Predicting Risk Factors for COPD Hospital Readmission: A Big Data Analysis

Abstract

SESSION TITLE: COPD 3 SESSION TYPE: Original Investigation Poster PRESENTED ON: Wednesday, November 1, 2017 at 01:30 PM - 02:30 PM PURPOSE: The national data on COPD readmission rate is about 20%. Patients are usually admitted for acute exacerbation. Since the co-morbidity in this population is high, exacerbations of other diseases that arise from co-morbidities can also contribute to re-admission. The LACE hospital readmission tool and Charlson comorbidity index (CCI) are often used for predicting readmission although their reliability is unproven. Risk stratification tools suggest that CHF, CVD, COPD, and diabetes were strong predictors for high-risk readmission. The purpose of this study is to identify factors that potentially contribute to hospital readmission. The goal of this study is to provide evidence based strategies to decrease the number of avoidable admissions. METHODS: A large database on COPD patients over the year of 1/1/2014 to 12/31/2015 from a large university health care system in the Southeastern United States was used for this study. Factors including social-demographic, physiological factor (CCI score, first and last Creatinine, Potassium, Hemoglobin, WBC, INR, albumin, AST, ALT, Total bilirubin), medications (Coumadin, ACEI, beta-blockers, aspirin, statin, antibiotics, prednisone, nebulizer therapy, use of non-invasive ventilation (CPAP/BiPAP), and length of stay are identified variables for this study. Generalized multilevel modeling (binary response, logit-link function) was used to develop a predictive model for 30 day readmission adjusting for varying numbers of encounters per COPD patient. RESULTS: There were total of 6596 encounters/admissions during the data collection period with 4678 patients. 3657 patients only had one admission while 1021 patients had more than 1 admission. There were 11 patients who had more than 10 admissions. Only 6.2% met the criteria of 30 days readmission, which is, much less than 20% as reported by other studies. 12.7% of the patients had a diagnosis of COPD and CHF. Other demographic characteristics that were analyzed included gender (male 2375 (50.8%), female 2303 (49.2%), age (mean age: 67.5 +/- 12.8 years), race (63.4% White, 29.8% AA, 6.8% other). CCI (p=.079) and length of stay (p=.072) alone were not significant predictors for COPD readmission risk. The factors which significantly predicted a 30 day readmission were a lower initial hemoglobin level (p=.024), lower albumin (p=.048), treatment with prednisone (p=.004), patients who had a myocardial infarction (p=.043) and a CVD (p=.008). CONCLUSIONS: This university health care system is a biggest referral center in the area, which may explain that the 30 days readmission rates were lower than that of the national level. Comorbidities, especially associated with history of CVD and MI, can increase the risk for hospital readmission. CLINICAL IMPLICATIONS: Adequate nutritional support and prudent steroid management are possible additional interventions to reduce re-admissions rates, especially in the patients with the comorbidity of COPD and CHF. Further studies are needed to define predictive model for readmission and to determine if correction of anemia and nutrition can reduce re-admission in this cohort. This study further validated the need for combining cardiac and pulmonary rehabilitation programs in the community rather than treating both systems separately. DISCLOSURE: The following authors have nothing to disclose: Weihua Zhang, Melinda Higgins, Cherry Wongtrakool, Ruxana Sadikot No Product/Research Disclosure Information