Key result
Local LASSO model improves 30-day psychiatric readmission prediction over LACE index to ~0.65 AUC.
Why the study?
Schizophrenia and psychotic disorders-related hospitalizations are the second leading cause of 30-day readmission, but the effect of patient and hospital-level factors on readmission costs is not well understood.
Cohort (n=14,994)
No
Effect estimate: ΔAUC 0.092 (95% CI 0.623-0.678)
Absolute Event Rate: 0.649% vs 0.557%
p-value: p=<0.001
Encourages site-specific readmission models; hypothesis-generating and requires prospective validation before practice change.
BACKGROUND: Schizophrenia and psychotic disorders (SPDs)-related hospitalizations are the second leading cause of 30-day readmission. This study assessed the effect of patient and hospital-level factors on readmission costs following index hospital discharges for SPDs. METHODS: 2014 Nationwide Readmissions Database was used to identify SPD-related discharges between January 1, 2014, and November 30, 2014. Multivariable logistic regression was used to estimate patient and hospital-level predictors for readmissions. A two part model was used to estimate the predictors of readmission and associated cost for index hospital discharges with SPDs. RESULTS: A total of 77 625 of 343 579 (22%) index hospital discharges for SPDs resulted in readmissions. The average index and readmission costs were $9285 and $8593, respectively. Being 25 to 44 years old (odds ratio: 1.14), with nonmental comorbidities (odds ratio: 1.52), and admitted in private hospitals (odds ratio: 1.24) had significantly higher odds of readmission rates. Being males ($105), 25 to 44 years of age ($99), urban residents ($312), Medicare enrollees ($713 over privately insured), high-income area residents ($393), having multiple comorbidities ($923), and those admitted in large metropolitan ($680) and government-owned hospitals ($417) had higher costs of readmission. CONCLUSION: The high SPD-related readmission rates can be reduced by providing integrated behavioral health services for this high-risk cohort.
No takes yet. Share an insight, caveat, or question.
Wani et al. (2019) conducted a cohort in Psychiatric disorders (n=14,994). Locally derived LASSO prediction model vs. LACE index was evaluated on 30-day psychiatric readmission prediction model discrimination (AUC) (ΔAUC 0.092, 95% CI 0.623-0.678, p=<0.001). A locally derived LASSO prediction model for 30-day psychiatric readmission achieved an AUC of 0.649, significantly outperforming the LACE index in a United Arab Emirates cohort.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: