Toxic epidermal necrolysis and Stevens Johnson syndrome in South Africa: a 3-year prospective study

Transcription

1 Q J Med 2012; 105: doi: /qjmed/hcs078 Advance Access Publication 28 April 2012 Toxic epidermal necrolysis and Stevens Johnson syndrome in South Africa: a 3-year prospective study S.M.H. KANNENBERG 1, H.F. JORDAAN 1, C.F.N. KOEGELENBERG 2, F. VON GROOTE-BIDLINGMAIER 2 and W.I. VISSER 1 From the 1 Division of Dermatology and 2 Division of Pulmonology, Department of Medicine, University of Stellenbosch & Tygerberg Academic Hospital, Western Cape Province, Cape Town, South Africa Address correspondence to Dr S.M.H. Kannenberg, Division of Dermatology, Department of Medicine, University of Stellenbosch, P O Box 19063, Tygerberg, 7505, Cape Town, South Africa. surethakannenberg@hotmail.com Received 4 January 2012 and in revised form 28 March 2012 Summary Background: Toxic Epidermal Necrolysis (TEN) and Stevens Johnson syndrome (SJS) remain feared medication-related reactions. HIV infection and tuberculosis predispose to drug eruptions, yet there is a paucity of data on TEN/SJS in populations with high prevalences of both diseases. Aim: The aim of this prospective observational study was to describe the features and outcomes of patients admitted with TEN/SJS at a large academic hospital in South Africa. We aimed to identify poor prognostic indicators and to validate the use of the TEN-specific severity-of-illness score (SCORTEN) in this population. Methods: All patients admitted with TEN/SJS over a 3-year period were enrolled. Disease severity was graded according to percentage skin involved and SCORTEN. Co-morbid diagnoses, clinical features, investigations, complications and outcomes were noted. Results: 75 patients ( years, 16 males, 59 HIV positive) were classified as TEN (n =42),TEN/SJS overlap (n = 11) and SJS (n = 22). Twenty-four percent died, most from refractory septic shock. Non-survivors had a higher mean SCORTEN on Days 1 and 3 (1.89 vs. 1.04, P = and 2.27 vs. 0.90, P < 0.001). ASCORTEN52 ondays1and3predictednonsurvival (OR = 2.94, P = 0.047; OR = 7.45, P < 0.001). Other predictors of non-survival included HIV infection (OR = 6.01, P = 0.058), HIV tuberculosis co-infection (OR = 8.5, P < 0.001), 540% skin involvement (OR = 20.27, P < 0.001), anaemia (OR = 4.68, P = 0.005), hypoalbuminemia (OR = 8.5, P = 0.001) and severe sepsis (OR = 71.09, P < 0.001). Conclusions: Most patients with TEN/SJS were HIV positive and female. We validated the use of SCORTEN and identified several prognostic indicators, most significant being HIV-tuberculosis co-infection, 540% skin involvement and severe sepsis. Introduction Toxic Epidermal Necrolysis (TEN) and Stevens Johnson syndrome (SJS) remain two of the most serious and feared medication-related cutaneous adverse reactions. 1 TEN and SJS form part of a spectrum of unpredictable reactive disorders, with SJS at the more benign end and TEN at the other. 2 The latter, in fact, has a mortality rate of 30%. 3 Both are fortunately rare afflictions with an incidence of approximately two per million population per year in Europe. 1,4 There is a myriad of drugs that can cause TEN and SJS, with several antibiotics, anti-tuberculosis drugs,! The Author Published by Oxford University Press on behalf of the Association of Physicians. All rights reserved. For Permissions, please journals.permissions@oup.com

2 840 S.M.H. Kannenberg et al. anti-convulsants, nevirapine, nonsteroidal anti-inflammatory drugs (NSAIDS) and allopurinol being some of the most frequent offending agents. 5 7 Furthermore, human immunodeficiency virus (HIV) infection has been linked to an increased risk for TEN and SJS, but whether this association is due to an increased exposure to drugs or immunomodulation remains unclear. 8 Bastuji-Garin et al. 9 developed and validated a TEN-specific severity-of-illness score (SCORTEN) more than a decade ago. The SCORTEN, which should be calculated on Days 1 and 3, have been shown to be remarkably accurate in predicting mortality in patients from Europe and North America. 