To determine (a) the prevalence of patients supposedly allergic to penicillin who have a positive radioallergosorbent test (RAST) result for penicillin G or V and (b) the predictive power of family physicians' clinical judgement that a patient who is supposedly allergic to penicillin will have a positive RAST result.Design
Prospective multicentre cross-sectional observational study.Setting
Eleven primary care practices in Newfoundland; 10 were in a rural setting.Patients
Of 110 consecutive adult patients with a supposed allergy to penicillin 97 agreed to participate in the study; 92 underwent RAST.Interventions
Patients helped physicians complete a questionnaire and had a venous blood sample taken for the RAST. Physicians examined the clinical history and judged whether the patient was likely to have a positive RAST result.Mean outcome measures
Rates of positive and negative RAST results for penicillin V and G.Results
Of the 92 patients 8 had a positive RAST result and 84 a negative one. The positive predictive power of a "good" clinical history (e.g., urticaria, swollen eyes, tongue or lips, or an anaphylactic reaction witnessed by a physician) was low (10%); the negative predictive power of a "poor" clinical history (e.g., nausea, vomiting, diarrhea, fever, nonspecific rash or fainting) was 92%.Conclusions
Less than 10% of primary care patients with a supposed allergy to penicillin will have a positive RAST result. In addition, physicians' predictions of allergy in such patients are imprecise.