PT - JOURNAL ARTICLE AU - Jeffrey F. Mather AU - Gilbert J. Fortunato AU - Jenifer L. Ash AU - Michael J. Davis AU - Ajay Kumar TI - Prediction of Pneumonia 30-Day Readmissions: A Single-Center Attempt to Increase Model Performance. AID - 10.4187/respcare.02563 DP - 2013 Aug 13 TA - Respiratory Care PG - respcare.02563 4099 - http://rc.rcjournal.com/content/early/2013/08/13/respcare.02563.short 4100 - http://rc.rcjournal.com/content/early/2013/08/13/respcare.02563.full AB - Background: Existing models developed to predict 30 days readmissions for pneumonia lack discriminative ability. This study is an attempt to increase model performance with the addition of variables found of benefit in other studies. Methods: From 133,368 admissions to a tertiary-care hospital from Jan 2009 to March 2012, the study cohort consisted of 956 index admissions for pneumonia using the CMS definition. We collected variables previously reported to be associated with 30-day all-cause readmission, including vital signs, comorbidities, laboratory values, demographics, socio-economic indicators, and indicators of hospital utilization. Separate logistic regression models were developed to identify the predictors of all-cause hospital readmission 30 days after discharge from the index pneumonia admission for pneumonia-related readmissions, and for pneumonia-unrelated readmissions. Result: Of the 965 index admissions for pneumonia, 148 (15.5%) patients were readmitted within 30 days. Variables in the multivariate-model significantly associated with 30-day all cause readmission include male gender (odds ratio (OR)=1.59, 95% confidence interval(CI) = 1.03-2.45), 3 or more previous admissions (OR=1.84, 95% CI = 1.22-2.78), chronic lung disease (OR=1.63, 95% CI = 1.07-2.48), cancer (OR= 2.18, 95% CI = 1.24-3.84), median income <= $43,000 (OR=1.82, 95% CI = 1.18-2.81), history of anxiety or depression (OR=1.62, 95% CI = 1.04-2.52) and hematocrit < 30% (OR=1.86, 95% CI = 1.07-3.22). The model performance as measured by the c-statistic was 0.71 (0.66-0.75), with minimal optimism according to bootstrap re-sampling (optimism corrected c-statistic = 0.67). Conclusions: The addition of variables to measure socio-economic status (SES), health-care utilization significantly improved model performance when compared to the model using CMS variables alone.