An integrated model of functional status and socioeconomic factors affecting hospital length of stay and 30-day readmission risk in individuals with heart failure

Background: Acute management of heart failure (HF) is a high-cost enterprise. Cost-effective management of acute HF hinges upon hospital outcomes such as length of stay (LOS) and 30-day readmission (30dRA) rate. Literature has produced regression models that predict these hospital outcomes; however, existing models use a limited scope of traditional medical predictors resulting in weak predictive ability. Functional status (FS) and socioeconomic factors (SEFs) have been found to predict various hospital outcomes in patients with HF; however, existing model performance is limited in its scope. Therefore, a modernized, holistic approach through the integration of FS and SEFs into existing medical predictor based regression models may better predict LOS and 30dRA rate in HF. Purpose: To determine the predictive utility of a model utilizing FS, SEFs, and traditional medical variables on hospital LOS and 30-day re-admission rate (30dRA) in individuals with HF. Methods: Secondary data for 2016 to 2020 calendar years was gathered from a Trauma Level I, safety-net hospital. Hospital admissions with a primary diagnosis of HF were included. Subjects under 18 years old at admission and death during hospitalization were excluded. A total of 2204 medical records were analyzed using hierarchical linear regression on log-transformed LOS data and 1953 records were analyzed using logistic regression on 30dRA data. Results: A LOS model utilizing FS, SEFs, and traditional medical factors was found to be significant (r2 = .207, adjusted r2 = .204, F(8, 2195) = 71.579, p < .001). A 30dRA model utilizing SEFs and traditional medical factors was found to be significant (χ2(10) = 43.185, df = 10, p < .001). Within the 30dRA model, FS was found to be not statistically significant (OR: .996, 95% CI [.985 to 1.007], p = .449). Conclusion: A model utilizing FS, SEFs, and traditional medical factors can predict hospital LOS. FS appears to be less contributory to a 30dRA model compared to SEFs and traditional medical factors.