Predicting psychiatric readmission: sex-specific models to predict 30-day readmission following acute psychiatric hospitalization
Journal Articles
Overview
Research
Identity
Additional Document Info
View All
Overview
abstract
PURPOSE: Psychiatric readmission is a common negative outcome. Predictors of readmission may differ by sex. This study aimed to derive and internally validate sex-specific models to predict 30-day psychiatric readmission. METHODS: We used population-level health administrative data to identify predictors of 30-day psychiatric readmission among women (n = 33,353) and men (n = 32,436) discharged from all psychiatric units in Ontario, Canada (2008-2011). Predictor variables included sociodemographics, health service utilization, and clinical characteristics. Using derivation data sets, multivariable logistic regression models were fit to determine optimal predictive models for each sex separately. Results were presented as adjusted odds ratios (aORs) and 95% confidence intervals (CI). The multivariable models were then applied in the internal validation data sets. RESULTS: The 30-day readmission rates were 9.3% (women) and 9.1% (men). Many predictors were consistent between women and men. For women only, personality disorder (aOR 1.21, 95% CI 1.03-1.42) and positive symptom score (aOR 1.41, 95% CI 1.09-1.82 for score of 1 vs. 0; aOR 1.44, 95% CI 1.26-1.64 for ≥ 2 vs. 0) increased odds of readmission. For men only, self-care problems at admission (aOR 1.20, 95% CI 1.06-1.36) and discharge (aOR 1.44, 95% CI 1.26-1.64 for score of 1 vs. 0; aOR 1.79, 95% CI 1.17-2.74 for 2 vs. 0), and mild anxiety rating (score of 1 vs. 0: aOR 1.30, 95% CI 1.02-1.64, derivation model only) increased odds of readmission. Models had moderate discriminative ability in derivation and internal validation samples for both sexes (c-statistics 0.64-0.65). CONCLUSIONS: Certain key predictors of psychiatric readmission differ by sex. This knowledge may help to reduce psychiatric hospital readmission rates by focusing interventions.
