Patient-reported high suicide risk despite non-high clinician ratings is associated with later suicidal ideation after adolescent psychiatric hospitalization: a retrospective real-world longitudinal cohort study

<title>Abstract</title> <p>Background Suicide-risk formulation during adolescent psychiatric hospitalization integrates patient self-report with physician and nursing assessments, but these sources may disagree. Whether high patient-reported risk remains clinically relevant when clinical ratings are not high is uncertain. Methods We retrospectively analyzed routine inpatient assessments and naturalistic post-discharge patient-reported outcomes at one center. The baseline-eligible cohort included adolescents aged 11–18 years with complete BSI-CV, physician-recorded C-SSRS-based, nurse-recorded NGASR, age, sex, and depression-severity data. The platform-generated BSI-CV high-risk category defined patient risk. C-SSRS was the primary clinical anchor, with NGASR examined in parallel. Post-discharge questionnaires were completed during routine care rather than scheduled research visits. We retained one assessment nearest each 1-, 3-, and 6-month target. Random-intercept linear mixed models estimated associations with follow-up BSI-CV suicidal ideation severity, adjusting for follow-up month, age, sex, and baseline depression severity. We also compared baseline characteristics by post-discharge questionnaire availability. Results Of 200 baseline-eligible adolescents, 129 (64.5%) had a valid post-discharge questionnaire and 71 (35.5%) had none. 126 adolescents contributed 201 assessments within the analytic windows. Adolescents with observed follow-up were younger (mean 14.72 vs 15.34 years; standardized mean difference [SMD] = − 0.35; p = 0.020). Baseline BSI-CV high risk was also more frequent in this group (68.2% vs 54.9%; SMD = 0.28; p = 0.067). Other measured differences were small to modest. In the analytic cohort, BSI-CV classification disagreed with C-SSRS in 29.4% (κ = 0.379) and NGASR in 26.2% (κ = 0.408). Relative to concordant non-high risk, patient-only high risk was associated with higher follow-up BSI-CV scores in both C-SSRS (β = 1.50, 95% CI 0.06–2.95) and NGASR analyses (β = 1.63, 95% CI 0.05–3.21). Conclusions Among adolescents with observed naturalistic follow-up, high BSI-CV self-assessed risk was associated with greater subsequent self-reported suicidal ideation when clinical classifications were not high. Measured baseline characteristics did not indicate greater initial risk among adolescents without questionnaires. However, 35.5% had no observed outcome, and unmeasured selection remains possible. Discordance should prompt reconciliation across information sources rather than dismissal of the patient report.</p>