Hierarchical Bayesian Analysis of Suicidal Ideation Among Middle‐Aged and Older Adults in South Korea

BACKGROUND: Research on suicidal ideation has largely relied on population-average models that implicitly generalize group-level associations into individual outcomes. This study examines the extent to which risk and protective factors for suicidal ideation operate differently across individuals among middle-aged and older adults. METHODS: Data from the Korean Welfare Panel Study (2012-2024) included 2500 adults aged 51 and above. We employed hierarchical Bayesian logistic regression with an individual-specific random intercept and random slopes for time-varying predictors. Demographic characteristics (sex, age, education) accounted for part of the heterogeneity in the individual-specific coefficients, with the remaining heterogeneity captured by person-specific random effects. RESULTS: Substantial individual heterogeneity emerged in the effects of risk and protective factors on suicidal ideation. Demographic characteristics explained part of this heterogeneity. Specifically, spousal presence was more protective for men, and higher education buffered material hardship effects. Remaining heterogeneity was reflected in person-specific random slopes (sensitivity parameters), which varied substantially across individuals even after accounting for demographic characteristics. Scenario analyses demonstrated that high-sensitivity individuals experienced greater risk increases than low-sensitivity individuals under the same stressor changes across psychosocial conditions, and an illustrative case showed substantial predicted risk elevation at modest symptom levels in an individual with high depression sensitivity. CONCLUSIONS: These findings highlight the importance of shifting from group-average effects towards individualized risk stratification that accounts for individual-specific vulnerability profiles. The hierarchical Bayesian approach shows promise for generating person-specific risk estimates that may support the identification of higher-risk subgroups for targeted screening and prevention planning.