Table 2

Design of included studies

Study	Evaluation type	Measure of health benefit	Evaluation details	Data source	Quality/bias considerations
Boath et al 28	CEA	Recovery from PND (no longer fulfilling Research Diagnostic Criteria)	Trial or model: trial Perspective: health service Time horizon: 6 months Price year: 1992/1993 Currency: British £	Observational study—healthcare utilisation self-reported and obtained from medical records	Treatment allocation was non-randomised. Reported that no significant differences in sociodemographic characteristics or outcome measures between groups at baseline. No loss to follow-up reported.
Petrou et al 29	CEA	Months of postnatal depression avoided (SCID-II)	Trial or model: trial Perspective: health and social services Time horizon: 18 months Price year: 2000 Currency: British £	RCT—health and social care utilisation was self-reported by participants	Structured clinical interviews were used to identify depression in both treatment groups. The numbers/characteristics of those declining to participate were not reported.
Morrell et al 30	CUA	QALYs (derived from the SF-6D) EPDS	Trial or model: trial Perspective: health and social services Time horizon: 18 months Price year: 2003/2004 Currency: British £	RCT—health and social care utilisation obtained from medical records (up to 6 months) and participant self-report (at 12 and 18 months)	Data were collected on women declining to take part but differences with sample were not discussed. Sample was broadly representative of general population. Missing economic data were significant at 12 and 18 months, 6 months was used as the primary time horizon.
Stevenson et al 31	CUA	QALYs (derived from EPDS mapped onto SF-6D)	Trial or model: model (mathematical) Perspective: health and social services Time horizon: 12 months Price year: not reported Currency: British £	Published data sources and expert opinion informed the model. EPDS, SF-36 and costs from published RCTs	As the model was mathematical, no structure was reported in the paper. Probabilistic sensitivity analyses were conducted.
Dukhovny et al 35	CEA	Cases of PND averted at 12 weeks post partum	Trial or model: trial Perspective: societal Time horizon: 12 weeks Price year: 2011 Currency: $C	Multiregion RCT—resource utilisation was self-reported by participants	Only two people did not complete healthcare utilisation questionnaires and fewer than 0.01% of individual resource utilisation items were missing at random.
Ride et al 34	CEA; CUA	Prevalence of depression and anxiety (DSM-IV criteria) QALYs (from the EQ-5D)	Trial or model: trial Perspective: health and social services Time horizon: 20 weeks Price year: 2013/2014 Currency: $A	Cluster RCT—health and social care utilisation self-reported by participants	Differences between the treatment groups were adjusted for in the analysis. The intracluster coefficients were small but non-negligible for QALYs, which may have reduced the ability to detect an effect of the intervention.
Grote et al 32	CEA	Depression severity (SCL-20) Depression-free days PTSD checklist	Trial or model: trial Perspective: health plan or insurer Time horizon: 18 months Price year: 2013 Currency: US$	RCT—health and social care utilisation self-reported by participants	The costs included only related to mental healthcare. The perspective was ’public health' and so could have also included primary and community healthcare services. Those with partial cost data (n=12/164) were more likely to have probable PTSD and to have been randomly assigned to the intervention.
Wilkinson et al 33	CEA; CUA	QALYs (derived from published literature) EPDS	Trial or model: model (decision tree) Perspective: health plan (Medicaid) Time horizon: 2 years Price year: 2014 Currency: US$	Systematic review of existing literature to inform the model. Some cost parameters estimated from Medicaid data	Some parameters were from studies of anxiety/depression outside of the perinatal period. Probabilistic sensitivity analyses were conducted. The model structure is pragmatic, but perhaps over simple in terms of suicide risk—only women who discontinue treatment are at risk of suicide, women who do not seek help or those who screen negative are not deemed to be at risk of suicide.

CBT, cognitive behavioural therapy; CEA, cost-effectiveness analysis; CUA, cost-utility analysis; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, fourth edition; EPDS, Edinburgh Postnatal Depression Scale; PTSD, post-traumatic stress disorder; QALY, quality-adjusted life year; RCT, randomised controlled trial; SCID-II, Structured Clinical Interview for Depression, second edition; SCL-20, 20-item Symptom Checklist Depression Scale.