Depressive symptoms and risk of incident hip fracture in older adults: A harmonized multinational cohort study

Hip fracture is a major cause of disability and mortality in older adults. Prior studies examining depression and hip fracture have reported inconsistent findings. We investigated the association between depressive symptoms and incident hip fracture among adults aged 60 years or more across multiple countries.

We conducted a harmonized longitudinal analysis of five population-based cohorts: CHARLS, ELSA, HRS, SHARE, and KLoSA. We fitted cohort-specific Cox proportional hazards models and pooled fully adjusted estimates using fixed- and random-effects meta-analysis. We additionally applied a longitudinal modified treatment policy with a sequential doubly robust estimator and SuperLearner to estimate counterfactual risks under emulated baseline depressive-symptom scenarios.

The analytic sample included 46,177 participants (CHARLS 2550; ELSA 4073; HRS 8650; SHARE 25,515; KLoSA 5389), with 1887 incident hip fractures during follow-up. In fully adjusted models, depressive symptoms were associated with higher hip fracture risk in CHARLS (HR 1.78; 95% CI 1.20-2.64), ELSA (HR 1.81; 95% CI 1.22-2.68), HRS (HR 1.62; 95% CI 1.29-2.04), and SHARE (HR 1.33; 95% CI 1.17-1.51), whereas estimates in KLoSA were attenuated (HR 1.12; 95% CI 0.84-1.48). Pooled estimates were HR 1.36 (95% CI 1.23-1.50) under the fixed-effects model and HR 1.41 (95% CI 1.19-1.63) under the random-effects model. LMTP analyses yielded consistently higher odds of subsequent hip fracture under the "all depressive symptoms" scenario in four cohorts, whereas KLoSA showed a nonsignificant increase in hip fracture odds.

Depressive symptoms were associated with a higher subsequent risk of hip fracture in four of five cohorts of older people, with modest pooled effects. Findings in KLoSA were heterogeneous and statistically uncertain despite directionally positive estimates, underscoring the need to clarify cohort- and context-specific sources of variability.