U.K. guidelines for the management of Stevens–Johnson syndrome/toxic epidermal necrolysis in adults 2016

Purpose and scope

The overall objective of the guidelines is to provide up‐to‐date, evidence‐based recommendations for the diagnosis and management of the full spectrum of Stevens–Johnson syndrome (SJS), toxic epidermal necrolysis (TEN) and SJS–TEN overlap in adults during the acute phase of the disease.

The document aims to: (i) offer an appraisal of all relevant literature up to February 2016, focusing on any key developments; (ii) address important, practical clinical questions relating to the primary guideline objective, i.e. accurate diagnosis and identification of cases and suitable treatment; (iii) provide guideline recommendations; and (iv) discuss areas of uncertainty, potential developments and future directions

SJS/TEN is rare and few healthcare professionals are confident in the recognition and management of the disorder. There is widely divergent practice among different specialities and healthcare settings, and limited information on outcomes. These guidelines aim to provide recommendations on the diagnosis and management of SJS/TEN, to inform clinical decision‐making and, when justified by evidence, to standardize practice. The breadth of this document should be sufficient to assist clinicians of all relevant specialities in the management of patients with SJS/TEN. The recommendations will also inform pathways of care to optimize healthcare delivery and highlight key areas of uncertainty for future research.

In these guidelines, the term SJS/TEN encompasses the full spectrum of the disease, i.e. SJS, TEN and SJS–TEN overlap (see section section 10 for clinical definition of the separate entities). The guideline is presented as a detailed review with highlighted recommendations for practical use (see section section 64), in addition to the development of a new patient information leaflet [available on the British Association of Dermatologists’ (BAD) website, www.bad.org.uk]. Unless otherwise specified, recommendations apply to all forms of the disease.

Exclusions

This guideline does not cover paediatric patients. An addendum to this guideline addressing the needs of paediatric patients with SJS/TEN is planned. The evidence for treatment of children differs from that of adult patients, and will be considered separately.

The management of long‐term sequelae of SJS/TEN is not considered in this document.

Stakeholder involvement and peer review

The guideline development group (GDG) consisted of consultant dermatologists (D.C., S.A.W., H.Y.L., J.S., C.B.B., M.R.A.‐J., K.M.T.W., G.A.E.W., C.H.S.), including oral (J.S.) and urogenital specialists (S.A.W., C.B.B.), a histopathologist (M.P.), burns/plastic surgeon specialists (P.D., G.W., M.S.), a burns anaesthestist (R.V.M.), intensive care specialists (A.V., R.V.M.), an ophthalmologist (J.K.G.D.), a dermatological clinical nurse specialist (D.B.) and a patient (P.W.). The draft document went out for consultation to the BAD membership, British Dermatological Nursing Group, Primary Care Dermatological Society, British Burn Association, British Association of Plastic, Reconstructive and Aesthetic Surgeons, Royal College of Ophthalmologists, Association of Anaesthetists of Great Britain and Ireland, Intensive Care Society and SJS Awareness U.K. The comments received were actively considered by the GDG. Following further review, the amended draft was recirculated to the stakeholders for comments and the finalized version peer‐reviewed by the Clinical Standards Unit of the BAD (made up of the Therapy & Guidelines Subcommittee) prior to publication.

Methodology

This set of guidelines has been developed using the BAD's recommended methodology,1 and with reference to the Appraisal of Guidelines Research and Evaluation (AGREE II) instrument (www.agreetrust.org).2 Recommendations were developed for implementation in the National Health Service using a process of considered judgement based on the evidence. A series of specific questions was developed within the scope of the guideline; for each question, PubMed, MEDLINE and Embase, and the Cochrane Library were searched for meta‐analyses, randomized and nonrandomized controlled clinical trials, open studies and case series to February 2016. The literature was searched for evidence to address the questions. The search terms and strategies are detailed online (Data S1; see Supporting Information). Additional relevant references were also isolated from citations in the reviewed literature, as well as specific targeted searches for hypersensitivity testing, causative drugs, diagnostic and prognostic indicators, topical treatments, oral/mouth, growth factors and granulocyte colony‐stimulating factor (G‐CSF), and analgesia. The lead authors screened the identified titles, and those relevant for first‐round inclusion were selected for further scrutiny. All co‐authors then reviewed the abstracts for the shortlisted references and the full papers of relevant material were obtained; disagreements in the final selections were resolved by discussion with the entire GDG. In section section 47, active interventions were assessed only if studies recruited at least eight patients with SJS/TEN into the treatment group. The structure of the guidelines was then discussed, with headings and subheadings decided; different co‐authors were allocated separate subsections. Each co‐author then performed a detailed appraisal of the selected literature, and all subsections were subsequently collated and edited to produce the final guideline.

Limitations of the guideline

This document has been prepared on behalf of the BAD and is based on the best data available when the document was prepared. It is recognized that under certain conditions it may be necessary to deviate from the guidelines and that the results of future studies may require some of the recommendations herein to be changed. Failure to adhere to these guidelines should not necessarily be considered negligent, nor should adherence to these recommendations constitute a defence against a claim of negligence.

Plans for guideline revision

The proposed revision date for this set of recommendations is scheduled for 2021; where necessary, important interim changes will be updated on the BAD website.

Background

SJS and TEN are severe mucocutaneous reactions, usually to drugs, characterized by blistering and epithelial sloughing.3 The two terms describe phenotypes within a severity spectrum, in which SJS is the less extensive and TEN the more extensive form. The incidence of SJS/TEN is approximately 1–2 cases per million per year.4  5 Although rare, SJS/TEN is a devastating disease; in severe cases the acute phase may be accompanied by a variety of systemic complications, including multiorgan failure. The mortality for SJS is < 10%, with the figure rising to 30% for TEN; overall SJS/TEN mortality is about 22%.6 As well as carrying an appreciable mortality, survivors of the acute illness often develop significant long‐term sequelae.7

SJS/TEN is characterized by widespread epithelial keratinocyte apoptosis and necrosis, a process initiated by drug‐induced cytotoxic T lymphocytes (CTLs).8 Major histocompatibility complex class I‐restricted drug presentation leads to clonal expansion of CD8⁺ CTLs,9 which infiltrate the skin, while soluble factors induce keratinocyte apoptosis.10 Proapoptotic molecules, including tumour necrosis factor (TNF)‐α, interferon (IFN)‐γ and inducible nitric oxide synthase, may link drug‐induced immune responses to keratinocyte damage.11 Although soluble Fas ligand, perforin and granzyme have all been implicated in triggering keratinocyte programmed cell death,12  13 current evidence favours granulysin as the key mediator of apoptosis in SJS/TEN.14 A study by Chung et al. demonstrated the presence of high concentrations of secretory 15‐kDa granulysin in TEN blister fluid, while injection of 15‐kDa granulysin into mouse skin induces keratinocyte apoptosis, mimicking SJS/TEN.14

Diagnosis, initial assessment, drug causality and prognosis in Stevens–Johnson syndrome/toxic epidermal necrolysis

What are the clinical features of Stevens–Johnson syndrome/toxic epidermal necrolysis?

SJS/TEN is an acute, severe dermatosis characterized by epidermal loss and multisite mucositis, accompanied by systemic disturbance. A study by Revuz et al. of 87 consecutive patients managed in a specialist unit gives a detailed description of the clinical features observed in SJS/TEN.15 In general, a prodrome of fever, malaise and upper respiratory tract symptoms precedes the eruption by several days. Ocular inflammation may also develop before the skin signs. Cutaneous pain is a prominent early feature in SJS/TEN, and the presence of this symptom should alert the physician to incipient epidermal necrolysis. Clinical observation in the series of Revuz et al. demonstrated a wide variation in the type of lesion and degree of skin involvement. The earliest lesions are atypical targets and/or purpuric macules (Figs 1 and 2). Initial sites of involvement are commonly the upper torso, proximal limbs and face. Thereafter, lesions spread to involve the rest of the trunk and distal limbs (Fig. 3). Involvement of the palms and soles is often prominent (Fig. 4). Large areas of confluent erythema develop in severe cases (Fig. 5). Lesional skin is tender to touch; minimal shearing forces will cause the epidermis to peel back (Fig. 6). This fragility is demonstrated by the Nikolsky sign, in which gentle lateral pressure causes lesional, detachable epidermis (as opposed to detached epidermis) to slide over the dermis. Although not specific for SJS/TEN (it is also positive in pemphigus), the Nikolsky sign is a helpful clinical indicator of epidermal necrolysis. Involved areas expand and coalesce, reaching a maximum 5–7 days after disease onset. Blistering ensues in which necrotic epidermis separates from underlying dermis, producing flaccid bullae (Fig. 7). Extensive necrolysis results in the detachment of sheets of epidermis, leaving areas of exposed dermis (Fig. 8). Denuded dermis exudes serum, becomes secondarily infected and readily bleeds. Involvement of the mucous membranes of the eyes, mouth, nose and genitalia is usually an early feature and leads to an erosive and haemorrhagic mucositis (Figs 9  10  11). In the series described by Revuz et al. of 87 patients with SJS/TEN, 97% developed erosive mucous membrane lesions; oral involvement was observed in 93% of patients, ocular in 78%, genital in 63% and all three sites in 66%.15 Respiratory tract epithelial necrolysis can occur, resulting in bronchial obstruction and ventilatory compromise; necrolysis of gastrointestinal epithelium leads to profuse diarrhoea. Owing to hypoperfusion and acute tubular necrosis, acute kidney injury may occur in the early phase of SJS/TEN. Mild elevation of liver enzymes is common, but significant hepatic impairment is rare.

Figure 1

Atypical targets. On the palm of this patient with Stevens–Johnson syndrome/toxic epidermal necrolysis there are numerous circular lesions; most are characterized by a dark red centre surrounded by a pink ring. In some areas, the lesions are confluent.

Figure 2

Purpuric macules. This patient with Stevens–Johnson syndrome/toxic epidermal necrolysis developed numerous dark red, flat lesions on the torso and back. Over a few days the lesions had joined to produce large areas of dusky erythema.

Figure 3

Scattered lesions. Discrete lesions in Stevens–Johnson syndrome/toxic epidermal necrolysis may be disseminated widely over the torso and limbs. In this case, the patient developed numerous dusky and blistering lesions on the neck, chest and abdomen. She also has severe mouth involvement and is being fed through a nasogastric tube.

Figure 4

Involvement of plantar skin. Involvement of palms and soles in Stevens–Johnson syndrome/toxic epidermal necrolysis can be prominent and, as at other sites, may blister.

Figure 5

Confluent erythema. Confluent erythema is covering the back; there are small eroded areas of epidermal detachment on the upper back.

Figure 6

Skin shearing. The epidermis within an area of lesional skin on this man's thigh has peeled away following minimal trauma.

Figure 7

Blisters. Numerous large, flaccid blisters developed on the back of this woman who had Stevens–Johnson syndrome–toxic epidermal necrolysis overlap. In her case there was epidermal detachment of 20% of her body surface area.

Figure 8

Epidermal loss. Extensive necrolysis results in the detachment of sheets of epidermis and exposure of large areas of denuded dermis. This patient with toxic epidermal necrolysis had epidermal detachment of 60% body surface area.

Figure 9

Eye involvement. (Top) Early in the acute phase of Stevens–Johnson syndrome/toxic epidermal necrolysis there is a purulent keratoconjunctivitis and eyelid oedema. (Bottom) The same patient 3 days later, there is persistent ocular inflammation and involvement of eyelid skin.

Figure 10

Mouth involvement. (Top) Severe exudative and erosive cheilitis is typical. (Bottom) Erosions may occur at any buccal site within the mouth including, as in this case, the palate.

Figure 11

Genital involvement. Involvement of the urethral meatus is visible, along with lesions elsewhere on the penis and confluent involvement of the scrotum. In this patient, the mucositis also affected the distal urethra.

Despite the striking clinical presentation of SJS/TEN, a number of disorders can present with blistering of the skin and mucous membranes; the clinical differential diagnosis includes pemphigus, pemphigoid, other immunobullous disorders and erythema multiforme major (EMM) (see Table 1).

Table 1

Differential diagnosis of Stevens–Johnson syndrome/toxic epidermal necrolysis

Erythema multiforme major

Pemphigus vulgaris

Mucous membrane pemphigoid

Bullous pemphigoid

Paraneoplastic pemphigus

Bullous lupus erythematosus

Linear IgA bullous dermatosis

Generalized bullous fixed drug eruption

Bullous acute graft‐versus‐host disease

Staphylococcal scalded skin syndrome

Acute generalized exanthematous pustulosis

What are the recognized clinical phenotypes in Stevens–Johnson syndrome/toxic epidermal necrolysis?

In a study by Bastuji‐Garin et al.,16 a group of physicians experienced in managing SJS/TEN reviewed the clinical photographs of > 200 patients and categorized the cases according to type of cutaneous lesion and extent of maximal epidermal detachment: (i) SJS [epidermal detachment < 10% body surface area (BSA) plus widespread purpuric macules or flat atypical targets]; (ii) overlap SJS–TEN (detachment of 10–30% BSA plus widespread purpuric macules or flat atypical targets); (iii) TEN with spots (detachment > 30% BSA plus widespread purpuric macules or flat atypical targets); (iv) TEN without spots (detachment > 30% BSA with loss of large epidermal sheets without purpuric macules or target lesions).

