Skip to main content

Stevens–Johnson syndrome/toxic epidermal necrolysis and erythema multiforme drug-related hospitalisations in a national administrative database



Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN) and erythema multiforme (EM) are immunologically-mediated dermatological disorders commonly triggered by drug exposure and/or other external agents. We aimed to characterise SJS/TEN- and EM-drug-related hospitalisations in a nationwide administrative database, focusing on demographic and clinical characteristics, and in the most frequently implicated drug classes.


We analysed all drug-related hospitalisations with associated diagnosis of SJS/TEN or EM in Portuguese hospitals between 2009 and 2014. We compared gender, age, comorbidities, length of stay, and in-hospital mortality and estimated the number of episodes per million packages sold of drug classes. Predictors of in-hospital mortality were investigated in both conditions by logistic regression.


There were 132 SJS/TEN-related and 122 EM-related hospitalisations. Incidence and in-hospital mortality of SJS/TEN episodes (24.2%) were consistent with previous studies. HIV co-infection was more common among SJS/TEN hospitalisations (9 vs. 2% with EM; P = 0.009). Liver disease, advanced age, and a TEN diagnosis, were significantly associated with higher risk of mortality in patients with SJS/TEN. The highest numbers of SJS/TEN and EM episodes per million drug packages sold were observed for antivirals (8.7 and 1.5, respectively), antineoplastic/immunosuppressive drugs (5.6 and 3.9, respectively) and hypouricaemic drugs (5.0 and 2.4, respectively).


SJS/TEN in-hospital mortality is high, and its risk factors include advanced age, liver disease, and TEN diagnosis. The drug classes most frequently associated with these conditions include antivirals, hypouricaemic drugs and antineoplastic/immunosuppressive drugs. Administrative databases seem useful in the study of SJS/TEN drug-related hospitalisations, yielding results consistent with previous studies and on a nationwide basis.


Severe cutaneous adverse reactions (SCARs) are an example of severe type B adverse drug reactions, and are associated with high morbidity and mortality [1]. SCARs encompass three distinct clinical entities: (1) the spectrum of Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN), (2) acute generalised exanthematous pustulosis (AGEP), and (3) drug reactions with eosinophilia and systemic symptoms (DRESS) [2, 3]. Diagnosis of these conditions is further complicated by the existence of overlap syndromes, characterised by the coexistence of features from different entities [4]. The SJS/TEN spectrum is the most common and lethal of all SCARs. It associates with a mortality of up to 40%, versus just under 5% for AGEP and 10% for DRESS [5, 6]. SJS/TEN is characterised by cutaneous detachment and blister formation—in SJS, skin detachment affects less than 10% of the body surface area, while TEN requires involvement of over 30% [7, 8]. Cases with 10–30% of body surface area involvement are classified as SJS–TEN overlap syndrome [7, 8].

Until recently, there was widespread belief that erythema multiforme (EM) major was a milder form of SJS/TEN spectrum [9, 10]. However, this assumption has now been largely abandoned—most cases of EM are associated with herpes virus infections, while only a minority are deemed to be caused by drugs [9,10,11,12]. Distinction between EM and SJS/TEN is crucial since the latter is associated with much greater severity and higher mortality [9], and as these conditions have different treatment approaches. Clinically, SJS/TEN is characterised by macules or flat atypical target lesions with widespread distribution or preferential trunk involvement, which rapidly evolve as a blistering disorder of the skin and mucosal surfaces. Conversely, EM typically presents with predominantly acral target lesions [11, 13]. Nevertheless, atypical presentations can make it difficult to distinguish between the two entities [13], especially at the beginning of the clinical presentation and when there is a history of previous drug exposure.

An improved knowledge on the risk factors associated with SJS/TEN might thus facilitate the distinction between SJS/TEN and EM, as well as provide clues concerning the pathophysiology of this condition [11, 14, 15]. However, is spite of its severity, the epidemiology of SJS/TEN remains insufficiently studied [16], in part because its rarity renders traditional case–control or cohort studies particularly time- and resource-consuming. On the other hand, administrative databases are being increasingly used in the assessment of such rare and very rare conditions [16]. Therefore, in this study, we analysed a nationwide administrative database with the aim of characterising drug-related hospitalisations in patients with SJS/TEN, with a focus on gender and age, comorbidities, length of hospital stay, in-hospital mortality, and responsible drug classes. We compared these results to those observed for patients with a diagnosis of drug-related EM, so that we could infer whether in administrative databases SJS/TEN cases are mostly distinguished from other often confused conditions.


