Understanding Individual and Contextual Characteristics of Early Readmission in Patients with Heart Failure: A Multilevel Survival Approach and Spatial Analysis

Abstract

Background: Heart failure (HF) is a growing health concern and costly condition. The increasing medical and economic burden of hospitalizations for HF is a significant problem for patients, their families, and the US health care system. Excessive readmission to the hospital imposes a tremendous burden to patients with HF but also on the health care system. During the transition from hospital to home, some patients might be at higher risk of future episodes of decompensation and early (30-day) readmissions. There has been an extensive amount of research focused on understanding individual characteristics and clinical risk factors in predicting 30-day readmissions, such as increasing age, being male, severe illness, and more extended hospital stays. However, additional important factors, such as those at the patients’ contextual-level that may contribute to the high risk of readmission have not been included in risk prediction models. When considering patients’ recovery and transitions of care following hospitalization for HF, it is important to identify the role of the post-discharge environment in the readmission of HF patients. The post-discharge environment is defined as the interaction between individual, family, and environmental factors and how these factors relate to each other. Purpose: The overall purpose of this dissertation is to address the gap in the literature in understanding how contextual-level factors influence the likelihood of 30-day readmission in patients with HF. This dissertation comprises the following studies: Study 1 summarizes and critically analyzes the current evidence on risk prediction models that examine individual- and contextual-level risk factors associated with 30-day readmission in patients with HF based on a conceptual framework adapted from the Andersen’s Behavioral Model of Health Services Utilization (ABM); Study 2 categorizes and tests individual- and contextual-level factors associated with the likelihood of 30-day readmission among HF patients and measures the impact of neighborhood socioeconomic disadvantage (a contextual factor) on 30-day readmission for patients previously hospitalized with HF. Methods: Study 1 used a systematic scoping review methodological framework followed by a thematic analysis to assess, summarize, and interpret evidence on risk factors associated with 30-day readmission in patients with HF during the post-discharge vulnerable phase. For the purposes of this paper, the vulnerable phase is defined as the early post-discharge period during which patients with HF are more vulnerable to readmissions within a 30-day period following an HF hospitalization. The review used the ABM as a conceptual framework to select risk factors that may influence the post-discharge vulnerable phase after hospitalization for HF (the health outcome). Study 2 is a retrospective secondary data analysis of an existing HF dataset. This study was implemented based on the Strengthening Reporting of Observational Studies in Epidemiology criteria to improve observational research reporting. We used hierarchical linear modeling and a multilevel survival approach to model 30-day HF readmission risk as a function of fixed and random effects that combine individual- and contextual-level factors. In addition, the study used a spatial analysis technique to evaluate the effect of neighborhood socioeconomic disadvantage on? 30-day readmission among patients diagnosed with HF. Results: Study 1 results show that risk prediction models of 30-day readmission used risk factors related to the individual predisposing domain (such as demographics); few risk prediction models used risk factors related to the individual enabling domain (social support); and a majority of risk prediction models examined risk factors related to the individual needs domain (such as the presence of multiple comorbid conditions) to discriminate patients readmitted with HF within 30 days from those not readmitted. At the contextual level, very few risk prediction models included factors related to the health systems and environmental domains (such as patients residing in urban or rural areas or access to care) associated with 30-day readmission during the post-discharge vulnerable phase. Study 2 results show a variety of individual- and contextual-level risk factors related to 30-day readmission in patients with HF. With regard to individual-level risk factors, longer lengths of hospitalization, being in the surgical unit, and non-cardiac admissions were associated with a significantly shorter time to all-cause readmission (increased risk) during the post-discharge vulnerable phase. With regard to contextual-level risk factors, low household income ($24,999 annually on average) and households with only high school education had a significantly shorter time to all-cause readmission (increased risk) during the post-discharge vulnerable phase. Altogether, these findings indicate that the contribution of both individual-level and contextual-level risk factors simultaneously resulted in a better model fit to assess 30-day readmission risk or patients with HF. Lastly, those who lived in the most disadvantaged neighborhoods, as measured by the Area Deprivation Index (ADI), had a higher risk of all-cause readmission than their counterparts who lived in less disadvantaged neighborhoods. Conclusions: This dissertation adds to the growing literature on the contribution of both individual- and contextual-level risk factors on 30-day HF readmission. Living in a disadvantaged neighborhood brought a higher risk for 30-day readmission following hospitalization for HF. Thus, the findings of the scoping review and the testing of individual- and contextual-level risk factors associated with 30-day readmission highlighted the need to develop patient-centered health care interventions based on the patient’s social context to target the individual- (patient) and contextual-level (neighborhood) risk factors to reduce 30-day readmissions among patients with HF.