Comorbid Patterns in the Homeless Population: A Theoretical Model to Enhance Patient Care.

introductionFrom the perspective of social determinants, homelessness perpetuates poor health and creates barriers to effective chronic disease management, necessitating frequent use of emergency department (ED) services. In this study we developed a screening algorithm (checklist) from common comorbidities observed in the homeless population in the United States. The result was a theoretical screening tool (checklist) to aid healthcare workers in the ED, including residents, medical students, and other trainees, to provide more efficacious treatment and referrals for discharge.

methodsIn this retrospective cohort study we used the Nationwide Emergency Department Sample (NEDS) to investigate comorbidities and ED utilization patterns relating to 23 injury-related, psychiatric, and frequent chronic medical conditions in the US adult (≥18 years of age) homeless population. Cases were identified from the NEDS database for 2014-2017 using International Classification of Diseases, 9th and 10 revisions, and Clinical Classification Software diagnosis codes. We performed a two-step cluster analysis including pathologies with ≥10% prevalence in the sample to identify shared comorbidities. We then compared the clusters by sociodemographic and ED-related characteristics, including age, gender, primary payer, and patient disposition from the ED. Chi-square analysis was used to evaluate categorical variables (ie, gender, primary payer, patient disposition from the ED), and analysis of variance for continuous variables (age).

resultsThe study included 1,715,777 weighted cases. The two-step cluster analysis identified nine groups denominated by most prevalent disease: 1) healthy; 2) mixed psychiatric; 3) major depressive disorder (MDD); 4) psychosis; 5) addiction; 6) essential hypertension; 7) chronic obstructive pulmonary disease (COPD); 8) infectious disease; and (9) injury. The MDD, COPD, infectious disease, and Injury clusters demonstrated the highest prevalence of co-occurring disease, with the MDD cluster displaying the highest proportion of comorbidities. Although the addiction cluster existed independently, substance use was pervasive in all except the healthy cluster (prevalence 36-100%). We used the extracted screening algorithm to establish a screening tool (checklist) for ED healthcare workers, with physicians as the first point of contact for the initial use of the screening tool.

conclusionHealthcare workers in the ED, including residents, medical students, and other trainees, provide services for homeless ED users. Screening tools (checklists) can help coordinate care to improve treatment, referrals, and follow-up care to reduce hospital readmissions. The screening tool may expedite targeted interventions for homeless patients with commonly occurring patterns of disease.