Although evidence-based treatments for Prolonged Grief Disorder (PGD) exist, pretreatment characteristics associated with differential improvement trajectories have not been identified. To identify clinical factors relevant to optimizing PGD treatment outcomes, we used unsupervised and supervised machine learning to study treatment effects from a double-blinded, placebo-controlled, randomized clinical trial. Participants were randomized into four treatment groups for 20 weeks: citalopram with grief-informed clinical management, citalopram with prolonged grief disorder therapy (PGDT), pill placebo with PGDT, or pill placebo with clinical management. The trial included 333 PGD patients aged 18-95 years (M age = 53.9; SD ± 14.4), predominantly female (77.4%) and white (84.4%). Symptom trajectories were assessed using latent growth mixture modeling based on Inventory for Complicated Grief scores collected at six time points every 4 weeks. The relationship between patient-level characteristics and assigned trajectories was examined using logistic regression with elastic net regularization based on the administration of citalopram, PGDT, and risk factors for developing PGD. Three distinct response trajectories were identified: lesser severity responders (60%, n = 200), greater severity responders (18.02%, n = 60), and non-responders (21.92%, n = 73). Differences between greater severity responders and non-responders emerged as statistically significant by Week 8. The elastic net model demonstrated acceptable discrimination between responders and non-responders (AUC = .702; accuracy = .684). Higher baseline depression severity, grief-related functional impairment, and absence of PGDT were associated with reduced treatment response likelihood. These findings underscore the importance of early identification of clinical factors to optimize individualized PGD treatment strategies.
Trial registration: clinicaltrials.gov Identifier: NCT01179568.
Keywords: Grief; citalopram; machine learning; therapy; trajectories.