There is acknowledgement that these phenotypic categories are not mutually exclusive; cases with features from two or more of the above groups may occur.17

At the less severe end of the spectrum, differentiation of SJS from EMM can be difficult. In EMM, there is mucous membrane involvement and cutaneous blistering with epidermal detachment of < 10% BSA. However, in contrast to SJS, the skin lesions in EMM consist of typical targets or raised atypical targets, predominantly localized on the limbs and extremities. Distinguishing EMM from SJS/TEN has causality implications: EMM is mostly related to herpes simplex virus (HSV) reactivation and rarely to drugs; SJS/TEN is usually triggered by a drug, rarely by an infection.17  18 Mycoplasma‐induced SJS is reported and in some cases, mostly children, may be characterized by a phenotype of predominantly mucous membrane involvement with little or no cutaneous lesions. This clinically atypical form of SJS has been termed ‘Mycoplasma pneumoniae‐associated mucositis’.19

What are the histopathological features of Stevens–Johnson syndrome/toxic epidermal necrolysis?

Although a diagnosis of SJS/TEN is suggested by the physical signs, histopathology of a skin biopsy is necessary to support the clinical assessment and exclude other blistering dermatoses (see Table 1). Histologically, there is variable epidermal damage, ranging from individual cell apoptosis to confluent epidermal necrosis. Epidermal changes are associated with basal cell vacuolar degeneration and subepidermal vesicle or bulla formation. Adnexal structures such as sweat ducts and hair follicles are occasionally involved. Within the dermis, there is often only a mild, predominantly perivascular infiltrate of lymphocytes and histiocytes, with small numbers of eosinophils present in some cases (Fig. 12).20

Figure 12

Histopathology of Stevens–Johnson syndrome/toxic epidermal necrolysis. There are multiple apoptotic keratinocytes throughout the full thickness of the epidermis, and a subepidermal split forming a bulla. There is a perivascular lymphocytic infiltrate within the dermis.

What initial assessment should be undertaken in a patient presenting with Stevens–Johnson syndrome/toxic epidermal necrolysis to secondary/tertiary care?

If SJS/TEN is suspected, discontinue any potential culprit drug immediately. The patient should undergo an evaluation of the critical disease components. Clinical examination includes an appraisal generic to any acutely ill patient, as well as assessments specific to SJS/TEN. Institution of an immediate management protocol is necessary as soon as the patient has been assessed.

Recommendations [strength of recommendation D (good practice point); level of evidence 4]

Take a detailed history from the patient (and/or relatives) with specific reference to the following points. (i) Ask about symptoms suggestive of SJS/TEN, including a prodromal illness (fever, malaise, upper respiratory tract symptoms); onset of a painful rash, initially on the face and chest; and involvement of mucosal sites (eyes, mouth, nose, genitalia). (ii) Note the date when the rash first appeared and document progression of the eruption. (iii) Ask about symptoms indicating involvement of the respiratory tract: cough, dyspnoea, bronchial hypersecretion and haemoptysis. (iv) Ask about symptoms indicating bowel involvement: diarrhoea and abdominal distension. (v) Determine the index date (date of onset of the adverse reaction) by asking when the patient developed the first symptom or sign of the disorder, for example sore throat, rash, skin pain and sore eyes/mouth. (vi) Record previous or ongoing medical problems; ask specifically about a history of recurrent HSV infections and chest infections. (vii) Record all medicines taken over the previous 2 months, including over‐the‐counter (OTC) and complementary/alternative therapies; document the date treatments were started and the date, when appropriate, of dose escalation; document the date when drugs were stopped. Note if there has been a brand switch or any medication errors. (viii) Record any previous history of drug allergies, including details of the reaction type. (ix) Actively consider other causes of severe skin disease characterized by blistering and involvement of mucus membranes (see Table 1). (x) Perform a full physical examination, recording baseline body weight. Record the vital signs and measure oxygen saturation with a pulse oximeter. (xi) Look for target lesions (particularly atypical targets), purpuric macules, blisters and areas of epidermal detachment. (xii) Examine all mucosal sites, looking for mucositis, blisters and erosions. (xiii) Record the extent of erythema and the extent of epidermal detachment separately on a body map (Fig. 13); for each parameter estimate the percentage of BSA involved using the Lund and Browder chart. Detachment should include detachable epidermis (i.e. Nikolsky positive), as well as detached epidermis; it is this figure, rather than the amount of erythema, that has prognostic value.

Figure 13

Body map schematics of skin involvement in SJS/TEN. (a) Extent of skin erythema (in pink) = 65% body surface area (BSA); extent of epidermal detachment (in red) = 10% BSA. (b) Extent of erythema (in pink) = 90% BSA; extent of epidermal detachment (in red) = 45% BSA.

The following investigations should be undertaken: (i) full blood count, erythrocyte sedimentation rate, C‐reactive protein, urea and electrolytes, magnesium, phosphate, bicarbonate, glucose, liver function tests, coagulation studies, and mycoplasma serology; (ii) chest X‐ray; (iii) a biopsy from lesional skin, just adjacent to a blister, sent for routine histopathology, and a second biopsy taken from periblister lesional skin should be sent unfixed for direct immunofluorescence, to exclude an immunobullous disorder; (iii) swabs from lesional skin for bacteriology; (iv) organize photographs of the skin to show type of lesion and extent of involvement.

Initiate a primary management plan. (i) Discontinue any potential culprit drug causing SJS/TEN immediately (see Table 2). (ii) Establish peripheral venous access; where possible, insert the cannula through nonlesional skin; commence appropriate intravenous (IV) fluid resuscitation if clinically indicated (see section section 23). A fluid chart should be initiated. (iii) Ascertain if the patient can maintain adequate hydration and nutrition orally; if this is not possible, insert a nasogastric tube and institute nasogastric feeding (see section section 25). (iv) Insert a urinary catheter when urogenital involvement is causing significant dysuria or retention. A urinary catheter will also permit accurate output monitoring to assist fluid replacement.

Table 2

Most common drugs causing Stevens–Johnson syndrome/toxic epidermal necrolysis

Allopurinol

Carbamazepine

Lamotrigine

Nevirapine

Oxicam NSAIDs

Phenobarbital

Phenytoin

Sulfamethoxazole and other sulfur antibiotics

Sulfasalazine

NSAIDs, nonsteroidal anti‐inflammatory drugs.

Is there a clinical method of determining drug causality in Stevens–Johnson syndrome/toxic epidermal necrolysis?

SJS/TEN is primarily a drug‐induced phenomenon, with a culprit drug being demonstrated in approximately 85% of cases.21 Identification of the causative agent may be straightforward in cases where a single drug is implicated, but difficulties are posed by a patient who has been exposed to multiple drugs. Early withdrawal of the suspected agent is mandatory, as this decreases the risk of death.22 The patient's other regular medicines should be continued.

Notoriety of specific drugs in causing SJS/TEN can be determined from population pharmacovigilance data, which are useful in identifying a likely culprit.6  23  24  25  26  27  28 Two multinational case–control studies have evaluated drug causation risk: the first, conducted from 1989 to 1995, included 372 cases and 1720 controls;23 the second, carried out between 1997 and 2001, consisted of 379 cases and 1505 controls.6 From case–control studies, a list of drugs that are strongly associated with the induction of SJS/TEN has been drawn up; these medicines are responsible for one‐half of all cases (see Table 2). Paracetamol, aspirin, ibuprofen and corticosteroids have an unclear association but are likely to be confounders used to treat prodromal symptoms of SJS/TEN.

An algorithm, termed ALDEN (ALgorithm of Drug causality in Epidermal Necrolysis), has been developed to help define drug causality in SJS/TEN.21 ALDEN is generally used as a tool for retrospective assessment of drug causality, and not for use in the acute phase of illness. However, the key parameters described in ALDEN can be applied as a useful framework for determining drug culpability in clinical practice.

Recommendations (strength of recommendation D; level of evidence 3)

List all medications (including OTC preparations) taken by the patient over a period of 2 months prior to the onset of symptoms. Use multiple sources [including patient, relatives, general practitioner (GP), pharmacist] to obtain a full drug history.

Delineate a timeline for each drug, identifying the date the drug was commenced and, when appropriate, discontinued. Tools to facilitate timeline analysis include web‐based freeware such as www.drugrash.co.uk. A latent period between the initial drug intake and onset of SJS/TEN always occurs; 5–28 days following drug initiation is the most likely period, unless there is a history of a previous reaction to the same drug, in which case the latency may be shorter.

Estimate the probability that the drug was present in the body at the onset of the reaction by taking into consideration pharmacokinetic parameters of the drug (e.g. half‐life), any renal or hepatic dysfunction, and possible drug interactions, which might lead to higher or lower levels of drug in the body.

For each drug taken by the patient, document any previous exposures and any previous adverse reactions. Implication of a particular medicine is more likely if the patient gives a past history of a drug hypersensitivity reaction with the same or similar drug.

For each drug taken by the patient, assess notoriety to cause SJS/TEN (see Table 2).

When a causative drug cannot be identified, consider other possible aetiological factors, such as mycoplasma infection.

What is the value of prognostic scoring in Stevens–Johnson syndrome/toxic epidermal necrolysis?

In severe cases of SJS/TEN the acute systemic sequelae lead to multiorgan failure and death. Papers investigating potential prognostic markers in SJS/TEN have identified several factors associated with death, including delayed transfer to a specialist unit, increasing patient age, increasing total BSA involvement, presence of septicaemia and occurrence of granulocytopenia.29  30

In 2000 Bastuji‐Garin et al. published a validated prognostic scoring system for SJS/TEN, called SCORTEN, which uses seven clinical parameters to predict probability of hospital mortality (see Tables 3 and 4).31 In SCORTEN, one point is attributed to each of the seven parameters, with increasing scores predicting higher mortality rates.

Table 3

SCORTEN calculation

Age > 40 years

Presence of malignancy

Heart rate > 120 beats min−1

Epidermal detachment > 10% BSA at admission

Serum urea > 10 mmol L−1

Serum glucose > 14 mmol L−1

Bicarbonate < 20 mmol L−1

BSA, body surface area.

Table 4

SCORTEN predicted mortality

Number of parameters	Predicted mortality (%)
0	1
1	4
2	12
3	32
4	62
5	85
6	95
7	99

A longitudinal assessment of 144 patients demonstrated that SCORTEN rises slightly during hospitalization, with a significant difference between day 1 and day 4.32 Since the publication of the paper by Bastuji‐Garin et al.,31 retrospective analysis of several SJS/TEN case series has confirmed SCORTEN's ability to predict mortality accurately.33  34  35  36  37

Recommendation (strength of recommendation C; level of evidence 2+)

SCORTEN should be calculated in all patients with SJS/TEN within the first 24 h of admission.

Initial management and supportive care

What care setting is needed for the management of patients with Stevens–Johnson syndrome/toxic epidermal necrolysis?

Extensive epidermal detachment in SJS/TEN is accompanied by thermoregulatory dysfunction, large insensible fluid loss and haemodynamic instability. Other sequelae of acute skin failure include anaemia, leucopenia, renal impairment, liver dysfunction and systemic sepsis, the last being the most frequent cause of death. It is therefore recommended that patients with large areas of epidermal loss (> 10% BSA) are admitted to a specialist intensive care unit (ICU) for critical care management and specialist nursing (see Fig. 14). As the cutaneous defect in SJS/TEN is analogous to a superficial burn, many patients are transferred to a burn centre that can deliver both intensive supportive management and skin‐directed therapy. Staff, facilities and standard operating procedures in a burns centre are well defined for the management of skin failure and extensive skin loss. Three studies and a systematic review of TEN cases have demonstrated that rapid admission to a burns centre is associated with improved survival, while delayed transfer is accompanied by increased mortality.29  38  39  40

Figure 14

Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN) pathway of care. MDT, multidisciplinary team; ICU, intensive care unit.

Provision of the many specialist services necessary for managing SJS/TEN cases engenders a need for multidisciplinary team (MDT) working. The SJS/TEN MDT should be coordinated by a specialist in skin failure, usually dermatology and/or plastics, and should include clinicians from intensive care, ophthalmology and skincare nursing. Within an ICU (specialist or burn centre) the patient should be nursed in a side room to facilitate barrier nursing and to allow the immediate environment to be heated. Studies in burns patients have demonstrated that, at room temperature, energy expenditure increases by 40% of basal metabolic rate (BMR) with skin loss of 10% BSA, while at 80% BSA skin loss energy expenditure increases by 120% of BMR.41  42 The same thermoregulatory dysfunction occurs in extensive epidermal necrolysis and therefore a raised ambient temperature is necessary in SJS/TEN patient care to reduce energy consumption and associated metabolic stresses.

Recommendations [strength of recommendation D (good practice point); level of evidence 4]

Patients with SJS/TEN with > 10% BSA epidermal loss should be admitted without delay to a burn centre or to an ICU with experience of treating patients with SJS/TEN and facilities to manage the logistics of extensive skin loss wound care.

The SJS/TEN MDT should be coordinated by a specialist in skin failure, usually dermatology and/or plastic surgery, and should include clinicians from intensive care, ophthalmology and specialist skincare nursing. Additional clinical input to the MDT may be required from respiratory medicine, gastroenterology, gynaecology, urology, oral medicine, microbiology, pain team, dietetics, physiotherapy and pharmacy.

Patients with SJS/TEN must be barrier‐nursed in a side room controlled for humidity, on a pressure‐relieving mattress, with the ambient temperature raised to between 25 °C and 28 °C.

What skin management regimen should be followed in Stevens–Johnson syndrome/toxic epidermal necrolysis?

In SJS/TEN, necrotic epidermis is prone to detach from underlying dermis; therefore, careful handling of the skin in these patients is essential. In particular, care must be taken to minimize shearing forces applied to the skin, especially when moving and positioning the patient (antishear handling).43 Day‐to‐day bedside care is preferably undertaken by specialist nurses familiar with skin fragility disorders.43 Other attending clinicians, who are unfamiliar with the problems of epidermal detachment, should be warned before examining the patient. Despite careful nursing, lesional epidermis in SJS/TEN often peels away, especially at pressure areas, to leave zones of denuded dermis.

