Penicillin Allergy Assessment and Delabeling

A penicillin allergy label and a true, current penicillin allergy are not the same thing, and the gap between them is enormous. Roughly 10% of the general population carries a documented penicillin or beta-lactam allergy in their chart — yet when that history is formally evaluated, 90-95% of those patients turn out not to be truly allergic at all.^[1] For the underlying immunologic mechanisms behind a genuine reaction (the Gell-Coombs classification of hypersensitivity types) and how to recognize a severe cutaneous adverse reaction (SCAR) — knowledge this chapter assumes — see the Drug Allergy chapter; this chapter picks up from there and focuses specifically on what to actually do about a penicillin allergy label: how to risk-stratify it, how to safely remove it in most patients, and how to build a service that does this at scale.

This chapter covers, in order: what actually sits behind most penicillin allergy labels; the validated PEN-FAST risk score and what real programs have found when pharmacists apply it; the current, quantitative, side-chain-driven understanding of cross-reactivity with cephalosporins, carbapenems, and aztreonam; the direct oral amoxicillin challenge protocol; referral pathways for higher-risk histories; and — because this is as much a service-design problem as a clinical one — how pharmacist-led delabeling programs have actually been built, staffed, and scaled in real hospitals and health systems.

What actually sits behind a penicillin allergy label

Most labels reflect one of a handful of recurring situations, and distinguishing them is most of the diagnostic work:

A true, currently relevant allergy — what actually distinguishes it

The features that separate a genuine, still-relevant IgE-mediated (Type I) penicillin allergy from the situations above are exactly the criteria PEN-FAST turns into a score (see Assessment & Diagnosis): how recently the reaction occurred, whether it involved anaphylaxis, angioedema, or a SCAR feature, and whether it was severe enough to require treatment.^[3] Any SCAR feature specifically — blistering, mucosal involvement, or fever with a widespread rash appearing one to several weeks into a course — changes every downstream decision in this chapter and is covered in full in the Drug Allergy chapter; treat any such history as a hard stop, not a scoring input (see Red Flags & Refer).

Factors that predict the label is inaccurate

• The reaction was never formally characterized at the time — no clear description of timing, morphology, or severity exists anywhere in the record.^[1]
• The reaction occurred more than 10 years ago, particularly in childhood.^[1][3]
• The reported symptoms are intolerance-type (nausea, vomiting, diarrhea, headache) rather than an allergic phenotype (hives, swelling, breathing difficulty, rash).
• The "reaction" was actually a family history or screening-based avoidance with no personal exposure at all.
• The reaction occurred during a course of amoxicillin or ampicillin given for an undifferentiated sore throat or viral-type illness in a child or young adult — raise infectious mononucleosis as a possible alternative explanation (see the Mononucleosis chapter).

Factors that predict a currently relevant true allergy

• The reaction occurred within the last 5 years.^[3]
• Anaphylaxis, angioedema, or any SCAR feature (see Recognize the Condition and Red Flags & Refer).^[3]
• The reaction was severe enough to require treatment — an emergency department visit, epinephrine, or a corticosteroid.^[3]
• Repeated or parenteral (especially intravenous) exposure to the same beta-lactam, which carries more opportunity for true sensitization than a single oral course.^[1]
• Multiple, unrelated drug allergy labels ("multiple drug intolerance syndrome") — a marker of a more complex history that generally warrants formal allergy referral rather than a simple direct challenge.^[1]

• Any history of blistering, skin detachment, or mucosal (oral/ocular/genital) involvement with a drug reaction, at any point in the past, however long ago.
• A history of anaphylaxis, laryngeal edema, or any reaction requiring epinephrine — refer for formal skin testing before any future beta-lactam course.
• An unclear, inconsistent, or undocumented history that cannot be confidently risk-stratified.
• Multiple, unrelated drug allergy labels.
• A pregnant patient with an unresolved penicillin allergy label who has not yet been assessed — penicillin is the preferred agent for intrapartum group B streptococcus prophylaxis, and this assessment should happen well before labour, not be discovered as a problem during it (see Special Populations).

