions meaning specific comparisons
studies reported here which may also constrain of group CBT with RPC were not possible. BDI
interpretations. These included difficulties of scores for those receiving individual treatment
ascertaining the use and effects of concurrent were significantly better than for those receiving
treatments, the reporting and effects of the time a group treatment, although it should be noted
postpartum of the participants, and the definitions that this was a combined score relating to various
of the CBT aspects of each treatment. These issues different forms of group treatment which included
are outlined for each of the included studies. A a CBT group. The paper gave an indication (in
summary of the clinical effectiveness findings is the presented figures) that individual treatment
provided in Table 10. was similarly effective to group CBT treatment;
however, the statistics were not available to
Environment
Social
comparison
facilitators Practicalities
group members and knowledge
Adverse effects/
positive effects
Community
Partners
session
Allows for
Facilitates
confirm this. Furthermore, there was no baseline conducted in the UK, therefore it is likely to best
comparison, and it was unclear whether compliance reflect current UK practice. When assessing the
with interventions was reliable as it was unclear CBT component of the treatment it appeared that
whether participants were receiving other forms of CBT was one of three core components of the
treatment, such as antidepressants, concurrently. treatment. The treatment was also predefined and
The intervention used in the Milgrom et al.59 clearly reported. All participants were <12 months
study was clearly defined as group CBT and was a postpartum, indicating a genuine PND group.
manualised treatment. Therefore, generalisations
may be made to the PND population. All The Meager and Milgrom study,62 again not
participants were <12 months postpartum, an RCT, appeared to have a good-quality
indicating a genuine PND group. Whilst the design, although other aspects were poor. The
method of randomisation was not ideal (one coded reporting of the results was very unclear, failing
slip of paper drawn from a bag containing multiple to report estimates of random variability, making
slips coded in equal number for each of the four interpretations very difficult. Therefore, although
treatment conditions), it was preferable to studies they report significant improvements on the EPDS
that did not attempt to randomise the participants. and BDI in the intervention group compared
to the control group it was difficult to make any
The Honey43 paper was shown to be high quality conclusions based on these findings. The authors
and showed that at the end of treatment the did report that there were no significant differences
EPDS score for those receiving group CBT had between the groups on medication usage, giving
been reduced by 3.48 points (95% CI 0.23 to an indication that the intervention group was not
6.73) more than those receiving RPC. At follow- confounded. The CBT aspect of the treatment was
up this difference was even more apparent, with ill-specified and was one of eight components of
the intervention group reporting an EPDS score the treatment; therefore it was difficult to make any
of 4.48 points lower (95% CI 1.01 to 7.95) than interpretations regarding the effects of group CBT.
those receiving RPC. However, both values had Further, the time postpartum was not reported,
wide CI ranges. Honey43 also reported that these making it difficult to ascertain whether the
differences were not related to antidepressant experimental group was a genuine PND group.
use. Honey43 was the only included study that was
32
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
The critical appraisal of Rojas et al.58 showed that However, these differences were not significant
it was generally a high-quality paper, although on the BDI. The treatment groups did not differ.
baseline characteristics were not compared Again, this study was not UK based; therefore
across groups and it did have further significant generalisations should be made with caution with
limitations relating to our objectives/analyses. The regard to UK practice. The definition and use of
compliance with the intervention may not have CBT in the treatment was ill-specified and it was
been reliable as participants in both the control impossible to ascertain whether CBT was a core
and intervention groups were taking concurrent component of the treatment. The time postpartum
medications which were not controlled for in the ranged from 1 to 24 months, therefore it was
analyses. Participants in the MCI group received unclear whether all participants could be defined
antidepressants as part of the intervention, and as having PND, and generalisations to PND should
this resulted in more participants in this group be made with caution.
taking antidepressants at both time points (59%;
36%) compared to those in the UC group (17%; The final non-RCT, Highet and Drummond,60 was
11%). This greater use of antidepressants made of low quality. All aspects of reporting were poor,
interpretations relating specifically to the group particularly the reporting of the results which was
CBT aspect of the intervention difficult. However, in part due to the retrospective, quasi-experimental
as most other studies do not report concurrent nature of the study. As such, group CBT was not
medication use and it is likely that in reality directly compared against the WLG, or against
women with PND will be offered antidepressants individual treatment. However, they did report
in addition to psychological interventions we comparisons of those treated in groups, either
concluded that this study provided important alone or with individual treatment in addition,
information on the effectiveness of the group CBT against those treated individually only, showing
aspect of the intervention. This limitation should that those receiving only individual treatment had
be kept in mind when making inferences regarding significantly better depression outcomes than those
group CBT. Rojas et al.58 demonstrated that at receiving group only or a combined treatment.
the end of treatment the EPDS score for those The design of the study also resulted in difficulties
receiving the MCI had been reduced by nearly five ascertaining compliance with interventions as most
points more than for those receiving UC. At follow- participants were receiving more than one type
up this difference was still apparent, although of treatment. They also did not report baseline
to a lesser degree, and this difference was not comparisons. No definitions were provided for
significant with the MCI group reporting an EPDS the group CBT intervention. It is likely that CBT
score of 2.2 points lower than those receiving UC, was a core component as it was termed group
which may be of questionable clinical significance. CBT. However, it was impossible to ascertain
The Rojas et al.58 study was based in Chile, this information because of poor reporting.
therefore this may have further implications for the Further, time postpartum was not reported for
interpretations of the findings given that current the participants, making generalisations to PND
practice may differ from that available in the UK. difficult.
The definition and use of CBT in the treatment was
ill-specified and it was unclear whether CBT was a Milgrom et al.,59 Meager and Milgrom,62 and
core component of the treatment. Furthermore, the Highet and Drummond60 were all Australian
number of participants taking part in each group studies constraining generalisations to UK practice.
session was 20, a much higher number than would
be expected for group CBT, and much higher The strongest evidence on which to base an
than used in the other studies presented here. assessment of clinical effectiveness was the data
All participants were <12 months postpartum, comparing group psycho-education incorporating
indicating a genuine PND group. CBT treatment with RPC, UC or WLG. However,
a number of caveats need to be put in place prior
The design of the Clark et al.61 study was found to to making any assertions. Honey43 and Rojas
be of high quality. The only significant problem et al.58 supported the idea that group psycho-
noted was with ascertaining compliance with education incorporating CBT is more effective
interventions as it was unclear whether participants than UC, although the interpretations by Rojas et
were receiving concurrent treatment. The findings al.58 may relate to concurrent group therapy and
demonstrated that both group and individual antidepressant use and the level of CBT in the
treatment resulted in lower BDI and CES-D intervention was very unclear. Honey43 seemed
scores post treatment than for the control group. more likely to reflect a group CBT treatment as
33
CBT was one of three pre-defined core components was some indication that individual treatment was
of the treatment. Meager and Milgrom62 also as effective as group CBT although the statistics
provided an indication that group CBT is more were not available to confirm this. The Honey43
effective than UC; however, because of the study was of high quality and demonstrated that
low quality of the reporting of the results, the those receiving group CBT had lower depression
uncertainty that the treatment accurately reflects scores than RPC, but wide CIs made interpretations
CBT and the fact that time postpartum was not difficult. However, as the only UK study it may be
reported, this interpretation should be treated most relevant. Meager and Milgrom62 reported
with caution. It should also be noted that Honey43 a high-quality design although the reporting of
was the only UK study, making the applicability of the results was poor. Significant improvements in
findings to practice in the UK particularly relevant. depression scores compared to the control group
The Clark et al. study61 reported that group were ascertained, but it was difficult to make firm
treatment was more effective than UC, although it interpretations because of poor reporting. The
did not differ from individual treatment. However, Rojas et al. study58 found lowered depression
because of difficulties in ascertaining levels of scores in the intervention group compared to
concurrent treatment and the wide range of time UC but the study had significant limitations.
postpartum, these findings should be treated with Clark et al.61 reported a high-quality study but
caution. there were problems ascertaining compliance
with interventions. Both group and individual
The Milgrom et al.59 study did not provide evidence treatment resulted in lower depression scores than
that group CBT was more effective than UC, for control, but interventions did not differ. Highet
although it was difficult to ascertain whether this and Drummond60 reported a low-quality study with
was the case because of reporting. No comparisons poor quality reporting. Therefore, although they
against UC were made for the treatments examined demonstrated that individual treatment resulted
in the Highet and Drummond study.60 in better outcomes than group or combined
treatments, any interpretations had to be treated
There was very little evidence to compare with caution.
group CBT with individual treatment, and any
interpretations should be treated with caution. Qualitative review
Clark et al.,61 a study of reasonable quality, and The two qualitative studies included in the
Milgrom et al.,59 which had a design of high quality, review52,53 differed in quality. Neither study could
showed that although intervention was more be classed as a qualitative research study, but both
effective than UC, individual counselling was not were evaluations that incorporated a qualitative
superior to group CBT. The findings reported approach. The Davies and Jasper study52
by Highet and Drummond,60 a poor-quality appeared fairly well conducted and reported, and
study, showed that individual treatment was more considerations of reflexivity and ethical issues were
effective than either group or combined individual dealt with, the data analysis showed some rigor
and group treatment. Therefore, overall it was not and findings were clearly stated. The Morgan et
appropriate to make firm assertions about group al.53 study failed to report the design and methods
CBT compared with individual interventions more of data collection clearly. They did not report that
generally or individual CBT more specifically. reflexivity and ethical issues had been addressed,
and the data analysis was not rigorous, although
Patient preferences, satisfaction and acceptability findings were clearly stated. A positive aspect was
were reported by only one study.60 There were that in the main the evidence reported was from
no preferences for CBT over medication, and no direct quotes or author interpretation supported by
preference for group CBT over group behaviour direct quotes.
therapy; however, individual treatment was
preferred to group treatment (combined score). The studies also differed in the extent to which
However, because of the poor reporting we could we could be confident that the group treatment
not be certain that this treatment did accurately included a CBT component. Davies and Jasper52
reflect group CBT, also the postpartum status of stated that the treatment was predefined and
the participants was not reported, therefore it was based on a CBT model, whereas Morgan et al.53
difficult to make interpretations with any certainty. stated that cognitive behavioural strategies were
employed as part of the group programme,
In summary, the Milgrom et al.59 study was of high and these were not reported as predefined or
quality although the reporting was unclear. There manualised and did not give an indication of the
34
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
level of CBT employed. Furthermore, neither PND. Participants found the caring and supportive
study restricted the sample to those who were environment served to facilitate the sharing of
<12 months postpartum. Included participants feelings, and the development of a community
were <18 months postpartum in the Davies and served to reduce isolation; the practical aspects
Jasper study52 and between 2 and 24 months in of the group were also acknowledged as being
the Morgan et al. study.53 This raised issues of important in the treatment process.52 It was also
whether the treatment groups in these studies were reported by some participants that it was difficult
representative of a PND population. to talk openly in the group setting, whilst others
found difficulty in applying CBT techniques.
There were also inherent limitations in the Participants also reported that partner sessions
studies described here and some of these were were helpful.
acknowledged by some authors. There seemed
to be a tendency for participants to give overly In comparing the CBT groups with support
optimistic views of an episode of care. This may groups, other than issues around the use of CBT
have influenced the reporting of results toward techniques there appeared to be little difference
the positive aspects of the group treatment. In in user perspectives. The findings suggested that
the main, data were only reported for those the effects of support groups may rely more heavily
participants who had been referred as suitable for on common factors rather than specific factors
the group, had attended and had found the group relating to the particular theoretical basis of the
a positive experience. Again this would influence techniques being applied.
findings in a positive direction. Extraneous factors
may have also had an impact on the findings. In summary, the Davies and Jasper study52 was
Cotherapy was not consistently reported across of reasonable quality and demonstrated that
the studies, making it difficult to disentangle the participants found the group treatment helped
effects of the psychological intervention and other them to develop better relationships with their
treatment the women may have been receiving. baby and facilitated social comparisons relating to
It was also difficult to assess whether the natural PND. Although participants in the Morgan et al.
remission of depression may have occurred study53 reported that cognitive components of the
during the study period. Authors may have under- course were particularly helpful as the study details
reported negative opinions as the objective of the were not clearly reported, interpretations had to be
studies was to identify the positive aspects of group treated with caution.
treatment. For these reasons it was important to
assess the interpretations made on the basis of
these studies with caution. Discussion
Women reported that the group environment Group psycho-education incorporating CBT
enabled them to develop better relationships with appeared to be clinically effective when compared
their baby.52 Women with PND used the groups to RPC, UC or WLG in three studies.43,58,62 The
to make social comparisons allowing them to reduction in depression scores was not consistent
understand that their feelings associated with PND across time: one58 demonstrated a significant
were normal, and the group allowed the women to reduction in depression scores at end of treatment
assess their roles as wife and mother.52 The findings but did not report this effect at follow-up, whilst
that related specifically to the CBT content of another43 did not find a significant reduction
the group programmes included reports that at end of treatment but did at follow-up. The
some women had difficulty in applying the CBT remaining three studies5961 could not demonstrate
techniques;52 however, others found the cognitive such reductions specific to group CBT. Further,
components of the course particularly helpful and interpretations should be made with caution as
were able to put them into practice.53 There were a number of the included studies may include
also common factors which may be attributed to concurrent therapy, the effects of which are difficult
any type of psychological intervention rather than to separate from the group treatment. There
specifically to CBT. It was reported that every is also uncertainty surrounding how accurately
session had been helpful in ways they had not the treatment reflects CBT in some studies, and
expected.52 uncertainty around whether generalisations
can be made to the PND population because of
There were also findings which may relate participants being at different times postpartum
generally to depression rather than specifically to in some studies. Furthermore, only one of the
35
included was conducted in the UK making the the sharing of feelings, the development of a
applicability of findings from the other studies community which served to reduce isolation, and
questionable.43 There is not enough evidence the practical aspects of the group which were also
to adequately compare group treatment with acknowledged as being important in the treatment
individual treatments or other comparators. process. Participants also reported that partner
sessions were helpful, and it was reported that men
The qualitative review showed that participants also benefited from group sessions, resulting in
had positive views of group treatment. However, increased practical help for women.
it is important to note that it is difficult to assess
how accurately the groups reflected group A further consideration related to both the
CBT and, further, whether the participants quantitative and qualitative studies surrounds
reflected a genuine PND population. Specific therapist variability. As previously noted, variability
benefits of CBT were reported, with participants in therapist effectiveness can account for variance
commenting that cognitive components were in treatment outcomes.69 As a relatively small
helpful. Findings specific to PND included number of therapists were involved in facilitating
comments that participants were able to develop the group interventions it is likely that the
better relationships with their baby, understand performance of the therapist could have had a
their feelings associated with PND and assess significant affect on the results.
their roles as wife and mother. Some negative
aspects were also reported although these were It is acknowledged that although all eight included
in the minority; these included difficulty in studies record that at least an element of CBT was
applying the CBT techniques and difficulty in employed in the group treatment, they may not
talking openly in the group setting. It is unclear fully represent CBT in its most widely recognised
whether CBT in particular or factors common form. However, owing to the high level of common
to any group activities are instrumental in the factors operating in psychological therapies, such
treatment. More general findings demonstrated as engendering hope, the conclusions we draw
that group members appreciated the caring and may be applicable to many psychological group
supportive environment which served to facilitate treatments.