1,10 It is currently widely utilized by health-care workers in the developing world, despite the fact that it has as yet never been validated in a population with a high HIV and tuberculosis (TB) prevalence. 8 Furthermore, practically all other proposed prognostic factors were described in cohorts from non-endemic areas. 11 The South African population comprises only 0.7% of the global population but carries 17% of the HIV burden. 12 Approximately 15% of all pregnant women attending antenatal clinics in the Western Cape Province were HIV positive in Furthermore, the incidence of pulmonary TB in the Western Cape Province is 940 cases per 100,000 population, the highest recorded incidence according to the World Health Organisation. 13 Both diseases predispose to drug reactions, yet there is a paucity of published data on TEN and SJS in this population. The main aim of this 3-year prospective observational study was to describe the clinical features, risk factors, severity, complications and outcomes of patients admitted with TEN and SJS at a large academic hospital in Cape Town, South Africa. Furthermore, we aimed to identify potential poor prognostic indicators and to validate the SCORTEN specifically in this population. Methods Study population All patients referred to the Division of Dermatology of Tygerberg Academic Hospital with probable TEN or SJS during the 3-year period July 2008 to June 2011 were potential candidates for this prospective observational study. Our institution is a 1200-bed academic hospital in Cape Town, South Africa. It is one of two referral centres and renders a tertiary service to a population of 1.5 million people. The Health Research Ethics Committee of Stellenbosch University approved the study (project number N08/ 09/242). We screened all relevant referrals and included all patients who required admission and who fulfilled the diagnostic criteria for either TEN or SJS (see below), provided that informed consent could be obtained from either the study subject or, in the case of minors, from a legal guardian. Clinical criteria for TEN or SJS We considered TEN and SJS to be clinical diagnoses and performed biopsies only for confirmation if there was no clear history of exposure to a potential precipitating agent (in the preceding 2 months), atypical clinical features or at the discretion of the treating physician. We adapted the diagnostic criteria suggested by Bastuji-Garin and co-workers 2 and stratified patients into three main categories: (i) SJS: epidermal detachment of <10% of the body surface in association with widespread erythema and/or dusky red macules. (ii) SJS/TEN overlap: epidermal detachment of 10 30% of the body surface plus widespread dusky red macules. (iii) TEN: epidermal detachment of >30% of the body surface coupled with widespread dusky red macules. Demographic data, comorbid diagnosis and risk factors Apart from general demographic data, comorbid diagnoses were documented, including HIV status. CD 4 counts were not specifically requested for study purposes, but all available results were collected (including results from the three months prior to admission). A detailed drug history was obtained from each patient, and the potential precipitating agent(s), based on temporal relation, were noted. Skin involvement and disease severity The duration of the skin eruption (prior to admission), morphology, the presence of mucosal involvement and the percentage of body surface involved were documented in each case. Moreover, the SCORTEN was ascertained on all subjects on Days 1 and 3. One point was scored for each of seven criteria present [age >40 years, presence of a malignancy, heart rate >120, percentage of epidermal detachment >10%, serum urea level >10 mmol/l (>27 mg/dl), serum glucose level >14 mmol/l and serum bicarbonate level <20 mmol/l]. 9 Full blood counts, basic biochemistry and C-reactive protein levels were routinely determined on admission.

