Depression and Hospital Utilization in Patients with 30-Day Readmissions: Implications for Integrated Sdoh Screening

Background: Thirty-day hospital readmissions, emergency department (ED) visits, and length of stay (LOS) are healthcare quality metrics influenced by medical and non-medical factors. Social determinants of health (SDOH) are recognized as key non-medical drivers of health outcomes. The Centers for Medicare and Medicaid (CMS) require healthcare systems to conduct SDOH screenings for adult inpatient admissions. While the link between depression and chronic conditions is well-established, measures of mental health are often excluded from studies examining the impact of SDOH and healthcare quality metrics. This pilot study aims to (1) assess how including depression impacts ED and LOS and (2) test the feasibility of the protocol for a more extensive study.

Methods: A retrospective chart review study was used to analyze data from 220 randomly selected charts of patients with a 30-day hospital readmission from January 2024 through December 2024. The predictive variables of interest were SDOH domains required by CMS (e.g., food insecurity, housing instability, unreliable transportation, disconnected utility services, interpersonal violence). Age, gender, and race were also included. The outcome variables were LOS, measured as the total number of admitted hospital days in 2024, and the total number of ED visits in 2024.

Analysis: Descriptive statistics were used to assess differences in patient demographics and to assess the extent of missing data. Hierarchical regression analysis examined how the inclusion of depression influenced the prediction of length of stay (LOS) and emergency department visits (ED) across three models: (1) demographic factors (age, gender, race), (2) demographic factors combined with SDOH (e.g., housing instability, food insecurity, transportation), and (3) all aforementioned variables with the addition of depression. Utility assistance and interpersonal safety were excluded due to excessive missing data.

Results: Regression results indicate an overall model of seven the predictors that significantly predict the number of ED visits [R² = .13, F(7,148) = 3.12, p <.05]. This model accounted for 13% of the variance in ED visits. Depression b=.248,t(148) = 3.14, p = .002 and housing instability b= -.223,t(148) = -2.22, p = .028 contributed significantly to the model. Models excluding depression were not statistically significant. For LOS, the overall model of seven predictors accounted for 12% of the variance in LOS [R² = .12, F(7,151) = 3.01, p <.05]. Age b =.246,t(151) = 2.83, p = .005, race Age b=.174,t(151) = 2.14, p = .034, and depression b =.194,t(151) = 2.47, p = .015 contributed significantly to the model.

Implications and Conclusions: Depression was the only variable significantly impacting both LOS and ED visits across all models. This challenges the widespread practice of focusing on primarily on material hardship in SDOH screening in acute care settings. Results highlight the importance of equipping multidisciplinary team members responsible for discharge planning with the skills to address mental and physical health needs.

This exploratory study validates the feasibility of expanding the study to explore additional domains for a more holistic approach to SDOH that can be incorporated into existing workflows. Patterns of missing data will also be investigated further.