Policies and practices addressing medication and substance use in recovery housing: findings from a nationally representative survey

Background: Harm reduction and recovery housing approaches are often thought to be in conflict, yet each share goals of reducing negative impacts of substance use.Objectives: Examining this, we surveyed substance use and medication policies and practices among a state-stratified, nationally representative sample of 838 residences (male = 479; female = 263; co-ed = 94; unknown = 2).Methods: Survey items assessed residence characteristics and practices as well as 12-step orientation and adherence to social model principles (valuing peer support, lived experience, and shared responsibility). We generated weighted estimates of policies and practices; logistic regression analyses tested differences by program orientation.Results: Nationally, nearly all (97%) residences require abstinence from alcohol and illegal drugs, implement drug testing (99%), evict residents who break substance use rules (94%), keep naloxone on site (98%), and have individuals trained to administer naloxone (94%). Fewer (56%) accept residents taking psychiatric medications. Primarily 12-step oriented residences (vs. not) were less likely to accept residents taking methadone (OR = 0.41, p = .003), psychiatric medications (OR = 0.52, p = .006), and medications for pain (OR = 0.37, p < .001) but more likely to keep naloxone on site (OR = 10.38, p = .010). Residences with greater social model adherence (vs. not) were more likely to accept residents taking medications for substance use (OR = 2.55, p = .015) and to provide naloxone training (OR = 2.55, p = .001).Conclusion: While largely abstinence-based, the majority of residences engage in harm reduction practices to safeguard against drug overdose. Efforts are needed to increase acceptance of residents on medications for substance use and other psychiatric conditions. Supporting social model principles may help in decreasing overdoses and increasing medication acceptance.