There are differing approaches to local management of lesional skin with no good evidence as to which is superior.

In the conservative approach, detached epidermis is left in situ to act as a biological dressing for the underlying dermis. In cases where bullae are prominent, blister fluid should be aspirated or expressed, and the blister roof allowed to settle onto the underlying dermis.44  45  46 Frequent application of a bland emollient to the whole skin is helpful during the acute phase of SJS/TEN to support barrier function, reduce transcutaneous water loss and encourage re‐epithelialization.44 The use of an appropriate dressing on exposed dermis will reduce fluid and protein loss, limit microbial colonization and help pain control. Covering the denuded skin may also accelerate re‐epithelialization. Once active blistering and epidermal detachment ceases, re‐epithelialization commences. Healing may occur within a few days or may be protracted and take a number of weeks to attain completion.

A surgical approach involves debridement of detached epidermis to remove potentially infected material followed by physiological wound closure using biosynthetic dressings, xenograft or allograft.40  47 This more aggressive approach can be considered following failure of conservative management, characterized by clinical deterioration, extension of epidermal detachment, local sepsis/subepidermal pus, delayed healing and wound conversion (the spontaneous progression of superficial skin loss into a deeper cutaneous defect).

Denuded dermis in SJS/TEN exudes serum and becomes coated with necrotic debris. The exposed dermis and haemorrhagic crust act as a substrate for microbial colonization, initially by Staphylococcus aureus and later by Gram‐negative rods from the digestive flora, especially Pseudomonas aeruginosa.7 Cutaneous infection will impair re‐epithelialization and may lead to systemic sepsis, which is the most common cause of death in SJS/TEN. Indiscriminate administration of prophylactic systemic antibiotics may increase skin colonization, particularly with Candida albicans; therefore, antimicrobial therapy should only be instituted if there are clinical signs of infection.7 The SJS/TEN disease process may be accompanied by a fever, which complicates detection of secondary sepsis; therefore, patients should be monitored carefully for other signs of systemic infection such as confusion, hypotension, reduced urine output and reduced oxygen saturation.48 Cutaneous infection may be accompanied by an increase in skin pain. The detection of sepsis may also be indicated by a rise in C‐reactive protein and neutrophilia. If a monoculture of organisms is detected on culture of swabs taken from multiple sites, which had previously showed mixed growth, this sign indicates that one particular strain of organism is becoming predominant and increases the likelihood of invasive infection.48 Consider activation of HSV in eroded or vesicular areas that are slow to heal, particularly in genital and oral sites.

Recommendations [strength of recommendation D (good practice point); level of evidence 4]

The following general principles are applicable to all patients with SJS/TEN and in all settings: (i) employ strict barrier nursing to reduce nosocomial infections; (ii) careful handling of the patient's skin and reduction of shearing forces will lessen the extent of epidermal detachment; (iii) limit epidermal trauma by avoiding the use of sphygmomanometer cuffs, adhesive electrocardiography leads, adhesive dressings and identification wrist tags; (iv) take swabs for bacterial and candidal culture from three areas of lesional skin, particularly sloughy or crusted areas, on alternate days throughout the acute phase of SJS/TEN; (v) take viral swabs if herpes virus infection is suspected; (vi) administer systemic antibiotics only if there are clinical signs of infection (the choice of systemic antibiotic should be guided by local microbiological advice); (vii) in patients with diarrhoea and who are immobile consider a faecal management system to prevent faecal soiling of wounds; (viii) pay strict attention to background and procedural pain/sedation requirements.

Skin management may involve a conservative and/or surgical approach based on the specialist MDT's daily review of the individual needs of the patient

Institute a conservative approach in all patients as follows: (i) Regularly cleanse wounds and intact skin by irrigating gently using warmed sterile water or saline or an antimicrobial such as chlorhexidine (1/5000). (ii) Apply a greasy emollient, such as 50% white soft paraffin with 50% liquid paraffin, over the whole epidermis, including denuded areas; consider using aerosolized formulations to minimize shearing forces associated with topical applications. Avoid preparations containing sensitizers or irritants. (iii) Apply a topical antimicrobial agent to sloughy areas only. The choice of topical antibiotic should be guided by local microbiological advice. Consider silver‐containing products/dressings (owing to risk of absorption, use of silver‐containing products should be limited if extensive areas are being treated). (iv) The detached, lesional epidermis may be left in situ to act as a biological dressing. Blisters should be decompressed by piercing and expression or aspiration of tissue fluid. (v) Apply nonadherent dressings to denuded dermis [suitable dressings include Mepitel^™ (Mölnlycke Health Care, Dunstable, U.K.) or Telfa^™ (Covidien, Mansfield, MA, U.S.A.)]. (vi) A secondary foam or burn dressing should be used to collect exudate (suitable dressings include Exu‐Dry^®; Smith & Nephew, London, U.K.).

Consider transfer to a burn centre in patients with TEN (> 30% BSA epidermal loss) and evidence of the following: clinical deterioration, extension of epidermal detachment, subepidermal pus, local sepsis, wound conversion and/or delayed healing. In a burn centre, conservative measures may be supplemented with a surgical approach: (i) remove necrotic/loose infected epidermis and clean wounds using a topical antimicrobial agent (e.g. betadine or chlorhexidine) under general anaesthetic; (ii) consider debridement with Versajet^™ (Drytac, Bristol, U.K.); (iii) physiological closure with Biobrane (Smith & Nephew)/allograft/xenograft skin in patients with early presentation involving noninfected and large confluent areas.

What fluid replacement regimen should be followed in Stevens–Johnson syndrome/toxic epidermal necrolysis?

A major goal of supportive care in SJS/TEN is fluid resuscitation to prevent end‐organ hypoperfusion and shock. Extensive epidermal detachment will result in large insensible, transcutaneous fluid losses, which is compounded by decreased oral intake due to disease involvement of the mouth. In patients with burns the predicted volume of fluid replacement is proportional to the surface area of burn involvement: the Parkland formula is generally used, with fluid resuscitation in adults being commenced at 15% BSA involvement.49 In SJS/TEN, requirements are lower than those predicted by Parkland; overaggressive fluid resuscitation may be associated with pulmonary, cutaneous and intestinal oedema. A study by Shiga and Cartotto of 21 patients with TEN with extensive epidermal loss recorded fluid requirements over the first 3 days of admission and estimated that replacement volumes can be determined by the following formula: 2 mL kg⁻¹ body weight/% BSA epidermal detachment.50

Recommendations (strength of recommendation D; level of evidence 3)

Insert peripheral and central venous lines through nonlesional skin, whenever possible, and change peripheral venous cannulas every 48 h, if possible.

Monitor fluid balance carefully; catheterize if appropriate/necessary.

Fluid replacement can be guided by urine output and other end‐point measurements (see below). Individualized fluid management should be adjusted on a daily basis

When necessary, use continuous invasive haemodynamic monitoring through a central line to guide fluid resuscitation. Where estimation of fluid balance is challenging in severely affected patients, use central venous saturation and flow monitoring based on pulse contour analysis of arterial waveforms. Serial serum lactate measurements may also help to detect tissue hypoperfusion.

After establishing adequate IV fluid replacement initially, oral administration of fluids should be progressively increased, if tolerated.

What nutrition regimen should be followed in Stevens–Johnson syndrome/toxic epidermal necrolysis?

SJS/TEN is characterized by a hypermetabolic response, with energy expenditure approximately twice the predicted resting value.51  52  53 Extensive epidermal detachment is also associated with loss of large amounts of albumin and protein from blister fluid. Therefore, in SJS/TEN cases with significant areas of skin involvement, a nutritional regimen must be initiated early to support metabolic disturbances, minimize protein losses and promote healing. As in other intensive care situations, enteral nutrition is preferable to parenteral nutrition to reduce peptic ulceration and limit translocation of gut bacteria. As buccal mucositis in SJS/TEN often precludes normal oral intake, nasogastric feeding with a silicone tube should be instituted when necessary. General principles of intensive care nutrition are applicable and these are summarized in The European Society for Clinical Nutrition and Metabolism guidelines.42  54

Recommendations (strength of recommendation C; level of evidence 3)

Provide continuous enteral nutrition throughout the acute phase of SJS/TEN, either by the oral route or via nasogastric feeding if the former is precluded by buccal mucositis.

Deliver up to 20–25 kcal kg⁻¹ daily during the early, catabolic phase of SJS/TEN. During the anabolic, recovery phase, the aim should be to provide between 25 and 30 kcal kg⁻¹ daily.

What analgesia is required in Stevens–Johnson syndrome/toxic epidermal necrolysis?

SJS/TEN is characterized by cutaneous pain, which is most severe at sites of epidermal detachment. There are no studies investigating different analgesic regimens in SJS/TEN. In the absence of disease‐specific evidence, patient comfort should be ensured using the principles of the World Health Organization's analgesic ladder (www.who.int/cancer/palliative/painladder/en/).55

Patients should receive adequate background simple analgesia to ensure comfort at rest, with the addition of opiates, as required, delivered enterally, or by patient‐controlled analgesia (PCA), or via infusion. Involvement of the skin of the hands by SJS/TEN may limit the ability of the patient to operate a PCA device. If the patient is in moderate‐to‐severe pain, which is uncontrolled by simple analgesia, then an opiate‐based regimen using morphine should be initiated. Careful monitoring of level of consciousness, respiratory rate and oxygen saturation is essential for safe delivery of opiate infusions. Additional analgesia is often needed to address increased pain associated with patient handling, repositioning, dressing changes and physiotherapy.56 Intranasal diamorphine or sublingual fentanyl can be useful for more limited procedures, unless extent of nose and mouth involvement precludes use of these administration routes.

Adjuvants, including γ‐aminobutyric acid analogues, may have an opiate‐sparing role. Topical anaesthesia of mucous membranes may facilitate placement of nasogastric tubes and urinary catheters.

Recommendations [strength of recommendation D (good practice point); level of evidence 4]

Use a patient‐appropriate validated pain tool to assess pain in all conscious patients at least once a day.56 If the score is mild, pain control with regular paracetamol (acetaminophen) should be introduced, supplemented if required with oral opiate‐based therapy such as codeine, or a synthetic opiate such as tramadol. Nonsteroidal anti‐inflammatory drugs should be avoided because of the potential for renal and gastric injury.

If the pain score equates to moderate or severe pain then prescribe regular opiate‐based analgesia (e.g. morphine or fentanyl) delivered enterally, or by PCA, or via infusion. IV opiates may be safely delivered in dedicated skin failure burn centres or ICUs, with appropriate nursing levels. In patients needing opiate‐based analgesia, pain should be re‐evaluated using the pain score on a 4‐hourly basis and prior to any interventions. Involve the hospital acute pain team early.

Procedures such as dressing changes and bathing may require supplementary analgesia. Entonox^™ (50% nitrous oxide, 50% oxygen; BOC Healthcare, Manchester, U.K.) may be useful for this purpose when the area of cutaneous involvement is small (< 10% BSA). Patients with significant areas of epidermal detachment will require additional pain relief for procedures. Practitioners trained in sedation techniques should deliver this in a safe, monitored environment, such as an ICU, operating theatre or burn centre. Techniques to consider include target‐controlled remifentanil infusions or bolus ketamine‐based analgo‐sedation. Some patients may require general anaesthesia.

What additional supportive medication is advisable in a patient with Stevens–Johnson syndrome/toxic epidermal necrolysis?

As in other critical care situations, patients with SJS/TEN are subject to stress‐related gastric or duodenal ulceration and, if immobile, at risk of venous thromboembolism. Gastric protection with a proton pump inhibitor is recommended, in the minority in whom enteral nutrition cannot be established.57 Prophylactic anticoagulation with low molecular weight heparin is necessary, unless contraindicated.

Anaemia and leucopenia are common complications of the acute phase of SJS/TEN. Neutropenia will increase the risk of life‐threatening sepsis and therefore administration of recombinant human G‐CSF has been used to resist infectious complications.58 It has also been suggested that the use of G‐CSF in SJS/TEN may be immunomodulatory and enhance re‐epithelialization.58

Recommendations (strength of recommendation C; level of evidence 3)

Patients with SJS/TEN who are immobile in bed should receive low molecular weight heparin as prophylactic anticoagulation against venous thromboembolism.

During the acute phase of SJS/TEN, patients in whom enteral nutrition cannot be established may benefit from a proton pump inhibitor to protect against upper gastrointestinal stress ulceration.

Patients with SJS/TEN who are neutropenic may benefit from the administration of recombinant human G‐CSF.

Treatment of eye involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis

Acute ocular involvement in SJS/TEN usually occurs concurrently with skin disease but may develop before or after the appearance of cutaneous signs. A longitudinal study by Gueudry et al. of 159 patients with SJS/TEN demonstrated that 74% (n =117/159) suffered eye involvement during the acute phase of the illness.59 Inflammation of the mucosal surfaces of the eye and eyelids is accompanied by chemosis, conjunctivitis, pseudomembrane formation, and corneal and conjunctival epithelial defects. Power et al. described a grading system for ocular SJS/TEN, which characterizes the acute eye complications (see Table 5).60 Of the 159 patients in the study by Gueudry et al.,59 58% (n =92/159) had mild involvement, 8% (n =12/159) moderate and 8% (n =13/159) severe; in the remaining 26% (n =42/159) of patients there was either no acute eye involvement or no recording of acute eye features.59 Patients with TEN had more frequent, but not more severe, acute ocular involvement than patients with SJS. Sixty‐three percent (n =31/49) of those with acute ocular involvement went on to develop chronic complications.59 Yip et al. also used the Power et al. classification to assess the outcomes of 117 patients with SJS/TEN of whom 81 (69%) had acute ocular involvement.61 This was mild in 40%, moderate in 25% and severe in 4%. Forty‐four patients had a minimum of 6 months of follow‐up; 50% developed late complications, of which the most common were severe dry eyes and trichiasis. There was no difference in the severity of acute eye involvement or late complications when patients with SJS and TEN were compared. In addition, the severity of the acute ocular disease was not a risk factor for late complications. The conclusions that can be drawn from these studies are that neither the severity of the systemic disease nor the grade of the acute ocular disease are predictive for late ocular complications. Because the latter may not become apparent until late after the acute episode, clinicians should be aware of this and make appropriate follow‐up arrangements. However, effective management of ocular inflammation in the acute stages of SJS/TEN, summarized in Table 5, will prevent or mitigate acute blinding, corneal complications and may reduce the severity of the chronic eye disease.