The real, quantified cost of leaving the label in place

This is not a theoretical concern. In a UK population-based matched cohort study of 301,399 adults (64,141 with a documented penicillin allergy, 237,258 matched comparators), penicillin-allergic patients had an adjusted hazard ratio of 1.69 for MRSA infection and 1.26 for C difficile infection compared with matched comparators without the label.^[2] More than half of the excess MRSA risk and over a third of the excess C difficile risk was directly attributable to the alternative, non-beta-lactam antibiotics (vancomycin, fluoroquinolones, clindamycin) those patients received instead of a beta-lactam.^[2] Beyond these two specific organisms, the same substitution pattern is associated more broadly with longer hospital stays, higher treatment costs, and — when a non-beta-lactam is used for surgical prophylaxis in a penicillin-labeled patient — a higher risk of surgical-site infection.^[1]

The harm runs in the other direction too: a genuinely missed anaphylactic reaction can progress to airway obstruction and death, and continued exposure during an unrecognized SCAR sharply worsens outcomes. The point of this chapter is not to remove labels indiscriminately — it is to get the label right, in either direction.

The structured history — the actual starting point

For most patients, a systematic history is what establishes risk, not a laboratory test. Characterize the index reaction's timing, morphology, and severity, and specifically screen for any SCAR feature before applying any score (see the Drug Allergy chapter for the detailed "what actually happened" interview framework, which applies here unchanged).

PEN-FAST — the validated bedside tool

PEN-FAST converts that history into a probability of a currently relevant penicillin allergy using four criteria.^[3]

• 0 points — very low risk (about 0.6% probability of a positive allergy test)
• 1-2 points — low risk (about 5%)
• 3 points — moderate risk (about 20%)
• 4-5 points — high risk (about 50%)

A cutoff below 3 points was validated with a negative predictive value above 96% for a clinically significant penicillin allergy, supporting its use to identify patients appropriate for direct oral challenge without preceding skin testing.^[3]

How PEN-FAST has actually performed once pharmacists put it to work

Validation statistics are one thing; what matters operationally is how the tool performs when it is actually deployed. In a five-hospital British Columbia program spanning October 2018 to May 2022, pharmacists applied PEN-FAST or an equivalent institution-specific algorithm as the initial risk-stratification step for 711 patients across adult, pediatric, obstetric, and immunocompromised populations, successfully delabeling 98.0% of them overall (individual site rates ranged from 95% to 99.2%).^[4] In a separate Australian tertiary hospital, a pharmacist-led ward-round program using a structured phenotyping tool found that essentially all patients approached (96.5%, 82 of 85) consented to either direct delabeling or allergy testing on the spot, and pharmacist recommendations were followed by the medical team 81.5% of the time.^[5] These are real, published operational numbers, not projections — direct evidence that risk-stratified delabeling is not solely an allergist-driven process.

Referral pathway for higher-risk or unclear presentations

Patients who do not meet low-risk criteria — a PEN-FAST score of 3 or more, an unclear or inconsistent history, or a reaction within 5 years that required treatment — should be referred for formal skin testing by an allergist: percutaneous testing first, followed by intradermal testing if the percutaneous result is negative, generally followed by a supervised graded oral challenge if the skin test itself is negative.^[1] A positive skin test confirms current IgE-mediated sensitization; the drug should be avoided, or reserved for desensitization if it is truly needed (see the Drug Allergy chapter for the full desensitization procedure). Because timely access to formal allergist skin testing is not uniformly available, the risk-stratified direct-challenge pathway for low-risk patients described in this chapter is precisely what allows the large majority of patients to be safely delabeled without ever needing that appointment.^[6]

Skin prick testing — the first step of formal allergist evaluation for a higher-risk or unclear penicillin allergy history, generally followed by intradermal testing if the prick test is negative. (Rhoda Baer / National Institutes of Health; public domain, Wikimedia Commons)

Serum tryptase

A tryptase drawn during a suspected immediate reaction (rising within 15 minutes to 3 hours, peaking at 1-2 hours) can support an IgE/mast-cell-mediated mechanism retrospectively — see the Anaphylaxis chapter for the full protocol. A normal tryptase does not rule out a true reaction.

• Distinguish an inaccurate penicillin allergy label from a currently relevant, true IgE-mediated allergy using a structured history and a validated risk score.
• Safely delabel low-risk patients through a direct oral amoxicillin challenge, restoring first-line access to penicillins and narrow-spectrum cephalosporins.
• Apply current, quantitative, side-chain-based cross-reactivity data — for cephalosporins, carbapenems, and aztreonam alike — rather than the outdated flat "10% cross-reactivity" teaching.
• Identify and permanently protect patients with a true SCAR history from any future re-exposure.
• Refer higher-risk or unclear presentations for formal allergy skin testing rather than guessing.
• Update every record — chart, pharmacy dispensing system, and patient-held record — the moment a label is confirmed, refined, or removed, so it does not silently reappear at the next encounter.
• Understand and be able to explain the real, quantified antimicrobial-stewardship harm of an unresolved label.
• Know what a pharmacist-led penicillin allergy delabeling service actually looks like in practice, and what it takes to build one.

