The effectiveness of psychosocial interventions for reducing problematic substance use, mental ill health, and housing instability in people experiencing homelessness in high income countries: A systematic review and meta‐analysis

Abstract Background Adults experiencing homelessness in high income countries often also face issues of problematic substance use, mental ill health, in addition to housing instability, so it is important to understand what interventions might help address these issues. While there is growing evidence of the effectiveness of psychosocial interventions for the general population, limited evidence exists specifically for those experiencing homelessness. Objectives To summarise the existing evidence of whether psychosocial interventions work in reducing problematic substance use, mental ill health, and housing instability for adults experiencing homelessness in high income countries. Search Methods We used searches undertaken for the Homelessness Effectiveness Evidence and Gap Map (EGM) 5th edition. These were supplemented with hand searches of key journals and a call for evidence. Selection Criteria We included all Randomised Control Trials and non‐randomised studies where a comparison group was used and which examined psychosocial interventiONS for adults experiencing homelessness. ‘Psychosocial intervention’ is a broad term and covers several interventions, including cognitive behavioural therapy (CBT), contingency management, and motivational interviewing. We focused on studies that measure at least one of three outcomes: reduction in problematic substance use (alcohol and/or drugs); reduction in mental ill‐health; reduction in housing instability. Data Collection and Analysis For included studies sourced from the EGM, we used the risk of bias assessments reported in the EGM. For included studies sourced from our own searches, we used the same tools used in the EGM to undertake our own assessments. We carried out meta‐analysis where possible, and where not possible, presented included studies narratively. Findings We included 26 papers covering 23 individual intervention studies. All of the included studies were from the United States. Of the 26 papers, 14 were assessed as having medium or high risk of bias, with main issues being lack of masking/blinding, lack of power calculations, and high levels of drop‐out. Effectiveness of Psychosocial Interventions We found that psychosocial interventions overall were better than standard care (−0.25 SD, 95% confidence intervals [CI] [−0.36, −0.13]). This finding covered six different interventions and was subject to a high level of between‐study differences (heterogeneity). We also found that psychosocial interventions were more effective than standard care in relation to all three of our outcomes of interest, although were statistically significant only for substance abuse and mental ill‐health. For substance use, we found an average effect size of (−0.34 SD, 95% CI [−0.48, −0.21]); for mental ill health of (−0.18 SD, 95% CI [−0.34, −0.01]); and for housing instability of (−0.10 SD, 95% [−0.90, 0.70]). Effectiveness of Individual Psychosocial Interventions We were able to undertake five meta‐analyses (statistical summaries) with respect to four types of intervention: CBT, Contingency Management, Motivational Interviewing, and Brief Motivational Interventions, in relation to specific outcomes. Of these five analyses, we found significant effects for the effectiveness of Contingency Management in reducing problematic substance use (−0.49 SD, 95% CI [−0.85, −0.14]), and of Motivational Interviewing in reducing mental ill‐health (−0.19 SD, 95% CI [−0.26, −0.12]). We also found non‐significant effects in relation to CBT and reducing mental ill health (−0.30 SD, 95% CI [−0.61, 0.002]), Motivational Interviewing and reducing problematic substance use (−0.27 SD, 95% CI [−0.56, 0.01]), and Brief Motivational Interventions and reducing problematic substance use (−0.24 SD, 95% CI [−0.61, 0.13]). Meta‐analysis was not possible for any other interventions or outcomes. Author Conclusions This systematic review sought to understand the effectiveness of psychosocial interventions for adults in high income countries experiencing homelessness, for reducing problematic substance use, reducing mental ill‐health, and increasing housing stability. The review shows potential benefits of these interventions, with some encouraging results for some interventions and outcomes. Where we could calculate effect sizes, these were often small and, in many cases, crossed the line of no effect (i.e., there is a chance that they are equally or less effective than treatment as usual). Significant heterogeneity between studies and high rates of drop‐out in many studies reduces the confidence in the interventions. There are some limitations with the evidence base. The included studies were entirely from the United States. There was a clear gender bias in the included studies, with nearly two‐thirds of participants being men. (This is despite 4 of the 26 included studies focusing on women only.) We also found that the theoretical basis for the approach of interventions was not sufficiently considered, so it was difficult to understand why the intervention expected the outcomes they measured. Finally, many of the studies included were assessed as having high or medium risk of bias.