Table 5

Classification of acute ocular involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis

Mild	Eyelid oedema
	±mild conjunctival injection
	±chemosis
Moderate	Membranous conjunctivitis
	±corneal epithelial defects (> 30% healing with medical therapy)
	±corneal ulceration
	±corneal infiltrates
Severe	Symblepharon formation
	±nonhealing corneal epithelial defects
	±visual loss
	±conjunctival fornix foreshortening

Recently, a simplified system for grading ocular involvement has been published (Table 6).62 The grading is related both to predisposing factors for severe ocular complications and to poor late outcomes of visual disturbance and dry eye. It seems to be more predictive than the system of Power et al.60  62  63

Table 6

Grading scores for acute ocular severity of Stevens–Johnson syndrome/toxic epidermal necrolysis

Grade	Acute ocular manifestations
0 (none)	No ocular involvement
1 (mild)	Conjunctival hyperaemia
2 (severe)	Either ocular surface epithelial defect or pseudomembrane formation
3 (very severe)	Both ocular surface epithelial defect and pseudomembrane formation

Table reproduced from Sotozono et al.62

These long‐term ocular manifestations, which constitute the most significant sequelae of SJS/TEN, include corneal and conjunctival scarring, severe dry eye due to aqueous tear deficiency, secondary to both cicatrization of the lacrimal ductules and accessory lacrimal glands, as well as meibomian gland dysfunction, mucin deficiency secondary to goblet cell destruction, distichiasis, entropion, trichiasis and ocular surface failure.59 Patients with chronic eye involvement require lifelong management for dryness, conjunctival inflammation and ocular discomfort.

Management of the acute ophthalmological consequences of SJS/TEN demands the minimization of destructive ocular surface and lid margin inflammation, the management and prevention of conjunctival adhesions, infection prophylaxis, and the prompt identification and management of blinding complications of corneal exposure, ulceration and infection.

Is topical therapy effective in treating ocular involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis?

Local ocular measures are supportive and empirical. Ocular surface lubrication and conjunctival hygiene must be maintained throughout the acute illness. If a patients with SJS/TEN in an ICU or a burn centre is semiconscious or unconscious, prevention of corneal exposure is essential.64  65 Compromise of corneal defences, or proven ocular infection, are indicators for the use of a topical antibiotic. Topical corticosteroids are commonly used to ameliorate conjunctival inflammation. A retrospective study of SJS/TEN cases compared 33 patients receiving ocular topical corticosteroid during the acute phase with 31 patients treated with lubricant alone;66 no information was given concerning the dose, length of treatment or type of topical corticosteroid used. Visual outcomes, assessed at an unspecified time after the acute episode, were significantly better in the group receiving topical corticosteroid compared with the lubricant group.66

Recommendations [strength of recommendation D (good practice point); level of evidence 4]

The eyes should be examined by an ophthalmologist as a part of the initial assessment of a patient with SJS/TEN. Thereafter, daily ophthalmological review is necessary during the acute illness. A direct ophthalmoscope to provide illumination, magnification and a cobalt blue light source for fluorescein examination is useful. Two‐hourly application of a lubricant (e.g. nonpreserved hyaluronate or carmellose eye drops) should be started immediately and continued through the acute illness.

Ocular hygiene, to remove inflammatory debris and break down conjunctival adhesions, must be carried out each day by an ophthalmologist or ophthalmically trained nurse, and can be performed using saline irrigation, a squint hook and forceps. Scissors may be needed when adhesions are well developed and cannot be removed with forceps alone. Blind sweeping of the fornices with a cotton bud or glass rod is not recommended and may potentially cause damage. The application of a topical local anaesthetic (e.g. proparacaine or tetracaine) is necessary prior to the procedure.

In the unconscious patient, prevention of corneal exposure is essential to reduce the risk of ulceration and infection. Establishing a moisture chamber with polyethylene film should be used to maintain corneal epithelial integrity. On healthy eyelid skin, paper tape can be used to affix the moisture chamber; care must be taken if there is sloughing of the eyelid skin. All ICUs and burn centres should have corneal exposure prophylaxis protocols in place.

Broad‐spectrum topical antibiotic prophylaxis is recommended by the authors in the presence of corneal fluorescein staining or frank ulceration, when microbial keratitis has been excluded. The choice of antibiotic should be guided by local knowledge of antimicrobial resistance patterns, which vary widely in different countries. In the U.K. a quinolone preparation is recommended, such as moxifloxacin or levofloxacin, which has a wide range of activity against Gram‐positive and Gram‐negative organisms, used four times daily. Other U.K. experts recommend conjunctival cultures on admission and prophylaxis guided by the sensitivity results.

For suspected corneal infection, which may manifest as corneal stromal loss in the absence of an infiltrate, culture‐guided treatment is mandatory following initial hourly use of broad‐spectrum topical antibiotic therapy, according to local protocols, and may be modified by microbial sensitivity results when these become available. Candida keratitis is relatively common in patients with surface disease so that cultures for both bacteria and fungus are required.

The use of topical corticosteroid drops (e.g. nonpreserved dexamethasone 0.1%), supervised by an ophthalmologist, may reduce ocular surface damage in the acute phase of SJS/TEN. Topical corticosteroids can mask the signs of corneal infection and should be used with caution in the presence of a corneal epithelial defect.

Nonpreserved (as opposed to preserved) topical eye drops are recommended; these are available in the U.K. for all the preparations mentioned above.

Is systemic therapy effective in treating ocular involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis?

Two studies have attempted to assess the efficacy of systemic therapy on ocular outcomes in SJS/TEN. A study by Araki et al. reported five patients with acute ocular SJS/TEN receiving pulsed IV methylprednisolone, followed by a tapering course of oral prednisolone;67 there was no control group. Treatment was started at a mean of 1.2 days after SJS/TEN onset. At 12 months, the best‐corrected visual acuity (VA) was 20/20 or better; there was slight discomfort in 10/10 eyes, necessitating artificial tears; there was no evidence of epithelial defects, neovascularization, opacification or keratinization.67

In a retrospective case series from Yip et al. the ocular outcomes in eight patients with TEN treated with IV immunoglobulin (IVIg) were compared with a group of eight historical controls, five of whom received oral prednisolone.68 The patients in the IVIg‐treated group were all survivors (patients who died were excluded from study). A total of 25% (n =2/8) of patients had mild eye involvement, 50% (n =4/8) were moderately affected and 25% (n =2/8) had severe eye involvement. All patients received 2 g kg^–1 IVIg over 2 days. In total, 87% (n =7/8) received oral prednisolone prior to IVIg; 37% (n =3/8) received a tapering course of oral prednisolone after IVIg.68 The outcome measures were ocular complications and VA 6 weeks post‐IVIg. There was no difference in severity of visually significant ocular complications between the IVIg and control groups.68

A further retrospective paper describes the effect of systemic immunomodulatory treatment on ocular outcomes in 43 patients from three centres treated with five systemic therapies: steroids (n =18), IVIg (n =5), steroids with IVIg (n =14), systemic pulse steroids (n =3) and supportive care only (n =3). This paper concluded that the grading system used had prognostic value for poor ocular outcomes but that their study could not demonstrate therapeutic benefit from the use of systemic immunomodulatory treatment to mitigate the ocular complications.62  63

When given in the acute phase of SJS/TEN, there is no robust evidence for the benefit of systemic corticosteroids or IVIg to improve ocular outcomes. Further studies are required to establish the role of either intervention.

Is the use of amniotic membrane transplantation effective in treating ocular involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis?

There are a number of published cases, and one case–control trial, reporting good ocular outcomes following amniotic membrane transplantation (AMT) during the acute phase of SJS/TEN.69  70  71  72  73  74  75  76 The suggested benefits of AMT include reduced inflammation, enhanced re‐epithelialization, reduction of scarring and less symblepharon formation.69  70  71  72 One case report of bilateral AMT suggests that early AMT may provide better outcomes than later application.77 Cryopreserved amniotic membrane (obtained from the National Health Service Blood and Transplant Tissue Services, Liverpool, U.K.) is now available in a large (5 × 5 cm) size, as well as smaller sizes. The large size is ideal for lining the entire ocular surface. AMT may be sutured onto the ocular surface, usually under general anaesthetic or deep sedation. The technique is described both by Muqit et al. (using a procedure also favoured by the authors, incorporating absorbable sutures that do not require a further anaesthetic for removal)78 and by Gregory,70 who recommends a similar procedure, but incorporating nonabsorbable sutures and bolsters, which adds steps to the operation that may be unnecessary. Videos are attached to both papers. For sutured AMT the risks are low, without complications reported in the available case series, and the procedure is not complex but time consuming. It takes about 90 min to perform on each eye, and is easier to carry out in an operating room under general anaesthesia, although, as with most ophthalmological procedures, it can be carried out under local anaesthesia. After the procedure, a symblepharon ring or conformer may be inserted. Symblepharon rings and conformers are available in the U.K. (Orbital Prosthetic Supplies, Crawley, U.K.; www.orbitalprosthetic.com): their use and sizing has been described elsewhere.79 AMT performed at the bedside under topical anaesthesia has been published in four reports of eight SJS/TEN cases.72  73  74  75 In these patients, amniotic membrane was applied using a proprietary device and the tissue clipped into a plastic symblepharon ring (Prokera Bio‐Tissue, Miami, FL, U.S.A.).

It is possible to retain large sheets of amniotic membrane in the conjunctival sac using symblepharon rings or conformers only.80 The procedure can be performed at the bedside relatively easily. A large retrospective study comparing the results of AMT in 39 sutureless, and 36 sutured cases, demonstrated better outcomes in sutureless than in sutured AMT.80

A retrospective study from one unit compared 17 patients with SJS/TEN (33 eyes) receiving medical eye management, with 13 patients (25 eyes) treated with AMT performed in the acute phase of the disease.76 VA assessed within 3 months of treatment demonstrated significantly better outcomes in the AMT group.76

Recommendations (strength of recommendation D; level of evidence 3)

For patients in whom ocular hygiene is impossible without general anaesthesia, and in those with extensive loss of ocular surface epithelia unresponsive to conservative measures, the use of AMT should be considered. It may improve outcomes if this procedure is carried out earlier rather than later in cases with severe ocular surface ulceration. Sutureless AMT using a symblepharon ring, and performed at the bedside is a new, promising, low‐risk technique for delivering this therapy.

Treatment of mouth involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis

Oral involvement in SJS/TEN is characterized by painful mucosal erythema with subsequent blistering and ulceration. Similar involvement of the vermillion of the lips progresses to haemorrhagic sloughing with the development of dark adherent crusts. The tongue and palate are frequently affected, while in severe cases, mucosal involvement may extend to the oropharynx, larynx, respiratory tract and oesophagus. Drinking and eating are usually severely compromised by oral involvement in acute SJS/TEN. If tolerated, ingested foods need to be soft, moist and low in acidity. However, fluids usually need to be given intravenously and nutrition supplied via a soft, fine‐bore nasogastric tube (see section section 25).

A long‐term complication of acute oral involvement is labial and intraoral scarring, which may restrict mouth opening and cause difficulty with eating or speaking.81 Sicca syndrome, caused by damage to minor salivary glands, develops as a chronic problem in up to 40% of patients.82

Is topical therapy effective in treating oral involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis?

Local measures for treating oral involvement in acute SJS/TEN are supportive and empirical. Along with regular examination of the mouth and lips, attention should be given to regular emollients, topical analgesia and topical antiseptics. In an uncontrolled series of patients with a variety of blistering conditions affecting the mouth, including SJS/TEN, topical corticosteroids were shown to reduce oral inflammation.83 Topical corticosteroids are widely prescribed in oral SJS/TEN, but there is limited directly applicable evidence for their use.

Recommendations [strength of recommendation D (good practice point); level of evidence 4]

The mouth should be examined as a part of the initial assessment of a patient with SJS/TEN. Thereafter, daily oral review is necessary during the acute illness.

Apply white soft paraffin ointment to the lips immediately, and then every 2 h throughout the acute illness. Protect ulcerated mucosal surfaces with a mucoprotectant mouthwash, used three times a day (e.g. Gelclair^®; Alliance Pharmaceuticals, Chippenham, U.K.). Clean the mouth daily with warm saline mouthwashes or an oral sponge, sweeping the sponge gently in the labial and buccal sulci to reduce the risk of fibrotic scars.

Use an anti‐inflammatory oral rinse or spray containing benzydamine hydrochloride every 3 h, particularly before eating. If pain is inadequately controlled with benzydamine, then a topical anaesthetic preparation, e.g. viscous lidocaine 2%, 15 mL per application, may be used as an alternative. Cocaine mouthwashes 2%–5% can be used for severe oral discomfort three times daily.

