Determinants of healthcare use by homeless people with schizophrenia or bipolar disorder: results from the French Housing First Study.

Objectives: There is limited evidence available on the health-seeking behaviours of individuals in relation to determinants of healthcare use. This study aimed to analyse the determinants of healthcare use (including both hospital and outpatient services) among homeless people with severe mental health illnesses.

Study Design: The study used data from a multicentre, randomised, controlled trial conducted in four large French cities (the French Housing First Study).

Methods: Data were drawn from 671 homeless people enrolled in the study between August 2011 and April 2014. Mobile mental health outreach teams recruited homeless individuals with severe mental health illnesses who were living on the street or in emergency shelters, hospitals or prisons. Data collection was performed during face-to-face interviews. Healthcare service use included hospitalisations, mental health and regular emergency department (ED) visits and outpatient visits to healthcare facilities or physicians' offices over a 6-month follow-up period. The data were analysed with zero-inflated (ZI) two-part models.

Results: In total, 61.1% of participants had at least one hospitalisation stay over the previous 6 months, with a mean of 25 (+/- 39.2) hospital days, and the majority (51%) had visited the ED (either for regular or mental health issues) during the same time period. The results confirmed the role of financial barriers (resources and health insurance) in seeking hospital care (P < 0.05). The main predictors for hospital use in the study population were a better social functioning score (odds ratio [OR]: 1.03; P < 0.001) and having schizophrenia (OR: 1.39; P < 0.01). Higher mental health scores (assessed by the Medical Outcomes Study 36-item Short Form Health Survey) (OR: 1.03, P < 0.01) and alcohol dependence (OR: 2.13; P < 0.01) were associated with not using ED healthcare services. Being 'absolutely homeless' predicted an increased use of the ED and a zero use of outpatient services. Inversely, no association with factors related to the homelessness trajectory was found in hospital ZI negative binomial models.

Conclusion: This study is important because a comprehensive understanding of the determinants of healthcare use enables healthcare systems to adapt and develop. The efficiency of medicosocial interventions targeting the homeless population with mental health illnesses must also be assessed.

Clinical Trial Number: NCT01570712.