ID Practice Update

Short lecture / 15-20 minutes

Penicillin allergy

Does this history prevent us from using a beta-lactam?

A label raises two separate questions: what caused the original reaction, and whether the antibiotic needed today is likely to cause the same reaction. Sometimes the second question can be answered before the first.

Resident core / Optional fellow detail / Edited November 10, 2026

Optional fellow discussion is below

The resident lecture includes the bedside workflow and PEN-FAST. The fellow section covers the evidence and its limits.

This lecture is intended as a supplementary resource for clinicians and trainees to review evidence and discuss clinical cases.

This lecture is intended as a supplementary resource for clinicians and trainees to review evidence and discuss clinical cases.

The reaction history determines the next step

A 52-year-old with a susceptible bacterial infection has an amoxicillin allergy label. At age 8, a rash appeared several days into treatment. There were no mouth sores, blisters, breathing symptoms, or hospitalization. The patient has avoided penicillins since then.

The record says “rash.” What can the patient recall about its appearance, its timing after a dose, and how it was treated? Is there a reliable record of taking the culprit or another beta-lactam since?

About 10% of U.S. patients report a penicillin allergy, but most do not have allergy confirmed after evaluation. Confirmation rates vary with the setting and patient selection. Inaccurate labels, infection-associated rashes, and loss of sensitization over time contribute to the discrepancy. The history still matters. A label of "rash" is not enough information. [1] [2]

An inaccurate label often leads to broader or less effective therapy. That is associated with more adverse drug effects, C. difficile infection, resistant organisms, longer hospitalization, and higher cost. An accurate history can preserve access to an effective antibiotic while keeping a serious past reaction visible in the record. [2]

Classify the reaction before choosing another beta-lactam

Timing and phenotype matter more than the word allergy.
PatternTypical timingTypical findingsResident implication
Immediate, IgE-mediatedUsually within minutes to about 1 hour, but may occur within several hoursUrticaria, angioedema, bronchospasm, hypotension, anaphylaxisA past immediate reaction calls for a history of the culprit, timing, and symptoms. If anaphylaxis is suspected now, give intramuscular epinephrine immediately and activate the emergency response. Allergy testing is a later question. [13]
CytotoxicDays to weeksImmune hemolytic anemia or thrombocytopeniaDo not use a routine challenge pathway. Avoid the culprit and obtain specialist guidance.
Other systemic or organ-specific reactionsUsually days to weeks; varies by syndromeSerum sickness or serum sickness-like reaction, drug fever, interstitial nephritis, or drug-induced liver injuryThese histories require separate assessment because the mechanisms differ. A negative immediate-reading skin test does not establish tolerance. A history of serious drug-induced organ injury warrants culprit avoidance and specialist assessment; it falls outside routine PEN-FAST and direct-challenge pathways.
Delayed, T-cell mediatedUsually days to weeksBenign morbilliform eruption, or severe disease such as SJS/TEN, DRESS, or AGEPSeparate a benign eruption from a severe cutaneous reaction. Never use PEN-FAST or routine challenge to clear a severe delayed reaction.
Intolerance or nonallergic historyVariableIsolated nausea, diarrhea, headache, family history, or a label without a patient reactionWhen an isolated side effect or family history fully explains the label and there was no personal allergic reaction, penicillin allergy testing is usually unnecessary. An unclear past reaction still needs assessment. [2] [4]

The classic Gell-Coombs categories help organize the phenotype, but real reactions do not always fit neatly into one box. Ask specifically about mucosal injury, skin peeling, facial edema, eosinophilia, kidney or liver injury, cytopenias, treatment required, and subsequent beta-lactam tolerance. [1] [4]

PEN-FAST estimates test positivity; phenotype determines whether the score applies

First consider blistering or mucosal injury, systemic illness with rash, or drug-induced organ or blood-cell injury. Those histories fall outside routine direct challenge, even if the arithmetic gives a low score. A point estimate is not an estimate of this patient’s risk of anaphylaxis. [1] [5]

Adult PEN-FAST score. The outcome predicted is a positive penicillin-allergy test.
CriterionPoints
Five years or less since the reaction, or timing unknown2
Anaphylaxis or angioedema, or Severe cutaneous adverse reaction2
Treatment required, or treatment unknown1
ScorePositive testing in the derivation cohort, not anaphylaxis risk
0Very low, less than 1%
1-2Low, about 5%
3Moderate, about 20%
4-5High, about 50%

Among Melbourne patients scoring below 3, 96.3% had negative allergy testing. That proportion ranged from 84.9% to 98.4% in the external cohorts. A low score therefore helps identify adults for further assessment; it neither removes a label nor establishes that a challenge is safe. A remote severe cutaneous reaction can score below 3. Isolated drug fever, serum sickness, interstitial nephritis, and drug-induced liver injury were excluded from the study. Open the PEN-FAST calculator. [5]

