Date of Award


Document Type


Degree Name

MA in Psychology



First Advisor

Elizabeth Brondolo

Second Advisor

Wilson McDermut


Healthcare service utilization (HSU) describes how individuals engage with healthcare systems. Studies examining differences in rates of HSU among acute care-seeking patients have identified disparities in access to appropriate treatments. Alcohol-use disorders (AUDS) are increasingly prevalent among patients presenting for treatment in acute care settings. AUDS are emblematic of a broader trend in acute care; disproportionate rates of acute care encounters by patients with heavy socioeconomic (e.g., socioeconomic status, homelessness) burden, primary psychiatric/behavioral disorders (e.g. schizophrenia, AUDs), and reliance on emergency rooms for seeking treatment. Numerous studies have linked these risk factors with both AUDs and with high HSU. Despite this, few studies have examined HSU specifically within this patient population.

Given these findings in the literature, we expected to find that factors related to patients’ sociodemographic characteristics and emergency room use would better predict future HSU than patients’ primary diagnosis, levels of disease burden, or comorbidities. A binary classification algorithm was used to model the impact of access and need on future HSU. A widely-used metric for efficacy of HSU, 30 Day readmissions, was treated as the outcome..

The classification model indicated that the strongest predictor of a patient’s future HSU was their past HSU. Patients with 3 or more past year ED visits had over 9 times greater higher odds of readmission within 30 days (point estimate=2.285; 95% CI: 1.99, 2.42; p<.0001) compared with patients with no past year ED use. This effect was linear; e.g. Patients with 2 past year ED admissions were also significantly more likely to return within 30 days compared to patients with no past year ED visits (point estimate=0.915, 95% CI: 0.645, 1.19; OR: 2.5, p<.00001). Other risk factors, such as smoking history, also conferred significantly higher odds of 30 day readmission. Patients with documented smoking histories were over 2 times more likely to return within 30 days (point estimate=0.716; 95% CI: 0.51, 0.92; p<.00001) compared with nonsmokers.