36
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Chapter 4
Assessment of cost-effectiveness
To retrieve papers on cost-effectiveness the search Unfortunately this trial did not incorporate an
terms for clinical effectiveness were rerun on individual CBT arm; therefore this comparison was
MEDLINE, the Cumulative Index to Nursing and not explicitly modelled. Only the cost-effectiveness
Allied Health Literature (CINAHL) and EMBASE of group CBT compared with RPC was evaluated.
using an economics filter. The economics filters
used are provided in Appendix 1. The literature When modelling the cost-effectiveness of group
retrieved in the searches on the NHS Economic CBT it was assumed that benefit would be
Evaluation Database (NHS EED) and the Health accrued only on initiation of the treatment. Once
Economic Evaluations Database (HEED) was also treatment had commenced it was assumed that
reviewed. The searches were undertaken in January there would be a linear increase in the benefit of
2008. Databases were searched from 1950 to 2008 group CBT compared with RPC, peaking at the
with the actual date range for each of the databases end of treatment. Although the CIs were wide, data
searched depending on coverage of the individual from the Honey RCT43 provided relatively strong
database. The searches were not restricted by evidence that any gain would persist throughout a
language. 6-month follow-up period (Figures 4 and 5).
Pertinent economic literature was planned to We assumed that any gain would be maintained
be assessed using the Drummond and Jefferson over the 6-month period and would then be
checklist.70 However, no applicable publications followed by a linear decline in the advantage of
were found. group CBT compared with RPC that was assumed
to be reduced to zero 1 year after treatment.
That is, a linear decline over a 14-week period.
Independent economic The duration of this decline was chosen as the
assessment authors understand that after 12 months PND
would be reclassified as general depression. This
No existing models of the cost-effectiveness may be conservative as the focus of CBT is on
of group CBT for PND were identified in the developing skills that may provide longer benefits;
systematic review of the literature. As such, a de longer time periods are evaluated in sensitivity
novo economic model was constructed. analyses. A linear decline was chosen as it appeared
reasonable, other distributions may be applicable,
Methods however, given the large uncertainty in the model
parameters, particularly length of comparative
Given the scarcity of the data identified within the advantage associated with group CBT, it was
clinical effectiveness section, a pragmatic approach believed that fitting other distributions would be
was taken when constructing the mathematical introducing unnecessary complexity.
model, with the intention to provide indicative
estimations of likely cost-effectiveness ratios rather These assumptions would lead to a gain in EPDS
than a definitive answer. scores associated with group CBT compared with
RPC as depicted in Figure 6; the base-case values
A conceptual model was constructed which would have been used in this figure. Because of the small
investigate the benefits associated with group CBT time period of the model neither benefits nor costs
for PND. The design of the conceptual model were discounted.
was influenced by the data that were available to
37
4 2 0 2 4
Favours Favours
experimental control
Honey (2002)43 2.32 6.75 22 6.8 4.94 23 100.0% 4.48 (7.95 to 1.01)
4 2 0 2 4
Favours Favours
experimental control
4.5
Comparative advantage in EPDS
4.0
associated with group CBT
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0 4 8 12 16 20 24 28 32 36 40 44 48 52 56
Time since randomisation (weeks)
FIGURE 6 The conceptual model of the effects of group CBT on EPDS compared with RPC.
In the Figure 6, 8 weeks relates to the end of the The assumed effectiveness of group CBT
treatment period, 26 weeks the assumed time compared with RPC
at which maximum comparative advantage As previously noted, the CIs for the effects of group
declines and 52 weeks the period at which there CBT compared with RPC at end of treatment
was assumed to be no comparative advantage of and follow-up in the Honey RCT43 were large. In
group CBT compared with RPC. The duration order to reduce the uncertainty and to provide an
of comparative advantage was altered within estimation of the constant benefit assumed from
sensitivity analyses. the end of treatment to 6 months thereafter all
38
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
data points for Honey43 were pooled together to (i.e. became lower) the SF-6D score improved (i.e.
produce a single estimate. It is acknowledged that became higher). It was also noted that regardless of
this will remove any correlation between the two any change in EPDS score, the utility of a woman
time points, but this was deemed a worthwhile was 0.0625 higher at 6 months than at 6 weeks.
sacrifice. The assumed efficacy is depicted in This result was not surprising given that the EPDS
Figure 7. does not include a sleep component and it is likely
that women would be achieving more hours of
Thus it was expected that a woman who received sleep at 6 months than at 6 weeks.
group CBT would have at the end of treatment
and for the following 6-month period, on average, A plot of the residuals versus the fitted values
an EPDS score that was 3.98 lower than a similar is provided in Figure 9 and visually displays no
woman who received only RPC. The 95% CI marked bias within the fit.
ranged from a reduction of 4.69 to a reduction of
3.27. Tests for heteroskedasticity were conducted in
stata version 9 (StataCorp LP, College Station,
Mapping from changes in EPDS scores TX) using the BreuschPagan/CookWeisberg test.
to changes in utility This showed that the variance was not constant
In order that the cost-effectiveness ratios calculated (p=0.008) and therefore robust standard errors
can be compared with those estimated for were used when sampling from the regression
other interventions in other disease areas NICE equation.
recommends that QALYs be used as the metric
for health gain.71 A methodology was thus needed The stata output is provided in Table 11.
to translate between changes in EPDS scores and
changes in utility. This was achieved by using data In order that the correlation between the slope
from the PoNDER trial13 which had recorded both and constant of the regression was maintained
EPDS scores and utility scores [as measured using the variancecovariance matrix was identified.
the Short Form questionnaire-6 Dimensions (SF- Cholesky decomposition techniques73 were used,
6D)] for women following childbirth at 6 weeks and assuming that the coefficients for both the EPDS
6 months. The SF-6D is a preference-based scoring change and the constant were normally distributed,
system that provides a utility value for a patient.72 in order to preserve correlation. The variance
covariance matrix is provided in Table 12.
Data were taken for those woman (n=401),
regardless of arm in the RCT, who had an EPDS Sampling parameters for the slope and
score of 12 or greater at 6 weeks following constant of the regression equation and
childbirth and who had values for both EPDS and for the efficacy of group CBT
SF-6D at both 6 weeks and 6 months. The change In order to estimate the overall utility gain
in EPDS and SF-6D between 6 weeks and 6 months associated with group CBT, PSA were conducted.74
was recorded; these data are plotted in Figure 8. One thousand Monte Carlo estimations of the
distribution of the efficacy of group CBT were
A moderate relationship was observed (r2=0.27) sampled along with 1000 pairs of correlated
that indicated that as the EPDS score improved slope and constant coefficients describing the
4 2 0 2 4
Favours Favours
experimental control
0.6
y = 0.0113x + 0.0625
R2 = 0.2687
0.5
0.4
0.3
Change in SF-6D
0.2
0.1
0.0
30 25 20 15 10 5 0 5 10 15
0.1
0.2
Change in EPDS
0.4
0.2
Residuals
0.0
0.2
FIGURE 9 A plot of the residuals versus the fitted values from the change in SF-6D versus change in EPDS regression.
40
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
TABLE 11 stata output when regressing change in utility with change in EPDS score
Robust 95% CI
TABLE 12 The variancecovariance matrix associated with the regression of change in utility with change in EPDS score
linear relationship between change in EPDS and The estimated costs per woman
change in utility, thus forming 1000 parameter completing a group CBT course
configurations. The gain in utility for each woman The costs for two scenarios of delivering group
associated with each configuration was calculated CBT for PND were explored: one if the Honey
algebraically using the assumptions depicted in RCT43 regime was to be replicated; and the second
Figure 6. The range of the 1000 utility estimates is being the delivery methods deemed by the authors
provided in Figure 10. to be most likely were group CBT to become widely
available. The resources expected to be associated
The mean value of the utility gain was 0.032 with with each strategy are detailed in Table 13 and
a 95% CI, using a percentile method of 0.025 to Table 14, respectively. These values were relatively
0.041. similar, being 1317 and 1246 per woman.
0.20
Percentage of samples
0.10
0.00
0.020 0.025 0.030 0.035 0.040 0.045 0.050
Utility gain
FIGURE 10 The distribution of sampled utility gains per woman receiving group CBT.
41
TABLE 13 The resources required to duplicate the group CBT regimen used in Honey43
Source
A Number of weekly sessions 8 Honey43
B Length of sessions (hours) 2 Honey43
C Number of health visitors required 2 Honey43
D Preparation time required per session 2 Assumed 1 hour per session per health visitor
E Additional time required in excess of the 2 Assumed 1 hour per session per health visitor
session
F Average number of participants 5 Honey43
G Time for initial assessment per 2 Assumed 2 hours per participant
participant (hours)
H Health visitor time required 74 (ABC)+A(D+E)+FG
I Cost per hour of health worker time () 89 Morrell et al.13 cost per hour of client time including
training costs for psychological therapies (79 in
20034, this has been amended using inflation indices
to represent current prices)
J Total cost of health visitor 6586 HI
K Total cost per person () 1317 J/F
TABLE 14 The resources required using the delivery methods deemed by the authors of this report to be most likely were group CBT
to become widely available (see Chapter 1 for more detail)
Source
A Number of weekly sessions 12 Authors
B Length of sessions (hours) 2 Authors
C Number of facilitators required 2 Authors one health visitor and one newly qualified
clinical psychologist (same salary assumed)
D Preparation time required per session 2 Authors assumed 1 hour per session per facilitator
E Additional time required in excess of the 2 Assumed 1 hour per session per facilitator
session
F Average number of participants 8 Authors
G Time for initial assessment per 2 Assumed 2 hours per participant
participant (hours)
H Health visitor time required 112 (ABC)+A(D+E)+FG
I Cost per hour of health worker time () 89 Morrell et al.13 cost per hour of client time including
training costs for psychological therapies (79 in
20034, this has been amended using inflation indices
to represent 20078 prices)
J Total cost of facilitators 9968 HI
K Total cost per person () 1246 J/F
The costs presented in Tables 13 and 14 may be estimated that 1500 per woman completing a
underestimates as they did not include any set-up group CBT course would be approximately correct
costs or additional running costs, such as room regardless of the calculation method.
hire and crche facilities, which may have been
incurred. With the additional likelihood that Base-case results
women receive initial assessment (hence incurring
costs) but do not progress to group CBT and other The estimated cost per QALY result for the base
miscellaneous costs that would be incurred we case is provided in Table 15.
42
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
TABLE 15 The estimated cost per QALY of group CBT compared with RPC
Mean cost per Mean QALY gain per Mean cost per 95% cost per QALY using
woman () woman QALY () a percentile method
1500 0.032 46,462 37,008 to 60,728
These results are displayed in Figure 11 with a was particularly of importance when analysing the
CEAC.45 The mean cost per QALY results were length of time that group CBT would provide a
high when compared with recommended NICE benefit compared with RPC, as the duration used
thresholds of 20,000 and 30,000 per QALY,71 within the base case (of 1 year after initiating group
indicating that group CBT was unlikely to be cost- CBT) was particularly uncertain. If the benefits of
effective based on present assumptions. group CBT persisted for longer periods then the
cost per QALY estimated in the base case would be
Sensitivity analyses unfavourable to group CBT when compared with
There was uncertainty in the assumptions RPC.
regarding the modelled results that were explored
in univariate sensitivity analyses (Table 16). These A further sensitivity analysis was conducted
included altering the costs per woman of running where combinations of parameter values that
the service, changing the estimated utility gain per were plausible but favourable to group CBT were
woman and extending the length of time during selected. This gave a value below 20,000, a
which a woman would receive a utility benefit to a common threshold of cost-effectiveness used by
period of 18 months. A further sensitivity analysis NICE.71 This indicated that whilst the base case did
was also undertaken assuming arbitrarily that an not appear to be cost-effective there were plausible
additional 0.02 QALYs were gained as a crude scenarios that were cost-effective, and a definitive
exploratory analysis of estimation of potential answer could only be made once there was more
utility gains associated with the womans partner or certainty in the costs of conducting group CBT, in
by the baby. the efficacy of pure group CBT and in the duration
of residual benefit.
It was seen that each altered variable had the ability
to alter markedly the cost per QALY ratios. This
1.0
0.9
0.8
Probability of being cost-effective
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60 70 80
Cost per QALY threshold (000)
Mean cost per woman Mean QALY gain per Mean cost per QALY
Sensitivity analysis () woman ()
Base case 1500 0.032 46,462
Cost per woman decreased to 750 750 0.032 23,231
Cost per woman increased to 2000 2000 0.032 61,948
Lower 95% of efficacy assumed (EPDS 1500 0.027 56,626
decrease of 3.27)
Upper 95% of efficacy assumed (EPDS 1500 0.038 39,481
decrease of 4.69)
Linear decline in advantage extended to 1500 0.044 34,382
18 months
Additional QALY gain of 0.02 applied 1500 0.052 28,846
Cost per woman decreased to 1000, 1000 0.047 19,230
EPDS decrease of 4.3 assumed, linear
decline in advantage extended to
18 months
Exploration of the expected case, as the mean cost was 1418 compared with
value of information within the 1500 and the duration of comparative advantage
decision problem was 16 months compared with 12 months. The
As discussed previously there was considerable mean of the probabilistic values were not surprising
uncertainty within the decision problem. This given that it was commented that the comparative
uncertainty was initially explored using sensitivity advantage in the base case was likely to be
analyses assuming remaining parameters in the conservative, and that were group CBT to become
base case remained constant. Such analyses showed more widespread the cost per participant would be
that there were plausible scenarios where group likely to fall.
