Between hospice and harm reduction: mixed methods analysis of barriers to palliative care for people who inject drugs

People who inject drugs (PWID) experience a high burden of life-limiting illness and barriers to palliative care. Inpatient hospices may reduce these inequities, but little is known about their policies, practices, and preparedness to care for PWID. We examined Ontario hospices and considered how local institutional practices may function as micro-level drug policy. We conducted a convergent mixed-methods jurisdictional study of adult inpatient hospices in Ontario, Canada. Between June and October 2024, leaders from 33 of 53 eligible organizations (62%) completed an interviewer-administered structured survey and semi-structured interview. Survey data were summarized using frequencies and percentages descriptively. Interview data underwent team-based inductive descriptive thematic analysis, and the two strands were integrated during interpretation. Leaders from 29/33 hospices (88%) reported admitting people with substance use disorders, but 5/33 (15%) reported admitting PWID and 10/33 (30%) reported a formal written admission policy addressing substance use. No hospice reported a supervised or designated consumption space, and 2/33 (6%) reported availability of a safer supply of opioids. Seventeen (52%) reported a naloxone protocol, while responses to suspected opioid toxicity varied. Qualitative findings described an aspiration to provide inclusive care alongside gaps in policy and training; concerns about safety, space, and legality; uncertainty regarding toxicity response; logistical questions about drugs and equipment; and perceived reputational risk. Leadership reports suggest a substantial gap between inclusive hospice values and organizational preparedness to support PWID. Local admission criteria and bedside protocols may function as de facto drug policy by shaping access to end-of-life care. Hospice-specific guidance, staff education, individualized toxicity-response planning, and evaluation of harm-reduction-informed models are needed.