Use an antiseptic oral rinse twice daily to reduce bacterial colonization of the mucosa. Agents available include 1.5% hydrogen peroxide mouthwash (e.g. Peroxyl^® mouthwash, 10 mL twice daily; Colgate, New York, NY, U.S.A.) or 0.2% chlorhexidine digluconate mouthwash (e.g. Corsodyl^® mouthwash, 10 mL twice daily; GSK, Brentford, U.K.). Diluting 0.2% chlorhexidine mouthwash by up to 50% will reduce the soreness that can accompany this treatment.

Oral and lip swabs should be taken regularly if bacterial or candidal secondary infection is suspected. Candidal infection should be treated with nystatin oral suspension 100 000 units four times daily for 1 week, or miconazole oral gel (e.g. Daktarin^® oral gel; Johnson & Johnson, New Brunswick, NJ, U.S.A.) 5–10 mL held in the mouth after food four times daily for 1 week. Slow healing of the oral mucosa may reflect secondary infection by, or reactivation of, HSV.

Consider using a topical corticosteroid four times daily (e.g. betamethasone sodium phosphate 0.5 mg in 10 mL water as a 3‐min rinse‐and‐spit preparation). A more potent preparation, clobetasol propionate 0.05%, mixed in equal amounts with Orabase^® (Colgate), can be applied directly to the sulci, labial or buccal mucosae daily during the acute phase.

Is systemic therapy effective in treating oral involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis?

There are no controlled trials of systemic corticosteroid in the management of mouth involvement in SJS/TEN; however, steroids are commonly used by oral physicians in severe EMM. In a series comprising eight EMM cases and four cases of SJS, patients were treated with fluocinolone, prednisolone or methylprednisolone in a variety of doses for 5–7 days during the acute illness.84 Oral assessment revealed a complete remission of buccal lesions in all patients 7–10 days after onset of systemic corticosteroid treatment.84 Nonetheless, there is currently insufficient evidence to recommend systemic corticosteroids for the treatment of the oral manifestations of acute SJS/TEN. There are no published studies of oral outcomes in acute SJS/TEN treated with systemic ciclosporin or IVIg.

Treatment of urogenital involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis

Involvement of the urogenital tract in SJS/TEN is characterized by mucosal erythema, blistering and erosions. During the acute phase pain is prominent and urinary dysfunction (dysuria or retention) is common. Secondary infection by bacteria or candida is a frequent complication of urogenital involvement. HSV activation may also occur. Erosions of the genital mucosae may persist for many weeks after the acute phase has resolved, ultimately healing with scarring.85  86 In the long term, serious morbidity can ensue in the form of strictures and stenosis of the urethra, phimosis in males and vaginal synechiae in females, with resultant urinary and sexual dysfunction.87

Is topical therapy effective in treating urogenital involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis?

Local measures for treating urogenital involvement in acute SJS/TEN are supportive and empirical. Along with regular examination of the urogenital tract, attention should be given to regular emollient, appropriate dressings and topical antimicrobial treatment. Topical corticosteroids may be useful to reduce urogenital inflammation.88  89

Recommendations [strength of recommendation D (good practice point); level of evidence 4]

Examine the urogenital tract as a part of the initial assessment of a patient with SJS/TEN. In women, early assessment by a vulval specialist is recommended for consideration of dilators to prevent vaginal synechiae. Speculum examination needs to be undertaken with plastic speculums so that all vaginal walls can be fully assessed. Uncircumcised male patients should be checked for preputial retractability. Thereafter, daily documented urogenital review is necessary during the acute illness.

Apply white soft paraffin ointment to the urogenital skin and mucosae immediately and thereafter every 4 h through the acute illness.

Use Mepitel^™ (Mölnlycke Health Care) dressings to eroded areas in the vulva and vagina to reduce pain and prevent adhesions. A dilator or tampon wrapped in Mepitel should be inserted into the vagina to prevent formation of synechiae.

Consider applying a potent topical corticosteroid ointment once daily to the involved, noneroded, urogenital surfaces.

Catheterize all patients to prevent strictures forming in the urethra.

Is systemic therapy effective in treating urogenital involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis?

There are no published studies of urogenital outcomes in patients with acute SJS/TEN treated with systemic corticosteroids, ciclosporin or IVIg.

Treatment of airway involvement in Stevens–Johnson syndrome/toxic epidermal necrolysis

Pulmonary complications are an underappreciated manifestation of SJS/TEN, and may be a marker of disease severity and mortality. Respiratory tract involvement does not seem to correlate with the extent of epidermal detachment.

A single‐centre, prospective study characterized airway involvement in SJS/TEN into three groups based on whether hypoxaemia and respiratory symptoms were present, and the time course of the development of hypoxaemia.90 Patients without hypoxaemia during the acute phase (group 1) suffered no pulmonary complications or sequelae, and had low mortality.90 One‐quarter of patients presented with early pulmonary manifestations (group 2) characterized by dyspnoea and an increased respiratory rate – chest radiographs were typically normal on admission, with later diffuse pulmonary infiltrates.90 Bronchial hypersecretion occurred in 70% of these patients, and fibreoptic bronchoscopy revealed a pattern of diffuse loss of bronchial epithelium in the proximal airways, with evolving epithelial detachment caused by epithelial necrosis. There was no evidence of airway infection in this group, and normal bronchial mucosa began to recover at the same time as skin recovery in survivors. Mechanical ventilation was required in 90% of cases, and a high mortality rate (70%) was reported. In patients with bronchial epithelial necrolysis, airway sloughing may occur, and can cause sudden airway obstruction and death (De Prost, personal communication). A small percentage of patients with early pulmonary manifestations go on to develop chronic respiratory problems, which include bronchiolitis obliterans, bronchiectasis and chronic bronchitis. The prognosis is poor, with a mortality rate of approximately 40%.91 Delayed pulmonary manifestations (group 3) occurred in 19% of cases and consisted of a heterogeneous group of conditions, including atelectasis, bacterial pneumonia and fluid overload.90 Biopsy confirmed the absence of pulmonary epithelial detachment in these patients; there was no requirement for ventilation in this group, and all patients recovered.90

Recommendations [strength of recommendation D (good practice point); level of evidence 4]

Respiratory symptoms and hypoxaemia on admission should prompt urgent discussion with an intensivist and rapid transfer to the ICU or burn centre, as deterioration requiring mechanical ventilation is likely. Relatives should be counselled as to the prognostic significance of this development.

Fibreoptic bronchoscopy should be undertaken to identify bronchial involvement, evaluate prognosis and investigate the presence of pneumonitis by bacterial sampling. Bronchoscopy may have a role in preventing atelectasis and airway obstruction by allowing the mechanical removal of sloughed bronchial epithelium.

Patients with ongoing respiratory symptoms should be closely monitored with pulmonary function testing and high‐resolution computed tomography scanning.

Active therapy in Stevens–Johnson syndrome/toxic epidermal necrolysis

Currently, no active therapeutic regimen with unequivocal benefit exists for SJS/TEN. Only one randomized controlled trial has been conducted in TEN: the anti‐TNF agent thalidomide was compared with placebo; however, the study was discontinued prematurely because of an excess of deaths in the thalidomide treatment group.92

The immunological basis for SJS/TEN has led physicians to prescribe immunomodulating drugs, several of which have been studied in uncontrolled series. In this guideline, active interventions were assessed if studies recruited at least eight patients with SJS/TEN into the treatment group. Of all active interventions (excluding thalidomide) only three drugs fulfilled the inclusion criteria: IVIg, systemic corticosteroid and ciclosporin.

Is treatment with intravenous immunoglobulin effective in Stevens–Johnson syndrome/toxic epidermal necrolysis?

Evidence for possible efficacy of IVIg in SJS/TEN came from a study indicating a role for Fas–Fas ligand (Fas–FasL) interaction in TEN keratinocyte apoptosis.12 This study demonstrated that high concentrations of normal immunoglobulin inhibited Fas–FasL interaction and apoptosis through anti‐Fas activity. The authors then reported an uncontrolled, prospective open trial of 10 patients with TEN treated with IVIg, none of whom died.12 As per the 1998 report of Viard et al.,12 a further 16 studies of IVIg treatment in SJS/TEN met the inclusion criteria for assessment (data summarized in Table 7).28  34  46  93  94  95  96  97  98  99  100  101  102  103  104  105 Of the 17 studies identified, 11 are retrospective cohort studies and six are prospective. Pooling the results yields a total of 374 SJS/TEN receiving IVIg; however, one study contained patients reported in previously published series;95 thus, the total number of patients is lower. In all studies the primary outcome measure is mortality. In a systemic review and meta‐analysis, published in 2012 by Huang et al. (all studies of at least eight patients with SJS/TEN receiving IVIg),106 a pooled estimate of mortality risk was determined, comparing IVIg and supportive care in patients with TEN and SJS–TEN overlap. Statistical analysis was also performed on the raw data to compare the clinical differences between high‐ and low‐dose treatment in adult patients, and between paediatric and adult patients treated with IVIg. The overall mortality rate of 221 patients with TEN and SJS–TEN overlap (patients with SJS were not included) treated with IVIg was 19.9%. The pooled odds ratio (OR) for mortality from six observational controlled studies comparing IVIg and supportive care was 1.00 [95% confidence interval (CI) 0.58–1.75; P =0.99].106 Paediatric patients treated with IVIg had significantly lower mortality than adults (0% vs. 21.6%, respectively; P =0.01). Adults treated with high‐dose IVIg exhibited significantly lower mortality than those treated with low‐dose IVIg (18.9% vs. 50%, respectively; P =0.02); however, multivariate logistic regression model adjustment indicated that IVIg dose does not correlate with mortality (high‐dose vs. low‐dose OR 0.494, 95% CI 0.106–2.300; P =0.37).104 However, these findings should be interpreted with caution given the major methodological limitations of the original cohorts. Since the publication of the meta‐analysis by Huang et al.,106 a further study by Firoz et al., which included 23 patients with TEN treated with IVIg, demonstrated no improved survival in patients receiving IVIg vs. supportive care alone.103 In 2013, Lee et al. published a retrospective analysis of 64 patients with SJS–TEN overlap or TEN treated with IVIg at a single specialized referral centre in Singapore.104 The mortality in their patients, when compared with predicted outcome from SCORTEN, showed no benefit from IVIg.104 In addition, when stratified according to dosage, there was no mortality difference between patients who received high‐dose (> 3 g kg⁻¹) vs. low‐dose (< 3 g kg⁻¹) IVIg.104

Table 7

Summary of intravenous immunoglobulin (IVIg) studies in Stevens–Johnson syndrome (SJS)/toxic epidermal necrolysis (TEN)

Study	Setting	Design	Patients (n)	Intervention: IVIg dose	Outcome: mortality	Comparator	Reported benefit	Level of evidencea	Additional notes
Viard et al. (1998)12	Multiple centres	Prospective; case series; noncontrolled	TEN (10)	0.2–0.75 g kg−1 daily for 4 days	0/10 (0%)	NA	Effective	3	Variable IVIg dosing
Stella et al. (2001)93	Single burns unit	Prospective; case series; noncontrolled	SJS (1)Overlap SJS–TEN (7)TEN (1)	0.6–0.7 g kg−1 daily for 4 days	1/9 (11%)	NA	Effective	3
Bachot et al. (2003)96	Single dermatology HDU	Prospective case series with SCORTEN‐predicted mortality as comparator	SJS (9)SJS–TEN overlap (5)TEN (20)	2 g kg−1 given over 2 days	11/34 (32%)	SCORTEN‐predicted mortality 8.2/34 (23%)	Ineffective	3
Campione et al. (2003)97	Single dermatology unit	Retrospective case series with SCORTEN‐predicted mortality as comparator	TEN (10)	2 g kg−1 given over 5 days	1/10 (10%)	SCORTEN‐predicted mortality 3.5/10 (35%)	Effective	3
Prins et al. (2003)95	Multiple centres	Retrospective; case series (including previously reported cases, from three other studies); noncontrolled	Overlap SJS/TEN (7)TEN (41)	Mean 0.7 g kg−1 daily for 4 days	6/48 (12%)	NA	Effective	3	Duplicated cases from Viard et al. (1998),12 Campione et al. (2003)97 and Trent et al. (2003)94
Trent et al. (2003)94	Single dermatology unit	Retrospective case series with SCORTEN‐predicted mortality as comparator	SJS–TEN overlap (6)TEN (10)	1 g kg−1 daily for 4 days (n =15); 0.4 g kg−1 daily for 4 days (n =1)	1/16 (6%)	SCORTEN‐predicted mortality 5.81/16 (36%)	Effective	3
Al‐Mutairi et al. (2004)98	Single dermatology unit	Retrospective, noncontrolled	TEN (12)	0.5–1.0 g kg−1 for 4–5 days	0/11 (0%)	NA	Effective	3
Brown et al. (2004)99	Single burns unit	Retrospective, case–control	TEN (24)	0.4 g kg−1 for 4 days	10/24 (42%)	6/21 (29%) mortality in steroid‐treated controls	Ineffective	2−
Shortt et al. (2004)100	Single burns unit	Retrospective; case–control	SJS–TEN overlap (16)	Mean dose 0.7 ± 0.2 g kg−1 daily for 4 ± 1 days	4/16 (25%)	6/16 (37%) mortality in historic controls	Equivocal	2−
Kim et al. (2005)34	Dermatology unit	Retrospective case series with SCORTEN‐predicted mortality as comparator	TEN (14)	IVIg: 1.6–2.0 g kg−1	1/14 (7%)	SCORTEN‐predicted mortality: 2.4/14 (17%)	Effective	3
Tan et al. (2005)101	Dermatology unit	Retrospective; case series; noncontrolled	SJS–TEN overlap (4)TEN (8)	2 g kg−1 over 2 days (n =10); 1.5 g kg−1 over 2 days (n =2); 9/12 patients had corticosteroid prior to IVIg	1/12 (8%)	No comparator group	Effective	3	Cases validated but histopathological evidence in only 4/12 cases; co‐treatment with corticosteroid
Gravante et al. (2007)102	Burns unit	Retrospective; case series	SJS (1)TEN (16)	0.4 g kg−1 daily for 5 days	7/17 (41%)	No comparator group or SCORTEN‐predicted mortality given	Effective	3
Stella et al. (2007)46	Burns unit	Retrospective case–control series; control group – supportive care only	IVIg group: SJS (2); SJS/TEN overlap (16); TEN (5)	0.7 g kg−1 daily for 4 days + methylprednisolone 250 mg daily for 2 days	6/23 (26%)	SCORTEN‐predicted mortality 8.2/23 (36%)	Effective	3	This series included nine cases published in Stella et al. (2001).93 Control group TBSA involvement significantly lower in IVIg group compared with controls
			Control group: TEN (8)
Schneck et al. (2008)28	Burns and dermatology units	Retrospective; case–control	SJS (9)SJS–TEN overlap (11)TEN (15)	Mean dose 1.9 g kg−1 over 1–7 days	12/35 (34%)	Mortality for supportive care alone 22/87 (25%)	Ineffective	2−	Well‐validated cases. Study may have included cases published in Bachot et al. (2003).96 Heterogeneous care settings
Firoz et al. (2012)103	Single burns unit	Prospective; case‐controlled SCORTEN‐based comparison	TEN (23)	4 g kg−1 over 3 days	No significant difference in survival between IVIg group and control	Supportive care only (n =51)	Ineffective	2−
Lee et al. (2013)104	Single specialized centre	Retrospective SCORTEN‐based comparison	SJS–TEN overlap (28)TEN (36)	Mean dose 2.4 g kg−1 over 4 days	20/64 (31%)	SCORTEN‐predicted mortality 18.22/64 (28%)	Ineffective	2−	No mortality difference when stratified according to high‐ or low‐dose IVIg
Aihara et al. (2015)105	Multiple dermatology centres	Prospective; case series; no comparator group	SJS (5)TEN (3)	400 mg kg−1 daily IVIg for 5 consecutive days in conjunction with systemic corticosteroid therapy (variable dose)	0/8 (0%); 7/8 patients classed as ‘responders’ according to a severity‐of‐illness score designed by the authors	None	Effective	3	Co‐treatment of all patients with corticosteroid – variable dosing. Only 8/41 cases included in the final report – reporting bias