Testing, challenge, and desensitization answer different questions

Skin testing looks for sensitization. A challenge tests tolerance of a particular drug. Desensitization makes a necessary course possible despite suspected or established allergy.
ApproachPurposeClinical considerations
Immediate-reading penicillin skin testingValidated reagents and controls assess immediate IgE sensitization. In a compatible history, a negative skin test generally leads to an observed amoxicillin challenge to confirm tolerance. A negative skin test by itself is not the same as tolerating the drug.Most useful after anaphylaxis or a recent suspected IgE-mediated reaction. A negative immediate-reading test does not establish safety after SJS/TEN, DRESS, AGEP, interstitial nephritis, serum sickness, hemolytic anemia, or other serious non-IgE reactions. Specialist delayed skin or patch testing is a different evaluation; it does not authorize routine culprit challenge after a severe reaction. [2] [4] [12]
Observed drug challengeTests current tolerance of the selected drug. “Direct” means preceding skin testing is omitted in an appropriately selected patient. A graded challenge uses more than one step; neither approach induces temporary tolerance.For a low-risk history, observation by trained staff and immediately available rescue treatment are part of a standardized challenge. A severe delayed reaction or drug-induced organ or blood-cell injury excludes routine direct challenge. [1]
DesensitizationCreates temporary tolerance when a particular drug is necessary and immediate allergy is established or strongly suspected. Tolerating the course under desensitization does not demonstrate that the original label was wrong.Requires a supervised protocol and uninterrupted dosing; it does not remove the allergy label. SJS/TEN, DRESS, AGEP, immune cytopenias, and serum sickness are contraindications to the usual immediate-allergy desensitization pathway. Serious drug-induced organ injury also requires avoidance and specialist assessment. [4] [9]

Consider Allergy/Immunology when the label is forcing inferior therapy, when an immunocompromised patient is likely to need repeated antibiotics, or when multiple antimicrobial allergy labels substantially limit treatment. [2] [4]

A penicillin allergy does not automatically exclude cefazolin

There is no single cross-reactivity rate for every penicillin/cephalosporin pair. Risk depends on confirmed versus unverified allergy and R1 similarity. Cefazolin has very low cross-reactivity. With confirmed aminopenicillin allergy, assess any aminocephalosporin with a shared R1 side chain separately. [4]

With an unverified, nonanaphylactic penicillin history and no severe delayed features, cephalosporins can generally be given without preceding testing or extra allergy precautions. After penicillin anaphylaxis, the 2022 parameter conditionally supports a cephalosporin with a dissimilar R1 side chain, such as cefazolin, without preceding testing or extra precautions. Neither situation clears an independent cefazolin allergy. For a carbapenem, a history of immediate penicillin or cephalosporin allergy, including anaphylaxis, likewise does not itself require preceding testing or extra precautions. Severe delayed skin reactions or drug-induced organ injury are a separate problem; these recommendations do not establish safety in those patients. [4]

The opening patient’s remote, uncomplicated rash makes a supervised direct amoxicillin challenge a reasonable option if the full history remains low risk. If cefazolin is the appropriate antibiotic now, selecting it need not wait for an amoxicillin challenge. Tolerating cefazolin would document cefazolin tolerance; it would not resolve the original amoxicillin label. [4] [6]

Compare the side chains on both axes

The matrix compares the R1 side chains of 15 drugs. It helps explain why two cephalosporins of different generations may share a relevant structure. An identical or similar side chain does not prove clinical allergy; an unmarked cell does not prove tolerance. The culprit, reaction phenotype, and desired drug still determine whether an alternative or a supervised evaluation makes sense. [4] [7]

IIdentical R1 side chain. Potential cross-reactivity.
SSimilar R1 features. The entire side chain is not identical; clinical implications vary by pair and reaction phenotype.
·No highlighted R1 pairing in this selected chart. This does not mean zero allergy risk.
†Carbapenem comparison; this symbol gives no risk estimate. See the clinical note below.
=The same antibiotic. This is not an alternative to the culprit.
Rows show the suspected culprit; columns show the proposed drug. Symbols describe structural relationships, not permission to challenge. The chart does not clear a drug after a severe skin, organ, or blood-cell reaction.
Proposed antibiotic →

