Analyzing 30-day readmission rate for heart failure using different predictive models

The Center for Medicare and Medical Services in the United States compares hospital's readmission performance to the facilities across the nation using a 30-day window from the hospital discharge. Heart Failure (HF) is one of the conditions included in the comparison, as it is the most frequent and the most expensive diagnosis for hospitalization. If risk stratification for readmission of HF patients could be carried out at the time of discharge from the index hospitalization, corresponding appropriate post-discharge interventions could be arranged. We, therefore, sought to compare two different risk prediction models using 48 clinical predictors from electronic health records data of 1037 HF patients from one hospital. We used logistic regression and random forest as methods of analyses and found that logistic regression with bagging approach produced better predictive results (C-Statistics: 0.65) when compared to random forest (C-Statistics: 0.61).