CBT would be deemed cost-effective compared
with RPC. The estimated cost per QALY result having fitted
distributions to data on comparative advantage
Further analyses were undertaken using formal and cost of group CBT per woman is provided in
expected value of perfect information (EVPI) Table 17 with a CEAC presented in Figure 12.
techniques, which indicate the most that a decision-
maker would pay to remove all uncertainty from It was seen that the cost per QALY value had
the decision problem.46,47 This analysis required fallen to 36,062, but this value still fell outside
that distributions were assigned to variables subject the recommended cost-effectiveness threshold.
to uncertainty. The uncertainty in the efficacy However, some scenarios fell below a value of
had previously been estimated; however, the 30,000 per QALY,71 which may be deemed a more
uncertainties in the costs of group CBT per woman appropriate threshold than 20,000 as only utility
and the duration of comparable advantage had gains relating to the woman (neither the partner
not addressed using scenario analyses rather than or baby) were considered, indicating that there was
a distribution. It was deemed that a triangular uncertainty in the correct decision.
distribution for costs ranging from 750 to 2000
with a mode of 1500 was not unreasonable The EVPI methodology evaluates in monetary
considering potential economies of scale and terms the cost of potentially making the wrong
also that a triangular distribution for duration of decision using a net benefit approach.75 Given
comparative advantage ranging from 1 to 2 years our chosen parameter distributions, the EVPI per
with a mode of 1 year was also not unreasonable. woman receiving group CBT was calculated to be
The authors recognised that these distributions 53.50.
were arbitrary but believed that the exploratory
results provided from this analysis would provide The number of births in 2003 in the UK was
an indication of the likely value of information. 695,500;76 assuming that 17.3% of women had
It was also commented that these values did not an EPDS score of 12 or over13 this equates to an
match identically those in the deterministic base estimated 120,000 women suffering from PND
44
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
TABLE 17 The estimated cost per QALY of group CBT compared with RPC having fitted statistical distributions to uncertain
parameters
Mean cost per Mean QALY gain per Mean cost per 95% cost per QALY using
woman () woman QALY () a percentile method
1418 0.039 36,062 20,464 to 76,293
per annum. It was assumed that CBT may be the benefit of removing all uncertainty from one of
most appropriate treatment for the forthcoming four variables: the assumed efficacy of group
10 years. If the birth rate and the prevalence of CBT in increasing EPDS values; the assumed cost
PND stay constant this would equal 1,200,000 per woman treated of group CBT; the assumed
women with an incident case of PND over the next duration of comparative advantage of group CBT;
10 years. For simplicity, we have assumed that each and the assumed gradient in the relationship
case of PND represents a new episode. Therefore, between EPDS values and the SF-6D. Figure 13
1,200,000 women were estimated to benefit from depicts the EVPI and the EVPPI of the four
increased knowledge regarding the efficacy, cost selected variables.
and duration of comparative advantage of group
CBT compared with RPC. Combining the number It was seen that variables with the biggest influence
of women who could benefit and the EVPI per on the cost-effectiveness of group CBT were
woman would mean that decision-makers would be the cost of treating a woman and the assumed
willing to pay a maximum of 64M to remove all relationship between EPDS values and the SF-
uncertainty in the decision problem. This amount 6D. By contrast, there was less to be gained by
appeared more than sufficient to adequately researching the increase in EPDS associated with
fund an RCT to assess the value of the uncertain group CBT and the length of comparative benefit.
parameters as well as to explicitly incorporate However, even the variable with the lowest EVPPI
individual CBT within the RCT. value per woman (the efficacy in terms of EPDS)
when multiplied by the number of women affected
Furthermore, the expected value of partial perfect would still equate to an estimated maximum cost of
information (EVPPI)48 was used to estimate the 500,000 to remove all uncertainty in this variable.
1.0
0.9
0.8
Probability of being cost-effective
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60 70 80 90
Cost per QALY threshold (000)
FIGURE 12 The CEAC for group CBT compared with RPC having fitted statistical distributions to uncertain parameters. 45
60
53.50
Value of perfect information per woman
50
40
treated ()
30 26.59
22.70
20
10
4.10
0.42
0
All variables Efficacy of Cost of Duration of Gradient of
group CBT in group CBT per comparative relationship
EPDS terms woman benefit of between
group CBT EPDS and SF-6D
FIGURE 13 The value of perfect information associated with parameters in the model.
by the lack of a sleep component in the SF-6D. woman, could not be analysed using PSA, and only
Furthermore, the exploratory analyses have selected combinations were tested. Furthermore,
indicated those parameters to which the cost- utility measurements were not recorded in the
effectiveness ratio is most sensitive, allowing future appropriate RCT, thus relying on a regression
research to be more targeted. of EPDS to SF-6D, which introduces further
uncertainty. The sensitivity analyses undertaken
The primary limitation of our analysis is caused by indicate that the cost per QALY can change
the uncertainty in key model inputs. Whilst some, markedly with plausible combinations of values,
such as the efficacy of group CBT could be tested which means that a definitive answer on whether
using PSA, others, such as the residual benefit group CBT is likely to be cost-effective cannot be
of group CBT or the costs of group CBT per provided given current data.
47
Chapter 5
Assessment of factors relevant to
the NHS and other parties
49
Chapter 6
Discussion
reported. Finally, the value of information analyses health visitors. User perspectives assessed in the
guide the development of future research agendas. qualitative review may have been biased toward
positive comments, although this was difficult to
The primary limitation of our analyses is caused ascertain because of the limited detail provided on
by the uncertainty in key model inputs. Whilst the methods incorporated.
some inputs, such as the efficacy of group CBT,
could be tested using PSA, others, such as the Although NICE guidelines for antenatal and
residual benefit of group CBT or the costs of postnatal care exist,5 these provide little detail
group CBT per woman, could not be analysed on the referral process and the content of
using PSA, and only selected combinations treatment programmes. Therefore, it was also
were tested. Furthermore, utility measurements difficult to ascertain whether group treatments
were not recorded in the appropriate RCT, thus and the comparators reflect current practice in
relying on a regression of EPDS to SF-6D which the UK. Only two of the studies43,52 assessed in
introduces further uncertainty. The sensitivity the review had a UK setting and both were pilot
analyses undertaken show that the cost per QALY investigations.
can change markedly with plausible combinations
of values, which means that a definitive answer on Impacts on the family and child have been
whether group CBT is likely to be cost-effective highlighted as important outcomes in the
cannot currently be provided. EVPPI analyses have treatment of PND. However, they could not be
indicated where further research should initially be assessed here because of limited available data.
focused.
Based on the evidence presented here it is unclear
whether drop-out and withdrawal rates have
Uncertainties implications for group interventions. Although
reasons for loss to follow-up are presented in some
Clinical effectiveness
cases, it is unclear whether patient acceptability of
There was little quantitative or qualitative RCT group treatment is a causal factor in the drop-out
evidence to assess the effectiveness of group CBT rates reported.
for PND. The evidence that was available was of
low quality in the main because of poor reporting It has been reported that variability in therapist
of the results. Furthermore, little information was effectiveness can account for variance in treatment
reported on concurrent treatment used in the outcomes, and is independent of both the
studies, which was controlled for in only two of the therapists professional background and patient
studies.43,62 factors at the start of treatment.69 Given the small
number of participants, and therefore the small
The evidence from the clinical effectiveness review number of therapists involved in facilitating the
provides inconsistent and low-quality information interventions reported here, it is possible that a
on which to base any interpretations for service particularly good or poor therapist could have
provision. Although three of the included markedly affected the results. As such there may
studies43,58,62 provide some indication that group be severe limitations in generalising the results
psycho-education incorporating CBT is effective observed in the RCTs to other settings.
compared with RPC, there is enough doubt in
the quality, the level of CBT implemented in Cost-effectiveness
the group programmes, and the applicability
to a PND population to significantly limit any The cost per QALY ratio for group CBT in PND
interpretations. Some studies lacked important is uncertain because of gaps in the evidence base.
detail of the intervention, making it difficult Research is urgently needed to populate key
to assess whether the treatment did genuinely parameters in the model including the effectiveness
reflect group CBT. Further, the time postpartum of group CBT compared with both RPC and
of the participants varied to a great extent individual CBT in terms of a utility measure rather
across the studies, making generalisations to a than EPDS, the costs of conducting CBT courses
PND population problematic. Furthermore, the and the duration of residual benefit associated
potentially small number of health visitors involved with CBT treatment. The cost-effectiveness ratio
in delivering the group CBT in the RCT assumed reported should be treated with caution until more
applicable to the UK setting may provide severe robust data become available.
limitations in generalising the results to other
52
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Chapter 7
Conclusions
53
Acknowledgements
55
References
1. Milgrom J, Martin PR, Negri LM. Treating postnatal approaches for depression in postnatal women:
depression. A psychological approach for health care pragmatic cluster randomised trial in primary care.
practitioners. England: Wiley; 1999. BMJ 2009;338:110.
2. Cox JL, Murray D, Chapman G. A controlled study 13. Morrell CJ, Warner R, Slade P, Dixon S, Walters S,
of the onset, duration and prevalence of postnatal Paley G, et al. Psychological interventions for
depression. Br J Psychiatry 1993;163:2731. postnatal depression: Cluster randomised trial and
economic evaluation. The PoNDER trial. Health
3. World Health Organization. Tenth revision of the Technol Assess 2009;13(30).
international classification of diseases and related health
problems. Clinical descriptions and diagnostic guidelines. 14. Gaynes BN, Gavin N, Meltzer-Brody S, Lohr KN,
Geneva: WHO; 1992. Swinson T, Gartlehner G, et al. Perinatal depression:
prevalence, screening accuracy, and screening
4. American Psychiatric Association. Diagnostic and outcomes. Evid Rep Technol Assess (Summ) 2005:18.
statistical manual of mental disorders, fourth edition. New
York, NY: American Psychiatric Association; 1994. 15. OHara MW, Swain AM. Rates and risk of
postpartum depression-a meta-analysis. Int Rev
5. National Institute for Health and Clinical Psychiatry 1996;8:3754.
Excellence. Antenatal and postnatal mental health:
clinical management and service guidance (Brief record). 16. Beck AT, Steer RA, Brown GK. Beck Depression
London: NICE; 2007. Inventory Manual. San Antonio, TX: Psychological
Corporation; 1996.
6. Jomeen J, Martin C. Confirmation of an occluded
anxiety component within the Edinburgh Postnatal 17. Wieck A, Kumar R, Hirst AD, Marks MN,
Depression Scale (EPDS) during early pregnancy. Campbell IC, Checkley SA. Increased sensitivity
J Reprod Infant Psychol 2005;23:14354. of dopamine receptors and recurrence of affective
psychosis after childbirth. BMJ 1991;303:61316.
7. National Institute for Health and Clinical
Excellence, Scottish Executive Health Department, 18. Beck CT. Postpartum depression: it isnt just the
Department of Health, and Social Services and blues. Am J Nurs 2006;106:4050.
Public Safety, Northern Ireland. Why mothers die
19971999. The fifth report of the confidential enquiries 19. Kendell RE, Chalmers JC, Platz C. Epidemiology of
into maternal deaths in the United Kingdom. London: puerperal psychoses. Br J Psychiatry 1987;150:
RCOG Press; 2001. 66273. [Erratum published in Br J Psychiatry
1987;151:135.]
8. Cooper P, Murray L. Course and recurrence of
postnatal depression: Evidence for the specificity 20. OHara M. The nature of postpartum depressive
of the diagnostic concept. Br J Psychiatry disorders. In Murray L, Cooper P, editors.
1995;166:1915. Postpartum depression and child development. New York,
NY: Guildford; 1997: 331.
9. Cooper PJ, Murray L. Postnatal depression. BMJ
1998;316:18846. 21. Beck C. A meta-analysis of predictors of postpartum
depression. Nurs Res 1996;45:297303.
10. Beck C. Predictors of postpartum depression: An
update. Nurs Res 2001;50:27585. 22. OHara M, Schlechte J, Lewis D, Wright E.
Prospective study of postpartum blues: Biologic
11. Robertson E, Grace S, Wallington T, Stewart D. and psychosocial factors. Arch Gen Psychiatry
Antenatal risk factors for postpartum depression: 1991;48:8016.
a synthesis of recent literature. Gen Hosp Psychiatry
2004;26:28995. 23. Cooper P, Murray L, Hooper R, West A. The
development and validation of a predictive
12. Morrell CJ, Slade P, Warner R, Paley G, Dixon S, index for postpartum depression. Psychol Med
Walters SJ, et al. Clincial effectiveness of health 1996;26:62734.
visitor training in psychologically informed
57
24. Murray L, Cooper P. The impact of postpartum 36. Dennis CL, Chung-Lee L. Postpartum depression
depression on child development. Int Rev Psychiatry helpseeking barriers and maternal treatment
1996;8:5563. preferences: a qualitative systematic review. Birth
2006;33:32331.
25. Murray L. The impact of postnatal depression
on infant development. J Child Psychol Psychiatry 37. Royal College of Psychiatrists. Perinatal Maternal
1992;33:54361. Mental Health Services. Council Report CR88. London:
Royal College of Psychiatrists; 2000.
26. Hay D, Pawlby S, Sharp D, Asten P, Mills A,
Kumar R. Intellectual problems shown by 38. Curtis L. Unit costs of health and social care.
11-year-old children whose mothers had Canterbury: University of Kent; 2007.
postnatal depression. J Child Psychol Psychiatry
2001;42:87189. 39. Department of Health. Treatment choice in
psychological therapies and counselling. Evidence based
27. Carter A, Garrity R, Chazan C, Little C, Briggs G. clinical practice guideline. London: DoH; 2001.
Maternal depression and comorbidity: Predicting
early parenting, attachment security, and toddler 40. National Institute for Health and Clinical
social-emotional problems and competencies. J Am Excellence. Depression (amended). Management of
Acad Child Adolesc Psychiatry 2001;40:1826. Depression in primary and secondary care. NICE Clinical
Guideline 23 (amended). London: NICE; 2007.
28. Campbell SB, Cohn JF, Meyers T. Depression in
first-time mothers motherinfant interaction and 41. Cuijpers P, van Straten A, Andersson G, van
depression chronicity. Dev Psychol 1995;31:34957. Oppen P. Psychotherapy for depression in adults:
A meta-analysis of comparative outcome studies.
29. Escribe-Aguir V, Gonzalez-Galarzo MC, J Consult Clin Psychol 2008;76:90922.
Barona-Vilar C, Artazcoz L. Factors related to
depression during pregnancy: are there gender 42. Dennis CL, Hodnett E. Psychosocial and
differences? J Epidemiol Community Health psychological interventions for treating
2008;62:41014. postpartum depression. Cochrane Database Syst Rev
2007;CD006116.
30. Areias MEG, Kumar R, Barros H, Figueiredo E.
Correlates of postnatal depression in mothers and 43. Honey KLB. A brief psycho-educational group
fathers. Br J Psychiatry 1996;169:3641. intervention for postnatal depression. Br J Clin
Psychol 2002;41:4059.
31. Healthcare Commission towards better births. A
review of maternity services in England. Service Review. 44. Ades AE, Cliffe S. Markov Chain Monte Carlo
London: Commission for Healthcare Audit and estimation of a multi-parameter decision model:
Inspection; 2008. consistency of evidence and the accurate assessment
of uncertainty. Med Decis Making 2002;22:35971.