Suspected culprit ↓
Amoxicillin Ampicillin Cefadroxil Cefprozil Cephalexin Cefaclor Cefazolin Cefuroxime Ceftriaxone Cefotaxime Cefpodoxime Cefepime Ceftazidime Aztreonam Meropenem
Amoxicillin = I I · · · · · · · · †
Ampicillin = I I · · · · · · · · †
Cefadroxil I = I · · · · · · · · †
Cefprozil I I = · · · · · · · · †
Cephalexin I = I · · · · · · · · †
Cefaclor I I = · · · · · · · · †
Cefazolin · · · · · · = · · · · · · · †
Cefuroxime · · · · · · · = †
Ceftriaxone · · · · · · · = I I I †
Cefotaxime · · · · · · · I = I I †
Cefpodoxime · · · · · · · I I = I †
Cefepime · · · · · · · I I I = †
Ceftazidime · · · · · · · = I †
Aztreonam · · · · · · · I = †
Meropenem † † † † † † † † † † † † † † =

Read the important pairings

Amoxicillin shares R1 with cefadroxil and cefprozil; ampicillin shares R1 with cephalexin and cefaclor. These two amino side-chain groups are related, which is why amoxicillin–cephalexin is marked S rather than I. Ceftriaxone, cefotaxime, cefpodoxime, and cefepime share R1; cefuroxime has related methoxyimino features. Ceftazidime and aztreonam share R1. Their comparisons with the five methoxyimino cephalosporins are marked S for partial structural resemblance, not an identical side chain or a uniform clinical risk. The DGAKI structural discussion describes occasional ceftazidime cross-reactivity within this group. Aztreonam's S cells extrapolate the structural comparison from its identical R1 with ceftazidime; they do not establish clinical cross-reactivity for each pair. Published charts disagree about some related pairs. [4] [8] [11]

Cefazolin and carbapenems

Cefazolin has no shared R1 side chain with the other drugs shown. A penicillin allergy alone does not establish cefazolin allergy. For meropenem (†), the 2022 parameter permits a carbapenem in patients with penicillin or cephalosporin allergy, including prior anaphylaxis, without routine preceding testing or additional precautions. That recommendation does not establish tolerance in someone with a carbapenem allergy or a severe delayed reaction. The aztreonam–meropenem cell is outside that specific recommendation. For immediate penicillin or cephalosporin allergy, the same parameter permits aztreonam without preceding testing except when the history involves ceftazidime allergy. The S cells do not override that clinical recommendation or clear a prior aztreonam allergy or severe delayed reaction. [4]

Anaphylaxis to a cephalosporin differs from anaphylaxis to penicillin. After cephalosporin anaphylaxis, the 2022 practice parameter conditionally suggests confirming a negative cephalosporin skin test before giving a parenteral cephalosporin with a nonidentical R1 side chain. A negative test alone does not prove tolerance. Do not use this chart to justify an unsupervised challenge after anaphylaxis or a routine culprit challenge after severe cutaneous or organ injury. [4]

This is a teaching matrix, not a complete allergy clearance tool. It does not map R2 side chains, beta-lactamase inhibitor allergy, or other penicillins such as penicillin G/V, oxacillin, and piperacillin. Within-penicillin cross-reactivity can involve shared penicillin determinants. This matrix highlights selected R1 relationships, not every molecular resemblance. Published charts disagree about the clinical importance of partial similarities, including aztreonam–ceftriaxone. Neither S nor an unmarked cell supplies a patient-specific probability: use the phenotype and the clinical recommendations above. [4] [8]

Document what was tested and what the result means

When the completed evaluation rules out the suspected allergy, remove that label from the allergy list and explain the result. Tolerance of a different beta-lactam is recorded separately. Useful documentation includes the original reaction, the drug and dose administered during evaluation, the date, the observation period, and the outcome. The patient and the next clinician need to know which drug was tolerated and which label changed. Absence of immediate symptoms does not rule out a later rash. [2] [4]

If the patient instead recalls blistering and oral ulcers, the history now raises concern for a severe delayed reaction. The time elapsed does not make a routine test dose appropriate. Avoid the suspected culprit and seek specialist assessment; a low PEN-FAST score cannot clear this history. [1] [5]

Apply this on the next consult

In the original case, a remote, uncomplicated childhood rash may support a supervised amoxicillin challenge after assessment. In the blistering and oral-ulcer variant, routine challenge is inappropriate; the suspected culprit should be avoided pending specialist assessment. If an appropriate evaluation clears amoxicillin, remove that label and communicate exactly what was tested. Tolerance of cefazolin alone does not clear amoxicillin.

