Red flags
The whole assessment is one question: is this a simple exanthem or a severe cutaneous adverse reaction (SCAR)?**
- Mucosal involvement - eyes, mouth, genitals, nares -> SJS/TEN until proven otherwise
- Skin pain or tenderness out of proportion to appearance - precedes detachment
- Positive Nikolsky sign (lateral pressure shears epidermis), flaccid blisters, dusky targetoid macules, sheets of detachment
- Fever >38.5 C with a drug rash
- Facial oedema - the signature of DRESS
- Sterile pustules on oedematous erythema + fever - AGEP
- Lymphadenopathy, hepatitis, eosinophilia >1.5 x10^9/L, atypical lymphocytes - DRESS
- Erythroderma (>90% BSA)
- Purpura that does not blanch, retiform necrosis - vasculitis, DIC, purpura fulminans
- Hypotension, tachycardia, hypoxia, stridor, angioedema - anaphylaxis
- Anuria, rising creatinine - DRESS interstitial nephritis, drug-induced AIN
- *"It is only a rash" is the commonest fatal misjudgement in this area*
Differential by mechanism1 exam ›
By mechanism
| Mechanism | Timing after start | Examples |
|---|---|---|
| Type I - IgE mast cell | Minutes-1 h | Urticaria, angioedema, anaphylaxis. Beta-lactams, NSAIDs, contrast, NMBAs |
| Type II - cytotoxic IgG | Days-weeks | Drug-induced immune haemolysis, thrombocytopenia |
| Type III - immune complex | 7-14 days | Serum sickness-like, drug-induced vasculitis, urticarial vasculitis |
| Type IVa/b - Th1/Th2 + eosinophils | 2-8 weeks | DRESS/DiHS |
| Type IVc - cytotoxic T cell | 4-28 days | SJS/TEN, fixed drug eruption |
| Type IVd - neutrophilic (IL-8) | 1-3 days | AGEP |
| Non-immune / pharmacological | Variable | Pseudoallergy (vancomycin infusion reaction via MRGPRX2), NSAID-exacerbated respiratory disease, ACEi angioedema (bradykinin) |
The SCAR phenotypes - the table that matters
| SJS/TEN | DRESS | AGEP | |
|---|---|---|---|
| Latency | 4-28 days | 2-8 weeks (long) | 1-3 days (short) |
| Morphology | Dusky atypical targets -> flaccid bullae, detachment | Morbilliform -> confluent, facial oedema | Sterile non-follicular pustules on oedematous erythema, flexural onset |
| Mucosa | Severe, >=2 sites | Mild/absent | Absent |
| Fever | Yes | Yes | High |
| Bloods | Lymphopenia | Eosinophilia, atypical lymphocytes, deranged LFT | Neutrophilia |
| Organs | Skin, mucosa, lung, gut | Liver, kidney, heart, thyroid, lung | Rarely |
| Mortality | SJS ~5%, TEN 25-35% | ~5-10% | <5% |
| Classic drugs | Allopurinol, carbamazepine/lamotrigine/phenytoin, sulfonamides, nevirapine, NSAIDs (oxicams) | Allopurinol, carbamazepine, phenytoin, sulfasalazine, vancomycin, minocycline, dapsone | Aminopenicillins, macrolides, diltiazem, hydroxychloroquine, terbinafine |
- SJS/TEN classified by detachment: SJS <10% BSA, overlap 10-30%, TEN >30%
Non-drug mimics
- Viral exanthem (commonest confounder - EBV + amoxicillin), staphylococcal scalded skin syndrome (subcorneal split, no mucosa, children), erythema multiforme (HSV-driven, acral, typical 3-zone targets, NOT a drug reaction), Kawasaki, toxic shock, acute GVHD, paraneoplastic pemphigus, generalised pustular psoriasis (mimics AGEP)
Focused history
- Build a drug timeline against the rash onset - the single highest-yield act
- Every drug started in the preceding 8 weeks (not just this admission)
- Include: over-the-counter, herbal, "just a short course", contrast, recently ceased drugs
- Match latency to phenotype - a drug started 3 days ago cannot have caused DRESS
- Prior reaction to this or a related drug? - and what happened exactly
- "Allergic to penicillin" recorded as childhood rash: 90-95% of such labels are not true allergy
- Prodrome - fever, malaise, sore throat, stinging eyes 1-3 days before the rash = SJS/TEN
- Skin pain, eye grittiness/photophobia, dysuria, odynophagia - mucosal disease
- Systemic review - jaundice, dark urine, oliguria, dyspnoea, palpitations
- Personal/family history of atopy, HIV, autoimmune disease, prior SCAR
- Ancestry - HLA risk alleles (see Investigations)
- Recent infection (Mycoplasma, HSV), tattoo, vaccination
Focused examination
- Estimate % BSA of detached or detachable epidermis (rule of nines; palm+digits = 1%)
- Count only detached/Nikolsky-positive skin - not all the erythema
- Nikolsky sign and Asboe-Hansen sign (blister extends with pressure)
- Morphology: morbilliform / urticarial / targetoid / pustular / bullous / purpuric / desquamating
- All mucosal surfaces - always