32. National Institute for Health and Clinical
Excellence. Antenatal and postnatal mental 45. Fenwick E, Claxton K, Sculpher M. Representing
health: Clinical management and service guidance. uncertainty: the role of cost-effectiveness
2006. URL: www.nice.org.uk/nicemedia/pdf/ acceptability curves. Health Econ 2001;10:77987.
APMHConsFullGuideline.pdf (accessed 1 April
2009). 46. Claxton K, Posnett J. An economic approach to
clinical trial design and research priority-setting.
33. Appleby L, Warner R, Whitton AF, Faragher B. Health Econ 1996;5:51324.
A controlled study of fluxetine and cognitive-
behavioral counselling in the treatment of postnatal 47. Ades AE, Lu G, Claxton K. Expected values of
depression. BMJ 1997;314:9326. sample information calculation in medical decision
making. Med Decis Making 2004;24:20727.
34. National Institute for Health and Clinical
Excellence. Commissioning antenatal and postnatal 48. Felli JC, Hazen JB. Sensitivity analysis and the
mental health services. 2007. URL: www.nice.org. expected value of perfect information. Med Decis
uk/usingguidance/commissioningguides/apmh.jsp Making 1998;18:95109.
(accessed 1 April 2009).
49. Cox JL, Holden JM, Sagovsky R. Detection of
35. Family & Parenting Institute. Health visitors: A postnatal depression: Development of the 10-
progress report. 2009. URL: www.familyandparenting. item Edinburgh Postnatal Depression Scale.
org/Filestore//Documents/Our_work/Campaigns/ Br J Psychiatry 1987;150:7826.
HealthVisitors_a_progress_report.pdf (accessed 1
April 2009).
58
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
50. Beck AT, Steer RA. Manual for the Beck Depression 63. Wampold BE, Brown GS. Estimating variability in
Inventory. San Antonio, TX: Psychological outcomes attributable to therapists: A naturalistic
Corporation; 1993. study of outcomes in managed care. J Consult Clin
Psychol 2005;73:91423.
51. Centre for Reviews and Dissemination. Systematic
reviews: CRDs guidance for undertaking reviews in 64. Popay J, Rogers A, Williams G. Rationale and
health care. University of York, York: CRD; 2009. standards for the systematic review of qualitative
literature in health services research. Qual Health
52. Davies S, Jasper M. A first-stage evaluation of Res 1998;8:34151.
a group programme for PND. Community Pract
2004;77:42633. 65. Duskin C. Experiences of social comparison in a
postpartum depression support group. Dissertation
53. Morgan M, Matthey S, Barnett B, Richardson C. A Abstracts International: Section B: The Sciences and
group programme for postnatally distressed women Engineering 2006;67:539.
and their partners. J Adv Nurs 1997;26:91320.
66. Beck C. Teetering on the edge: A substantive theory
54. Deeks JJ, Dinnes J, DAmico R, Sowden AJ, of postpartum depression. Nurs Res 1993;42:428.
Sakarovitch C, Song F, et al. Evaluating non-
randomised intervention studies. Health Technol 67. Pitts F. Professional. Health visiting: the Monday
Assess 2003;7(27). group: postnatal depression revisited. Community
Pract 1999;72:3279.
55. Downs SH, Black N. The feasibility of creating a
checklist for the assessment of the methodolgical 68. Eastwood P. Group work 1. Promoting peer group
quality both of randomised and non-randomised support with postnatally depressed women. Health
studies of health care interventions. J Epidemiol Visit 1995;68:14850.
Community Health 1998;52:37784.
69. Najavits LM, Weiss RD. Variations in therapist
56. National CASP collaboration for qualitative effectiveness in the treatment of patients with
methodologies. Critical Appraisal Skills Programme substance use disorders: An empirical review.
(CASP) making sense of evidence. 10 questions to help Addiction 1994;89:67988.
you make sense of qualitative research. England: Public
Health Resource Unit; 2006. 70. Drummond MF, Jefferson TO. Guidelines for
authors and peer reviewers of economic submissions
57. Sandelowski M, Barroso J. Handbook for synthesizing to the BMJ. BMJ 1996;313:27583.
qualitative research. New York, NY: Springer
Publishing Company, Inc.; 2007. 71. National Institute for Health and Clinical
Excellence. Guide to the methods of technology
58. Rojas G, Fritsch R, Solis J, Jadresic E, Castillo C, appraisals. 2008. URL: www.nice.org.uk/media/
Gonzlez M, et al. Treatment of postnatal depression B52/A7/TAMethodsGuideUpdatedJune2008.pdf
in low-income mothers in primary-care clinics in (accessed 1 April 2009).
Santiago, Chile: a randomised controlled trial.
Lancet 2007;370:162937. 72. Brazier JE, Roberts J, Deverill M. The estimation of
a preference-based measure of health from the SF-
59. Milgrom J, Negri LM, Gemmill AW, 36. J Health Econ 2002;21:27192.
McNeil M, Martin PR. A randomized controlled
trial of psychological interventions for postnatal 73. Briggs A, Claxton K, Sculpher M. Decision modelling
depression. Br J Clin Psychol 2005;44:52942. for health economic evaluation. Oxford: Oxford
University Press; 2006.
60. Highet N, Drummond P. A comparative
evaluation of community treatments for post- 74. Claxton K, Sculpher M, McCabe C, Briggs A,
partum depression: implications for treatment Akehurst R, Buxton M. Probabilistic sensitivity
and management practices. Aust N Z J Psychiatry analysis for NICE technology assessment: not an
2004;38:21218. optional extra. Health Econ 2005;14:33947.
61. Clark R, Tluczek A, Wenzel A. Psychotherapy for 75. Stinnett A, Mulahy J. Net health benefits: a
postpartum depression: a preliminary report. new framework for the analysis of uncertainty
Am J Orthopsychiatry 2003;73:44154. in cost-effectiveness analyses. Med Decis Making
1998;18:S68S80.
62. Meager I, Milgrom J. Group treatment for
postpartum depression: A pilot study. Aust N Z 76. Office for National Statistics. Births and deaths.
J Psychiatry 1996;30:85260. London: Office for National Statistics; 2004.
59
77. Schafer JL. Analysis of incomplete multivariate data. 79. Tanner BA. A multidimensional client satisfaction
London: Chapman and Hall; 1997. instrument. Eval Program Plann 1982;5:1617.
60
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Appendix 1
Literature search strategies
A list of the electronic Copies of the search strategies
bibliographic databases used in the major databases
searched
Search strategy used on MEDLINE
1. MEDLINE and MEDLINE In-Process & Other
2. MEDLINE In-Process & Other Non-Indexed Non-Indexed Citations
Citations 1. Depression, Postpartum/
3. CINAHL 2. post-partum depression.tw.
4. Cochrane Database of Systematic Reviews 3. post partum depression.tw.
(CDSR) 4. postpartum depression.tw.
5. Cochrane Central Register of Controlled Trials 5. depression, post partum.tw.
(CENTRAL) 6. depression, post-partum.tw.
6. EMBASE 7. depression, postpartum.tw.
7. Database of Abstracts of Reviews of Effects 8. post-natal depression.tw.
(DARE) 9. post natal depression.tw.
8. NHS EED 10. postnatal depression.tw.
9. NHS Health Technology Assessment (HTA) 11. depression, post natal.tw.
10. PsycINFO 12. depression, post-natal.tw.
11. Science Citation Indexes 13. depression, postnatal.tw.
12. Social Sciences Citation Indexes 14. post pregnancy depression.tw.
13. Applied Social Sciences Index and Abstracts 15. postpregnancy depression.tw.
(ASSIA) 16. post-pregnancy depression.tw.
14. BIOSIS 17. or/1-16
15. British Nursing Index 18. Depression/
16. Social Care Online 19. depress$.tw.
17. Office of Health Economics Economic 20. 18 or 19
Evaluations database. 21. Postpartum Period/
22. post-partum.tw.
A list of additional sources 23. post partum.tw.
24. postpartum.tw.
1. National Research Register (NRR) 25. postnatal$.tw.
2. Research Findings Register (ReFeR) 26. post natal$.tw.
3. Current Controlled Trials and its links 27. post-natal$.tw.
4. Health Services Research Projects in Progress 28. postpregnancy.tw.
(HSRProj) and index to theses 29. post pregnancy.tw.
5. health service research and guideline 30. child birth.tw.
producing bodies (e.g. Scottish Intercollegiate 31. childbirth.tw.
Guidelines Network, NICE, National 32. labor and delivery.tw.
Guidelines Clearinghouse, etc.) have been 33. labour and delivery.tw.
consulted via the internet and other key 34. puerperal.tw.
organisations (e.g. Association for Postnatal 35. or/21-34
Illness, Postnatal illness-Support & Help 36. 20 and 35
Association) have been contacted 37. 17 or 36
6. grey literature has been identified from 38. antenatal depression.tw.
searches of databases including dissertation 39. ante-natal depression.tw.
abstracts. 40. ante natal depression.tw.
41. or/38-40
61
Appendix 2
Data abstraction tables quantitative review
Outcome
Study Funding Methods Participants Interventions measures
Milgrom et al. National Health Design: RCT; Sample size: Intervention Depression: BDI
(2005)59 & Medical three intervention 192 (group group: group- Anxiety: BAI
Research council, arms and one CBT=46), (group based CBT, group-
Austin Hospital control arm counselling=47), based counselling, Social support:
Medical Research (individual individual SPS
Tool of
Foundation identification: counselling=66), counselling
EPDS, DSM-IV (RPC=33) Control group:
minor or major Diagnosed condition: RPC
depression depression
Method of diagnosis:
DSM-IV
Honey Wales Office of Design: RCT with Sample size: 45 (23 Intervention Depression:
(2002)43 Research and a treatment arm controlled PEG), (22 group: controlled EPDS
Development for and a control arm RPC) PEG Social support:
Health and Social Tool of Diagnosed condition: Control group: Duke UNC;
Care identification: PND RPC DAS; WCC-R
EPDS Method of diagnosis:
EPDS
Rojas et al. Fondo de Ciencia Design: RCT; one Sample size: 230 Intervention Depression:
(2007)58 y Tecononlogia intervention arm (MCI=114, UC=116) group: MCI EPDS; SF-36
(FONDECYT- and one control Diagnosed condition: Control group: UC
Chile) Grant arm major depression
Tool of Method of diagnosis:
identification: DSM-IV
EPDS
BAI, Beck Anxiety Inventory; DAS, Dyadic Marital Adjustment Scale; Duke UNC, Duke UNC Social Support
questionnaire; PEG, psycho-educational group; SF-36, Short Form questionnaire-36 items; SPS, Social Provisions Scale;
WCC-R, Ways of Coping Checklist-Revised.
65
Outcome
Study Funding Methods Participants Interventions measures
Highet and NR Design: Sample size: Intervention group: Depression: EPDS
Drummond community-based 146=136 eight different Physiological and
(2004)60 study; between treatment group, treatment conditions, psychological
groups for 10 WLG participants may be anxiety: State Trait
treatment vs wait Diagnosed included in one or Anxiety Inventory;
list; within groups condition: PND more groups but this GHQ
across treatments is not clearly stated
Method of which participants are Social support:
Tool of diagnosis: Social Support
identification: in which groups; CBT,
not detailed CBT and medication, Scales
pretreatment considered
questionnaire, medication only,
by health-care group CBT only,
EPDS provider to have group and individual
PND CBT, individual CBT
only, group cognitive
and behaviour
therapy, group
behaviour therapy
only
Control group: WLG
Meager and NR Design: between Sample size: 20 Intervention group: Depression: EPDS;
Milgrom groups, two (group=10), group treatment BDI
(1996)62 groups (WLG=10) (including CBT) Self-esteem:
Tool of Diagnosed Control group: WLG Coopersmith Self-
identification: condition: PND esteem inventory
EPDS, BDI Method of Mood: Profile of
diagnosis: EPDS, Mood States
BDI Social support: SPS
Parenting: PSI
Relationship
adjustment: DAS
Clark et al. Perinatal Design: between Sample size: Intervention group: Depression: BDI;
(2003)61 Foundation, groups; three 39=13 MITG, MITG CES-D
Madison, WI, and groups 15 IPT, 11 WLG Individual therapy Stress: PSI
the Research and Tool of Diagnosed group: IPT
Development Child: BSID;
identification: condition: major Control group: WLG PCERA
Fund, Department DSM-IV, BDI depression
of Psychiatry,
University of Method of
Wisconsin Medical diagnosis: DSM-
School IV
BSID, Bayley Scales of Infant Development; CES-D, The Center for Epidemiological Studies Depression Scale; DAS, Dyadic
Marital Adjustment Scale; GHQ, General Health Questionnaire; NR, not reported; PCERA, The ParentChild Early
Relational Assessment; PSI, Parenting Stress Index; SPS, Social Provisions Scale.
66
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Co-therapy or
Study Description of treatment medication Comparator Sample size
Milgrom et al. Group-based CBT designed to NR RPC: the routine care 192: 52 of those
(2005)59 address specific target behaviours provided via the states allocated to
within the context of general universal Maternal and a treatment
components recognised as Child Health Service condition did
important in determining the not attend; 121
success of cognitive behavioural completed post-
intervention. Each session intervention
involved psycho-education, measures
review of homework exercises,
role playing and discussion
Group-based counselling
designed for depression
Individual counselling
Honey PEG; educational information on Antidepressant use, RPC: further details not 45 (four dropped
(2002)43 PND; strategies for coping; use of details not given provided out of PEG group
cognitive-behavioural techniques; but were followed
relaxation up)
Rojas et al. MCI included PEGs and Participants were UC: all services 230: 101
(2007)58 structured pharmacotherapy if excluded if receiving normally available in participants in
needed or had received the clinics, including MCI completed
treatment for antidepressant dugs, assessment at
depression during brief psychotherapeutic 3 months and
current postnatal interventions, medical 106 completed
period, but were consultation or external assessment at
offered medication referral for speciality 6 months; 108
as part of the treatment (although participants in UC
intervention and psychotherapy and group completed
control groups speciality treatments rarely assessment at
numbers given in offered) 3 months and
results section 102 completed
assessment at 6
months
67
Co-therapy or
Study Description of treatment medication Comparator Sample size
Highet and Varied by GP/health visitor. Not Various: see WLG: participants 188 participants
Drummond detailed. Community sample description of who had to wait initially involved in
(2004)60 treatment at least 3 weeks the study, 42 were
to receive group excluded from the
intervention final sample leaving
146 participants
Meager and Group treatment programme Participants could WLG: had the 20: four
Milgrom consisting of targets which take receive any other opportunity to participants
(1996)62 into consideration the risk factors treatments at any participate in dropped out of
for postpartum depression. time. Eight of the 20 the treatment each group leaving
An environment of social and participants were on programme once 12 participants (six
emotional support, an educational medication but it is the participants per group). These
component, a cognitive behavioural not stated how many in the treatment participants did not
component, encouragement of of these eight were group had complete follow-up
networking, examination of patterns in the experimental completed the measures
of communication, normalising of group. Post hoc programme
feelings, involvement of spouse in the analyses revealed no
group, practical homework significant differences
between the groups on
medication usage
Clark et al. MITG mothers therapeutic NR WLG those 39: four
(2003)61 intervention and peer support waiting to receive participants in the
group and infant development group MITG MITG were lost
occurred simultaneously, followed by to follow-up
motherinfant dyadic group. Based
on interpersonal, psychodynamic,
family systems, and cognitive
behavioural approaches
IPT group individual therapy,
relating to partners, children and
others
68
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Number Length
of Number of Professional background of
Study Recruitment sessions in group sessions therapist
Milgrom Recruitment was via a community Nine, 510 90 One of two senior therapists
et al. screening programme conducted weekly minutes delivered the interventions,
(2005)59 at 47 maternal and child health supported by cotherapists with
centres in northern metropolitan professional registrations and
Melbourne and rural eastern VIC, backgrounds in clinical psychology,
Australia postgraduate psychology research
and nursing with postgraduate
qualifications in counselling and/
or psychology. All received one-
to-one instruction in use of the
therapy manuals and regular,
intensive supervision from the
principal investigator
Honey Women were referred by Eight, Four to six 2 hours Health visitors
(2002)43 their health visitor if they were weekly per group
attending mother and baby clinics
in Gwent, scoring above 12 on
the EPDS
Rojas Recruited mothers at any stage Eight, Maximum 50 Midwives or nurses with 8 hours
et al. during first postnatal year from weekly 20 minutes of training and supervision every
(2007)58 three clinics in Santiago, Chile. week. A medical doctor was
Approached whilst waiting for responsible for the group
health-related consultations.