For syphilis in pregnancy, penicillin G is the only proven treatment for fetal infection and prevention of congenital syphilis. A reported penicillin allergy warrants prompt allergy assessment. Give stage-appropriate penicillin G once allergy is excluded, or after supervised desensitization when indicated. Severe delayed reactions require specialist assessment outside routine testing or challenge. Allergy guidance · Syphilis guidance

Know what the validation studies did and did not establish

PEN-FAST estimates positive allergy testing, not an individual's risk of anaphylaxis. In the 622-person Melbourne derivation/internal-validation cohort, 17 of 460 people with scores below 3 had positive testing, giving a negative predictive value of 96.3% (95% CI, 94.1–97.8%). The three external cohorts included 945 people; their negative predictive values ranged from 84.9% to 98.4%. The score bands above are derivation estimates, not fixed risks in every referral population. [5]

A later Singapore referral-clinic validation, published online in 2025 and in a 2026 issue, found positive testing in 16 of 215 patients with PEN-FAST below 3: negative predictive value 92.6% (95% CI, 89.2–94.9%). Those reactions included positive skin testing, urticaria, and mild delayed rashes. This retrospective study reinforces the importance of phenotype and setting; it does not establish a new universal cutoff. [10]

PALACE randomized 382 adults at six specialist outpatient centers and analyzed 377: 187 assigned to direct oral challenge and 190 to skin testing followed by challenge. Physician-verified immune-mediated reactions within one hour occurred in 1/187 versus 1/190. The risk difference was 0.0084 percentage points (90% CI, −1.22 to 1.24), meeting the prespecified 5-percentage-point noninferiority margin. The trial used a 60-minute observation and day-5 follow-up to measure immediate and delayed outcomes. These were study procedures, not a universal observation or follow-up schedule. Delayed rash or urticaria occurred in 6/187 versus 3/190; no serious adverse events were reported. The sample does not exclude rare severe reactions. [6]

PALACE enrolled adults with PEN-FAST below 3, but more than 94% in each arm scored 0 or 1. Key exclusions included a history of anaphylaxis with any drug, severe delayed skin or organ reactions, chronic spontaneous urticaria, mast cell disease, and pregnancy. These are limits of this trial, not a claim that every excluded group can never undergo specialist allergy evaluation. The results support a supervised pathway in carefully selected low-risk adults; they do not support bypassing phenotype review or giving an unmonitored test dose. [6]

Side-chain charts are structural aids, not outcome calculators. Published charts disagree about some partial similarities. Useful recognized pairings include amoxicillin with cefadroxil, ampicillin with cephalexin or cefaclor, and ceftazidime with aztreonam. Cefazolin has no shared R1 side chain with the other drugs shown. The reaction phenotype remains essential even when a cell is unmarked. [4] [8]

When the preferred drug is being withheld because of an allergy label, the consultant should state whether the concern is immediate allergy, severe delayed hypersensitivity, side-chain cross-reactivity, or simple uncertainty. That distinction determines whether the next step is routine administration, supervised challenge, skin testing, desensitization, or avoidance.

Sources and further reading

1. CDC. Clinical features of penicillin allergy. 2025.

2. AAAAI. Penicillin allergy evaluation should be performed proactively. 2023 position statement.

3. CDC. Managing persons who have a history of penicillin allergy. STI Treatment Guidelines, 2021.

CDC. Syphilis during pregnancy.

4. Khan DA, et al. Drug allergy: A 2022 practice parameter update. J Allergy Clin Immunol. Published December 2022. Full final practice parameter.

5. Trubiano JA, et al. Development and validation of a penicillin allergy clinical decision rule. JAMA Intern Med. Published March 16, 2020.

6. Copaescu AM, et al. PALACE randomized clinical trial. JAMA Intern Med. Published July 17, 2023.

7. Northwestern Medicine Antimicrobial Stewardship Program. Beta-lactam cross-reactivity side-chain chart. May 2024 revision; accessed September 27, 2026. The source chart combines R1 and R2 information. This teaching matrix uses an original layout and legend for selected R1 relationships, checked against sources 4 and 11.

8. Hutten EM, et al. Discrepancies in beta-lactam antibiotics cross-reactivity: implications for clinical practice. Allergy. 2025;80:3108-3114.

9. Alvarez-Cuesta E, et al. Standards for practical intravenous rapid drug desensitization & delabeling: a WAO committee statement. World Allergy Organ J. 2022;15:100640.

10. Lim XR, et al. Validation of the PEN-FAST penicillin allergy clinical decision rule in an Asian cohort. Clin Exp Allergy. 2026;56:402-410. Published online October 25, 2025.

11. Wurpts G, et al. Guideline on diagnostic procedures for suspected hypersensitivity to beta-lactam antibiotics. Allergo J Int. 2019;28:121-151. Side-chain chemistry sections; current management recommendations above follow the 2022 practice parameter.

12. Barbaud A, et al. Updated EAACI statement on drug hypersensitivity skin testing: methodology and non-irritative concentrations. Allergy. Published August 19, 2026.

13. Cardona V, et al. World Allergy Organization anaphylaxis guidance 2020. Acute treatment and Figure 4.

Teaching cases are fictional. Recommendations are paraphrased from the cited public sources. Confirm the active local pathway before testing, challenging, or desensitizing a patient.