Systematic Review and Meta-Analysis: Predictors of Relapsing, Recurrent, and Chronic Depression in Young People

Objective: Youth depression disrupts the social and vocational transition into adulthood. Most depression burden is caused by recurring or chronic episodes. Identifying young people at risk for relapsing, recurring, or chronic depression is critical. We systematically reviewed and meta-analyzed the literature on prognostic factors for relapsing, recurrent, and chronic depression in young people. Method: We searched the literature up (MEDLINE, PsycINFO, CINAHL, Embase, CENTRAL, WHO ICTRP, ClinicalTrials.gov, bioRxiv, MedRxiv) to March 6, 2024, and included cohort studies and randomized trials that assessed any prognostic factor for relapse, recurrence, or chronicity of depression in young people (aged 10-25 years at baseline) with a minimum of a 3-month follow-up. We assessed individual study risk of bias using the QUIPS tool and the certainty of evidence via the GRADE approach. We conducted random-effects meta-analyses with Hartung–Knapp–Sidik–Jonkman adjustment when 3 or more estimates on the same prognostic factor were available. Qualitative synthesis was conducted to identify promising prognostic factors that could not be meta-analyzed. Results: A total of 76 reports of 46 studies (unique cohorts or trials) were included that tested 388 unique prognostic factors in 7,488 young people experiencing depression. The majority of the reports were at high risk of bias (87%). We conducted 22 meta-analyses on unadjusted, and 7 on adjusted, prognostic factors of a poor course trajectory (ie, combined relapse, recurrence, and chronicity). Female sex (adjusted; odds ratio [95% CI] = 1.49 [1.15, 1.93], p = .003), higher severity of depressive symptoms (unadjusted; standardized mean difference [95% CI] = 0.53 [0.33, 0.73], p < .001), lower global functioning (unadjusted; standardized mean difference [95% CI] = −0.35 [−0.60, −0.10], p = .005), more suicidal thoughts and behaviors (unadjusted; standardized mean difference [95% CI] = 0.52 [0.03, 1.01], p = .045), and longer sleep-onset latency (unadjusted; mean difference [95% CI] = 6.96 [1.48, 12.44] minutes, p = .013) at baseline predicted a poor course trajectory of depression. The certainty of the evidence was overall very low to moderate. Promising prognostic factors that could not be meta-analyzed included relational/interpersonal factors (friend relationships and family relationships/structure). Conclusion: Our findings demonstrate the prognostic value of demographic and clinical factors for poor course trajectories of depression in young people. More research is needed to confirm the potential value of relational/interpersonal factors in predicting poor depression course. Limitations of the literature include the high risk of bias of included studies, which indicates that future studies should include large sample sizes and wider diversity of prognostic markers (eg, genetic and neurobiological) in multivariable models. The critical next step is to combine the identified prognostic factors and to evaluate their clinical value in identifying individuals at risk for a poor course trajectory of depression during youth, a life stage in which most of the disability and burden attributable to depression can be averted. Study preregistration information: Prognostic factors for relapsing, recurrent or chronic depression in youth: a systematic review with meta-analysis; https://www.crd.york.ac.uk/PROSPERO/view/CRD42023458646.