3 TEN and SJS in South Africa 841 Management, course and complications All patients received maximal supportive therapy (including fluid resuscitation, environmental control, sepsis surveillance, wound dressings, dietary support and physiotherapy), as the evidence for the general use of other medical interventions (e.g. intravenous immunoglobulin) are lacking. 1 Patients were referred to a high dependence unit or intensive care unit at the discretion of the attending clinician. We specifically documented the development of acute respiratory distress syndrome (ARDS), severe sepsis, acute renal failure and hospital acquired pneumonia, all defined according to accepted criteria Finally, patients were classified as either survivors (alive at the time of hospital discharge) or non-survivors. In the absence of a post-mortem investigation, we accepted the opinion of the department s morbidity and mortality committee on the cause of death. Statistical aspects Descriptive statistics and chi-square (not Yates corrected) or Fisher s exact tests (where indicated) were performed on dichotomous categorical variables, and t-tests on continuous data. Unless stated otherwise, data are displayed as means and standard deviation (SD). Results Study population and clinical criteria Between July 2008 and June 2011, a total number of 75 patients ( years, range 3 57 years) were enrolled. Of the 75 patients, 53 were of black and 22 of mixed race. The majority was female (n = 51, 68.0%) and HIV positive (n = 59, 78.7%). In total, 42 patients were categorized as TEN ( years), 11 as SJS/TEN overlap ( years) and 22 as SJS ( years). Eighteen patients (24.0%, years) did not survive their Table 1 Comorbid diagnoses and risk factors Total (n = 75), n (%) (n = 57), n (%) admission: 17 patients with TEN died (mortality 40.5%), as opposed to one patient with SJS/TEN overlap (mortality 9.1%, P = 0.075) and no patients with SJS (mortality 0%, P = 0.001). Comorbid diagnoses and risk factors The comorbid diagnoses and clinical risk factors for the development of TEN/SJS are summarized in Table 1. Of note is the fact that non-survivors were more likely to have underlying TB (OR = 8.5, P < 0.001) than survivors. In fact, the mortality of TEN in patients on TB treatment was 60% compared to 15% in the rest of the cohort (P = 0.001). All TB and pregnant/postpartum patients were HIV co-infected. HIV-positive patients had a mortality rate of 29.8%, whereas HIV-negative patients had a mortality rate of 6.3% (P = 0.058). Furthermore, the proportion of HIV-infected individuals among survivors and non-survivors differed (P = 0.097; Table 1). CD 4 counts were available in 55 of the 59 HIV-infected individuals, and the mean counts of patients who survived ( cells/ml) did not differ significantly from those who died ( cells/ml; P = 0.442). In comparison to survivors, non-survivors tended not to have epilepsy (P = 0.056). Only one patient (survivor) had an underlying malignancy (glioblastoma). Co-trimoxazole and Nevirapine were the most common precipitants of TEN/SJS (Table 2). Precipitating agents were commenced 3 58 days (mean 25.5 days) prior to the onset of TEN and/or SJS, ranging from a mean of 31.0 days for anti-tuberculosis treatment to 27.8 days for co-trimoxazole, 24.9 days for Nevirapine and 18.8 for all other drugs. The three cases of NSAID-induced disease occurred 3 14 days (mean 8.0 days) after exposure, which was significantly less than the other precipitants (P = 0.037). Among the pregnant/postpartum patients, Nevirapine was responsible for 9 of the 11 cases. In only three cases were no known precipitant identified. Of note is the fact that Carbamazepine was Non-survivors (n = 18), n (%) OR (95% CI) P-value HIV positive 59 (78.7) 42 (73.7) 17 (94.4) 6.01 ( ) 0.097* TB 15 (20) 6 (10.5) 9 (50) 8.50 ( ) <0.001 Epilepsy 11 (14.7) 11 (19.3) 0 (0) NA 0.056* Pregnant/postpartum 11 (14.7) 9 (15.8) 2 (11.1) 0.67 ( ) 0.725* Other 11 (14.7) 7 (12.3) 4 (22.2) 2.04 ( ) 0.444* *Fisher s exact test. CI: confidence interval; OR: odds ratio; NA: not applicable; TB: tuberculosis.