Screened using EPDS, those
scoring 10 or above were asked
to return for another assessment
2 weeks later. Those still scoring
10 or above were invited to
a baseline clinical assessment
(DSM-IV)
69
Reasons
Length of Numbers lost for loss to
Study Study site follow-up to follow-up follow-up Inclusion criteria Exclusion criteria
Milgrom Northern 12 weeks, 52 did not NR DSM-IV diagnosis of Depression affecting
et al. metropolitan and 12 attend; 121 depression; 37- to competence to give
(2005)59 Melbourne months after completed 42-week pregnancy; informed consent
and rural treatment postintervention infant birth weight (e.g. psychotic
eastern VIC, began measures 2.5kg and above; depression); risk
Australia no congenital requiring crisis
abnormality; no management;
major health participation in
problem; no other psychological
concurrent major programmes and
psychiatric disorder significant difficulty
with English
Honey Gwent, Wales, 8 weeks Three in each NR Attending mother Exhibiting psychotic
(2002)43 UK (after PEG condition and baby clinics. symptoms
finished) and (equals six) at >12 on EPDS.
6 months time three (6 Most recent child
after first months) <12 months
assessment
Rojas Santiago, Chile Baseline, At three months NR Mothers within their Women who had
et al. 3 months, 21 (13 from MCI, first postnatal year. received any form
(2007)58 6 months 8 from UC), at Meeting criteria for of treatment for
6 months 22 (8 major depression on depression during
from MCI, 14 DSM-IV their current
from UC) postnatal period;
those who were
pregnant; or those
with psychotic
symptoms, serious
suicidal risk, history
of mania, or alcohol
or drug abuse
70
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
71
Methods for
Diagnosed diagnosis of Education/socioeconomic Baseline
Study condition disorder Age Ethnicity background Patient history comparability
Milgrom et al. Condition: DSM-IV Mean=29.7 NR Family income AUS$41,400 Number of children [mean NR
(2005)59 depression years; SD=5.4 (SD=20,500). 62.7% with 12 (SD)]: 1.8 (0.8)
Time after years or more years of school. 30.5% Comorbidity: NR, but
diagnosis: NR with higher education exclusion criteria no
concurrent major psychiatric
disorder
Honey (2002)43 Condition: EPDS PEG: NR NR Number of children: PEG Groups did not
PND not mean=29.3 50% primiparous; RPC 59% differ significantly
confirmed years; primiparous on socio-
by diagnostic SD=5.36 years Comorbidity: NR demographic and
interview RPC: time 1 self-report
Time after mean=26.48 measures
diagnosis: NR years;
SD=5.68 years
Rojas et al. (2007)58 Condition: DSM-IV MCI: NR Number of years in education: Number of children [mean NR
depression mean=26.7 MCI, 0<8=20 (18%), (SD)]: MCI=2 (1), UC=2 (1.2)
Time after years; SD=6.4 812=82 (73%), >12=11 Comorbidity: NR, but
diagnosis: NR years (10%). UC, 0<8=17(15%), exclusion criteria no
UC: 812=87 (75%), >12=12 (10%) concurrent major psychiatric
mean=26.6 Main occupation: housewife, disorder
years; SD=7.4 MCI=94 (83%), UC=105
years (91%); student, MCI=3 (3%),
UC=4 (3%); employed,
MCI=16(14%), UC=5 (4%);
unemployed, MCI=1 (1%),
UC=2 (2%)
73
Appendix 2
Measurement
Study Outcomes Instruments periods ITT analysis
Milgrom Depression BDI Baseline, after Yes. Analyses were executed twice: once
et al. Anxiety BAI 12 weeks using only observed cases (121/192 possible
(2005)59 intervention, and cases), and once using multiple imputation
Social support SPS after 12 months under multivariate normal assumptions using
methods given by Schafer,77,78 employing
available demographic and psychometric data.
Conducted analyses to test the assumption
that missing data were missing at random (SAS
and winbugs)
Honey Sociodemographic NR Baseline Yes. Data missing for three participants in each
(2002)43 questionnaire condition at time 3. Missing data replaced by
Depression EPDS Baseline, after the group mean of each measure
BAI, Beck Anxiety Inventory; DAS, Dyadic Marital Adjustment Scale; Duke UNC, Duke UNC Social Support
questionnaire; NR, not reported; SAS, Statistical Analysis System; SF-36, Short Form questionnaire-36 items; SPS, Social
Provisions Scale; WCC-R, Ways of Coping Checklist-Revised.
74
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
BSID, Bayley Scales of Infant Development; CES-D, The Center for Epidemiological Studies Depression Scale; DAS, Dyadic
Marital Adjustment Scale; GHQ, General Health Questionnaire; PCERA, The ParentChild Early Relational Assessment;
PSI, Parenting Stress Index; SPS, Social Provisions Scale.
75
Milgrom Outcome Measure Baseline C1 (RPC vs C2 (CBT vs Changes in depression and anxiety immediately post intervention
et al. other three counselling) significantly differed between psychological intervention vs routine
(2005)59 interventions care. No evidence that CBT and counselling led to different outcomes
combined) in terms of depressive symptoms
Depression BDI (observed) Difference in 0.580 6.94 0.065 Intervention based on a counselling approach may be more effective
BDI scores when delivered on an individual basis. Percentages of women in
each treatment condition whose post-intervention BDI scores fell
SE 0.108 2.29 1.86 below the threshold for clinical depression (17) were: CBT 55%,
df 82 34 34 group counselling 64%, individual counselling 59% and RPC 29%.
Social support measure post intervention levels of perceived social
p-value <0.0001 0.005 0.97 support increased in each intervention group but fell for the RPC
BDI Difference in 0.510 4.06 0.75 group (although there appear to be no statistics)
(imputations) BDI Scores
SE 0.11 2.08 1.83
df 290.2 556.8 673.7
p-value <0.001 0.05 0.68
Honey Outcome Measure PEG RPC Statistical analysis For the EPDS there was a main effect of group F(1,36)=7.62,
(2002)43 (n=23) (n=22) p=0.01; main effect of time F(2,43)=12.06, p<0.001; and a significant
Mean (SD) interaction F(2,43)=3.16, p<0.05
Depression EPDS (time 19.35 (4.39) 17.95 (3.95) Main effect of group Simple effects; effect of time for the PEG F(2,86)=13.76, p<0.001; not
1 pre F(1,36)=7.62, p=0.01; main for RPC; marginally significant effect of group at time 3 F(1,107)=3.68,
intervention) effect of time F(2,43)=12.06, p=0.058
p<0.001; and a significant At time 3 there was a significant association between group and
EPDS (time 14.87 (5.97) 16.95 (5.44)
interaction F(2,43)=3.16, percentage scoring below cut-off [X 2 (1, N=45)=3.75, p<0.05],
2 post
p<0.05 significantly more women scored below cut-off in the PEG but not in
intervention 8
Simple effects; effect of time the RPC group. No differences at time 2. No differences between the
weeks)
for the PEG F(2,86)=13.76, groups on the social support measures
EPDS (time 3 12.55 (4.62) 15.63 (7.28) The brief PEG significantly reduced EPDS scores compared with
p<0.001; not for RPC;
6 months after RPC. This was not related to antidepressant use and was maintained
marginally significant effect of
intervention) 6 months after the group had ended. However, some women in the
group at time 3 F(1,107)=3.68,
p=0.058 PEG continued to show evidence of depressive symptomatology
6 months later. Improvements in mood were not accompanied
% women 35 27 NS
by changes in coping, perceptions of social support or the marital
scoring below
relationship
cut-off (time 2)
% women 65 36 2 (1, N=45)=3.75, p<0.05
scoring below
cut-off (time 3)
DOI: 10.3310/hta14440
ANOVA, analysis of variance; df, degrees of freedom; NS, not significant; PEG, psycho-educational group; SE, standard error; SF-36, Short Form questionnaire-36 items.
Health Technology Assessment 2010; Vol. 14: No. 44
77
78
Appendix 2
TABLE 31 Results of reported outcomes (psychological symptoms and interpersonal and social functioning) for non-RCTs
Pre treatment Post treatment Follow-up Receiving treatment resulted in significant decreases in depression
Treatment Clinical between pre and post treatment [grouptime interaction
Study condition scale n Mean SD n Mean SD n Mean SD F(1,137)=11.89, p<0.05]
Highet and Medication EPDS 15 19.27 4.38 15 14.47 6.80 Medication was no more effective than CBT. Participants treated with
Drummond only CBT (alone or in combination with medication) had greater decreases
(2004)60 in depression [grouptime interaction F(1,82)=11.08, p<0.05] and
psychological anxiety [grouptime interaction F(1,79)=5.98, p<0.05]
Group CBT EPDS 23 15.39 4.39 22 9.32 3.67 12 8.15 4.63 following treatment than those who received medication alone
only Comparison of subjects treated in groups (alone and in conjunction with
individual treatment) vs those treated individually revealed a significant
grouptime interaction [F(1,83)=16.98, p<0.05]. Depression was
significantly lower at post treatment in subjects treated individually
as opposed to those who received group or combined intervention
[t(84)=3.9, p<0.05]. At follow-up there was also a significant decrease
in depression [main effect for time, F(1,63)=11.36, p<0.05], particularly
in those treated in both group and individual settings [grouptime
interaction, F(1,63)=5.95, p<0.05]. Depression continued to decline
for those who had been treated in the combined setting [t(34)=5.26,
p<0.05], while there was no change for those treated in groups only
While CBT was no more effective than behavioural-based supportive
counselling, confounding effects of greater medication use and
greater treatment duration for those in the latter group may result in
underestimation of the efficacy and efficiency of CBT for this sample
Study Week
DOI: 10.3310/hta14440
Meager and Outcome Measure Group 0 (control n=10) 10 (control n=6) Least Actual difference
Milgrom (treatment n=10) (treatment n=6) significant
(1996)62 (mean) (mean) difference
Depression BDI Control 29.00 29.14
Treatment 29.70 16.80 10.27 12.90
(p<0.05)a
11.26 12.34
(p<0.10)b
EPDS Control 27.50 28.00
Treatment 24.80 15.80 7.92 9.0
(p<0.02)a
10.74 12.2
79
Appendix 2
80
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Appendix 3
Data abstraction tables qualitative review
81
82
TABLE 36 Study characteristics CBT-based group treatment
Total sample
DOI: 10.3310/hta14440
83
84
Appendix 3
Number
of Number Length of Therapist Professional background of
Study Recruitment sessions in group sessions contact therapist
Morgan et Women were Eight, Average of 2 hours NR Groups led by a female
al. (1997)53 referred from weekly six occupational therapist, with the
mothercraft units assistance of either a registered or
or from family enrolled female nurse, the couples
care cottages and session was led by these and a
community health male clinical psychologist
Davies Women were Twelve, Eight 90 minutes NR Health visitors with a registered
and Jasper referred by health weekly mental health nursing qualification
(2004)52 visitors and a family centre worker
facilitated the life skills group. A
primary mental health worker
provided the group leaders with
clinical supervision
Number Professional
of Number in Length of Therapist background of
Study Recruitment sessions group sessions contact therapist
Duskin Participants were NR NR NR NR Graduate students
(2006)65 recruited to take part in on a clinical
interviews from those psychology course
who already attended
the postpartum
depression support
group
Beck Participants were those Twice Number of Open- NR Nurse
(1993)66 attending the support monthly attendees ranged ended
group from 1 to 12
Pitts Women were referred NR NR NR NR Health visitor
(1999)67 to the group by health
visitors
Eastwood Women were referred 12 13 (eight NR NR Health visitors
(1995)68 to the group by health completed the led the group,
visitors course, only supervision was
four attended all provided by a clinical
sessions) psychologist
85
TABLE 40 Study site, follow-up and inclusion/exclusion criteria CBT-based group treatment
TABLE 41 Study site, follow-up and inclusion/exclusion criteria non-theoretically-based group treatment
Numbers Reasons
lost to for loss to Inclusion Exclusion
Study Study site Length of follow-up follow-up follow-up criteria criteria
Duskin CA, USA No follow-up NA NA Those taking NR
(2005)65 part in the
support group
Beck FL, USA Data collected during No follow-up NA Those taking NR
(1993)66 sessions, and during part in the
interviews conducted in support group
participants home
Pitts Southampton, Survey data collected 14 women did NR Those taking NR
(1999)67 UK only once during a not return the part in the
2-year period after survey support group
intervention
Eastwood Bexley, UK End of course and at a Five by end of NR An EPDS score Those
(1995)68 10-week recall course, seven of 13 or above suffering from
at 10-week psychotic
recall depression
were excluded
86
TABLE 42 Patient characteristics CBT-based group treatment
ESB, English speaking background; NESB, non-English speaking background; NR, not reported; SD, standard deviation.
87
Appendix 3
GHQ, General Health Questionnaire; NA, not applicable; NR, not reported.
HADS, Hospital Anxiety and Depression Scale; NA, not applicable; NR, not reported.
