Neighborhood Socioeconomic Deprivation and 30-Day Mortality and Readmission for Patients Admitted for Diabetes Management Free

Prior research has suggested an association between low individual-level and area-level socioeconomic status (SES) and complications for patients with diabetes, such as diabetic retinopathy (1). However, prior studies have been limited by small sample sizes and geographically limited cohorts and the inability to adjust for confounding factors. Therefore, we performed this study using a cohort of Medicare beneficiaries in the U.S. to investigate the association between neighborhood SES and 30-day mortality and readmission for patients admitted for diabetes management after adjusting for age, sex, race/ethnicity, medical comorbidities, and individual-level SES.

We used a 100% sample of Medicare fee-for-service claims from 2017 to 2019. Patients admitted with diagnosis-related groups (referring to the classification system used by the Centers for Medicare and Medicaid Services to identify principal reasons for admission) 637–639, which capture patients admitted for both type 1 and type 2 diabetes, were included. Exclusion criteria included age <65 years for both analyses, as 65 is the age at which patients qualify for Medicare in the absence of other qualifying criteria, such as end-stage renal disease, transfer-in status (i.e., excluding patients who were transferred from one acute-care hospital to another) for the mortality analysis, and transfer to another hospital for the readmission analysis. Concordant with Centers for Medicare and Medicaid Services methodology, we selected a single random hospitalization per person for analysis (2). The exposure of interest was the area deprivation index (ADI), which was used to classify patients into high (ADI 1–15), middle (ADI 16–85), and low (ADI 86–100) neighborhood SES. In brief, the ADI is a validated measure that synthesizes census block group-level information about SES, including factors such as educational attainment, income, unemployment, access to infrastructure, etc. (3). We accounted for clustering of patients within hospitals by using generalized estimating equation methods to estimate logistic regression models with 30-day mortality and 30-day readmission as primary outcomes. Unadjusted models estimated the association between ADI groups and outcomes; adjusted models controlled for age, sex, race/ethnicity, Medicare-Medicaid dual-eligibility status (a proxy for individual SES), end-stage renal disease status (as end-stage renal disease automatically qualifies patients for Medicare), year of discharge, and the 29 Elixhauser comorbid medical conditions (4). These conditions are used to classify the comorbidity burden of patients and have been validated as predicting hospital resource use and in-hospital mortality. The Elixhauser comorbidity indices for mortality and readmission were used for summary statistics, while the individual conditions were used for adjustment in multivariable models (4).

Our study included 130,151 patients in the mortality cohort and 129,498 in the readmission cohort. In the mortality cohort, mean age was 76.4 years (SD 7.8), 48.9% were male, and the mean Elixhauser mortality index score was 19.5 (SD 14.5). Characteristics of the readmission cohort were similar. Compared with patients from the highest SES neighborhoods, patients from the lowest SES neighborhoods were more likely to be female (54.7% vs. 48.8%), were younger (mean age 75.5 vs. 77.5 years), and were more likely to be dually eligible for Medicare and Medicaid (44.2% vs. 32.0%). Median Elixhauser mortality index scores were similar for both groups (19.5 vs. 19.8).

Our univariate logistic regression model showed that patients from low SES neighborhoods did not have higher rates of 30-day mortality than patients from high SES neighborhoods, but they did have higher rates of 30-day readmission (Table 1). However, after multivariable adjustment, patients from low SES neighborhoods had 22% higher 30-day mortality and 11% higher 30-day readmission rates. The fact that high SES areas had older populations likely explains why the association between neighborhood SES and mortality increased after adjustment, given the strong association between age and mortality.

A variety of hypotheses may explain the association between neighborhood socioeconomic deprivation and outcomes. First, it is possible that access to key health care resources that can modify the course of disease progression, such as comprehensive primary and specialty care, vary by location and that the consequences of disparate access are reflected in readmission and mortality rates. Similarly, it is likely that access to other important community resources, such as healthy, affordable food and safe areas to exercise, is lower in areas of low SES and that this likewise impacts the disease trajectory for patients (5). Other environmental risk factors, such as air pollution, may likewise contribute to this phenomenon (5). Alternatively, it is possible that health care systems and providers have bias against patients from low-SES neighborhoods, contributing to worsened outcomes after admission.

Our study’s large sample size with nationally representative data, focus on patients admitted for management of diabetes (as opposed to evaluating longitudinal outcomes in the community setting), and thorough adjustment strategy advance the literature on socioeconomic disparities in outcomes for patients with diabetes. Limitations of our study include its restriction to Medicare beneficiaries, meaning that results may not be generalizable to patients below the age of 65 who are insured through different mechanisms or who are uninsured. In conclusion, neighborhood socioeconomic deprivation is associated with higher mortality and 30-day readmission rates among Medicare beneficiaries with diabetes.