4 842 S.M.H. Kannenberg et al. responsible for one fatality in a non-epileptic patients (indication for its use was peripheral neuropathy), but that no single fatality was seen secondary to anticonvulsant therapy in epileptics. Skin involvement and disease severity Skin involvement was present for 7.0 (6.3) days in survivors and 8.1 (6.0) days in non-survivors (P = 1) prior to admission. For the study population as a whole, the body surface involvement ranged from 5% to 90% (mean 37.7%). Non-survivors had a mean involvement of 58.3% (16.9%), compared to survivors who had a mean of 31.2% (25.3%, P < 0.001). In total, 26 patients had 540% skin involvement: 17 (39.5%) of these patients died. The odds ratio (OR) of dying with this degree of involvement was (95% CI , P < 0.001). Non-survivors had a higher mean SCORTEN on Day 1 (1.89 vs. 1.04, P = 0.006) and on Day 3 (2.27 vs. 0.90, P < 0.001). Mortality rates according to SCORTEN are provided in Table 3. The combined mortality of patients with a SCORTEN 52 on Day 1 was 58.8%, whereas the combined mortality rates of patients with a SCORTEN 52 and 53 on Day 3 were 45.8 and 75.0%, respectively. The OR of dying with a SCORTEN 52 and 53 on Day 3 was 7.45 and 6.24, respectively (Table 4). All but one patient (with SJS) had mucosal involvement. Of the 42 patients with TEN, atypical features observed included diffuse bran-like desquamation (n = 7) and no visible denudation (n = 3). Atypical features observed with SJS/TEN overlap included desquamation (n = 2), no visible denudation (n = 1) and no erythema (n = 1), and with SJS these were desquamation (n = 3), no denudation (n = 6) and no erythema (n = 1). Histology Table 2 Precipitating therapeutic agents Total (n = 75), n (%) (n = 57), n (%) was obtained in seven patients, and the results were in all cases compatible with the diagnosis of TEN/SJS in various stages of evolution. Laboratory findings are summarized in Table 5. Of note is the fact that non-survivors had a significantly lower serum haemoglobin concentration (P < 0.001) and lower serum albumin (P < 0.001). A haemoglobin <10 g/dl was associated with mortality of 45.5% and an OR of 4.68 (95% CI , P = 0.005) of dying, whereas a serum albumin <25 g/l was associated with mortality of 60.0% and an OR of 8.50 (95% CI , P = 0.001) of non-survival. Course and systemic complications Proven or suspected severe sepsis was not only the most common complication (Table 6, n = 28), but also the only complication significantly associated with death (OR = 71.09, 95% CI ; Table 3 Mortality rates according to SCORTEN Total (n = 75) Non-survivor (n = 18), n (%) (n = 57) OR (95% CI) Non-survivors (n = 18) Mortality (%) SCORTEN Day SCORTEN Day 3 a a Seventy-two patients were alive on Day 3. P-value Co-trimoxazole 18 (24) 15 (26.3) 3 (16.7) 0.56 ( ) 0.534* Nevirapine 22 (29.3) 18 (31.6) 4 (22.2) 0.62 ( ) 0.184* Anti-tuberculosis treatment a 15 (20) 6 (10.5) 9 (50) 8.50 ( ) <0.001 Isoniazid (monotherapy) 1 (1.3) 1 (1.8) 0 (0) NA NA Anticonvulsants 12 (16) 11 (19.3) 1 (5.6) 0.25 ( ) 0.273* Antibiotics 4 (5.3) 3 (5.3) 1 (5.6) 1.06 ( ) 1 NSAIDS 3 (4) 3 (5.3) 0 (0) NA Other 3 (4) 2 (3.5) 1 (5.6) 1.62 ( ) 1 Unknown 3 (4) 2 (3.5) 1 (5.6) 1.62 ( ) 1 a Combination therapy (Rifampicin, Isoniazid, Pyrazinamide and Ethambutol). *Fisher s exact test. OR: odds ratio; CI: confidence interval; NSAIDS: non-steroidal anti-inflammatory drugs; NA: not applicable.