TABLE 46 Results of reported outcomes (psychological symptoms and interpersonal and social functioning) non-theoretically-based
group treatment
Study Results
Pitts EPDS of the 34 replies, 28 women had reduced scores, four had increased scores and two were
(1999)67 unchanged. 23 women scored below the cut-off of 12, and 11 above it
88
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Appendix 4
Summary of excluded trials
quantitative review
TABLE 47 Summary of reasons for excluding studies from the quantitative review
89
90
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
TABLE 48 Studies excluded from the quantitative review with rationale (continued)
continued
91
TABLE 48 Studies excluded from the quantitative review with rationale (continued)
92
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
TABLE 48 Studies excluded from the quantitative review with rationale (continued)
A further 17 were excluded at full paper sift, on the summary of the reasons for exclusion are shown in
basis of inclusion and exclusion criteria regarding Table 49. The name of the first author, year, journal
the CBT component of the intervention being and reason for exclusion are reported in Table 50.
investigated, or included only qualitative data. A
TABLE 49 Summary of reasons for excluding studies from the quantitative review because of CBT component
93
TABLE 50 Studies excluded from the quantitative review because of CBT component with rationale
94
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Appendix 5
Summary of excluded trials qualitative review
TABLE 51 Summary of reasons for excluding studies from the qualitative review
95
96
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
TABLE 52 Studies excluded from the qualitative review with rationale (continued)
continued
97
TABLE 52 Studies excluded from the qualitative review with rationale (continued)
98
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Appendix 6
References for excluded studies
References for Table 48 Boath E, Henshaw C. The treatment of postnatal
Counseling to prevent postnatal emotional problems. depression: A comprehensive literature review. J Reprod
Nurses Drug Alert 2007;31:40. Infant Psychol 2001;19:21548.
Groups dont help postnatal blues. Aust Nurs J 1996;3:16. Boath E, Major K, Cox J. When the cradle falls II: The
cost-effectiveness of treating postnatal depression in a
Psychological intervention for postpartum depression. psychiatric day hospital compared with routine primary
Nurses Drug Alert 2006;30:39. care. J Affect Disord 2003;74:15966.
Abramov L, Zohar AH. A controlled community-based Brugha T, Sharp HM, Cooper SA, Weisender C, Britto D,
family study of major depression with postpartum onset. Shinkwin R, et al. The Leicester 500 Project. Social
Am J Med Genet 1998;81:4556. support systems and the development of postnatal
depressive symptoms, a prospective cohort survey. Psychol
Ammerman R, Bodley A, Putnam F, Lopez W, Holleb L, Med 1998;28:6379.
Stevens J, et al. In-home cognitive behavior therapy for
a depressed mother in a home visitation program. Clin Brugha TSW. Pragmatic randomized trial of antenatal
Case Stud 2007;6:16180. intervention to prevent post-natal depression by reducing
psychosocial risk factors. Psychol Med 2000;30:127381.
Appleby L, Warner R, Whitton AF, Faragher B. A
controlled study of fluxetine and cognitive-behavioral Buist A, Westley D, Hill C. Antenatal prevention of
counselling in the treatment of postnatal depression. postnatal depression. Arch Womens Ment Health 1999;1:
BMJ 1997;314:9326. 16773.
Appleby L. Fluoxetine vs counselling for postnatal Buist A, Speelman C, Hayes B, Reay R, Milgrom J,
depression. N Z Med J 1997;110:221. Meyer D, et al. Impact of education on women with
perinatal depression. J Psychosom Obstet Gynaecol 2007;28:
Appleby L, Hirst E, Marshall S, Keeling F, Brind J, 4954.
Butterworth T, et al. The treatment of postnatal
depression by health visitors: impact of brief training on Camdeviren HA, Yazici AC, Akkus Z, Bugdayci R,
skills and clinical practice. J Affect Disord 2003;77:2616. Sungur MA. Comparison of logistic regression model
and classification tree: An application to postpartum
Austin MP, Frilingos M, Lumley J, Hadzi P, depression data. Expert Syst Appl 2007;32:98794.
Roncolato W, Acland S, et al. Brief antenatal cognitive
behaviour therapy group intervention for the prevention Carroll J, Reid A, Biringer A, Midmer D, Glazier R,
of postnatal depression and anxiety: A randomised Wilson L, et al. Effectiveness of the Antenatal Psychosocial
controlled trial. J Affect Disord 2008;105:3544. Health Assessment (ALPHA) form in detecting
psychosocial concerns: A randomized controlled trial.
Ayers S, Kenzie-McHarg K, Eagle A. Cognitive behaviour Can Med Assoc J 2005;173:2539.
therapy for postnatal post-traumatic stress disorder: case
studies. J Psychosom Obstet Gynaecol 2007;28:17784. Casiano ME. Outpatient medical management of
postpartum psychiatric disorders. NAACOGS Clin Issu
Berchtold N, Burrough M. Reaching out: depression Perinat Womens Health Nurs 1990;1:395401.
after delivery support group network. NAACOGS Clin Issu
Perinat Womens Health Nurs 1990;1:38594. Chung P, Yue X. Postpartum depression and social
support: A comparative study in Hong Kong. Psychologia
Bledsoe S, Grote N. Treating Depression During 1999;42:11121.
Pregnancy and the Postpartum: A Preliminary Meta-
Analysis. Res Soc Work Pract 2006;16:10920. Chun L-Q, Bo X, Wen J-X. A controlled clinical trial of
citalopram and citalopram combined with psychotherapy
Boath E, Cox J, Lewis M, Jones P, Pryce A. When the in the treatment of postpartum depression. Chinese
cradle falls: The treatment of postnatal depression in a Mental Health Journal 2005;19:8478.
psychiatric day hospital compared with routine primary
care. J Affect Disord 1999;53:14351. Cooper P, Murray L. The impact of psychological treatments
of postpartum depression on maternal mood and infant
99
development. New York, NY: Guilford Press; 1997. pp. Field T, Grizzle N, Scafidi F, Schanberg S. Massage and
20120. relaxation therapies effects on depressed adolescent
mothers. Adolescence 1996;31:90311.
Cooper PJ, Murray L, Wilson A, Romaniuk H. Controlled
trial of the short- and long-term effect of psychological Flynn H, Mahen H, Massey L, Marcus S. The impact
treatment of post-partum depression. I. Impact on of a brief obstetrics clinic-based intervention on
maternal mood. Br J Psychiatry 2003;182:41219. treatment use for perinatal depression. J Womens Health
2006;15:1195204.
Corral M, Wardrop AA, Zhang H, Grewal AK, Patton S.
Morning light therapy for postpartum depression. Arch Fones C. A postpartum support group. Birth 1984;11:
Womens Ment Health 2007;10:2214. 244.
Creedy D. Postnatal depression: improving the Free ML, Oei TPS, Sanders MR. Treatment outcome
experience of country women through professional and of a group cognitive therapy program for depression.
community awareness. Aust J Rural Health 1993;1:439. Int J Group Psychother 1991;41:53347.
Cuijpers P, Brannmark JG, Van SA. Psychological Gjerdingen D, Katon W, Rich DE. Stepped care
treatment of postpartum depression: a meta-analysis. treatment of postpartum depression a primary care-
J Clin Psychol 2008;64:10318. based management model. Womens Health Issues
2008;18:4452.
Currie J, Boxer E, Devlin E. Pramwalking as postnatal
exercise and support: an evaluation of the Stroll Your Grote N, Bledsoe S, Swartz H, Frank E. Feasibility
Way To Well-Being program and supporting resources in of providing culturally relevant, brief interpersonal
terms of individual participation rates and community psychotherapy for antenatal depression in an obstetrics
group formation. Aust J Midwifery 2001;14:215. clinic: A pilot study. Res Soc Work Pract 2004;14:397407.
[Erratum published in Aust J Midwifery 2001;14:5.]
Grote N, Bledsoe S, Swartz H, Frank E. Culturally
Dennis CL. The effect of peer support on postpartum relevant psychotherapy for perinatal depression in low-
depression: a pilot randomized controlled trial. income ob/gyn patients. Clin Soc Work J 2004;32:32747.
Can J Psychiatry 2003;48:11524.
Gruen D. A group psychotherapy approach to
Dennis CL. Preventing postpartum depression part postpartum depression. Int J Group Psychother 1993;43:
I: a review of biological interventions. Can J Psychiatry 191203.
2004;49:46775.
Gutteridge K. Postnatal depression -- an integrative
Dennis C. Counselling and cognitive behavioural therapy psychotherapeutic group approach. MIDIRS Midwifery
reduce anxiety and depression in women with postnatal Digest 2002;12:858.
depression. Evid Based Ment Health 2006;9:50.
Hagan RE. Preventing postnatal depression in mothers
Elliott S, Sanjack M, Leverton T. Parents groups in of very preterm infants: A randomised controlled trial.
pregnancy: A preventive intervention for postnatal BJOG 2004;111:6417.
depression? In Gottlieb B, editor. Marshalling social
support: formats, processes and effects. London: Sage; 1988. Halonen J, Passman R. Relaxation training and
pp. 86110. expectation in the treatment of postpartum distress.
J Consult Clin Psychol 1985;53:83945.
Elliott S, Leverton T, Sanjack M, Turner H,
Cowmeadow P, Hopkins J, et al. Promoting mental health Hayes B, Muller R, Bradley B. Perinatal depression: A
after childbirth: A controlled trial of primary prevention randomized controlled trial of an antenatal education
of postnatal depression. Br J Clin Psychol 2000;39:22341. intervention for primiparas. Birth 2001;28:2835.
Escobar GJ, Braveman PA, Ackerson L, Odouli R, Hayes B, Muller R. Prenatal depression: A randomized
Coleman-Phox K, Capra AM, et al. A randomized controlled trial in the emotional health of primiparous
comparison of home visits and hospital-based group women. Res Theory Nurs Pract 2004;18:16583.
follow-up visits after early postpartum discharge.
Pediatrics 2001;108:71927. Heh SS, Fu YY. Effectiveness of informational support in
reducing the severity of postnatal depression in Taiwan.
Fairchild M. Women with postpartum psychiatric illness: J Adv Nurs 2003;42:306.
A professionally facilitated support group. Soc Work
Groups 1995;18:4153. Holden JMS. Counselling in a general practice setting:
Controlled study of heath visitor intervention in
treatment of postnatal depression. BMJ 1989;298:2236.
100
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Honikman J. Role of self-help techniques for postpartum mood Maley B. Focusing on you: professional & personal
disorders. Washington, DC, US: American Psychiatric direction. Out of the blue: creating a postpartum
Association; 1999. pp. 195215. depression support group. AWHONN Lifelines
2002;6:625.
Horowitz JA, Cousins A. Postpartum depression
treatment rates for at-risk women. Nurs Res 2006; Markou P. Psychotherapeutic support group for the
55(Suppl. 2):S23S27. partners of women with post natal depression. Aust NZ
J Psychiatry 1999;33:A19.
Hynd S, Khan S. Identity and the experience of postnatal
depression: The use of Video Interaction Guidance. Matthey S. Calculating clinically significant
J Psychiatr Ment Health Nurs 2004;11:73841. change in postnatal depression studies using the
Edinburgh Postnatal Depression Scale. J Affect Disord
Johnston D, Tough S, Siever J. The Community Perinatal 2004;78:26972.
Care Study: home visiting and nursing support for
pregnant women. Zero Three 2006;27:1117. McClendon JM. Reexamining paroxetine and cognitive-
behavioral therapy in postpartum depression and
Jung V, Short R, Letourneau N, Andrews D. anxiety. J Clin Psychiatry 2005;66:14945.
Interventions with depressed mothers and their infants:
Modifying interactive behaviours. J Affect Disord Milgrom J, Negri L, Meager I. Group treatment for post-
2007;98:199205. partum depression. Int J Psychol 1996;31:1203.
Kersting A, Kuschel S, Reutemann M, Ohrmann P, Milgrom J, Martin PR, Negri LM. A randomized trial
Arolt V. Outpatient psychotherapy for mothers--first of psychological treatments for postnatal depression in
empirical results. Psychiatry 2003;66:33545. Australia objective. J Psychosom Res 2003;55:11920.
Klier CM, Muzik M, Rosenblum K, Lenz G. Milgrom J, Negri L, McNeil M, Martin P. A randomized
Interpersonal psychotherapy adapted for the group trial of psychological treatments for postnatal depression.
setting in the treatment of postpartum depression: A Int J Psychol 2004;39:509.
pilot study. Infant Ment Health J 2000;21:381.
Miller P, Chilvers C, Dewey M, Fielding K, Gretton V,
Kopelman R, Stuart S. Psychological treatments for Palmer B, et al. Counseling versus antidepressant therapy
postpartum depression. Psychiatr Ann 2005;35:55666. for the treatment of mild to moderate depression in
primary care: economic analysis. Int J Technol Assess
Lane B, Roufeil L, Williams S, Tweedie R. Its just Health Care 2003;19:8090.
different in the country: Postnatal depression and
group therapy in a rural setting. Soc Work Health Care Misri S, Kostaras X, Fox D, Kostaras D. The impact
2001;34:33348. of partner support in the treatment of postpartum
depression. Can J Psychiatry 2000;45:5548.
Lau Y. Establishing an interpersonal group therapy
program for Chinese women with postnatal depression Misri S, Reebye P, Corral M, Milis L. The use of
in Hong Kong. Hong Kong Nurs J 2005;41:717. paroxetine and cognitive-behavioral therapy in
postpartum depression and anxiety: a randomized
Lee D. Partner support reduced depressive symptoms controlled trial. J Clin Psychiatry 2004;65:123641.
in postpartum depression. Evid Based Ment Health
2001;4:51. Misri S, Reebye P, Milis L, Shah S. The Impact
of treatment intervention on parenting stress in
Lembke A. A psychosocial approach to postpartum postpartum depressed mothers: A prospective study.
depression. Psychiatr Times 2002;19:(6). Am J Orthopsychiatry 2006;76:11519.
Lockhart K. Karitane mothers post partum support Morrell CJ, Spiby H, Stewart P, Walters S, Morgan A.
group. Lamp 1988;45:1112. Costs and effectiveness of community postnatal support
workers: randomised controlled trial. BMJ 2000;321:
Loendersloot EWH. Management of post-partum 5938.
depression (PPD) in a special PPD clinic. J Psychosom
Obstet Gynaecol 1983;2:536. Morris JB. Group psychotherapy for prolonged postnatal
depression. Br J Med Psychol 1987;60:27981.
Magalhaes PV, Pinheiro RT, Faria AD, Osorio CM, da
Silva RA, Botella L. Impact of defense style on brief Murray L, Cooper PJ, Wilson A, Romaniuk H.
psychotherapy of postpartum depression. J Nerv Ment Dis Controlled trial of the short- and long-term effect of
2007;195:8703. psychological treatment of post-partum depression:
2. Impact on the motherchild relationship and child
outcome. Br J Psychiatry 2003;182:4207.
