Skin-cleaning among hospitalized people who inject drugs: a randomized controlled trial

To test the hypothesis that among hospitalized people who inject drugs (PWID), a brief intervention in skin-cleaning would result in greater reductions in follow-up emergency department (ED) or hospitalization rates compared with a usual care condition.

Randomized, two-group (intervention, n = 128; usual care, n = 124), single-site clinical trial with12-month follow-up.

Hospital inpatient services in Boston, Massachusetts, United States.

People who injected drugs on at least 3 days each week prior to hospital admission (n = 252). Participants averaged 37.9 (± 10.7) years of age; 58.5% were male, 59.3% were white and 61.1% had a diagnosis related to skin infection at enrollment.

Intervention was a skin hygiene education and skills-training behavioral intervention [short-term efficacy data on a behavioral intervention (SKIN)] consisting of two education- and skills-based skin-cleaning sessions, one during hospitalization and another 4 weeks later. The comparator was treatment as usual: an informational brochure about substance use treatment options and needle exchange programs in the area and follow-up clinical appointments as arranged by the inpatient medical staff.

Electronic medical records were reviewed and discharge diagnoses for each ED visit and hospital admission were categorized into injection-related bacterial events (e.g. cellulitis) and non-injection-related events. Negative binomial regression was used to test the intervention effects for the primary outcome and total ED visits, as well as the secondary outcomes, total number of hospitalizations, injection drug use-related (IDU-related) ED visits and IDU-related hospitalizations. We also tested whether the outcomes were moderated by whether the initial hospitalization was IDU-related.

Of people assigned to SKIN, 66 completed two sessions, 55 completed one session and seven completed zero sessions. Adjusting for baseline covariates, the mean rate of total ED visits in the next 12 months was non-significantly higher [incidence rate ratio (IRR) = 1.13, 95% confidence interval (CI) = 0.96, 1.33, P = 0.152] compared with usual treatment. The intervention did not significantly reduce total hospitalizations or IDU-related hospitalizations. Adjusting for baseline covariates, the mean rate of injection drug use-related ED visits in the next 12 months was lower (IRR = 0.57, 95% CI = 0.35, 0.91, P = 0.019) compared with treatment as usual.

A skin-cleaning intervention for people who inject drugs delivered during a hospitalization did not significantly reduce either overall emergency department use or hospitalization. There was some evidence that it may have reduced injection drug use-related emergency department visits.