Searching for underlying social determinants of health for thirty-day hospital readmissions

Objective: An evaluation of social factors associated with 30-day readmission was undertaken at our institution to determine which factors would be significantly associated with time to hospital readmission.Methods: Prospective observational study at an academic tertiary care hospital in the mid-Atlantic region of patients who were readmitted within 30 days of their last inpatient discharge. The electronic health record in conjunction with the regional hospital information system was used to generate a daily report to identify a convenience sample of readmitted patients. Using a standardized interview, data on 117 patients were collected for an exploratory analysis of social factors associated with readmission.Results: Regression modeling demonstrated poor correlation with prediction of time to readmission (R-squared = 0.2189). No individual social variables were found to be significant for influencing time to readmission (all p-values > .05). Common social factors were seen within the population affecting their utilization and access of healthcare. Poly-pharmacy was found in the majority of patients. Self-reported medication adherence was good, except with regards to mental health medication compliance. 97% of patients reported filling their prescriptions. 36% of the patients went to their follow-up appointment within 7 days although the vast majority of patients (92%) reported having a primary care doctor. 23% of patients expressed difficulty getting to their follow up appointments.Conclusions: At one single-center tertiary care hospital, there were some common underlying social determinants of health that may be related to readmission; however, no factors in isolation were predictive of hospital readmission. While there are common themes among readmitted populations, particularly in regard to factors driven by poverty, it is likely that the complex interaction of social factors with health continues to limit attempted administrative modeling of these data.