Penicillin Desensitization After a Confirmed Childhood Anaphylaxis
A single patient, two days after a joint-sparing debridement. The disagreement is whether the drug most likely to save his prosthesis is worth re-exposing him to, given a confirmed, not merely historical, IgE-mediated allergy.
Julian O., a 52-year-old man, has coached his neighborhood's under-12 soccer team every fall for nine years and ran his eleventh marathon four months before his right knee — replaced two years ago after decades of running finally wore out the joint itself — started swelling and aching in a way that had nothing to do with training. He came in with fever, a warm and effusive knee, and a joint aspirate growing Streptococcus agalactiae, fully susceptible to penicillin on sensitivities; orthopedics performed a debridement, antibiotics, and implant retention procedure two days ago, and the prosthesis itself looked salvageable on direct inspection.
The complication is a documented penicillin anaphylaxis at age nine — hives, throat tightness, and a documented epinephrine administration after a course of amoxicillin for strep throat — that has kept him on a permanent ‘penicillin allergy’ label for over four decades and, until this admission, on vancomycin whenever a beta-lactam might otherwise have been first-line. Formal skin testing, done at the allergy service's recommendation once this admission's organism came back penicillin-susceptible, confirmed a genuine IgE-mediated reactivity rather than a label carried forward on old paperwork alone. That distinction matters more than it might for a milder infection. The IDSA prosthetic joint infection guideline (Osmon et al., Clinical Infectious Diseases, 2013) recommends high-dose intravenous penicillin or ceftriaxone for streptococcal prosthetic joint infection, and it writes that recommendation for exactly the situation Julian is in — a susceptible streptococcus in a retained, not replaced, prosthesis — because bactericidal beta-lactam activity against this organism outperforms vancomycin, which kills streptococci more slowly and carries a real, if debated, higher relapse risk when hardware stays in. The one clause of that recommendation he doesn't satisfy is the assumption that a beta-lactam is available to give him.
This isn't the first infection vancomycin has carried him through — a soft-tissue infection eight years ago and a bout of cellulitis three years before that both resolved on it without incident, part of why nobody had revisited the allergy label until now. What's different this time is the organism and the stakes: Streptococcus agalactiae is a poor match for vancomycin's slower, time-dependent bactericidal activity against gram-positive organisms compared to a cell-wall-active beta-lactam, and a relapsed infection in a retained prosthesis usually means a second surgery, not just a second antibiotic course. The team's decision to formally test rather than simply continue treating around the old label was driven by that asymmetry — the drug that has quietly worked well enough before may not be the drug that reliably saves this particular joint.
On the ward, after the allergy consult
Pursue desensitization and treat with high-dose IV penicillin. Osmon's IDSA prosthetic joint infection guideline favors a bactericidal beta-lactam over vancomycin for susceptible streptococci specifically because vancomycin's activity against streptococci is comparatively slow and retained-hardware infections carry real relapse risk when definitive therapy underperforms. Rapid IV desensitization protocols, following the original Sullivan approach from 1982 and refined many times since, have an established safety record in monitored settings even for patients with confirmed IgE-mediated reactivity — his skin test result doesn't rule desensitization out, it's exactly the population this procedure exists for.
If his organism weren't reliably penicillin-susceptible, or if vancomycin's track record against it were comparable, I wouldn't push this — the case for desensitization rests entirely on the drug being genuinely better for him, not on principle.
I hear the efficacy argument, but ‘established safety record’ still means a real, if small, risk of a breakthrough reaction during the procedure itself, in a patient with a confirmed positive skin test and a documented anaphylaxis history — this isn't a mislabeled allergy we're clearing up, it's a real one we'd be deliberately re-exposing him to. Vancomycin isn't as fast against strep, but it has real outcomes data in prosthetic joint infection, and I'm not convinced the marginal efficacy gain justifies taking on anaphylaxis risk in a patient who's otherwise recovering well on the drug he's already tolerating.
You're both arguing from a correct premise. The efficacy case for penicillin over vancomycin here is real, and so is the residual risk of desensitizing someone with a confirmed positive skin test — that risk doesn't disappear just because desensitization protocols are generally safe in aggregate.
What resolves it is where and how we do this, not whether: desensitization in an ICU-level monitored setting, with epinephrine and full resuscitation equipment at the bedside, a slow graded-dose protocol rather than an accelerated one, and immunology present for the full infusion. Done that way, the residual risk is managed rather than simply accepted, which is different from either treating it as negligible or treating it as disqualifying.
Agreed after the allergy consult: desensitization proceeds, but in a monitored ICU-level setting with immunology present for the full infusion, and penicillin starts immediately on successful completion.
Not agreed: whether prosthetic joint infection patients with a confirmed beta-lactam allergy should routinely be referred for skin testing and desensitization going forward, or whether Julian's case — genuinely susceptible organism, salvageable prosthesis, reasonably fit patient — is closer to the ideal candidate than the general rule. The hospitalist's discomfort with the residual risk was not fully resolved by the staging plan, only made, in his words, ‘as acceptable as it's going to get.’