Primary Care Corner with Geoffrey Modest MD: Penicillin Allergy???

By Dr. Geoffrey Modest

A large concern in treating patients with infections is the very high prevalence of “penicillin allergy”, leading to the use of broad-spectrum antibiotics as well as 2nd or 3rd line medications, which are usually more toxic, along with their attendant effects on antimicrobial resistance as well as secondary infections such as C. difficile­­­­. A recent article looked at 2 methodologies to determine the safety of using beta-lactams in these “penicillin allergic” patients (see 10.1016/j.jaci.2017.02.005).


  • Of 1000 medicine in-patients with a noted penicillin allergy in a single Boston hospital, 625 were admitted with a presumed infection: mean age 66, 60% female, 70% white/16% black, reported penicillin allergy was rash or hives in 60%/angioedema 15%, anaphylaxis 8%
  • Patients were assessed during 3 different time periods: 148 patients in a standard-of-care group (SOC), 278 in a penicillin skin testing group (ST), and 199 in a group using a computerized guideline-based management app (APP) to predict real allergy
  • ST group: excluded patients with penicillin intolerance (such as GI upset), patients taking medications that might interfere with skin testing (such as antihistamines), and also patients with multiple beta-lactam allergies, penicillin anaphylaxis in the last 5 years, or type II-IV hypersensitivity reactions to penicillin
  • APP group: the clinical support basically divided people into low risk (benign delayed maculopapular rash); medium-to high-risk (urticaria, angioedema, anaphylaxis, recent or severe delayed maculopapular rash; and those who should avoid a beta-lactam (history of Stevens-Johnson syndrome, toxic epidermal necrolysis or exfoliative dermatitis; DRESS syndrome or acute interstitial nephritis; or serum sickness-like reaction)
  • Primary outcome was the actual use of penicillin or cephalosporin during the hospitalization


  • ST group: 179 (64%) were felt to be skin test eligible, but only 43 (24%) actually receive skin testing and none of those were allergic, defined as negative skin test and tolerance of an oral amoxicillin 250 mg test dose. As compared to the SOC group,
    • Nonsignificant 30% increased odds of use of penicillin or cephalosporin overall, adjusted OR 1.3 (0.8-2.0), but a highly-significant 5.7-fold increased use in a per protocol analysis, adjusted OR 5.7 (2.6-12.5), p<0.001 [the per protocol analysis limited the analysis to those few who actually got the skin test]
    • Of the ST per protocol patients, there was increased odds of penicillin or cephalosporin prescriptions for discharge treatment, with OR 2.5 (1.04-6.2)
  • APP group: 292 unique website views (averaging 26 seconds only), 112 users (38%) completed clinical decision support. Patients in the low or moderate-to-high risk groups as above were given test doses of beta-lactam antibiotics with an initial dose of 1/10 of an IV dose or 1/4 of an oral dose. The 2nddose was administered 30 minutes later, comprising the remainder of the therapeutic dose. Nurses assessed patients every 30 minutes for the duration of the challenge. As compared to the SOC group,
    • Significant 80% increased odds of use of penicillin or cephalosporin, adjusted OR 1.8 (1.1-2.9), p=0.03


  • Penicillin allergy is remarkably common, up to 15% of all inpatients are recorded as having a penicillin allergy, and 5-25% of inpatients who are treated for infections. Three quarters of patients with an alleged penicillin allergy would otherwise use a beta-lactam antibiotic in other studies, but in the SOC group only half of them received one.
  • Not using a beta-lactam antibiotic when that would otherwise be indicated leads to more treatment failures, adverse events, and antibiotic resistant organisms such as methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus.
  • There were few patients who actually had skin testing, mostly because of difficulty in coordinating the testing for those felt to be eligible, which the authors note would have been different if they hired an on-site clinician for that purpose (some patients also refused skin testing).
  • Another concern about skin testing is that fewer than 15% of US hospitals have the appropriate reagent on formulary. The per protocol analysis of the ST arm may be open to bias (vs the intention-to-treat analysis looking at the overall group), however the low numbers of patients getting skin testing clearly biased the results to negative.
  • However, it is quite remarkable how much effect the pretty simple computerized guideline and decision support provided. It should, however, be pointed out that this was not a clean randomized-controlled trial, so there are potential inherent biases (including the possibility that there were different sensibilities and approaches to treating infections, perhaps related to the different time periods above, or proclivities of the ID departments, ward attendings, residents, etc.)
  • These results parallel those in a prior blog regarding skin testing. See found that of 146 patients with history suggestive of IgE-mediated penicillin allergy (but excluding those with history of anaphylaxis), only one patient had a positive skin test, and the remaining 145 did fine with oral penicillin. Of note, as opposed to the prior blog, those with Type I (IgE- mediated) hypersensitivity reactions were excluded from the above study.
  • Why is “penicillin allergy” so rampant?? as a singular anecdote, 25 years ago one of my children had otitis media in the middle of the night when he was less than a year old, and as a really tired parent I decided to watch him instead of getting antibiotic treatment (he wasn’t really very sick appearing, and at that point in Europe most otitis media was not being treated with antibiotics), he remained relatively stable for the next day or 2, then developed a maculopapular rash. If he had been given a beta-lactam antibiotic, he would have been labeled as penicillin allergic perhaps for the rest of his life. I do realize that there are negatives to treating family members, however my tiredness won out….
  • So, what does this all mean? The combination of the current and prior blog strongly suggest that true penicillin allergy is really quite unusual (the number quoted is <5% of those with listed penicillin allergy). And the ability to use beta-lactams for common outpatient (and inpatient) infections is really useful, especially as we are trying so hard to protect our microbiome and decrease resistance. [And, by the way, adverse reactions, need for hospitalization, costs…]. It would be really great to have a large study looking at the computer-assisted app to see the real incidence of bad allergic reactions to beta-lactams in each of the low and moderate-to-high risk groups, with an eye to using beta-lactams, perhaps initially in a monitored setting (depending on the actual incidence of severe reactions in these cohorts in subsequent studies).

(Visited 1 times, 1 visits today)