101
OBrien LM, Boath E, Hodgson R, Cox JL. The long- Spinelli M, Endicott J. Controlled clinical trial of
term follow-up of women treated for postnatal depression interpersonal psychotherapy versus parenting education
at a specialist day hospital compared to routine primary program for depressed pregnant women. Am J Psychiatry
care. Int J Psychiatry Clin Pract 2002;6:199203. 2003;160:55562.
OHara MW, Rehm LP, Campbell SB. Predicting Stamp GE, Williams AS, Crowther CA. Evaluation of
depressive symptomatology: cognitive-behavioral antenatal and postnatal support to overcome postnatal
models and postpartum depression. J Abnorm Psychol depression: a randomised controlled trial. Birth 1995;
1982;91:45761. 22:13843.
OHara MW. Interpersonal psychotherapy for Steinberg SI, Bellavance F. Characteristics and treatment
postpartum depression. Womens Health Issues 1995; of women with antenatal and postpartum depression.
5:756. Int J Psychiatry Med 1999;29:20933.
OHara MWS. Efficacy of interpersonal psychotherapy Stuart S, Ohara MW. Treatment of postpartum
for postpartum depression. Arch Gen Psychiatry 2000; depression with interpersonal psychotherapy. Arch Gen
57:103945. Psychiatry 1995;52:756.
Olson MR, Cutler LA, Legault F. Bittersweet: a Stuart S, OHara M. Interpersonal psychotherapy
postpartum depression support group. Can J Public for postpartum depression: A treatment program.
Health 1991;82:1356. J Psychother Pract Res 1995;4:1829.
Pedrina F, Scmid T. Group therapy with mothers and Stuart S, OHara M. The efficacy of psychotherapy for
babies in postpartum crises: preliminary evaluation of a perinatal depression. Econ Neurosci 2001;3:4651.
pilot project. Group Analysis 2004;37:13751.
Tam WHL. A randomised controlled trial of educational
Prendergast J, Austin M. Early childhood nurse-delivered counselling on the management of women who have
cognitive behavioural counselling for post-natal suffered suboptimal outcomes in pregnancy. BJOG
depression. Australas Psychiatry 2001;9:2559. 2003;110:8539.
Reay R, Robertson M, Owen C. Interpersonal Tam WH, Lee D-T, Chiu H-F, Ma KC, Lee A,
psychotherapy for postnatal depression: A quality Chung T-K, et al. Educational counselling did not
improvement approach. Australas Psychiatry 2002; improve psychological wellbeing in women with
10:21113. suboptimal childbirth outcomes. Evid Base Obstet Gynecol
2004;6:1734.
Rees BL. Effect of relaxation with guided imagery on
anxiety, depression, and self-esteem in primiparas. Tezel A, Gzm S. Comparison of effects of nursing
J Holist Nurs 1995;13:25567. care to problem solving training on levels of depressive
symptoms in post partum women. Patient Educ Couns
Reid M, Glazener C, Connery L, Mackenzie J, Ismail D, 2006;63:6473.
Prigg A, et al. Two interventions for postnatal support.
Br J Midwifery 2003;11:2948. Thoppil J, Riutcel TL, Nalesnik SW. Early intervention
for perinatal depression. Am J Obstet Gynecol 2005;192:
Reid MG. A two-centred pragmatic randomised 14468.
controlled trial of two interventions of postnatal support.
BJOG 2002;109:116470. Ugarriza D. Group therapy and its barriers for women
suffering from postpartum depression. Arch Psychiatr Nurs
Ryding E, Wirn E, Johansson G, Ceder B, Dahlstrm A. 2004;18:3948.
Group counseling for mothers after emergency cesarean
section: A randomized controlled trial of intervention. Ugarriza DN, Schmidt L. Telecare for women with
Birth 2004;31:24753. postpartum depression. J Psychosoc Nurs Ment Health Serv
2006;44:3745.
Saltzberg M. A years experiences with a postpartum
depression group. Group 2003;27:219. Webster J, Linnane J, Roberts J, Starrenburg S, Hinson J,
Dibley L. IDentify, Educate and Alert (IDEA) trial: an
Spinelli M. Interpersonal psychotherapy for depressed intervention to reduce postnatal depression. BJOG
antepartum women: A pilot study. Am J Psychiatry 1997; 2003;110:8426.
154:102830.
Wickberg B, Hwang CP. Counselling of postnatal
Spinelli M. Interpersonal psychotherapy for antepartum depression: a controlled study on a population based
depressed women. New York, NY, US: Oxford University Swedish sample. J Affect Disord 1996;39:20916.
Press; 2001. pp. 10521.
102
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Wiggins MO. Postnatal support for mothers living in behavior in new mothers. J Child Psychol Psychiatry
disadvantaged inner city areas: A randomised controlled 1992;33:68598.
trial. J Epidemiol Community Health 2005;59:28895.
Harner KT. Effectiveness of professionally led postpartum
Wilkinson J, Phillips S, Jackson J, Walker K. Mad for support groups among depressed postpartum women. Fresno,
fitness: an exercise group to combat a high incidence of CA: Alliant International University; 2004.
postnatal depression. J Fam Health Care 2003;13:448.
Klier C, Muzik M, Rosenblum K, Lenz G. Interpersonal
Wheatley SL, Brugha TS, Shapiro DA, Berryman psychotherapy adapted for the group setting in the
JC. PATA PATA: midwives experiences of facilitating treatment of postpartum depression. J Psychother Pract
a psychological intervention to identify and treat mild Res 2001;10:12431.
to moderate antenatal and postnatal depression
Positive Attitude Towards Antenatal and Postnatal Kurzweil S. Relational-developmental therapy group
Adjustment, Training for All. MIDIRS Midwifery Digest for postnatal depression. Int J Group Psychother
2003;13:52330. 2008;58:1734.
Fleming AS, Klein E, Corter C. The effects of a social Amankwaa L. Postpartum depression among
support group on depression, maternal attitudes and African-American women. Issues Ment Health Nurs
2003;24:297316. 103
Andajani-Sutjahjo, Manderson L, Astbury J. Complex Campbell S, Cohn J. The timing and chronicity
emotions, complex problems: Understanding the of postpartum depression: Implications for infant
experiences of perinatal depression among new mothers development. In: Murray L, Cooper P, editors. Postpartum
in urban Indonesia. Culture 2007;31:10122. depression and child development. New York, NY: Guildford;
1997. pp. 16597.
Arborelius EU, Bremberg SG. Supportive and
nonsupportive qualities of child health nurses contacts Chan SW, Levy V, Chung TK, Lee D. A qualitative study
with strained infant mothers. Scand J Caring Sci of the experiences of a group of Hong Kong Chinese
2003;17:16975. women diagnosed with postnatal depression. J Adv Nurs
2002;39:5719.
Bgedahl-Strindlund. Parapartum mental illness:
An interview follow-up study. Acta Psychiatr Scand Chen CH, Wu HY, Tseng YF, Chou FH, Wang SY.
1997;95:38995. Psychosocial aspects of Taiwanese postpartum depression
phenomenological approach: a preliminary report.
Baker S, Choi P, Henshaw C, Tree J. I Felt as though Kaohsiung J Med Sci 1999;15:4451.
Id been in Jail: Womens experiences of maternity care
during labour, delivery and the immediate postpartum. Chen C, Wang S, Chung U, Tseng Y, Chou F. Being
Fem Psychol 2005;15:31542. reborn: the recovery process of postpartum depression in
Taiwanese women. Jan J Adv Nurs 2006;54:4506.
Beck C. The lived experience of postpartum depression:
A phenomenological study. Nurs Res 1992;41:16670. Clemmens DA. Adolescent mothers depression after the
birth of their babies: weathering the storm. Adolescence
Beck CT. Perceptions of nurses caring by mothers 2002;37:55165.
experiencing postpartum depression. J Obstet Gynecol
Neonatal Nurs 1995;24:81925. Creedy D. Postnatal depression and post traumatic stress
disorder: what are the links? Birth Issues 1999;8:12530.
Beck CT. Postpartum onset of panic disorder. Image J
Nurs Sch 1998;30:1315. Cubison J, Munro J. Acceptability of Using the EPDS as a
Screening Tool for Postnatal Depression. Psychology:
Bennett D, Slade P. Infants born at risk: Consequences Professional & Research Language. London: Jessica
for maternal post-partum adjustment. Br J Med Psychol Kingsley Publishers; 2005. pp. 152161.
1991;64:15972.
Edge D, Baker D, Rogers A. Perinatal depression among
Benoit C, Westfall R, Treloar A, Phillips R, Jansson SM. black Caribbean women. Health Soc Care Community
Social factors linked to postpartum depression: A 2004;12:4308.
mixed-methods longitudinal study. J Ment Health
2007;16:71030. Edge D. Perinatal depression: its absence among Black
Caribbean women. Br J Midwifery 2006;14:64650, 652.
Benvenuti P, Valoriani V, Degl I, Favini I, Hipwell A,
Pazzagli A. Postnatal depression and the impact on Edwards E, Timmons S. A qualitative study of stigma
infant-carer attachment strategies: A case report. Arch among women suffering postnatal illness. J Ment Health
Womens Ment Health 2001;3:15564. 2005;14:47181.
Berggren C. Out of the darkness and into the light: Engqvist I, Nilsson A, Nilsson K, Sjostrom B. Strategies
Womens experiences with depression after childbirth. in caring for women with postpartum psychosis--an
Can J Commun Ment Health 1998;17:10320. interview study with psychiatric nurses. J Clin Nurs
2007;16:133342.
Brown W, Shereshefsky P. Seven women: A prospective
study of postpartum psychiatric disorders. Psychiatry Everingham CR, Heading G, Connor L. Couples
1972;35:13959. experiences of postnatal depression: a framing analysis
of cultural identity, gender and communication. Soc Sci
Buchwald J, Unterman R. Precursors and predictors Med 2006;62:174556.
of postpartum depression: A retrospective study. J Prev
Psychiatry 1982;1:293308. Field T, Hernandez R, Feijo L. Breastfeeding in
depressed mother-infant dyads. Early Child Dev Care
Buultjens M, Liamputtong P. When giving life starts 2002;172:53945.
to take the life out of you: womens experiences of
depression after childbirth. Midwifery 2007;23:7791. Fisher J, Astbury J, Smith A. Adverse psychological
impact of operative obstetric interventions: a prospective
Campbell SB, Cohn JF, Meyers T. Depression in first- longitudinal study. Aust N Z J Psychiatry 1997;31:72838.
time mothers mother-infant interaction and depression
chronicity. Dev Psychol 1995;31:34957.
104
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
Fisher JR, Morrow MM, Ngoc NT, Anh LT. Prevalence, Abstracts International: Section B: The Sciences and
nature, severity and correlates of postpartum depressive Engineering 2001;62:2064.
symptoms in Vietnam. BJOG 2004;111:135360.
Lawler D, Sinclair M. Grieving for my former self: a
Fooladi MM. Therapeutic tears and postpartum blues. phenomenological hermeneutical study of womens
Holist Nurs Pract 2006;20:20411. lived experience of postnatal depression. Evidence Based
Midwifery 2003;1:3641.
Gaff-Smith M. The relationship between self-esteem,
social support and postnatal depression in adolescent Lesser J. From childhood maltreatment to motherhood:
mothers in rural New South Wales. Birth Issues An exploration of the triad of childhood abuse,
2003;12:6975. depression, and maternal role attainment in adolescent
mothers. Dissertation Abstracts International: Section B: The
Garel M, Dardennes M, Blondel B. Mothers Sciences and Engineering 1997;58:5329.
psychological distress 1 year after very preterm
childbirth. Results of the epipage qualitative study. Child Letourneau N, Duffett-Leger L, Stewart M,
Care Health Dev 2007;33:13743. Hegadoren K, Dennis CL, Rinaldi CM, et al. Canadian
mothers perceived support needs during postpartum
Giovannini M. The relevance of gender in postpartum depression. J Obstet Gynecol Neonatal Nurs 2007;36:4419.
emotional disorders. 1992:20931.
Leung E. Family support and postnatal emotional
Hall P. Mothers experiences of postnatal depression: adjustment. Bulletin of the Hong Kong Psychological Society
an interpretative phenomenological analysis. Community 1985;14:3246.
Pract 2006;79:25660.
Leung S, Arthur DG, Martinson I. Stress in women with
Hanley J, Long B. A study of Welsh mothers experiences postpartum depression: a phenomenological study. J Adv
of postnatal depression. Midwifery 2006;22:14757. Nurs 2005;51:35360.
Hanley J. The emotional wellbeing of Bangladeshi Lewis S, Nicolson P. Talking about early motherhood:
mothers during the postnatal period. Community Pract Recognizing loss and reconstructing depression. J Reprod
2007;80:347. Infant Psychol 1998;16:17797.
Holopainen D. The experience of seeking help for Luepker E. Joint admission and evaluation of
postnatal depression. Aus J Adv Nurs 2002;19:3944. postpartum psychiatric patients and their infants. Hosp
Community Psychiatry 1972;23:2847.
Horowitz JA, Chang SS, Das S, Hayes B. Womens
perceptions of postpartum depressive symptoms from Maloney DM. Postnatal depression: a study of mothers
an international perspective. International Nursing in the metropolitan area of Perth, Western Australia. Aust
Perspectives 2001;1:514. Coll Midwives Inc J 1998;11:1823.
Kane H. Emotion work, labeling, and gender in post- Mauthner NS. Postnatal depression the significance of
partum and post-adoptive depression. Dissertation social contacts between mothers. Womens Stud Int Forum
Abstracts International Section A: Humanities and Social 1995;18:31123.
Sciences 2006;67:3996.
Mauthner NS. Postnatal depression: how can midwives
Kazi A, Fatmi Z, Hatcher J, Kadir MM, Niaz U, help? Midwifery 1997;13:16371.
Wasserman GA. Social environment and depression
among pregnant women in urban areas of Pakistan: Mauthner N. Feeling low and feeling really bad about
Importance of social relations. Soc Sci Med feeling low: Womens experiences of motherhood and
2006;63:146676. postpartum depression. Can Psychol 1999;40:14361.
Ketler S. Separating mothering and motherhood: Mauthner N. Its a womans cry for help: A relational
Womens changing maternity experiences and the perspective on postnatal depression. Fem Psychol
normalization of postpartum depression in Cagliari, 1998;8:32555.
Italy, 19431994. Dissertation Abstracts International
Section A: Humanities and Social Sciences 1997;58:4327. Mauthner N. Re-assessing the importance and role
of the marital relationship in postnatal depression:
Kim H, Bracha Y, Tipnis A. Automated depression Methodological and theoretical implications. J Reprod
screening in disadvantaged pregnant women in an urban Infant Psychol 1998;16:15775.
obstetric clinic. Arch Womens Ment Health 2007;10:1639.