5 TEN and SJS in South Africa 843 Table 4 SCORTEN odds of non-survival Total (n = 75), n (%) (n = 57), n (%) Non-survivors (n = 18), n (%) OR (95% CI) P-value SCORTEN Day 1 n (36) 17 (29.8) 10 (55.6) 2.94 ( ) n (9.3) 5 (8.7) 2 (11.1) 1.30 ( ) 1* SCORTEN Day 3 n (32.0) 13 (22.8) 11 (61.1) a 7.45 ( ) <0.001 n (14.0) 2 (3.5) 6 (33.3) a 6.24 ( ) 0.011* a 15 of 18 non-survivors were alive on Day 3. *Fisher s exact test. OR: odds ratio; CI: confidence interval. Table 5 Laboratory investigations Total (n = 75) (n = 57) Non-survivors (n = 18) P-value Full blood count (mean SD) White cell count (10 9 /l) 6.24 (3.85) 6.10 (3.04) 6.64 (5.74) Neutrophils (10 9 /l) 3.91 (2.81) 4.12 (2.87) 4.12 (2.87) Lymphocytes (10 9 /l) 1.38 (1.34) 1.36 (0.97) 1.29 (2.20) Haemoglobin (g/dl) 11.0 (2.2) 11.5 (2.0) 9.5 (2.1) <0.001 Platelets (10 9 /l) 270 (136) 283 (141) 230 (114) C-Reactive Protein (mean SD) (94.2) (96.4) (88.1) Biochemistry (mean SD) Urea (mmol/l) 6.1 (4.4) 5.7 (4.5) 7.1 (3.9) Creatinine (mmol/l) 87.6 (38.2) 87.7 (39.4) 87.3 (35.5) Albumin (g/l) 28.7 (8.4) 31.9 (7.3) 22.2 (6.4) <0.001 SD: standard deviation. Table 6 Major systemic complications Total (n = 75), n (%) (n = 57), n (%) Non-survivors (n = 18), n (%) P-value Severe sepsis 28 (37.3) 11 (19.3) 17 (94.4) <0.001 Hospital acquired pneumonia 4 (5.3) 2 (3.5) 2 (11.1) 0.242* ARDS 2 (2.7) 0 (0) 2 (11.1) 0.055* Acute renal injury 2 (2.7) 0 (0) 2 (11.1) 0.055* *Fisher s exact test. ARDS: acute respiratory distress syndrome. P < 0.001). In total, 23 patients had positive blood cultures, with Staphylococcus aureus (n = 13), Acinetobacter species (n = 8) and Klebsiella pneumoniae (n = 4) being the most common isolates. Of the 17 patients with TEN who did not survive, all died of complications related to TEN or the hospitalization, namely refractory septic shock (n = 12), ARDS (n = 2), hospital-acquired pneumonia (n = 1) and pulmonary embolism (n = 1). The single mortality among the patients with SJS/TEN overlap was as a result of associated acute liver failure secondary to drug-induced hepatitis. Discussion In this prospective observational study, we found the majority of the patients with TEN and SJS to be

6 844 S.M.H. Kannenberg et al. young to middle-aged, female and HIV positive (76%). Co-trimoxazole and Nevirapine were the most common precipitants of TEN/SJS. We observed a mortality rate of 24.0%, with practically all deaths observed in patients with TEN and most secondary to refractory septic shock. The use of the SCORTEN was validated: non-survivors had higher mean SCORTEN values on Days 1 and 3 (1.89 vs. 1.04, P = and 2.27 vs. 0.90, P < 0.001). A SCORTEN 52 on Days 1 and 3 predicted non-survival (OR = 2.94, P = and OR = 7.45, P < 0.001). Several other predictors of non-survival were identified, including underlying HIV infection (but not CD 4 counts), HIV tuberculosis co-infection, 540% skin involvement, anaemia, hypoalbuminemia and severe sepsis. The increased prevalence of TEN/SJS in patients with HIV is well known. 1,8 As a risk factor for non-survival, however, HIV infection only narrowly missed statistical significance (P = 0.058). Interestingly, the mean CD 4 count was non-significantly higher in non-survivors. It is plausible that the observed differences in mortality rate in HIV- and non-infected individuals (29.8% vs. 6.3%) may have reached statistical significance had more patients been enrolled. The fact that 76% of our study population was HIV positive undoubtedly impacted on the general demographic profile of our study population. The mean age of patients was 39.9, which is lower than reported figures. 