NA, not applicable; HDU, high‐dependency unit; TBSA, total body surface area. ^aSee Appendix  1.

Is treatment with systemic corticosteroid effective in Stevens–Johnson syndrome/toxic epidermal necrolysis?

Corticosteroids have been used in the management of SJS/TEN for many years. Proponents emphasize the importance of high‐dose corticosteroid given early in the disease course to inhibit inflammation; opponents suggest that systemic corticosteroids increase the risk of sepsis. Ten published studies of corticosteroid treatment in SJS/TEN met the inclusion criteria for assessment: all are case series, none are randomized controlled trials and most are retrospective.28  29  30  34  107  108  109  110  111 Retrospective analysis of the EuroSCAR data indicated a lower mortality in German patients (but not French patients) treated with corticosteroids compared with controls receiving supportive care alone.28 Two studies investigated the effects of pulsed IV high‐dose corticosteroids.107  112 In the study by Kardaun and Jonkman,107 12 patients received 100 mg or 1.5 mg kg^–1 of IV dexamethasone for 3 days and were reported to have a decreased mortality compared with SCORTEN. Hirahara et al. presented a series of eight patients with SJS/TEN who received 1000 mg of IV methylprednisolone on three consecutive days, followed by either a tapering course of oral prednisolone or a further 2 days of half‐dose IV methylprednisolone.112 No patients died despite a SCORTEN‐predicted mortality of 1.6.112 Data from the corticosteroid studies are summarized in Table 8. Sources of potential bias are indicated in the table.

Table 8

Summary of corticosteroid studies in Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN)

Study	Setting	Design	Patients (n)	Intervention	Outcome: mortality	Comparator	Reported benefit	Level of evidencea	Additional notes
Murphy et al. (1997)111	Single burn centre	Retrospective; case series; noncontrolled	TEN (12)	Highly variable doses received, no details about specific dosing	8/12	NA	No comment made by authors	3	Timing, duration and dose of corticosteroid therapy highly variable
Schulz et al. (2000)29	Single burn centre	Retrospective; case series; noncontrolled	TEN (34)	No details about corticosteroid dosing	13/34	NA	Ineffective	3	No evidence of case validation; no details regarding steroid dosing
Tripathi et al. (2000)108	Single dermatology centre	Retrospective; case series; noncontrolled	SJS (13)	Methylprednisolone 160–240 mg daily; tapered when clinical response seen (no mean duration given)	1/13	NA	No comment	3	No evidence of case validation
Ducic et al. (2002)30	Single burn centre	Retrospective; case series; noncontrolled	TEN (29)	No information about dosage provided	13/29	NA	No comment made by authors	3	No evidence of case validation. TEN diagnosis based on ≥ 20% BSA detachment as opposed to 30%; therefore, some cases may be overlap. No information about corticosteroid dosing
Kim et al. (2005)34	Single dermatology centre	Retrospective; case series; noncontrolled	TEN (21)	Methylprednisolone 250–1000 mg daily, later switched to oral prednisolone	6/21 mortality	SCORTEN‐predicted mortality 5.97	Ineffective	3	Some attempt to validate cases. Variable doses of steroid used, no details about tapering
Kardaun and Jonkman (2007)107	Single dermatology centre	Retrospective; case series; noncontrolled	SJS (1) SJS–TEN overlap (4) TEN (7)	First four patients: IV dexamethasome 100 mg once daily for 3 days + 500 mg cyclophosphamide. Subsequent patients: 1.5 mg kg–1 IV dexamethasone for 3 days	1/12	SCORTEN‐predicted mortality: 4/12	Effective	3	Case validation unclear. Change of treatment protocol during series
Schneck et al. (2008)28	Multiple centres in Germany and France	Retrospective; case series; noncontrolled	SJS (57) SJS–TEN overlap (44)TEN (18)	Maximum steroid dose 250 mg prednisolone equivalent (IQR 100–500 mg). Given for median 4 days (2–12 days)	21/119	Mortality expressed as OR: corticosteroids compared with supportive care alone OR 0.4 (95% CI 0.1–1.7) in France and OR 0.3 (95% CI 0.1–1.1) in Germany. Overall: OR 0.4 (95% CI 0.2–0.9)	Ineffective (on multivariate analysis)	2−	Robust case validation. Steroid group had less severe disease than IVIg group
Yang et al. (2009)109	Single dermatology centre	Retrospective; case series; noncontrolled	SJS (10) and TEN (35) in corticosteroid arm (n =45) SJS (8) and TEN (12) in IVIg/corticosteroid arm (n =20)	1–1.5 mg kg−1 daily methylprednisolone or equivalent (details not given); IVIg 0.4 g kg−1 day for 5 days	10/45 (10 TEN) in corticosteroid group; 3/20 (two TEN and one SJS) in IVIg/corticosteroid group	SCORTEN‐predicted mortality 8.63 in corticosteroid group; 3.51 in IVIg/corticosteroid group	Corticosteroid alone ineffective, IVIg/corticosteroid may confer benefit (nonsignificant difference)	2−	Case validation unclear. Insufficient detail on corticosteroid dosing
Chen et al. (2010)110	Single dermatology centre	Retrospective; case series; noncontrolled	SJS (43) and TEN (15) in corticosteroid only arm (n =58) SJS (9) and TEN (15) in IVIg/corticosteroid group (n =24)	Hydorcortisone 100–700 mg daily or methylprednisolone 40–80 mg daily (cumulative dose equivalent to prednisolone 10–25 mg kg–1 daily for 7–14 days); IVIg mean dose 2.7 ± 1.5 g kg−1 over 3–15 days	2/58 in corticosteroid group; 3/24 in IVIg/corticosteroid group	SCORTEN‐predicted mortality 4.2 in corticosteroid group; 5.3 in IVIg/corticosteroid group	Effective	2−	Results suggesting benefit did not reach statistical significance. Patients in IVIg/corticosteroid group had higher TBSA involvement at baseline
Hirahara et al. (2013)112	Single dermatology centre	Retrospective; case series; noncontrolled	SJS (3) SJS–TEN overlap (2) TEN (3)	1000 mg IV methylprednisolone for 3 days	0/8	SCORTEN‐predicted mortality 1.6/8 (20%)	Effective	3	Subsequent doses of corticosteroid varied according to response

NA, not applicable; BSA, body surface area; IV, intravenous; IQR, interquartile range; OR, odds ratio; CI, confidence interval; IVIg, intravenous immunoglobulin; TBSA, total BSA. ^aSee Appendix  1.

Is treatment with ciclosporin effective in Stevens–Johnson syndrome/toxic epidermal necrolysis?

Effective inhibition of lymphocyte function identifies ciclosporin as a drug with theoretical benefit in SJS/TEN. Four cohort studies of ciclosporin treatment in SJS/TEN meet the inclusion criteria for assessment (summarized in Table 9).113  114  115  116 A study by Valeyrie‐Allanore et al. from the dermatology intensive therapy unit in Creteil, Paris, demonstrated that in 29 patients with SJS/TEN ciclosporin given at a dose of 3 mg kg⁻¹ daily for 10 days, and thereafter tapered, was effective.114 There were no deaths, despite a SCORTEN predicted mortality of 2.75/29.114 Singh et al. reported 11 patients with SJS/TEN treated with ciclosporin 3 mg kg⁻¹ daily for 7 days, and then tapered over a further 7 days. This group was compared with six historical SJS/TEN controls treated with systemic corticosteroids.115 There was a significantly enhanced speed of epithelialization and reduced length of hospital stay in the ciclosporin group.115 Comparison of SCORTEN‐predicted mortality demonstrated a benefit of ciclosporin over corticosteroids.115 Kirchhof et al. published a retrospective review from a single centre of 64 patients with SJS/TEN receiving either ciclosporin or IVIg (doses of each varied).116 Analysis of predicted SCORTEN mortality compared with actual mortality indicated a relative mortality benefit for the use of ciclosporin [standardized mortality ratio (SMR) 0.43] vs. IVIg (SMR 1.43).116 Data from the ciclosporin studies are summarized in Table 9. Sources of potential bias are indicated in the table.

Table 9

Summary of ciclosporin (CsA) studies in Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN)

Study	Setting	Design	Patients (n)	Intervention	Outcome: mortality	Comparator	Reported benefit	Level of evidencea	Additional notes
Arévalo et al. (2000)113	Burns unit	Case–control study (historic controls)	TEN (11)	Cases treated with 3 mg kg–1 CsA for 14 days, tapered by 10 mg daily for a further 2 weeks, then stopped	0/11	3/6 in cyclophosphamide group; cyclophosphamide 150 mg twice daily and methylprednisolone – variable doses (≥1 mg kg−1 daily	Yes	3	No evidence of case validation (but all cases have histopathological confirmation). Possible overestimation of BSA involvement. Co‐treatment with steroid and cyclophosphamide in control group at variable doses
Valeyrie‐Allanore et al. (2010)114	Dermatological ICU	Case series with SCORTEN‐based predicted mortality as comparator	SJS (10) SJS–TEN overlap (12) TEN (7)	CsA 3 mg kg–1 for 10 days tapered over 1 month	0/29	SCORTEN‐predicted mortality 2.75	Yes	2	Good validation. No control group. Consistent dosing among patients
Singh et al. (2013)115	Dermatology unit	Case–control study (historical controls)	SJS (5) SJS–TEN overlap (3) TEN (3)	CsA 3 mg kg−1 daily for 7 days, and tapered for 7 days	0/11	SCORTEN‐predicted mortality in control group treated with corticosteroid	Yes	3	No evidence of case validation
Kirchhof et al. (2014)116	Dermatology unit	Retrospective series: CsA vs. IVIg	SJS–TEN overlap (64)	Various doses of CsA and IVIg	CsA SMR 0.43; IVIg SMR 1.43	CsA‐treated group compared with IVIg‐treated group	Yes	3	No evidence of case validation. No control group

BSA, body surface area; ICU, intensive care unit; IVIg, intravenous immunoglobulin; SMR, standardized mortality ratio. ^aSee Appendix  1.

Other treatments used in the management of Stevens–Johnson syndrome/toxic epidermal necrolysis

Other therapies have been tried in SJS/TEN, but studies contain small numbers of patients and are, generally, uncontrolled. Plasmapheresis has been used in SJS/TEN that is refractory to treatment; reports suggest that this therapeutic modality may provide a rapid benefit.117  118 The bioregeneratory and immunomodulatory properties of G‐CSF have led some to claim efficacy (that transcends the management of leucopenia) in the active treatment of TEN by arresting the hypersensitivity and stimulating re‐epithelialization.58 More recently there has been interest in a therapeutic role for TNF‐α inhibitors. Paradisi et al. reported a series of 10 patients with SJS/TEN treated with a single 50‐mg subcutaneous dose of etanercept. There was no control group. All patients responded with a mean re‐epithelialization time of 8.5 days. None of the patients died, despite a mean SCORTEN‐predicted mortality rate of about 50%.119

Evidence synthesis

The GDG found interpretation of the reviewed studies challenging, given major ascertainment bias in reported cohorts, low numbers of patients, variation in the timing and nature of intervention, case mix and setting, and quality of supportive care. The GDG did not consider any of the data presented of sufficient quality or consistency to make specific recommendations either for or against the use of active interventions, and highlighted the need for future research (see section section 62). It was noted that there is a lack of consensus even among clinicians with experience in managing SJS/TEN, with, for example, strong advocates for use and avoidance of IVIg.

