Combining behavioral harm-reduction treatment and extended-release naltrexone for people experiencing homelessness and alcohol use disorder in the USA: a randomised clinical trial

The rate of alcohol-related mortality in people experiencing homelessness and alcohol use disorder is high and necessitates accessible and effective treatment for alcohol use disorder. However, typical abstinence-based treatments do not optimally engage this population. Recent studies have shown that harm-reduction treatment, which does not require abstinence, but instead aims to incrementally reduce alcohol-related harm and improve health-related quality of life, is acceptable to and effective for this population. The aim of this study was to test the efficacy of combined pharmacological and behavioural harm-reduction treatment for alcohol use disorder (HaRT-A) in people experiencing homelessness and alcohol use disorder.This randomised clinical trial was done at three community-based service sites (low-barrier shelters and housing programmes) in Seattle (WA, USA). Eligible participants were adults (aged 21-65 years) who met the DSM-IV-TR criteria for alcohol use disorder and who experienced homelessness in the past year. Participants were randomly assigned (1:1:1:1) by permuted block randomisation, stratified by site, to receive either HaRT-A plus intramuscular injections of 380 mg extended-release naltrexone (XR-NTX; HaRT-A plus XR-NTX group); HaRT-A plus placebo injection (HaRT-A plus placebo group); HaRT-A alone (HaRT-A alone group); or community-based supportive services as usual (services-as-usual control group). Patients assigned to receive HaRT-A attended sessions at baseline (week 0) and in weeks 1, 4, 8, and 12. XR-NTX and placebo injections were administered in weeks 0, 4, and 8. During the study, participants, interventionists, and investigators were masked to group assignment in the two injection arms. All participants were invited to follow-up assessments at weeks 4, 8, 12, 24, and 36. The primary outcomes were self-reported alcohol use quantity (ie, alcohol quantity consumed on peak drinking occasion, as measured with the Alcohol Quantity Use Assessment questionnaire) and frequency (measured with the Addiction Severity Index), alcohol-related harm (measured with the Short Inventory of Problems-2R questionnaire), and physical and mental health-related quality of life (measured with the Short Form-12 survey). Using piecewise growth modelling and an intention-to-treat model, we compared the effects of the three active treatment groups with the services-as-usual control group, and the HaRT-A plus XR-NTX group with the HaRT-A plus placebo group, over the 12-week treatment course and during the 24 weeks following treatment withdrawal. Safety analyses were done on an intention-to-treat basis. This trial is registered with ClinicalTrials.gov, NCT01932801.Between Oct 14, 2013, and Nov 30, 2017, 417 individuals experiencing homelessness and alcohol use disorder were screened, of whom 308 were eligible and randomly assigned to the HaRT-A plus XR-NTX group (n=74), the HaRT-A plus placebo group (n=78), the HaRT-A alone group (n=79), or the services-as-usual control group (n=77). Compared with the services-as-usual control group, the HaRT-A plus XR-NTX group showed significant improvements from baseline to 12 weeks post-treatment across four of the five primary outcomes: peak alcohol quantity (linear B -0·48 [95% CI -0·79 to -0·18] p=0·010; full model Cohen's d=-0·68), alcohol frequency (linear B -4·42 [-8·09 to -0·76], p=0·047; full model Cohen's d=-0·16), alcohol-related harm (linear B -2·22 [-3·39 to -1·06], p=0·002; full model Cohen's d=-0·56), and physical health-related quality of life (linear B 0·66 [0·23 to 1·10], p=0·012; full model Cohen's d=0·43). Compared with the services-as-usual control group, the HaRT-A plus placebo group showed significant improvements in three of the five primary outcomes: peak alcohol quantity (linear B -0·41 [95% CI -0·67 to -0·15] p=0·010; full model Cohen's d=-0·23), alcohol frequency (linear B -5·95 [-9·72 to -2·19], p=0·009; full model Cohen's d=-0·13), and physical health-related quality of life (linear B 0·53 [0·09 to 0·98], p=0·050; full model Cohen's d=0·35). Compared with the services-as-usual control group, the HaRT-A alone group showed significant improvements in two of the five primary outcomes: alcohol-related harm (linear B -1·58 [95% CI -2·73 to -0·42] p=0·025; full model Cohen's d=-0·40) and physical health-related quality of life (linear B 0·63 [0·18 to 1·07], p=0·020; full model Cohen's d=0·41). After treatment discontinuation at 12 weeks, the active treatment groups plateaued, whereas the services-as-usual group showed improvements. Thus, during the post-treatment period (weeks 12 to 36), the services-as-usual control group showed greater reductions in alcohol-related harm compared with both the HaRT-A plus XR-NTX group (linear B 0·96 [0·24 to 1·67], p=0·028; full model Cohen's d=0·24) and the HaRT-A alone group (linear B 1·02 [0·35 to 1·70], p=0·013; full model Cohen's d=0·26). During the post-treatment period, the services-as-usual control group significantly improved on mental health-related quality of life compared with the HaRT-A alone group (linear B -0·46 [-0·79 to -0·12], p=0·024; full model Cohen's d=-0·28), and on physical health-related quality of life compared with the HaRT-A plus XR-NTX group (linear B -0·42 [-0·67 to -0·17], p=0·006; full model Cohen's d=-0·27), the HaRT-A plus placebo group (linear B -0·42 [-0·69 to -0·15], p=0·009; full model Cohen's d=-0·27), and the HaRT-A alone group (linear B -0·47 [-0·72 to -0·22], p=0·002; full model Cohen's d=-0·31). For all other primary outcomes, there were no significant linear differences between the services-as-usual and active treatment groups. When comparing the HaRT-A plus placebo group with the HaRT-A plus XR-NTX group, there were no significant differences for any of the primary outcomes. Missing data analysis indicated that participants were more likely to drop out in the services-as-usual control group than in the active treatment groups; however, primary outcome findings were found to be robust to attrition. Participants in the HaRT-A plus XR-NTX, HaRT-A plus placebo, and HaRT-A alone groups were not more likely to experience adverse events than those in the services-as-usual control group.Compared with existing services, combined pharmacological and behavioural harm-reduction treatment resulted in decreased alcohol use and alcohol-related harm and improved physical health-related quality of life during the 12-week treatment period for people experiencing homelessness and alcohol use disorder. Although not as consistent, there were also positive findings for behavioural harm-reduction treatment alone. Considering the non-significant differences between participants receiving HaRT-A plus placebo and HaRT-A plus XR-NTX, the combined pharmacological and behavioural treatment effect cannot be attributed to XR-NTX alone. Future studies are needed to further investigate the relative contributions of the pharmacological and behavioural components of harm-reduction treatment for alcohol use disorder, and to ascertain whether a maintenance treatment approach could extend these positive outcome trajectories.National Institute on Alcohol Abuse and Alcoholism.