Self-administered home antibiotics matched other delivery models in a large UK cohort, with failure risk concentrated in older, multimorbid patients

A cohort study of 1,064 adult outpatient parenteral antimicrobial therapy (OPAT) episodes at a large UK teaching hospital, between January 2023 and January 2026, compared three ways of delivering the same treatment: self-administration by the patient (S-OPAT, 286 episodes, 26.9%), administration at home by a healthcare professional (H-OPAT) and clinic-based administration (C-OPAT) Primary Study (Durojaiye et al., PMID 42796749). Cure or clinical improvement occurred in 912 of 1,064 episodes (85.7%). Treatment failure occurred in 26 of 286 S-OPAT episodes (9.1%), with no significantly higher risk than H-OPAT (adjusted relative risk 0.72, 98.3% confidence interval 0.43–1.19) or C-OPAT (adjusted relative risk 0.94, 0.48–1.84). Unplanned hospitalisation and OPAT-related complications were broadly comparable across models, apart from potentially more vascular-access-related events with self-administration.

The finding that matters for service design is where failure clustered. Within S-OPAT, older age, higher comorbidity burden, diabetes, peripheral vascular disease and central venous access were each associated with treatment failure, and the authors suggest enhanced monitoring for those groups. This is a UK-sourced comparison of the three delivery models, and it supports self-administration for appropriately selected patients rather than for all.

It also partly tempers an earlier, non-UK result. A US retrospective cohort of 202 people with osteomyelitis of the foot, 77.7% of whom had diabetes, found self-administered OPAT associated with higher rehospitalisation (odds ratio 2.10) and lower 90-day wound healing (odds ratio 0.49) than healthcare-administered OPAT (Tarricone et al., PMID 42231791; discussed in the week ending 17 July synthesis). The two studies address different populations and infections, so neither overturns the other; the UK cohort shows the older, diabetic, multimorbid phenotype carrying the risk in a mixed OPAT caseload.

Three limits travel with the result. It is an observational cohort from one large teaching hospital’s service, so selection into each delivery model is not randomised. It did not analyse rurality, so whether the access benefit of self-administration outweighs the monitoring burden for dispersed rural patients remains open. And digital exclusion and carer capacity were not measured, so the study neither supports nor tests the published pattern that these, rather than clinical suitability alone, bind who benefits from home-based care.