- Eyes (conjunctival injection, pseudomembrane - ophthalmology same day), mouth, nares, genital, perianal
- Facial oedema (DRESS), lymphadenopathy (DRESS)
- Vital signs: fever, hypotension, tachypnoea, SpO2
- Hepatomegaly, jaundice
- Photograph and mark the extent - serial comparison is how progression is detected
Investigation strategy
Bedside and bloods - all suspected SCAR
- FBE + film - eosinophilia and atypical lymphocytes (DRESS); neutrophilia (AGEP); lymphopenia (TEN)
- LFT, UEC, CK, glucose, bicarbonate, lipase
- CRP, blood cultures, lactate if septic-appearing
- Urinalysis + microscopy - eosinophiluria, casts (AIN)
- Troponin and ECG in DRESS - myocarditis is a leading cause of death
- TFT at 6 weeks and 3 months in DRESS - late autoimmune thyroiditis
- Tryptase if anaphylaxis suspected (acute + baseline at >24 h)
Skin biopsy
- Lesional punch for histology + perilesional for DIF if immunobullous disease possible
- TEN: full-thickness epidermal necrosis, subepidermal split, sparse infiltrate
- AGEP: subcorneal/intraepidermal spongiform pustules
- Frozen section can separate TEN from SSSS within hours when the distinction changes management
Viral studies
- HHV-6/7, EBV, CMV PCR in DRESS - reactivation drives relapse and organ involvement
- Mycoplasma serology/PCR + HSV PCR if mucositis-predominant in a young person ("MIRM" - mucositis without much skin, not drug-induced)
Severity scoring
- SCORTEN at 24 h (age >40, malignancy, HR >120, BSA >10%, urea >10 mmol/L, glucose >14 mmol/L, bicarbonate <20 mmol/L)
- 0-1 -> ~3% mortality; >=5 -> >90%
- RegiSCAR score for DRESS probability
HLA pharmacogenomics
| Allele | Drug | Reaction |
|---|---|---|
| *HLA-B15:02** | Carbamazepine (also phenytoin, lamotrigine, oxcarbazepine) | SJS/TEN - screen before starting in South-East Asian, Han Chinese, Thai, Indian ancestry |
| *HLA-A31:01** | Carbamazepine | DRESS/hypersensitivity (European, Japanese) |
| *HLA-B58:01** | Allopurinol | SJS/TEN and DRESS - Han Chinese, Thai, Korean; risk inc in CKD |
| *HLA-B57:01** | Abacavir | Hypersensitivity - mandatory screening before prescribing |
| *HLA-B13:01** | Dapsone | DRESS |
Delayed confirmation - never in the acute phase
- Patch testing at 6 weeks-6 months (useful in AGEP, DRESS, fixed drug eruption; poor for TEN)
- Intradermal/skin prick for immediate reactions
- Graded oral challenge - the gold standard for delabelling; *absolutely contraindicated after SJS/TEN, DRESS or AGEP*
Management
1. Stop the drug - immediately, and stop all non-essential drugs
- Earlier cessation = lower mortality in SJS/TEN
- Where the culprit is uncertain, stop everything that can be stopped and substitute structurally unrelated agents
- *Do not "watch it" on the drug*
2. Decide the disposition
- TEN/SJS-overlap, erythroderma, or SCORTEN >=2 -> burns unit or ICU
- DRESS with organ involvement -> inpatient, often HDU
- Simple morbilliform exanthem, well, no red flags -> outpatient with clear return advice
3. Supportive care - this is what changes survival in SJS/TEN
- Fluid and electrolyte resuscitation (less than burns: ~2/3 of burns volumes), nutrition via NG early
- Ambient temperature 28-32 C, air-fluidised bed, minimal handling
- Non-adherent dressings; leave blister roofs intact as a biological dressing - do NOT debride routinely
- Analgesia - opioid infusion; procedural analgesia for dressings
- No prophylactic antibiotics; surveillance swabs, treat infection promptly (sepsis is the leading cause of death)
- Daily ophthalmology review - lubricants, topical steroid, amniotic membrane transplantation for severe disease (prevents symblepharon and blindness)
- Mouth, urogenital and anal care; vaginal dilators/topical steroid to prevent adhesions
- VTE prophylaxis; stress ulcer prophylaxis
4. Immunomodulation
### SJS/TEN
- No universally agreed agent; supportive care remains the backbone
- Ciclosporin 3-5 mg/kg/day - the most widely used; observational mortality benefit
- Etanercept 50 mg SC single dose - RCT-proven faster re-epithelialisation; meta-analyses suggest mortality benefit vs supportive care alone
- Short-course high-dose corticosteroid - used, contested
- IVIg monotherapy - no longer supported
- Choose one and start early - benefit is confined to the progressive phase
### DRESS
- Systemic corticosteroid 0.5-1 mg/kg/day prednisolone, tapered slowly over 6-12 weeks
- Rapid taper -> relapse; this is the commonest management error
