Statistical analysis

Demographic characteristics were compared using χ² tests for categorical variables. To examine patient, hospital, and clinical characteristics associated with readmission, we built multivariable logistic regression models to estimate the odds of readmission for each characteristic. Model variables were initially selected a priori based on clinical knowledge and review of existing literature and variable availability in NRD data. All estimates were generated using survey-weighted methods to account for the complex survey design of the NRD and generate national estimates for all variables. In particular, we used the full cohort to maintain all strata and yield proper variance estimates as intended by HCUP. The statistical analyses for our study were performed using SAS software, version 9.4 (SAS Institute, Inc., Cary, NC).