Introduction
Penicillin allergy records are often incorrect and may result in harm. We aimed to systematically review the effectiveness and safety of non-allergist healthcare worker delivery of penicillin allergy de-labelling (PADL).
Methods
We searched EMBASE/ MEDLINE/ CINAHL (Ovid), PsycInfo, Web of Science and Cochrane CENTRAL from inception to 21/01/22, and unpublished studies and the grey literature. The proportion of penicillin allergic patients de-labelled and harmed was calculated using random effects models.
Findings
Overall, 5019 patients were de-labelled. Using allergy history alone, 14% (95% CI, 9.0-21%) of 4350 assessed patients were de-labelled without reported harm. Direct drug provocation testing resulted in de-labelling 27%; (95% CI, 18-37%) of 4207 assessed patients. Of 1373 tested, 98% were de-labelled (95% CI, 97-99%), harm, none serious, was reported in 1% (95% CI, 0-2%). Using skin testing followed by drug provocation testing de-labelled 41% (95% CI, 24-59%) of 2890 assessed patients. Of 1294 tested patients 95.0% (95% CI, 90%-99%) were de-labelled, reported harm was low.(0%; (95% CI 0%-1%).
Interpretation
PADL by non-allergists is efficacious and safe. The proportion of assessed patients who can be de-labelled increases with complexity of testing method, but substantial numbers can be de-labelled without skin testing.
Funding
Source of funding: NIHR300542. Funder had no further role in the study.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)