International comparison of loneliness trajectories and its determinants: ELSA and SHARE studies

Abstract

Background Homelessness is associated with many risks such as increased morbidity, mortality and increased use of health services. However, there are few long-term follow-up studies on homeless. The aim of the study was to evaluate the prognosis of homeless shelter users in terms of morbidity and housing outcomes and to identify factors that predict being independently housed 10 years later. Methods This is a case-controlled study of 552 homeless men (mean age 49 years, SD 11) that stayed in a shelter during 2004, in Helsinki, Finland. By linking data from several registers the morbidity and housing situation of the study group at the end of 2014 was compared with an age-matched control group from the general population (946 men). Using logistic regression analysis we assessed the predictive effects of socioeconomic factors and health service use at baseline on being independently housed ten years later. Results By end of follow-up 6.0% of the previously homeless were independently housed, compared to 81.9% in the control group. Over half of the study group (52%) died during the follow-up, 38% lived in supported housing and 5% were still or again homeless after ten years. The respective shares in the control group were 14.6%, 3.4% and 0.1%. 78% of the study group had some psychiatric diagnosis, most commonly substance use disorder, compared with 16% in the control group. In the multivariate logistic regression analysis on the previously homeless we found that being married (OR 8.9, 95% CI 3.5–22.5), having less than 4 shelter nights in year 2004 (OR 10.4, 95% CI 3.1–34.6), more than base level education (OR 2.3, 95% CI 1.1–5.0) and having no emergency room visits (OR 3.6, 95% CI 1.2–10.8) prior to baseline predicted independent housing 10 years later. Conclusions Shelter users have a high mortality and morbidity and a vast majority of those surviving are in need of support in their everyday life also in the long run.