1 4,18 Moreover, a female predominance (3.7:1) was observed, which at least partially could be explained by the fact that 60% of the HIV-infected individuals in Southern Africa are female and by the routine use of Nevirapine to reduce mother-to-child transmission. 12,19 The development of TEN in the setting of TB with HIV co-infection was associated with excess mortality (OR = 8.50, 95% CI , P < 0.001). The fact that drug reactions are more common in co-infected patients is well documented, 20 but the present study is to the best of our knowledge the first to show a higher mortality in this setting. Drug reactions are often early manifestations of HIV infection and presumably due to a combination of (i) early altered cellular immunity and (ii) modifications in drug metabolism secondary to the induction of defective metabolic pathways due to HIV-induced production of interferons. 8 Th 1 /Th 2 cytokine imbalance is seen in practically all patients, as are reductions in the number and function of Langerhans cells, CD 4, NK cells, macrophages and monocytes Dermal dendritic cells are initially spared, and an increased synthesis of nitric oxide driven by HIV can be observed. 22 Exactly how these and other mechanisms interplay in the development of TEN and SJS, and what role the higher drug burden plays, remains to be elucidated. 8,21,22 Moreover, we found that the CD 4 count did not predict outcome in HIV-positive individuals, suggesting that degree of overall immunosuppression was less important. The reasons for the worse outcome in HIV-positive individuals are unclear but probably related to the impaired cellular immunity. 8,21,22 In the landmark study by Bastuji-Garin et al. 9 published in 2000, mortality was significantly higher in patients with a SCORTEN level of 2 and above. Numerous subsequent studies have confirmed the prognostic value of the SCORTEN. 18,25 29 Of note is the fact that practically all studies were performed in Europe and North America, enrolled older patients with a mean age >50 years and included very few patients with either HIV or TB. The present study is to the best of the author s knowledge the first to show that SCORTEN remains a valuable tool at predicting mortality in a young population with a high prevalence of HIV. A SCORTEN of 2 and above, both on Day 1 (OR = 2.94; P = 0.047) and even more so on Day 3 (OR = 7.45; P < 0.001) predicted non-survival. These findings are in keeping with the findings of Guegan et al. 10 who previously showed that the SCORTEN s predictive value peaked on Day 3. We observed a higher mortality in patients with a greater percentage of body surface involved. Moreover, patients with 540% involvement had an OR of (95% CI ; P < 0.001) of dying. Bastuji Garin reported an OR of 4.8 (95% CI ) with 530% involvement. 9 Other laboratory parameters not included in the SCORTEN, but known to be associated with increased mortality, include thrombocytopaenia, leucopaenia and neutropaenia. 1,30,31 Interestingly, we found anaemia and hypoalbuminemia to be significantly associated with non-survival. A serum albumin <25 g/l on admission was associated with mortality rate of 60.0% and an OR of 8.50 of non-survival, making one of the strongest predictors of adverse outcome. However, it is plausible that a low serum albumin is purely a surrogate of disease severity and duration and not an independent predictor. The mortality rate of 24% observed in our cohort of TEN and SJS was comparable to published figures. 1,9,10,32 The mortality rate of TEN (40.5%) was significantly higher than SJS (0%, P = 0.001) and on the higher end of the spectrum of reported mortality rates. 10,28,32,33 Refractory septic shock was shown to be the most common cause of death, and the diagnosis of severe sepsis was associated with an OR of of dying. Severe sepsis, per definition, implicates the presence of organ dysfunction,