Direct oral amoxicillin challenge

Confirm eligibility (no SCAR history, no anaphylaxis/angioedema history, PEN-FAST under 3 or an equivalent low-risk profile), have epinephrine and diphenhydramine immediately available, and obtain informed consent before the first dose.^[1] Two validated approaches exist: a single-step full therapeutic oral dose (typically 250-500 mg amoxicillin in an adult), observed for a minimum of 60 minutes; or a graded (2-step) challenge — 10% of a therapeutic dose first, then, if no symptoms after 30-60 minutes, the remaining 90%, followed by a minimum 60-minute observation period after the final dose.^[1]

No objective symptoms during the observation period means the challenge is negative — remove the label immediately from every system and counsel the patient. This is genuinely safe at scale: across 289 patients who underwent a direct oral challenge in the five-hospital British Columbia program, only 9 adverse events occurred (3.1%) — six were mild (Grade 1: urticaria or a delayed rash), two involved gastrointestinal symptoms with pruritus (Grade 2), and one had possible respiratory involvement (Grade 2-3) — none were fatal, and the overall program delabeling rate was 98.0%.^[4] If an objective reaction does occur, stop, treat by severity, document precisely, and refer to allergy/immunology; do not repeat the challenge.

Cross-reactivity — a quantitative, side-chain-tiered picture, not a flat percentage

The traditional teaching that cephalosporins carry roughly a 10% cross-reactivity risk with penicillin substantially overestimates true, clinically meaningful cross-reactivity, largely reflecting older studies confounded by manufacturing cross-contamination between penicillin and early cephalosporin production lines and by testing methods that did not account for side-chain specificity. A 2019 systematic review and meta-analysis pooling 21 observational studies (1,269 penicillin-allergic patients) found that cross-reactivity risk tracks closely with similarity of the R1 side chain, not the shared beta-lactam ring or drug class as a whole:^[7]

• Identical or near-identical side chain to amoxicillin/ampicillin (the aminocephalosporins — cefadroxil, cefprozil, cephalexin) — pooled cross-reactivity of 16.45% (95% CI 11.07-23.75%). Avoid without individualized allergy input in a confirmed aminopenicillin allergy.^[7]
• Intermediate side-chain similarity — pooled cross-reactivity of 5.60% (95% CI 3.46-8.95%). Individualize; a supervised first dose is reasonable.^[7]
• Low side-chain similarity (most modern cephalosporins, irrespective of generation, including cefazolin and ceftriaxone) — pooled cross-reactivity of only 2.11% (95% CI 0.98-4.46%). Current evidence supports giving these directly in most penicillin-allergic patients, including for surgical prophylaxis.^[7]

The same meta-analysis pooled 11 studies (1,127 penicillin-allergic patients) for carbapenem cross-reactivity and found a pooled risk of only 0.87% (95% CI 0.32-2.32%) — meropenem and ertapenem are, in practice, very rarely genuinely cross-reactive.^[7] The monobactam aztreonam does not meaningfully cross-react with penicillins or with most cephalosporins at all — the one specific, well-documented exception is ceftazidime, which shares an identical side chain with aztreonam; avoid assuming aztreonam is automatically safe in a patient with a confirmed ceftazidime allergy, and vice versa.^[1]

Practical consequence: a penicillin allergy label should never, by itself, be a reason to reach for a carbapenem or aztreonam over a structurally distinct cephalosporin — both are reasonable, evidence-supported options, and neither needs to be treated as a "safer" fallback than the other. See Drug Tables for the full grouping.

Confirmed true allergy, drug still needed

When a confirmed IgE-mediated allergy remains and no adequate alternative exists, rapid drug desensitization can provide temporary, course-specific tolerance — see the Drug Allergy chapter for the full procedure and its absolute contraindication in any SCAR history.

