Associations of mental health, substance use and housing instability with hospitalisation outcomes for adults with diabetes complications: a retrospective cohort study in Maryland, USA
Abstract
Abstract Introduction Mental health disorders (MH), substance use disorder (SUD) and housing instability (HI) are frequently co-occurring vulnerabilities that complicate diabetes self-management, contributing to healthcare utilisation. However, limited research has examined their combined associations with hospitalisation outcomes among patients admitted for diabetes-related complications. We evaluated the individual and joint associations of MH, SUD and HI with emergency department-initiated admission (ED-IA) and 30-day, 60-day and 90-day readmission. Research design and methods We conducted a retrospective cohort study using the Maryland statewide hospital discharge data repository (2016–2023). Patients aged ≥18 years with an index hospitalisation for diabetes complications were included. Multivariable logistic regression was used to estimate adjusted ORs (aORs) for the associations of MH, SUD and HI with ED-IA and readmission, adjusting for demographic, socioeconomic and clinical characteristics. Exposures were modelled individually and as composite categories to characterise associations for observed co-occurrence patterns. Results Among 48 308 patients, 45.2% had MH, 34.3% had SUD and 1.3% had HI. Readmission occurred in 4.8%, 7.8% and 10.3% at 30, 60 and 90 days; 87.9% of index admissions originated in the ED. MH was associated with ED-IA (aOR 1.31) and readmission (30-day aOR 1.61; 60-day aOR 1.56; 90-day aOR 1.53). SUD was not associated with ED-IA but was associated with readmission (aORs 1.51, 1.63 and 1.70). HI was associated with ED-IA (aOR 1.68) but not readmission. Patients with all three vulnerabilities had the highest odds of ED-IA (aOR 2.84) and readmission (aORs 2.93, 2.78 and 3.15). Conclusion MH, SUD and HI were independently and jointly associated with increased healthcare utilisation (ED-IA and readmission) among adults hospitalised for diabetes complications. These findings underscore the need for prospective research to evaluate whether integrated diabetes care models incorporating behavioural health interventions and housing support into comprehensive discharge planning reduce ED utilisation and hospital readmissions.