Predictive modeling pilot project for readmissions in heart failure patients with preserved ejection fraction

Access full-text files




Xiang, Pin

Journal Title

Journal ISSN

Volume Title



Objectives: To pilot a predictive model evaluating hospital readmissions for heart failure with preserved ejection fraction (HFpEF) patients and the association with management by a cardiologist, number of comorbidities, and type of treatment. Methods: This is a retrospective, observational study of claims data to evaluate the effect of various factors: age, gender, provider, baseline inpatient admissions, comorbidities and baseline drug treatment classes (e.g. antiarrhythmic, beta blocker, calcium channel blocker, diuretic, RAAS-inhibiting agents) on number of readmissions, time to readmission, and odds of readmission. Patients ≥18 years of age with an inpatient admission with a primary discharge diagnosis of HFpEF between October 1, 2011 and September 30, 2014 were identified and data were assessed 1-year pre- and post-hospitalization. Patient characteristics were described, and patients treated by a cardiologist were compared to those who were not. Multivariate regression and Cox proportional hazard models were used to assess the association of all-cause and heart failure-related readmissions adjusting for demographic and clinical covariates. Results: A total of 264 patients with HFpEF were identified (60.2% female; mean age of 79 years (SD 10.8) of which 77 [29%] did not see a cardiologist. Patients who saw a cardiologist were more likely to be male and had a greater number of comorbidities including diabetes, dyslipidemia, hypertension, coronary heart disease, cardiomyopathy, and valvular heart disease than those without cardiologist. Overall, 51% of the patients had an all-cause readmission and 15% had an HF-related readmission. Patients who had a cardiologist were associated with more all-cause readmissions (IRR of 2.21, p=0.0003) and a shorter time to all-cause readmissions (HR of 1.91, p=0.004). Being on diuretics was associated with more heart failure-related readmissions (IRR of 2.84, p=0.0301). A higher number of all-cause readmission was associated with patients having more comorbidities (IRR of 1.19, p=0.0038). Conclusion: This study demonstrated that all-cause and heart failure-related readmission is high in patients with HFpEF. The pilot predictive models show that various factors associated with higher risk patients, such as those with cardiologist management, more comorbidities, and use of diuretics, may be associated with increased hospital readmissions.


LCSH Subject Headings