Mauthner N, Doucet A. Reflections on a voice-centred
Lauer W. Postpartum depression: A phenomenological relational method: analysing maternal and domestic
exploration of the womans experience. Dissertation voices. In Ribbens J, Edwards R, editors. Feminist dilemas
105
in qualitative research: private lives and public texts. London: Rodrigues M, Patel V, Jaswal S, de Souza N. Listening to
Sage; 1998. pp. 11944. mothers: Qualitative studies on motherhood and
depression from Goa, India. Soc Sci Med 2003;57:
Mauthner N. Towards a feminist understanding of 1797806.
postnatal depression: I. Fem Psychol 1993;3:3505.
Ross LE, Steele L, Sapiro B. Perceptions of predisposing
Mayes L, Leckman J. Parental representations and and protective factors for perinatal depression in same-
subclinical changes in postpartum mood. Infant Ment sex parents. J Midwifery Womens Health 2005;50:e65e70.
Health J 2007;28:28195.
Roux G, Anderson C, Roan C. Postpartum depression,
Nahas V, Amasheh N. Culture care meanings and marital dysfunction, and infant outcome: a longitudinal
experiences of postpartum depression among Jordanian study. J Perinat Educ 2002;11:2536.
Australian women: a transcultural study. J Transcult Nurs
1999;10:3745. Scott D. Early identification of maternal depression as a
strategy in the prevention of child abuse. Child Abuse Negl
Nahas VL, Hillege S, Amasheh N. Postpartum 1992;16:34558.
depression. The lived experiences of Middle Eastern
migrant women in Australia. J Nurse Midwifery Scrandis D. Normalizing postpartum depressive
1999;44:6574. symptoms with social support. J Am Psychiatr Nurses Assoc
2005;11:22330.
Nicolson P. Counselling women with post natal
depression: Implications from recent qualitative research. Setse R, Grogan R, Cooper LA, Strobino D, Powe NR,
Couns Psychol Q 1989;2:12332. Nicholson W. Weight loss programs for urban-based,
postpartum African-American women: perceived
Nicolson P. Loss, happiness and postpartum depression: barriers and preferred components. Matern Child Health J
The ultimate paradox. Can Psychol 1999;40:16278. 2008;12:11927.
OHara MW, Rehm LP, Campbell SB. Postpartum Shakespeare J. Health visitor screening for PND using
depression. A role for social network and life stress the EPDS: a process study. Community Pract 2002;
variables. J Nerv Ment Dis 1983;171:33641. 75:3814.
Oates MR, Cox JL, Neema S, Asten P, Glangeaud F, Shakespeare J, Blake F, Garcia J. A qualitative study of
Figueiredo B, et al. Postnatal depression across countries the acceptability of routine screening of postnatal women
and cultures: A qualitative study. Br J Psychiatry 2004; using the Edinburgh Postnatal Depression Scale. Br J Gen
184:s10s16. Pract 2003;53:61419.
Regmi S, Sligl W, Carter D, Grut W, Seear M. A Small R, Johnston V, Orr A. Depression after childbirth:
controlled study of postpartum depression among the views of medical students and women compared.
Nepalese women: validation of the Edinburgh Birth 1997;24:10915.
Postpartum Depression Scale in Kathmandu. Trop Med
Int Health 2002;7:37882. Steinfeld G. Brief therapy with a postpartum depressive:
Integrating cognitive behavior and thought field
106
DOI: 10.3310/hta14440 Health Technology Assessment 2010; Vol. 14: No. 44
therapies within a spiritual framework. J Psychother Integr women and their 4-year-old children. Acta Psychiatr Scand
1999;9:33763. 1978;58:20112.
Stewart S, Jambunathan J. Hmong women and Ugarriza DN, Brown SE, Chang-Martinez C. Anglo-
postpartum depression. Health Care Women Int 1996; American mothers and the prevention of postpartum
17:31930. depression. Issues Ment Health Nurs 2007;28:78198.
Tammentie T, Paavilainen E, stedt-Kurki P, Tarkka MT. Ward C. Migrant mothers and the role of social support
Family dynamics of postnatally depressed mothers when child rearing. Contemp Nurse 2003;16:7482.
discrepancy between expectations and reality. J Clin Nurs
2004;13:6574. Wheatley SL, Brugha TS. Just because I like it
doesnt mean it has to work: personal experiences of
Templeton L, Velleman R, Persaud A, Milner P. The an antenatal psychosocial intervention designed to
experiences of postnatal depression in women from black prevent postnatal depression. International J Ment Health
and minority ethnic communities in Wiltshire, UK. Ethn Promotion 1999;1:2631.
Health 2003;8:20721.
White P. Heat, balance, humors, and ghosts: Postpartum
Thomas H. Womens postnatal experience following a in Cambodia. Health Care Women Int 2004;25:17994.
medically complicated pregnancy. Health Care Women Int
2004;25:7687. Williamson V, McCutcheon H. Postnatal blues. Singapore
Nursing Journal 2002;29:326.
Thurtle V. First time mothers perceptions of
motherhood and PND. Community Pract 2003;76:2615. Woollett A, Parr M. Psychological tasks for women
and men in the post-partum. J Reprod Infant Psychol
Uddenberg N, Englesson I. Prognosis of post partum 1997;15:15983.
mental disturbance: A prospective study of primiparous
107
No. 9 No. 13
No. 4
Implications of socio-cultural contexts
A costutility analysis of interferon beta Choosing between randomised and
for the ethics of clinical trials.
for multiple sclerosis. nonrandomised studies: a systematic
A review by Ashcroft RE, Chadwick
By Parkin D, McNamee P, Jacoby A, review.
DW, Clark SRL, Edwards RHT, Frith L,
Hutton JL. Miller P, Thomas S, Bates D. By Britton A, McKee M, Black N,
McPherson K, Sanderson C, Bain C.
No. 10 No. 5
A critical review of the role of neonatal Effectiveness and efficiency of methods No. 14
hearing screening in the detection of of dialysis therapy for end-stage renal Evaluating patient-based outcome
congenital hearing impairment. disease: systematic reviews. measures for use in clinical trials.
By Davis A, Bamford J, Wilson I, By MacLeod A, Grant A, Donaldson A review by Fitzpatrick R, Davey C,
Ramkalawan T, Forshaw M, Wright S. C, Khan I, Campbell M, Daly C, etal. Buxton MJ, Jones DR. 109
No. 7 No. 16
No. 26
The clinical effectiveness and cost- Screening for fragile X syndrome: a Can randomised trials rely on existing
effectiveness of routine dental checks: literature review and modelling. electronic data? A feasibility study to
a systematic review and economic By Song FJ, Barton P, Sleightholme explore the value of routine data in
evaluation. V, Yao GL, Fry-Smith A. health technology assessment.
By Davenport C, Elley K, Salas By Williams JG, Cheung WY,
C, Taylor-Weetman CL, Fry-Smith A, No. 17 Cohen DR, Hutchings HA, Longo MF,
Bryan S, etal. Russell IT.
Systematic review of endoscopic sinus
surgery for nasal polyps.
No. 8 No. 27
A multicentre randomised controlled By Dalziel K, Stein K, Round A,
Garside R, Royle P. Evaluating non-randomised
trial assessing the costs and benefits intervention studies.
of using structured information and By Deeks JJ, Dinnes J, DAmico R,
analysis of womens preferences in the No. 18 Sowden AJ, Sakarovitch C, Song F, etal.
management of menorrhagia. Towards efficient guidelines: how to
By Kennedy ADM, Sculpher MJ, monitor guideline use in primary care. No. 28
Coulter A, Dwyer N, Rees M, Horsley S,
By Hutchinson A, McIntosh A, A randomised controlled trial to assess
etal.
Cox S, Gilbert C. the impact of a package comprising a
patient-orientated, evidence-based self-
No. 9
No. 19 help guidebook and patient-centred
Clinical effectiveness and costutility consultations on disease management
of photodynamic therapy for wet Effectiveness and cost-effectiveness
and satisfaction in inflammatory bowel
age-related macular degeneration: of acute hospital-based spinal cord
disease.
a systematic review and economic injuries services: systematic review.
By Kennedy A, Nelson E, Reeves D,
evaluation. By Bagnall A-M, Jones L, Richardson Richardson G, Roberts C, Robinson A,
By Meads C, Salas C, Roberts T, G, Duffy S, Riemsma R. etal.
Moore D, Fry-Smith A, Hyde C.
No. 20 No. 29
No. 10
Prioritisation of health technology The effectiveness of diagnostic tests for
Evaluation of molecular tests for
assessment. The PATHS model: the assessment of shoulder pain due
prenatal diagnosis of chromosome
methods and case studies. to soft tissue disorders: a systematic
abnormalities.
By Townsend J, Buxton M, review.
By Grimshaw GM, Szczepura A,
Hultn M, MacDonald F, Nevin NC, Harper G. By Dinnes J, Loveman E, McIntyre L,
Sutton F, etal. Waugh N.
No. 21
No. 11 No. 30
Systematic review of the clinical
First and second trimester antenatal effectiveness and cost-effectiveness of The value of digital imaging in diabetic
screening for Downs syndrome: tension-free vaginal tape for treatment retinopathy.
the results of the Serum, Urine and of urinary stress incontinence. By Sharp PF, Olson J, Strachan F,
Ultrasound Screening Study (SURUSS). Hipwell J, Ludbrook A, ODonnell M,
By Cody J, Wyness L, Wallace S, etal.
By Wald NJ, Rodeck C, Hackshaw Glazener C, Kilonzo M, Stearns S, etal.
AK, Walters J, Chitty L, Mackinson AM.
No. 31
No. 22 Lowering blood pressure to prevent
No. 12
The effectiveness and cost-effectiveness The clinical and cost-effectiveness of myocardial infarction and stroke: a new
of ultrasound locating devices for patient education models for diabetes: preventive strategy.
central venous access: a systematic a systematic review and economic By Law M, Wald N, Morris J.
review and economic evaluation. evaluation.
By Calvert N, Hind D, McWilliams By Loveman E, Cave C, Green C, No. 32
RG, Thomas SM, Beverley C, Royle P, Dunn N, Waugh N. Clinical and cost-effectiveness of
Davidson A. capecitabine and tegafur with uracil for
No. 23 the treatment of metastatic colorectal
No. 13 The role of modelling in prioritising cancer: systematic review and economic
A systematic review of atypical and planning clinical trials. evaluation.
antipsychotics in schizophrenia. By Ward S, Kaltenthaler E, Cowan J,
By Chilcott J, Brennan A, Booth A,
By Bagnall A-M, Jones L, Lewis R, Brewer N.
Karnon J, Tappenden P.
Ginnelly L, Glanville J, Torgerson D,
etal. No. 33
No. 24 Clinical and cost-effectiveness of new
No. 14 Costbenefit evaluation of routine and emerging technologies for early
Prostate Testing for Cancer and influenza immunisation in people localised prostate cancer: a systematic
Treatment (ProtecT) feasibility study. 6574 years of age. review.
By Donovan J, Hamdy F, Neal D, By Allsup S, Gosney M, Haycox A, By Hummel S, Paisley S, Morgan A,
Peters T, Oliver S, Brindle L, etal. Regan M. Currie E, Brewer N. 115
130
DOI: 10.3310/hta14440 Health Technology Assessment 2010;Vol. 14: No. 44
Observers
Ms Kay Pattison, Dr Morven Roberts,
Section Head, NHS R&D Clinical Trials Manager,
Programme, Department of Medical Research Council
Health
131
Observers
Dr Tim Elliott, Dr Catherine Moody, Dr Ursula Wells,
Team Leader, Cancer Programme Manager, Principal Research Officer,
Screening, Department of Neuroscience and Mental Department of Health
Health Health Board
Observers
Ms Christine McGuire, Ms Kay Pattison Dr Caroline Stone,
Research & Development, Senior NIHR Programme Programme Manager, Medical
Department of Health Manager, Department of Research Council
Health
132
Current and past membership details of all HTA programme committees are available from the HTA website (www.hta.ac.uk)
DOI: 10.3310/hta14440 Health Technology Assessment 2010;Vol. 14: No. 44
Observers
Dr Phillip Leech, Ms Kay Pattison Dr Morven Roberts, Dr Ursula Wells
Principal Medical Officer for Senior NIHR Programme Clinical Trials Manager, MRC, PRP, DH, London
Primary Care, Department of Manager, Department of London
Health , London Health
133
Pharmaceuticals Panel
Members
Chair, Dr Peter Elton, Dr Dyfrig Hughes, Dr Martin Shelly,
Professor Imti Choonara, Director of Public Health, Reader in Pharmacoeconomics General Practitioner, Leeds,
Professor in Child Health, Bury Primary Care Trust and Deputy Director, Centre and Associate Director, NHS
University of Nottingham for Economics and Policy in Clinical Governance Support
Professor Robin Ferner, Health, IMSCaR, Bangor Team, Leicester
Deputy Chair, Consultant Physician and University
Dr Lesley Wise, Director, West Midlands Centre Dr Gillian Shepherd,
Unit Manager, for Adverse Drug Reactions, Dr Yoon K Loke, Director, Health and Clinical
Pharmacoepidemiology City Hospital NHS Trust, Senior Lecturer in Clinical Excellence, Merck Serono Ltd
Research Unit, VRMM, Birmingham Pharmacology, University of
Medicines & Healthcare East Anglia Mrs Katrina Simister,
Products Regulatory Agency Dr Ben Goldacre, Assistant Director New
Research Fellow, Division of Professor Femi Oyebode, Medicines, National Prescribing
Mrs Nicola Carey, Psychological Medicine and Consultant Psychiatrist Centre, Liverpool
Senior Research Fellow, Psychiatry, Kings College and Head of Department,
School of Health and Social London University of Birmingham Mr David Symes,
Care, The University of Service User Representative
Reading Dr Bill Gutteridge, Dr Andrew Prentice,
Medical Adviser, London Senior Lecturer and Consultant
Mr John Chapman, Strategic Health Authority Obstetrician and Gynaecologist,
Service User Representative The Rosie Hospital, University
of Cambridge
Observers
Ms Kay Pattison Mr Simon Reeve, Dr Heike Weber, Dr Ursula Wells,
Senior NIHR Programme Head of Clinical and Cost- Programme Manager, Principal Research Officer,
Manager, Department of Effectiveness, Medicines, Medical Research Council Department of Health
Health Pharmacy and Industry Group,
Department of Health
Observers
Ms Kay Pattison Dr Morven Roberts, Professor Tom Walley, Dr Ursula Wells,
Senior NIHR Programme Clinical Trials Manager, MRC, HTA Programme Director, Policy Research Programme,
Manager, Department of London Liverpool DH, London
Health
134
Current and past membership details of all HTA programme committees are available from the HTA website (www.hta.ac.uk)
DOI: 10.3310/hta14440 Health Technology Assessment 2010;Vol. 14: No. 44
135