Out-of-home care placement, psychiatric disorders, and risk of homelessness: a population-based cohort study

Abstract

Background: Family adversity and childhood abuse have been associated with an increased risk of homelessness in adulthood. However, to our knowledge, no population-based longitudinal studies have examined the association between out-of-home placement and subsequent homelessness, while accounting for psychiatric disorders and parental background. Thus, we aimed to investigate the association between out-of-home placement and psychiatric disorders during childhood or adolescence and homelessness later in life.

Methods: We used data from a nationwide, register-based birth cohort in Denmark from 1 January 1983 to 31 December 2003, who were alive at any point from 1 January 2001 to 31 December 2021. Children and adolescents (aged 0-17) were followed up from their 18th birthday for risk of homelessness. We investigated associations using national registers with information on homelessness and psychiatric disorders. We examined individuals with any out-of-home placement history with and without psychiatric disorders diagnosed up to age 18 according to the first episode of homelessness after age 18, defined as first contact with a shelter. We estimated the cumulative incidence function using an Aalen-Johansen estimator. Hazard ratios (HRs) for the association between out-of-home placement, psychiatric disorder, and homelessness were estimated by sex using Cox proportional-hazard regression analyses, adjusted for birth year, parental sociodemographic factors and psychiatric disorders. Additionally, we employed a stratified Cox model (stratified by mother) to examine the contribution of within-family covariates.

Findings: The study cohort included 1,226,130 Danish residents aged 18-39 years. Of these, 53,281 individuals (4·3%) had experienced out-of-home placement during childhood or adolescence. By age 39, 12·2% (95% CI 11·6-12·7) of males and 6·3% (95% CI 5·9-6·7) of females with a history of out-of-home placement experienced homelessness. This corresponds to a substantially increased risk compared to people without such a history, with adjusted HRs of 7·1 (95% CI 6·7-7·5) for males and 9·0 (8·2-9·8) for females. In sibling analyses restricted to 955,287 individuals and adjusted for birth year, out-of-home placement was associated with a 3-fold increased risk of homelessness compared to siblings without such a history (aHR 2·8 (95% CI 2·4-3·3) for males and aHR 3·2 (2·4-4·1) for females). Homelessness risk was even higher for individuals with a history of both out-of-home placement and psychiatric disorder, particularly when psychiatric diagnosis occurred after the first placement. In the full cohort, adjusted HRs were 11·8 (95% CI 10·8-13·0) for males and 19·3 (16·9-21·9) for females, compared to individuals with neither out-of-home placement nor psychiatric disorder.

Interpretation: Individuals with a history of out-of-home placement are at higher risk of experiencing homelessness, particularly those with co-occurring psychiatric disorders. These findings highlight the need for targeted interventions and coordinated efforts across medical and social services to prevent homelessness.

Funding: Funded by a grant from the Lundbeck Foundation to SFN, F-61171-23-50 and the Independent Research Fund to SFN 10.46540/3162-00051B.

Keywords: Epidemiology; Foster home care; Homelessness; Ill-housed persons; Out-of-home placement; Psychiatry; Public health.

Fig. 1

Flowchart. This flowchart illustrates the study cohort identified from the Danish registers. It also shows the derivation of the sub-cohort of individuals linked to their mother, which was used for sibling analyses. Additionally, it shows the cohort with links to both mother and father, used for a sensitivity analysis.

Fig. 2

Probability of first homelessness from age 18 in people with and without out-of-home placement and psychiatric disorder compared with people with neither out-of-home placement nor psychiatric disorders. Abbreviation: psych, any psychiatric disorder diagnosed before age 18. Exposed is defined as any out-of-home placement before the age of 18. The probability of first-time homelessness was analysed across groups categorised by their experiences with out-of-home placement and psychiatric disorders diagnosed before age 18. These groups were compared to individuals with neither out-of-home placement nor psychiatric diagnoses before age 18. The highest probability of first-time homelessness was observed in the following order: 1. Individuals diagnosed with a psychiatric disorder after their first out-of-home placement (“Exposed, psych”). 2. Individuals diagnosed with a psychiatric disorder before their first out-of-home placement (“Psych, exposed”). 3. Individuals with an out-of-home placement but no psychiatric disorder before age 18 (“Exposed”). 4. Individuals with a psychiatric disorder diagnosed before age 18 but no out-of-home placement. 5. Individuals with neither out-of-home placement nor psychiatric disorder before age 18 (Neither). Competing risks from emigration and death were accounted for using Aalen-Johansen estimator. The shaded areas around the lines illustrate 95% confidence intervals (CIs). The curves are smoothed to ensure that no individuals could be identified.

Fig. 3

Association of out-of-home placement and homelessness in the full sample and in the maternal sibling sample, stratified by sex. Analyses of first episode of homelessness in the full sample for males (a) and females (b) with and without out-of-home placement compared to individuals with no out-of-home placement before age 18. Model 1 is adjusted for birth year. Model 2 is additionally adjusted for maternal and paternal age at birth, maternal and paternal psychiatric disorder diagnosed prior to the cohort member's birth, maternal and paternal highest educational level at the cohort members 15th birthday, and unknown father. Analyses in the maternal sibling sample (c and d) are adjusted for birth year. Figures are presented on a logarithmic scale with 95% confidence intervals (CIs).

Fig. 4

Association of out-of-home placement, psychiatric disorder, and homelessness in the full sample and in the maternal sibling sample, stratified by sex. Abbreviation: OOHP = out-of-home placement. Analyses in the full sample in males (a) and females (b) with and without out-of-home placement and psychiatric disorder diagnosed before age 18 compared to individuals with neither out-of-home placement nor psychiatric disorder before age 18. Model 1 is adjusted for birth year. Model 2 is additionally adjusted for maternal and paternal age at birth, maternal and paternal psychiatric disorder diagnosed prior to the cohort member's birth, maternal and paternal highest educational level at the cohort members 15th birthday, and unknown father. Analyses in the maternal sibling sample in males (c) and females (d) are adjusted for birth year. Figures are shown on a logarithmic scale with 95% confidence intervals (CIs).