- Topical steroid alone if skin-limited with no organ involvement
- Ciclosporin or IVIg for steroid-refractory disease
- Treat HHV-6 reactivation-driven organ disease case by case
### AGEP
- Self-limiting - supportive care + potent topical steroid; resolves with desquamation in ~2 weeks
### Urticaria/angioedema/anaphylaxis
- IM adrenaline 0.5 mg (1:1000) into the anterolateral thigh for anaphylaxis, repeat every 5 min as needed
- Non-sedating antihistamines for urticaria; steroids are not first-line for anaphylaxis
5. Documentation and delabelling - the durable intervention
- Record the drug, the date, the phenotype and the outcome in the allergy field - "rash" is useless
- Absolute lifelong avoidance of the culprit and cross-reactive drugs after any SCAR
- Aromatic anticonvulsants (carbamazepine/phenytoin/phenobarbitone/lamotrigine) cross-react extensively
- Warn first-degree relatives after SJS/TEN with an HLA-linked drug
- Refer for delabelling where the label is a childhood "rash" and the phenotype was benign
- Give the patient a written record and a MedicAlert; notify the TGA (Blue Card)
Traps
- *Missing latency mismatch - the drug started yesterday is exonerated for DRESS; look back 8 weeks*
- *Calling a SCAR "a viral rash"* because the patient is febrile - fever supports SCAR, it does not exclude it
- *Not examining the mucosa or the eyes* - the finding that reclassifies the whole illness
- *Debriding blistered skin in TEN* - increases fluid loss and infection; leave the roof on
- *Fluid-resuscitating TEN to burns formulae* - over-resuscitation causes pulmonary oedema
- *Rapid steroid taper in DRESS* -> relapse, sometimes worse than the index episode
- *Forgetting DRESS's late sequelae* - autoimmune thyroiditis, type 1 diabetes and alopecia months later; check TFT at 6 weeks and 3 months
- *Treating erythema multiforme as a drug reaction* - it is HSV-driven; stopping an innocent drug mislabels the patient for life
- *Restarting the drug because "it was probably the infection"* - rechallenge after SCAR can be fatal
- *Recording "allergic to penicillin" after a morbilliform rash -> lifelong broad-spectrum antibiotics, more C. difficile, more MRSA, worse outcomes. Delabel.*
- *Allopurinol started at full dose in CKD* - the classic allopurinol hypersensitivity setting; start 50-100 mg and titrate
- *Assuming IVIg helps TEN* - it does not, and it delays the agent that might
Talk track
1. Separate benign from severe first
- "My first task is to decide whether this is a simple morbilliform exanthem or a severe cutaneous adverse reaction, because the two have completely different mortality. I look for mucosal involvement, skin pain, Nikolsky positivity, facial oedema, fever and organ dysfunction."
2. Use latency to identify the culprit
- "I construct a timeline of every drug started in the last eight weeks against the rash onset. Latency is phenotype-specific - one to three days for AGEP, four to twenty-eight days for SJS/TEN, and two to eight weeks for DRESS - so the timeline usually narrows the field to one or two agents."
3. Quantify severity objectively
- "I estimate the percentage of detached epidermis and calculate a SCORTEN at 24 hours, which both prognosticates and determines whether the patient needs a burns unit."
4. Stop the drug, then support
- "Early withdrawal of the culprit drug is the intervention with the clearest mortality benefit. After that, the evidence base is supportive: fluid and thermal management, wound care leaving blister roofs intact, analgesia, early ophthalmology review, and infection surveillance rather than prophylactic antibiotics."
5. Be honest about immunomodulation
- "There is no consensus agent for SJS/TEN. Ciclosporin and a single dose of etanercept have the best supporting data; IVIg monotherapy is no longer supported. DRESS is different - it needs prednisolone with a slow taper over six to twelve weeks, because rapid withdrawal reliably causes relapse."
6. Finish with the durable intervention
- "The lasting benefit is documentation. I record the exact drug, date and phenotype, counsel lifelong avoidance including cross-reactive agents, warn first-degree relatives where there is an HLA association, and report to the TGA. Equally, where the label is a childhood rash rather than a real reaction, I refer for delabelling - inaccurate allergy labels cause measurable harm."
8 of 8 sections written · drafted 2026-09-12