Withdrawal of culprit drug and meticulous attention to high‐quality, multidisciplinary supportive care is the priority.

Recommendations (strength of recommendation D; level of evidence 4)

There is no conclusive evidence to demonstrate the benefit of any one intervention over conservative management nor evidence to demonstrate harm from IVIg, systemic corticosteroids or ciclosporin in the context of SJS/TEN. The GDG considers that, ideally, such interventions should be practised under the supervision of a specialist skin failure MDT in the context of a clinical study or a case registry.

Discharge and follow‐up

As the patient recovers from the mucocutaneous and systemic manifestations of acute SJS/TEN, preparations can be made for discharge. Before leaving the hospital, information on the culprit drug should be relayed to the patient. The drug allergy should be documented in the patient's notes (or electronic patient record) and communicated to all physicians involved in the patient's care. Recent National Institute for Health and Care Excellence (NICE) guidelines highlight the importance of ensuring adequate documentation and information sharing regarding drug allergy.120 An adverse drug reaction should be reported to the pharmacovigilance authorities [the Medicines and Healthcare products Regulatory Agency (MHRA) in the U.K., www.mhra.gov.uk].

At discharge, an explanatory letter outlining the disorder should be sent to the patient's GP. The letter should state the culprit medication, outline the potential complications of SJS/TEN and specify the patient's follow‐up plan (see Appendix  2). NICE guidance suggests that severe nonimmediate cutaneous reactions are referred to a specialist drug allergy service for expert review, which in the context of SJS/TEN is likely to be a dermatology unit with appropriate subspecialty interest.120

Fatigue and lethargy is a major problem for several weeks following discharge; the patient, his/her relatives and employers need to be aware of this. Usually, a period of convalescence following hospital discharge is necessary. Psychological problems, including depression, are also common during this time.121 Chronic complications of SJS/TEN may develop weeks to months after the acute episode and are associated with significant morbidity and reduced quality of life. Ocular damage, arising from corneal scarring, is the most disabling complication and may develop at a variable period after the acute disease.122 Long‐term sequelae can involve other organs and survivors need monitoring for potential complications in multiple systems.

Recommendations [strength of recommendation D (good practice point); level of evidence 4]

Give the patient written information about the drug(s) to avoid. The name(s) of any related medication(s) that may cross‐react with the culprit should also be given to the patient. Educate the patient about the necessity to avoid the triggering drug(s).

At discharge, refer the patient to local social services, when appropriate, to arrange a needs assessment for support. If necessary, refer to the occupation therapy department.

Access support from SJS Awareness U.K. (http://www.sjsawareness.org.uk/) or appropriate national SJS/TEN support group.

Encourage the patient to wear a MedicAlert bracelet or amulet bearing the name of the culprit drug.

The drug allergy should be documented in the patient's notes. All doctors involved in the patient's care, especially the GP, should be informed about the drug allergy episode and the culprit.

Warn the patient to avoid OTC medications where precise constituents are unclear.

Report episodes of drug‐induced SJS/TEN to the national pharmacovigilance authorities (the MHRA's Yellow Card Scheme in the U.K., www.yellowcard.mhra.gov.uk).

If the patient had eye involvement during the acute phase, organize an outpatient clinic appointment with an ophthalmologist within a few weeks of discharge.

Organize an outpatient clinic appointment in dermatology or burns plastic surgery within a few weeks of discharge. Patients need to be monitored for complications in skin, mouth, urogenital tract, respiratory and gastrointestinal systems. Psychological evaluation and support should be considered.

Drug hypersensitivity tests in Stevens–Johnson syndrome/toxic epidermal necrolysis

Once a diagnosis of drug‐induced SJS/TEN has been made, investigations may be helpful in confirming or identifying the culprit medication. As T‐cell‐mediated mechanisms predominate in SJS/TEN, assays for drug‐specific IgE‐mediated immediate reactions (i.e. skin‐prick testing and specific IgE testing) are of no value. Appropriate tests are those that identify delayed type IV hypersensitivity reactions, most notably drug patch testing and in vitro assays, such as drug‐induced T‐cell proliferation (lymphocyte transformation tests) and drug‐induced lymphocyte cytokine production. Oral provocation studies are not ethical in SJS/TEN because of a risk of life‐threatening reactions. Although a lesser risk, induction of severe reactions has been recorded with patch testing.

What is the evidence that patch tests can identify a culprit drug in Stevens–Johnson syndrome/toxic epidermal necrolysis?

Reviews have commented on the value of patch testing in drug allergy and have suggested explanations for the variable sensitivity and specificity of this assay in different reaction patterns and with different drugs.123 A study by Wolkenstein et al. patch tested 22 cases of SJS/TEN (series n =59) and found only 9% (n =2/22) of the cases to be positive.124 Another study by Lin et al. demonstrated positive patch tests to carbamazepine in 62% (n =10/16) of cases of carbamazepine‐induced SJS/TEN compared with 0% (n =0/10) of controls.125 In this study, cross‐reactivity to other aromatic anticonvulsant drugs was observed.125 A recent French multicentre study reported 17 patients with SJS/TEN where positive patch tests were identified in 23% (n =4/17) of cases, but in this series patch testing showed positive reactions in 0% (n =0/5) of cases of carbamazepine‐induced SJS/TEN.126

What is the evidence that in vitro tests can identify a culprit drug in Stevens–Johnson syndrome/toxic epidermal necrolysis?

The method of isolation of peripheral blood mononuclear cells (PBMC) followed by in vitro challenge with the putative causal drug has been widely studied in the context of drug hypersensitivity testing. Although these assays involve the culture of PBMC with the drug in question, it is generally accepted that the responding cells are T lymphocytes.

Lymphocyte transformation tests (lymphocyte proliferation assay)

Lymphocyte proliferation, as measured by uptake of 3H‐thymidine in dividing cells, is widely referred to as the lymphocyte transformation test (LTT). However, variations in standard methodology mean that comparative interpretation is difficult. Overall, the number of reports specifically characterizing the value of LTT for SJS/TEN is low. Studies have reported positive LTT responses in 100% (n =4/4) of SJS cases to a variety of culprit drugs,127 and 75% (n =3/4) of SJS cases, again triggered by a range of medicines.128 However, neither of these reports compared drug LTT assays in a control population. A study by Kano et al. showed that in two cases of SJS, LTT results were positive in the acute phase of the reaction but weakened over 2 months.129 A report by Tang et al. suggested that LTT was not of value for the investigation of lamotrigine‐induced SJS/TEN as only 21% (n =3/14) cases showed a positive result.130

A study by Roujeau et al. reported 11 cases tested with LTT within 1 month of recovery from TEN.131 Although 44% of cases showed positive LTTs to the culprit drug, control groups also showed a similar frequency of positive drug LTT assays.131 The authors concluded that the assay is not useful in cases of TEN. However, other series have shown positive results in only 15% and 0% (n =0/18) of controls.132  133 In the series by Polak et al.,134 55% (n =5/9) of EMM/SJS/TEN cases demonstrated positive LTTs in the context of a low false‐positive rate of 4.9%, while another series showed the LTT to be positive in only 27% (n =4/15) of cases.133 Lymphocyte proliferation measured by other techniques such as the carboxyfluorescein succinimidyl ester dilution assay has not been widely examined in SJS/TEN. A recent report has suggested that blockade of inhibitory molecules to T‐cell proliferation during the LTT assay (with anticytotoxic T‐lymphocyte‐associated protein 4 and antiprogrammed death‐ligand 1), may increase positivity from 32% to 50%.135

Lymphocyte function assays

Drug‐induced lymphocyte production of cytokines or other mediators measured by enzyme‐linked immunosorbent assay, enzyme‐linked immunospot (ELISpot) or intracellular fluorescence has the advantage of utilizing a shorter assay time and does not require the use of radioisotopes. Furthermore, it does not rely on lymphocyte proliferation, which is likely to be more susceptible to direct drug toxicity in vitro. However, these assays are technically more challenging and require more complicated equipment.

Published reports have suggested the value of cytokine assays incorporating IFN‐γ, interleukin (IL)‐2, IL‐4, IL‐5, IL‐13, IL‐17, granzyme B, sFasL, granulysin and others to investigate drug hypersensitivity in vitro.134  136  137  138  139  140  141 In the study by Polak et al.,134 IFN‐γ drug ELISpot assays tested at a median ±interquartile range (IQR) of 19 ± 11 days after the onset of the reaction were shown to identify causal drugs in 78% (n =7/9) of cases of EMM/SJS/TEN.134 IL‐4 ELISpot was less sensitive and detected only 50% (n =4/8) of cases.134 A series of nine cases of SJS/TEN reported that 100% of cases had positive IFN‐γ ELISpot assays when tested after a median ±IQR of 12 ± 3 months and 67% (n =6/9) of cases also showed drug‐induced sFasL production.141 Detection of drug‐induced IFN‐γ has also been demonstrated in a recent series of 15 cases of SJS/TEN and, in this series, measurements of granzyme B and IL‐5 release following drug exposure were also significantly higher in patients compared with controls.133

Owing to ethical considerations, currently no diagnostic tests for SJS/TEN have been validated by deliberate re‐exposure challenge testing to the suspected culprit drug. Consequently, true sensitivity is not known and the possibility of false‐positive and false‐negative results exists. However, it is generally accepted that re‐exposure to a drug showing a positive result (from patch or blood tests) is likely to represent a significant risk of reinitiation of SJS/TEN. Diagnostic testing is not likely to be of benefit in cases where a culprit drug can be imputed with a high level of confidence from clinical grounds (e.g. where exposure was to only one medication). Furthermore, in cases where strict avoidance of possible culprit drugs and related compounds is of little consequence to the individual, the risk of false‐negative testing is likely to outweigh the benefits of the test. In cases where medication avoidance is detrimental to the individual or where accidental exposure is possible, further investigation may be warranted. Patch testing is generally safe but of low diagnostic value for the investigation of SJS/TEN causality. Patch testing specificity and sensitivity are drug‐dependent. Although LTT is of uncertain diagnostic value, lymphocyte drug‐induced IFN‐γ assays are of value for the investigation of SJS/TEN causality. However, evidence from several groups suggests that combination assays show increased sensitivity over one assay alone. The use of in vitro tests by clinicians will be determined by the availability of the relevant assays in specialist laboratories. Diagnostic testing should be interpreted by physicians experienced in the management of SJS/TEN who are familiar with the test system. It is to be emphasized that the tests described here have not been evaluated as tools for the diagnosis of SJS/TEN when the clinical picture is not typical.

Recommendations [strength of recommendation D (good practice point); level of evidence 4]

Routine drug hypersensitivity testing is not recommended following an episode of SJS/TEN.

Seek specialist advice on formal drug hypersensitivity testing where the culprit drug cannot be imputed with confidence and where medication avoidance is detrimental to the individual or where accidental exposure is possible (see also follow‐up recommendations above).120 Current evidence supports the use of a lymphocyte drug‐induced IFN‐γ assay as part of the investigative work‐up.

Future directions

Improving the management of patients with SJS/TEN requires attention to the processes of care delivery and to individual components of the therapeutic regimen. It has been suggested that SJS/TEN patient management should be carried out in a small number of regional centres, each equipped with appropriate intensive care facilities and specialist expertise.142 In this way, standardized care could be delivered to provide high‐quality care with improved clinical outcomes. Establishing a national network of SJS/TEN centres would enable the implementation of appropriate governance structures, including case registration on a national database; supraregional case conferences for case validation and standardization of care; regular audits against defined standards of care; opportunities for training and continuing professional development for the network members.

Future research into the management of the acute phase of SJS/TEN needs to answer a number of important questions: (i) What constitutes optimal supportive care? (ii) What, if any, are the optimum active, systemic therapies? (iii) What are the optimal investigations to identify the culprit drug in SJS/TEN?

Recommended audit points

For each patient with SJS/TEN in the last 5 years:

1. Has the patient had a SCORTEN performed on admission?

2. Has drug causality assessment been undertaken within the first 24 h of admission?

3. Has the patient been seen by an ophthalmologist within 24 h of admission? Have daily ocular assessments been made throughout the acute phase?

4. Has an initial assessment of mouth and urogenital tract involvement been undertaken within the first 24 h of admission? Have daily oral and urogenital assessments been made throughout the acute phase?

5. At discharge, has contact been made with the patient's GP? Has the patient been counselled about future avoidance of the culprit drug(s)? Has a MedicAlert bracelet/amulet been requested?

Summary

Table 10 summarizes the findings of this guideline. Full details of the evidence have been provided in the main text.