7 TEN and SJS in South Africa 845 hypoperfusion or hypotension, and may precede multi-organ dysfunction. 15 Although previous studies have shown no difference in the number of nosocomial infections in survivors and nonsurvivors, 1,9,28 our data suggest severe sepsis to be strongly associated with non-survival. TEN and SJS remain rare conditions, 1 3 and it is not surprising that practically all published data are retrospective in nature. 4,10,18,23 25,27 29 Our study population represents, to the best of our knowledge, one of the largest prospectively studied single centre cohorts, which is arguably the study s greatest strength. 9 Our study has certain limitations. We relied purely on a detailed drug history to identify offending agents (as there is no gold standard), and we did not differentiate between the various anti-tuberculous drugs. Furthermore, patients were generally not rechallenged while under our care. The data on the various precipitating therapeutic agents may therefore not be entirely accurate. In conclusion, in our population the majority of patients with TEN and SJS are young to middle aged, female and HIV positive. TEN has a mortality rate of 40%, which is comparable to reported rates. We validated the use of the SCORTEN, which predicted mortality (particularly on Day 3) and identified several other predictors of non-survival, most significant being HIV tuberculosis co-infection, 540% skin involvement and severe sepsis. Acknowledgements All authors contributed equally with study design, data collection, data analysis and data interpretation. SMH Kannenberg wrote the manuscript, which was revised by all other contributing authors. Conflict of interest: None declared. References 1. Pereira F, Mudgil A, Rosmarin D. Toxic epidermal necrolysis. J Am Acad Dermatol 2007; 56: Bastuji-Garin S, Rzany B, Stern RS, Shear NH, Naldi L, Roujeau JC. Clinical classification of cases of toxic epidermal necrolysis, Stevens Johnson syndrome, and erythema multiforme. Arch Dermatol 1993; 129: Ghislain PD, Roujeau JC. Treatment of severe drug reactions: Stevens Johnson syndrome, toxic epidermal necrolysis and hypersensitivity syndrome. Dermatol Online J 2002; 8:5. 4. Bastuji-Garin S, Zahedi M, Guillaume JC, Roujeau JC. Toxic epidermal necrolysis (Lyell syndrome) in 77 elderly patients. Age Ageing 1993; 22: Roujeau JC, Kelly JP, Naldi L, Rzany B, Stern RS, Anderson T, et al. Medication use and the risk of Stevens Johnson syndrome or toxic epidermal necrolysis. N Engl J Med 1995; 333: Caumes E, Bossi P, Katlama C, Bricaire F. Antiretroviral induced toxiderma in HIV-infected patients. Presse Med 2003; 32: Leone R, Venegoni M, Motola D, Moretti U, Piazzetta V, Cocci A, et al. Adverse drug reactions related to the use of fluoroquinolone antimicrobials: an analysis of spontaneous reports and fluoroquinolone consumption data from three Italian regions. Drug Saf 2003; 26: Todd G. Adverse cutaneous drug eruptions and HIV: a clinician s global perspective. Dermatol Clin 2006; 24: Bastuji-Garin S, Fouchard N, Bertocchi M, Roujeau JC, Revuz J, Wolkenstein P. SCORTEN: A severity-of-illness score for toxic epidermal necrolysis. J Invest Dermatol 2000; 115: Guegan S, Bastuji-Garin S, Poszepczynska-Guigne E, Roujeau JC, Revuz J. Performance of the SCORTEN during the first five days of hospitalization to predict the prognosis of epidermal necrolysis. J Invest Dermatol 2006; 126: Faye O, Wechsler J, Roujeau JC. Cell-mediated immunologic mechanism and severity of TEN. Arch Dermatol 2005; 141: Karim SS, Churchyard GJ, Karim QA, Lawn SD. HIV infection and tuberculosis in South Africa: an urgent need to escalate the public health response. Lancet 2009; 374: