Exploring Novel Medicare Readmission Risk Variables in Chronic Obstructive Pulmonary Disease Patients at High Risk of Readmission within 30 Days of Hospital Discharge.

RATIONALE: New risk factors for readmission within 30 days of hospital discharge for patients with chronic obstructive pulmonary disease (COPD) need to be identified in view of the lack of efficacy of current interventions for preventing readmission.
OBJECTIVES: To identify novel risk predictors for 30-day readmission among COPD index admissions at high risk of readmission.
METHODS: For this analysis, we used the fiscal year 2015 hospital-specific Medicare Hospital Readmissions Reduction Program workbook for Staten Island University Hospital (Staten Island, NY). We analyzed 41 variables, each with a risk-variable score. A predicted probability of readmission was calculated for each case by using the risk-variable regression coefficient and a hospital-specific effect. A predicted probability greater than 0.4 was used to identify patients with COPD with a high risk of readmission in both the readmitted and nonreadmitted groups. A percent ratio of the readmission percentage divided by the nonreadmission percentage was generated for each risk variable for patients with a predicted probability of readmission greater than 0.4. A percent ratio greater than 3 was used to identify high-risk variables predictive of readmission. A risk index was defined as the number of high-risk variables present for each index admission.
MEASUREMENTS AND MAIN RESULTS: Nine high-risk variables were identified. A risk index greater than 3 for all index admissions identified 54 (22.7%) of 238 readmitted patients versus 41 (6.5%) of 630 nonreadmitted patients (P < 0.0001; positive predictive value, 0.56; specificity, 0.93). A risk index greater than 2 for multiple-admission patients identified 56 (65.1%) of 86 readmitted patients versus 135 (40.7%) of 332 nonreadmitted patients (P < 0.0001; positive predictive value, 0.65; specificity, 0.86). Over 30% of readmitted patients meeting the risk index criteria were discharged to home without organized home care. Sleep apnea, vertebral fractures, and electrolyte and acid-base disorders were newly identified predictors of readmission.
CONCLUSIONS: This study developed a risk index based upon the 2015 Hospital Readmissions Reduction Program worksheet for one hospital to explore risk variables predictive of 30-day readmissions for patients with COPD at high risk of readmission (>0.4). Because most currently used interventions lack efficacy in preventing 30-day readmission, interventions based upon the newly identified variables should be validated with larger validation cohorts.