Table 10

Summary

Initial assessment on presentation	Take a detailed history from the patient and/or relatives
	Perform a full physical examination, including baseline body weight, and record the vital signs, including oxygen saturation
	Order a set of investigations: FBC, U&E, LFT, glucose, magnesium, phosphate, bicarbonate, mycoplasma serology, CXR, skin biopsy
	Initiate a primary management plan: (i) establish peripheral venous access; (ii) if patient cannot maintain adequate nutrition orally, insert a nasogastric tube and institute nasogastric feeding; (iii) insert a urinary catheter if urogenital involvement is causing significant dysuria/retention
	Strength of recommendation D (GPP)
Determination of drug causality	Identify causative agent and withdraw immediately
	Strength of recommendation D
Prognostic scoring	Calculate SCORTEN within the first 24 h
	Strength of recommendation C
Care setting	An MDT should be convened, coordinated by a specialist in skin failure, usually dermatology and/or plastic surgery, and including clinicians from intensive care, ophthalmology and skincare nursing
	Patients with > 10% BSA epidermal loss should be admitted without delay to a burn centre or ICU with experience of treating patients with SJS/TEN and facilities to manage the logistics of extensive skin loss wound care
	Patients must be barrier‐nursed in a side room controlled for humidity, on a pressure‐relieving mattress with the ambient temperature raised to between 25 °C and 28 °C
	Strength of recommendation D (GPP)
Skin management regimen 1 (applicable to all patients in all settings)	Employ strict barrier nursing to reduce nosocomial infections
	Take swabs for bacterial and candidal culture from three areas of lesional skin, particularly sloughy or crusted areas, on alternate days throughout the acute phase
	Administer systemic antibiotics only if there are clinical signs of infection
	Strength of recommendation D (GPP)
Skin management regimen 2 (this may involve a conservative and/or surgical approach based on the specialist MDT's daily review of the individual needs of the patient)	Institute a conservative approach in all patients as follows:
	Regularly cleanse wounds and intact skin by irrigating gently using warmed sterile water, saline or an antimicrobial such as chlorhexidine (1/5000)
	Apply a greasy emollient, such as 50% white soft paraffin with 50% liquid paraffin, over the whole epidermis, including denuded areas
	Apply a topical antimicrobial agent to sloughy areas only (choice should be guided by local microbiological advice). Consider silver‐containing products/dressings
	The detached, lesional epidermis may be left in situ to act as a biological dressing. Blisters should be decompressed by piercing and expression or aspiration of tissue fluid
	Apply nonadherent dressings to denuded dermis [suitable dressings include Mepitel™ (Mölnlycke Health Care, Dunstable, U.K.) or Telfa™ (Covidien, Mansfield, MA, U.S.A.)]
	A secondary foam or burn dressing should be used to collect exudate [suitable dressings include Exu‐Dry® (Smith & Nephew, London, U.K.)]
	Consider transfer to a Burn Centre in patients with TEN (> 30% BSA epidermal loss) and evidence of the following: clinical deterioration, extension of epidermal detachment, subepidermal pus, local sepsis, wound conversion and/or delayed healing. In a burn centre, conservative measures may be supplemented with a surgical approach:
	Remove necrotic/loose infected epidermis and clean wounds using a topical antimicrobial agent (e.g. betadine or chlorhexidine) under general anaesthetic
	Consider debridement with Versajet™ (Drytac, Bristol, U.K.)
	Physiological closure with Biobrane/allograft/xenograft skin in patients with early presentation involving noninfected and large confluent areas
	Strength of recommendation D (GPP)
Fluid replacement regimen	Site venous lines through nonlesional skin, whenever possible, and change peripheral venous cannulas every 48 h
	Monitor fluid balance carefully: catheterize if appropriate/necessary
	Establish adequate IV fluid replacement initially. Fluid replacement can be guided by urine output and other end‐point measurements. Individualized fluid management should be adjusted on a daily basis
	With improvement of SJS/TEN mouth involvement, oral administration of fluids should be progressively increased
	Strength of recommendation D
Nutrition regimen	Provide continuous enteral nutrition throughout the acute phase
	Deliver up to 20–25 kcal kg−1 daily during the early, catabolic phase, and 25–30 kcal kg−1 daily during the anabolic, recovery phase
	Strength of recommendation C
Analgesia	Use a patient‐appropriate validated pain tool to assess pain in all conscious patients at least once daily
	Patients should receive adequate analgesia to ensure comfort at rest, with the addition of supplementary opiates, as required
	Additional analgesia may be needed to address increased pain associated with patient handling, repositioning and dressing changes
	Strength of recommendation D (GPP)
Supportive therapeutic measures	Immobile patients should receive low molecular weight heparin
	Patients in whom enteral nutrition cannot be established should receive a proton pump inhibitor to reduce the risk of stress‐related gastrointestinal ulceration
	Neutropenic patients may benefit from recombinant human G‐CSF
	Strength of recommendation C
Treatment of eye involvement	Daily ophthalmological review is necessary during the acute illness
	Apply an ocular lubricant (e.g. nonpreserved hyaluronate or carmellose eye drops) every 2 h through the acute illness
	Ocular hygiene must be carried out each day by an ophthalmologist or ophthalmically trained nurse
	Application of topical corticosteroid drops (e.g. nonpreserved dexamethasone 0.1% twice daily) may reduce ocular surface damage
	Administer a broad‐spectrum topical antibiotic as prophylaxis (e.g. moxifloxacin drops four times daily) in the presence of corneal fluorescein staining or frank ulceration
	In the unconscious patient, prevention of corneal exposure is essential
	Strength of recommendation D (GPP)
Treatment of mouth involvement	Daily oral review is necessary during the acute illness
	Apply white soft paraffin ointment to the lips every 2 h through the acute illness
	Clean the mouth daily with warm saline mouthwashes or an oral sponge
	Use an anti‐inflammatory oral rinse or spray containing benzydamine hydrochloride every 3 h, particularly before eating
	Use an antiseptic oral rinse containing chlorhexidine twice daily
	Use a potent topical corticosteroid mouthwash (e.g. betamethasone sodium phosphate) four times daily
	Strength of recommendation D (GPP)
Treatment of urogenital involvement	Daily urogenital review is necessary during the acute illness
	Apply white soft paraffin ointment to the urogenital skin and mucosae every 4 h through the acute illness
	Use a potent topical corticosteroid ointment once daily to the involved, but noneroded, surfaces
	Use a silicone dressing (e.g. Mepitel™) to eroded areas
	Strength of recommendation D (GPP)
Treatment of airway involvement	Respiratory symptoms and hypoxaemia on admission should prompt early discussion with an intensivist and rapid transfer to an ICU or burn centre, where fibreoptic bronchoscopy should be undertaken
	Strength of recommendation D (GPP)
Active therapy	If active therapy is instituted it should be given, ideally, under the supervision of a specialist skin failure MDT in the context of clinical research and/or case registry
	Strength of recommendation D
Discharge and follow‐up	Give the patient written information about drug(s) to avoid
	Encourage the patient to wear a MedicAlert bracelet
	Drug allergy should be documented in the patient's notes; all doctors involved in the patient's care should be informed
	Report the episode to the national pharmacovigilance authorities
	Organize a dermatology outpatient clinic appointment, and, if required, a n ophthalmology outpatient appointment, within a few weeks of discharge
	Refer for review to unit with appropriate subspeciality interest
	Strength of recommendation D (GPP)
Diagnostic testing	Routine drug hypersensitivity testing is not recommended following an episode of SJS/TEN
	Seek specialist advice on hypersensitivity testing where (i) the culprit drug is not known or (ii) medication avoidance is detrimental to the individual or (iii) accidental exposure is possible
	Strength of recommendation D (GPP)

FBC, full blood count; U&E, urea and electrolytes; LFT, liver function tests; CXR, chest X‐ray; GPP, good practice point; BSA, body surface area; ICU, intensive care unit; SJS/TEN, Stevens–Johnson syndrome/toxic epidermal necrolyis; MDT, multidisciplinary team; IV, intravenous; G‐CSF, granulocyte colony‐stimulating factor.

Acknowledgments

We are very grateful to Miss Sara Haveron (British Association of Dermatologists Scientific Administrator). Dr Zainab Laftah (King's College Hospital) kindly provided the photograph used in Figure 8. We would like to thank everyone who commented on the draft during the consultation period.

1 Appendix

Levels of evidence

Level of evidencea	Type of evidence
1++	High‐quality meta‐analyses, systematic reviews of RCTs or RCTs with a very low risk of bias
1+	Well‐conducted meta‐analyses, systematic reviews of RCTs or RCTs with a low risk of bias
1−	Meta‐analyses, systematic reviews of RCTs or RCTs with a high risk of bias
2++	High‐quality systematic reviews of case–control or cohort studies. High‐quality case–control or cohort studies with a very low risk of confounding, bias or chance, and a high probability that the relationship is causal
2+	Well‐conducted case–control or cohort studies with a low risk of confounding, bias or chance, and a moderate probability that the relationship is causal
2−	Case–control or cohort studies with a high risk of confounding, bias or chance, and a significant risk that the relationship is not causal
3	Nonanalytical studies (e.g. case reports, case series)
4	Expert opinion, formal consensus

RCT, randomized controlled trials. ^aStudies with a level of evidence ‘−’ should not be used as a basis for making a recommendation.

Class	Evidence
A	At least one meta‐analysis, systematic review or RCT rated as 1++, and directly applicable to the target population
	A systematic review of RCTs or a body of evidence consisting principally of studies rated as 1+, directly applicable to the target population and demonstrating overall consistency of results
	Evidence drawn from a NICE technology appraisal
B	A body of evidence including studies rated as 2++, directly applicable to the target population and demonstrating overall consistency of results
	Extrapolated evidence from studies rated as 1++ or 1+
C	A body of evidence including studies rated as 2+, directly applicable to the target population and demonstrating overall consistency of results
	Extrapolated evidence from studies rated as 2++
D	Evidence level 3 or 4
	Extrapolated evidence from studies rated as 2+
	Formal consensus
D (GPP)	A GPP is a recommendation for best practice based on the experience of the GDG

RCT, randomized controlled trial; NICE, National Institute for Health and Care Excellence; GPP, good practice point; GDG, guidelines development group.

2 Appendix

Discharge letter

Dear Colleague,

Re: PATIENT'S NAME

Your patient was recently admitted to hospital following an episode of Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN). SJS/TEN is a rare, severe mucocutaneous reaction, usually to drugs, characterized by blistering and epithelial sloughing of skin and mucous membranes. During the acute phase, patients are extremely ill, often requiring admission to an intensive care unit or to a burn centre. The chronic phase of SJS/TEN is characterized by symptoms that develop insidiously over the weeks and months following presentation and usually persist after discharge from hospital. The most common sequelae are:

1. fatigue, especially in the initial weeks and months after discharge;

2. eye complications, including dryness, pain, photophobia, corneal scarring and visual impairment;

3. skin dyspigmentation (both hyper‐ and hypopigmentation);

4. loss of nails (sometimes with dystrophic regrowth);

5. urogenital problems, including dryness and complications from scarring;

6. mouth problems, including dryness, dental caries and complications from scarring;

7. post‐traumatic stress disorder, including nightmares, anxiety and depression;

8. other complications include bronchiectasis and other sequelae of lung involvement; stenosis of the gastrointestinal tract; loss of muscle mass.

Your vigilance in monitoring for these complications is vital; many of the long‐term outcomes, particularly ocular disorders, can be improved by early recognition and prompt intervention. In the event of any of the above sequelae please contact our department and we will refer the patient to the appropriate specialist.

SJS/TEN is most often caused by a medication. In the case of your patient the cause of SJS/TEN has been identified as ……………… . It is essential that your patient is not prescribed ………………, or drugs of the same class. We have informed him/her about which medication(s) to avoid. We would appreciate if you could order a MedicAlert bracelet or amulet identifying the culprit medication(s) above and indicating the reaction he/she suffered (SJS/TEN). Some patients develop a reluctance to take new medicines; however, recurrences of SJS/TEN are extremely rare and are almost always due to a repeat exposure to the offending drug or another in the same class.

If appropriate: In approximately 15% of SJS/TEN cases no drug can be implicated.

Please contact the department if you wish to discuss your patient's illness, or for further information about SJS/TEN.

Yours sincerely,

Author notes

Funding sources None.

Conflicts of interest J.K.G.D. has received grant/research support from Dompé and SiFi. M.R.A.‐J. has done commissioned work for Genus Pharmaceuticals; received sponsorship from AbbVie, Janssen‐Cilag, Pfizer, Galderma and Steifel to attend conferences; has performed clinical trials for Zymogenetics, Pfizer, Genentech, Johnson & Johnson, Centocor and Novartis; has received grant/research support from Emblation; and has developed a nonprofit website, www.drugrash.co.uk, to assist clinicians in the management of drug allergy. None of the authors has received commercial support from the manufacturers of any medication used in the management of Stevens–Johnson syndrome/toxic epidermal necrolysis.

D.C., S.A.W., P.D., H.Y.L., J.K.G.D., J.S., C.B.B., M.R.A.‐J., K.M.T.W., G.A.E.W., M.P., A.V., R.V.M., G.W., M.S., D.B., P.W. and C.H.S. are members of the guideline development group with technical support provided by L.S.E. and M.F.M.M.

This is a new set of guidelines prepared for the British Association of Dermatologists (BAD) Clinical Standards Unit, which includes the Therapy & Guidelines (T&G) Subcommittee. Members of the Clinical Standards Unit that have been involved are P.M. McHenry (Chairman T&G), J.R. Hughes, M. Griffiths, K. Gibbon, A.J. McDonagh, D.A. Buckley, I. Nasr, V.J. Swale, C.E. Duarte Williamson, N.J. Levell, T. Leslie, E. Mallon, S. Wakelin, S. Ungureanu, P. Hunasehally, M. Cork, K. Towers [British National Formulary (BNF)], J. Donnelly (BNF), C. Saunders (British Dermatological Nursing Group), L.S. Exton (BAD Information Scientist), A.G. Brain (BAD Clinical Standards Administrator) and M.F. Mohd Mustapa (BAD Clinical Standards Manager).

Produced in 2016 by the British Association of Dermatologists. NICE has accredited the process used by the British Association of Dermatologists to produce guidelines. Accreditation is valid for 5 years from May 2010 and has been extended by agreement to May 2016. More information on accreditation, and full details of our accreditation, can be viewed at www.nice.org.uk/accreditation.