World Health Organization report Global tuberculosis control- surveillance, planning, financing. Geneva, 2008:19 [ index.html.] Accessed 12 October Bernard GR, Artigas A, Brigham KL, Carlet J, Falke K, Hudson L, et al. The American European Consensus Conference on ARDS. Definitions, mechanisms, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 1994; 149: Bone RC, Balk RA, Cerra FB, Dellinger RP, Fein AM, Knaus WA, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest 1992; 101: Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P. Acute renal failure definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Critical Care 2004; 8:R American Thoracic Society. Guidelines for the management of adults with hospital-acquired, ventilator-associated, and healthcare-associated pneumonia. Am J Respir Crit Care Med 2005; 171: Rajaratnam R, Mann C, Balasubramaniam P, Marsden JR, Taibjee SM, Shah F, et al. Toxic epidermal necrolysis: retrospective analysis of 21 consecutive cases managed at a tertiary centre. Clin Exp Dermatol 2010; 35: Hunt GM, Coovadia A, Abrams EJ, Sherman G, Meyers T, Morris L, et al. HIV-1 drug resistance at antiretroviral treatment initiation in children previously exposed to single-dose nevirapine. AIDS 2011; 25: Yee D, Valiquette C, Pelletier M, Parisien I, Rocher I, Menzies D. Incidence of serious side effects from first-line antituberculosis drugs among patients treated for active tuberculosis. Am J Respir Crit Care Med 2003; 167:

8 846 S.M.H. Kannenberg et al. 21. Lehloenya RJ, Todd G, Badri M, Dheda K. Outcomes of reintroducing anti-tuberculosis drugs following cutaneous adverse drug reactions. Int J Tuberc Lung Dis 2011; 15: Cedeno-Laurent F, Gómez-Flores M, Mendez N, Ancer- Rodríguez J, Bryant JL, Gaspari AA, et al. New insights into HIV-1-primary skin disorders. Int AIDS Soc 2011; 14: Belsito DV, Sanchez MR, Baer RL, Valentine F, Thorbecke GJ. Reduced Langerhans cell Ia antigen and ATPase activity in patients with the acquired immunodeficiency syndrome. N Engl J Med 1984; 310: Tschachler E, Groh V, Popovic M, Mann DL, Konrad K, Safai B, et al. Epidermal Langerhans cells a target for HTLV-III/LAV infection. J Invest Dermatol 1987; 88: Trent JT, Kirsner RS, Romanelli P, Kerdel FA. Analysis of intravenous immunoglobulin for the treatment of toxic epidermal necrolysis using SCORTEN: The University of Miami Experience. Arch Dermatol 2003; 139: Trent JT, Kirsner RS, Romanelli P, Kerdel FA. Use of SCORTEN to accurately predict mortality in patients with toxic epidermal necrolysis in the United States. Arch Dermatol 2004; 140: Gerdts B, Vloemans AF, Kreis RW. Toxic epidermal necrolysis: 15 years experience in a Dutch burns centre. J Eur Acad Dermatol Venereol 2007; 21: Cartotto R, Mayich M, Nickerson D, Gomez M. SCORTEN accurately predicts mortality among toxic epidermal necrolysis patients treated in a burn center. J Burn Care Res 2008; 29: Sekula P, Liss Y, Davidovici B, Dunant A, Roujeau JC, Kardaun S, et al. Evaluation of SCORTEN on a cohort of patients with Stevens Johnson syndrome and toxic epidermal necrolysis included in the RegiSCAR study. J Burn Care Res 2011; 32: Westly ED, Wechsler HL. Toxic epidermal necrolysis: granulocytic leukopenia as a prognostic indicator. Arch Dermatol 1984; 120: Engelhardt SL, Schurr MJ, Helgerson RB. Toxic epidermal necrolysis: an analysis of referral patterns and steroid usage. J Burn Care Rehabil 1997; 18: Schulz JT, Sheridan RL, Ryan CM, MacKool B, Tompkins RG. A 10-year experience with toxic epidermal necrolysis. J Burn Care Rehabil 2000; 21: Haber J, Hopman W, Gomez M, Cartotto R. Late outcomes in adult survivors of toxic epidermal necrolysis after treatment in a burn center. J Burn Care Rehabil 2005; 26:33 41.