• The moment a label is delabeled, refined, or confirmed, update every relevant system — the chart/EHR allergy field, the community and hospital pharmacy dispensing system, and any patient-held record or medical-alert item. A delabeling that lives only in one system is functionally invisible the next time the patient presents somewhere else, and record fragmentation across systems is one of the most consistently reported real-world barriers to delabeling actually sticking.^[6]
• Counsel the patient in plain language on what actually happened and why — e.g., "the reaction you had as a child was a side effect, or a rash from a virus, not a true allergy, and it is safe for you to take penicillin-type antibiotics now." Patients who understand the distinction are far less likely to re-report the old label at a future visit.
• For a patient with a confirmed true allergy, provide clear, specific written documentation of exactly which drug(s) to avoid — and which structurally distinct beta-lactams remain safe based on side-chain cross-reactivity — rather than a blanket "avoid all antibiotics" instruction that unnecessarily narrows future options.
• Build a standardized intake questionnaire (timing, morphology, severity/treatment required, SCAR screening) into whatever workflow — pharmacy consult, admission history, preoperative clinic — first identifies a penicillin allergy label, so risk-stratification happens consistently rather than depending on who happens to ask.
• Never encourage or perform an unsupervised home rechallenge — even a "low risk" challenge should only be done with rescue medication immediately available and a clear escalation plan.
• Recognize documented barriers — limited delabeling skill or confidence, fear of provoking a reaction, unclear professional scope, and time pressure — as real, addressable obstacles to building a sustainable service, not just individual knowledge gaps.^[6]

Children

The same risk-stratified, direct-challenge approach used in adults applies to children, and has been used effectively in real programs — in the British Columbia program, 19 of 20 children (95%) referred for assessment at a dedicated pediatric hospital site were successfully delabeled, with no serious adverse events.^[4] A morbilliform rash appearing during an aminopenicillin course given for a presumed bacterial infection is very often actually a non-allergic viral exanthem — see the Mononucleosis chapter for the specific mechanism and incidence, and do not let a single childhood rash become a lifelong, unquestioned allergy label.

Pregnancy

An inaccurate penicillin allergy label is a particularly consequential problem in pregnancy: penicillin is the preferred, most effective agent for intrapartum group B streptococcus prophylaxis, and the usual non-beta-lactam alternatives are less effective and carry their own resistance concerns. There is no reason to defer an appropriate low-risk direct challenge or a clinically indicated referral simply because the patient is pregnant — in the British Columbia program, 178 of 180 pregnant patients (98.8%) referred through the obstetric-hospital site were successfully delabeled, demonstrating that this pathway is genuinely safe in this population.^[4] Assess proactively, well before labour, rather than discovering an unresolved label during an acute intrapartum presentation.

Immunocompromised patients

Patients who need frequent or lifelong antibiotic courses — including transplant recipients — accumulate a disproportionate lifetime exposure to broader-spectrum alternatives if a beta-lactam label is never reassessed, and this population is sometimes assumed to be too high-risk for a direct-challenge pathway. That assumption does not hold up: in the same British Columbia program, 54 of 56 bone-marrow-transplant recipients (96.4%) were successfully delabeled using the standard risk-stratified pathway.^[4] Accurate delabeling — or, where a true allergy is confirmed, a clear map of which structurally distinct beta-lactams remain safe — is especially valuable in this group.

Perioperative patients

A penicillin allergy label directly shapes surgical antibiotic prophylaxis choice, and an inaccurate label pushed toward a non-beta-lactam alternative is an independent risk factor for surgical-site infection.^[1] A preoperative assessment clinic is a natural, high-yield point in the care pathway to risk-stratify and, where appropriate, delabel a patient well before the day of surgery rather than defaulting to vancomycin or clindamycin prophylaxis by habit.

Multiple drug allergy labels

Patients carrying several unrelated drug allergy labels represent a more complex history and generally warrant formal allergy referral rather than a simple direct challenge, since the history itself is harder to risk-stratify reliably.

After a negative direct oral challenge, no further monitoring beyond the standard observation period is needed — the priority shifts entirely to making sure the delabeling actually sticks across every system and every future encounter (see Non-Pharmacologic Treatment).

