Purpose: To explore, within a UK setting, the cost-effectiveness of providing group cognitive behavioural therapy (gCBT) for women with postnatal depression (PND) compared with usual care (UC).
Method: A mathematical model was constructed that was populated with data from literature searches. Despite a high prevalence of PND (14%) data were scarce; only one RCT was deemed applicable for estimating clinical effectiveness, in which gCBT was shown to produce a reduction of 3.98 (95% CI 3.27 – 4.69) in the Edinburgh postnatal depression scale (EPDS) compared with UC. Full effectiveness was assumed to be reached after the treatment course (eight weeks) and to be maintained for six months. At treatment initiation and one year later, gCBT and UC were assumed to have equivalent EPDS scores. Linear changes in EPDS were assumed between these time points and full efficacy. Changes in EPDS were converted to changes in utility using primary data from women with PND (n=401) who completed both EPDS and SF-6D questionnaires, assuming a bivariate normal distribution. The cost-effectiveness of gCBT was estimated using probabilistic sensitivity analyses. The impact on cost-effectiveness of the length of residual benefit of gCBT and of the assumed cost per woman treated was explored in structural sensitivity analyses.
Result: In the base case the mean cost per QALY gained was £46,500; no simulations produced values below £30,000. However, cost-effectiveness was sensitive to structural uncertainties; the mean cost per QALY gained decreased to £24,800 using a cost of £800 per patient, and to £27,300 were the duration at which gCBT and UC were assumed to have equivalent EPDS scores extended by 12 months.
Conclusion: In the base case gCBT for PND would not be considered cost-effective under existing guidance, however there is considerable structural uncertainty in the model. Further data are needed for a definitive conclusion on cost-effectiveness.
Candidate for the Lee B. Lusted Student Prize Competition