The development and application of a prediction model for postpartum depression: optimizing risk assessment and prevention in the clinic

Abstract

Background: Preventive intervention can significantly reduce the human and economic costs of postpartum depression (PPD) compared with treatment post-diagnosis. However, identifying women with a high PPD risk and making a judgement as to the benefits of preventive intervention is a major challenge.

Methods: This is a retrospective study of parturients that underwent a cesarean delivery. Control group was used as development cohort and validation cohort to construct the risk prediction model of PPD and determine a risk threshold. Ketamine group and development cohort were used to verify the risk classification of parturients by evaluating whether the incidence of PPD decreased significantly after ketamine treatment in high-risk for PPD population.

Results: The AUC for the development cohort and validation cohort of the PPD prediction model were 0.751 (95%CI:0.700-0.802) and 0.748 (95%CI:0.680-0.816), respectively. A threshold of 19% PPD risk probability was determined, with a specificity and sensitivity in the validation cohort are 0.766 and 0.604, respectively. After matching the high-risk group and the low-risk group by propensity score, the results demonstrated that PPD incidence significantly reduced in the high-risk group following ketamine, versus non-ketamine, intervention (p < 0.01). In contrast, intervention in the low-risk group showed no significant difference in PPD outcomes (p > 0.01).

Limitation: Randomized trials are needed to further verify the feasibility of the model and the thresholds proposed.

Conclusion: This prediction model developed in this study shows utility in predicting PPD risk. Ketamine intervention significantly lowers PPD incidence in parturients with a risk classification threshold greater than 19%.

Keywords: Postpartum depression; Prediction model; Preventive intervention; Risk threshold.