Psychiatric disorders and mortality among people in homeless shelters in Denmark: a nationwide register-based cohort study

Summary

Background

The increased mortality of homeless people compared with non-homeless people might be linked to psychiatric disorders. However, homeless people are, because of their insufficient accommodation, difficult to sample and monitor, which has limited previous studies. We aimed to assess registered psychiatric disorders, mortality, and predictors of mortality in the homeless shelter population in Denmark.

Methods

We did a nationwide, prospective, register-based cohort study of homeless people aged 16 years and older who were registered in the Danish Homeless Register between Jan 1, 1999, and Dec 31, 2009. We calculated the proportion of registered psychiatric disorders, overall and cause-specific standardised mortality ratio (SMR), and life expectancy. Hazard ratios (HRs) were used to assess predictors of death.

Findings

32 711 homeless people (23 040 men and 9671 women) were included in the study population. 14 381 men (62·4%) and 5632 women (58·2%) had registered psychiatric disorders, and 11 286 men (49·0%) and 3564 women (36·9%) had a substance abuse diagnosis. During the study period, 3839 men (16·7%) and 951 women (9·8%) died. The overall SMR for men was 5·6 (95% CI 5·4–5·8) and for women was 6·7 (6·2–7·1), and external causes accounted for 1161 (27·9%) of 4161 deaths for which information on the cause was available. Remaining life expectancy at age 15–24 years was 21·6 years (95% CI 21·2–22·1) and 17·4 years (16·4–18·5) lower for homeless men and women, respectively, than the general population. Registered substance abuse disorder was associated with the highest mortality risk compared with no psychiatric contact registered (adjusted HR 1·4, 95% CI 1·3–1·5 for men; 1·7, 1·4–2·1 for women).

Interpretation

Health problems are extensive in the homeless shelter population and there is an urgent need for more sustained efforts to reduce the high morbidity and mortality, especially from external causes. Register data is an important resource to supplement existing knowledge on homeless people with more valid and detailed information.

Funding

The Danish Council for Independent Research.

Introduction

An increased mortality among homeless people compared with the general population in developed countries has been described.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17 Additionally, psychiatric morbidity in the homeless population is higher than in the general population in developed countries.18, 19, 20 Previous studies have linked mental illness, such as psychosis and schizophrenia, to lower mortality in homeless people.1, 3, 9, 11, 16 Although these findings only showed weak evidence of a difference in mortality, they do contrast with findings in the general population.²¹ Different types of substance abuse and dual diagnosis (a schizophrenia spectrum disorder and a substance abuse disorder) were identified as predictors of excess mortality among homeless people.1, 3, 4, 5, 12, 14, 16 The need for more prospective studies that investigate the factors associated with the excess mortality has been emphasised.9, 11, 12

Homeless people are a transient population and, because of their insufficient accommodation, are difficult to sample and monitor. As such, studies on this population have generally had small and selected sample sizes, which do not ensure representative findings. In Denmark, information on every contact with homeless shelters is contained in the Danish Homeless Register, which provides a new and unique opportunity to undertake detailed research of an unselected homeless population.

We aimed to estimate the proportion of registered psychiatric disorders among men and women in homeless shelters in Denmark, to calculate the overall and cause-specific standardised mortality ratio (SMR; the ratio between observed mortality in the studied population and the expected mortality) and life expectancy, and to identify predictors for mortality. We hypothesised that psychiatric disorders, especially substance abuse, are highly prevalent among homeless people.²⁰ Additionally, we expected that homeless people would have an increased SMR and a reduced life expectancy compared with the general population.1, 15 We also expected that substance abuse would be a strong predictor of death, and more harmful to health than both no psychiatric disorders and schizophrenia spectrum disorders.1, 3, 11 Finally, we hypothesised that the number of contacts with homeless shelters within the first year of contact would be a predictor of death, with increased risk associated with an increasing number of contacts.¹