• After a positive/reactive challenge or skin test: document the confirmed reaction precisely (timing, morphology, severity, treatment given) so the next clinician has an accurate, specific record rather than a bare "penicillin allergy" flag.
• Build in periodic reassessment: any allergy label that has never been formally evaluated is a reasonable, low-effort thing to flag for review at essentially any future pharmacist encounter, not something that needs a dedicated visit.
• If running a delabeling service rather than a single patient encounter, track program-level metrics deliberately — patients screened, delabeling rate, adverse event rate, and downstream change in restricted/broad-spectrum antibiotic use — the way both the Australian ward-round program (restricted-antibiotic prescribing fell from 42.5% to 17.9% post-intervention) and the British Columbia multi-site program did.^[5][4] These metrics are what sustain institutional support for the service over time.
• Watch specifically for the label reappearing after a documented delabeling — a genuine, reported failure mode caused by record fragmentation across systems rather than a clinical error, and one that periodic audit can catch.^[6]

1. An objective reaction occurs during a direct oral challenge: stop the drug immediately, treat according to severity (see the Pharmacologic Treatment Options table and, for anaphylaxis, the Anaphylaxis chapter), document the reaction precisely, and refer to allergy/immunology — do not repeat the challenge.
2. A patient's history does not clearly fit either the low-risk or clearly high-risk profile: default to referral for formal skin testing rather than proceeding with a direct challenge on an uncertain history.
3. A positive skin test: this confirms current IgE-mediated sensitization — avoid the drug going forward, or pursue desensitization if it is genuinely needed and no adequate alternative exists (see the Drug Allergy chapter).
4. A patient remains reluctant to have a label removed despite a negative challenge or negative skin testing: address the specific concern in plain language, provide clear written documentation the patient can keep, and respect their final decision while making sure the clinical record still reflects the objective result accurately.
5. A previously delabeled patient's allergy label reappears at a later encounter: this is usually a documentation/system-interoperability failure, not a new clinical event — confirm the original delabeling result, update every affected system again, and use it as a prompt to review whether the delabeling workflow itself needs a more durable record-update step.^[6]

• Accepting a decades-old, undocumented "penicillin allergy" label at face value without ever asking what actually happened.
• Treating nausea, vomiting, diarrhea, or headache as if it were an allergic reaction.
• Teaching or assuming a flat "10% cephalosporin cross-reactivity" figure rather than reasoning from the actual side-chain-tiered data (~16% for aminocephalosporins, ~6% intermediate, ~2% low similarity, well under 1% for carbapenems).^[7]
• Reflexively avoiding carbapenems or aztreonam in a labeled patient on the assumption that "any beta-lactam" is equally risky — the data does not support this.^[7]
• Attempting a direct oral challenge, a skin test intended to lead to challenge, or desensitization in any patient with a history of SJS/TEN, DRESS, or AGEP.
• Defaulting to vancomycin, a fluoroquinolone, or clindamycin for a "penicillin-allergic" patient without ever risk-stratifying the actual history first — this is precisely the substitution pattern behind the quantified MRSA and C difficile risk increase.^[2]
• Starting a direct oral challenge without epinephrine and diphenhydramine immediately available.
• Delabeling a patient's chart without also updating the pharmacy dispensing system and any patient-held record — an incomplete update is why the same label so often silently reappears at the next encounter.^[6]
• Assuming a penicillin allergy delabeling service must be allergist-led — the published pharmacist-led inpatient and multi-site programs described in this chapter demonstrate otherwise.^[5][4]

How to use this table: Doses are adult doses and require patient-specific confirmation (renal/hepatic function, weight, allergy history). Pediatric dosing is weight-based per the current product monograph unless stated. Drug pricing is intentionally omitted. Select any drug name to open its complete Pharmasist Drug Database page.

Direct Oral Challenge Agents

Used to actively confirm tolerance in a risk-stratified low-risk patient — this is a diagnostic/delabeling procedure, not empiric antibiotic therapy. See Choose Initial Treatment for the full protocol and eligibility criteria.

Beta-Lactams and Related Agents, Grouped by R1 Side-Chain Cross-Reactivity Risk

Cross-reactivity with penicillin is driven by R1 side-chain similarity, not by the shared beta-lactam ring or by drug class as a whole. Pooled figures from a 2019 systematic review/meta-analysis; see Choose Initial Treatment for the full explanation.

Anaphylaxis Rescue Drugs

Must be immediately available before any direct oral challenge begins — not agents to gather after a reaction starts.

Alternative Antibiotics Whose Overuse Reflects an Unresolved Label — Not First-Line Simply Because of a Penicillin Allergy

These agents are genuinely useful drugs, but their use specifically because a penicillin allergy was never risk-stratified is exactly the substitution pattern shown to independently raise the risk of MRSA and C difficile infection. Assess the label before reaching for any of these by default.