We used a database provided by the Portuguese Central Health System Administration containing data for all hospitalisations in mainland Portugal public hospitals. Anonymity was maintained for all hospitals and patients. For each episode, we had access to the main diagnosis (clinical condition responsible for the patient’s admission), up to 19 accompanying diagnoses, and up to 5 external causes of injury and poisoning (including adverse drug effects). Both diagnoses and external causes had been coded with ICD-9-CM codes after discharge; thus, both community cases requiring hospitalisation and in-hospital cases were identified. Coding in Portugal is standardised and performed by doctors with specific training, and internal and external auditing is regularly performed to ensure proper coding [17].

We analysed all hospitalisations with a main or supplementary diagnosis of SJS/TEN (ICD-9-CM codes 695.13–695.15) and an associated E code (ICD-9-CM codes E930.x–E949.x for adverse drug reactions—ICD-9-CM codes are listed in Additional file 1: Table S1—each code corresponds to the drug class deemed responsible for the reaction according to the responsible physician). We separately analysed hospitalisations with an associated diagnosis of SJS (695.13), SJS–TEN overlap (695.14), and TEN (695.15) to allow for comparison between these conditions. Since these three codes were introduced in October 2008 [18], we only analysed hospitalisations between January 2009 and December 2014. SJS/TEN episodes were compared with hospitalisations with main or supplementary diagnosis of EM (ICD-9-CM codes 695.10, 695.11, 695.12, and 695.19) and an associated E code.

We calculated the number of hospitalisations of SJS/TEN and EM per million inhabitants based on data published by the Portuguese National Institute of Statistics [19]. This rate probably provides a good estimation of the 6-year incidence of SJS/TEN in Portugal as the severity of this condition entails almost all patients to be hospitalised in public hospitals.

We compared gender, age, and comorbidities between episodes with a diagnosis of EM and hospitalisations with a diagnosis of SJS/TEN; the three clinical entities of the latter (SJS, SJS–TEN overlap, and TEN) were also compared with each other. We compared the frequency of comorbidities potentially associated with an increased risk of SJS/TEN (whether directly or indirectly), namely chronic kidney disease, hypertension, heart failure, diabetes, HIV infection, and liver disease (Additional file 1: Table S1). Chronic kidney disease is associated with an increased risk of allopurinol-induced SJS/TEN [20]. As the use of diuretics is also associated with the latter condition [21], we also assessed conditions whose treatment frequently requires the use of diuretics, namely hypertension, heart failure, and diabetes. HIV infection was assessed, as the risk of SJS/TEN is known to be higher among HIV+ patients [22]. Liver disease may also be a risk factor for SJS/TEN, particularly in regard to chronic viral hepatitis [23]. We also evaluated length of hospital stay, readmission rate, and in-hospital mortality. To study hospital readmissions, we identified individual patients admitted between 2009 and 2014 and followed them up until the end of the study period. Registries in our database had been anonymised and, therefore, individual patients were identified according to their gender, birthdate and residence—two episodes were deemed to have occurred with the same patient whenever the registered inpatient’s gender, birthdate and residence were equal, and the registered diagnoses were similar.

For each clinical entity, we compared the frequency of the drug classes recorded as implicated in the adverse drug reactions. In the dataset used, adverse drug reactions are identified by E codes, each of which corresponds to a different class of drugs. In addition, based on information provided by the Portuguese Authority of Medicines and Health Products (INFARMED), we estimated the number of EM and SJS/TEN episodes per million packages of drugs sold [24,25,26,27,28]. For this estimate, we excluded data from 2009 due to the risk of underreporting, as the ICD-9-CM SJS/TEN codes were introduced that year.

Categorical variables were compared using the Chi square test and the Fisher exact test. Continuous variables were analysed using the Mann–Whitney U test and the Kruskal–Wallis test. P values < 0.05 were considered statistically significant. To analyse risk factors significantly associated with in-hospital mortality (both for EM and SJS/TEN), we used logistic regression models. We performed univariable analyses assessing the association between in-hospital mortality and gender, age, length of hospital stay, SJS/TEN entities (SJS, SJS–TEN overlap, and TEN), hypertension, diabetes, heart failure, chronic kidney failure, liver disease, and HIV status. Variables with marginal association in the univariate analysis (P < 0.20) were included in multivariable models. The models were assessed by their area under the receiver operating characteristics curve (AUC-ROC) and by the Hosmer–Lemeshow goodness-of-fit test; multicollinearity was assessed using variance inflation factor. The results of the univariable and multivariable analyses are expressed as odds ratio (OR) with 95% confidence intervals (95% CI), and P values. All statistical analyses were performed using SPSS version 22.0 (IBM®SPSS® Statistics, Armonk, NY:IBM Corp.).

For this study, Ethics Committee Approval and Informed consent were not needed, as all data had previously been anonymised.


From 2009 to 2014, we recorded 122 hospitalisations with an associated diagnosis of EM (main diagnosis in 34 cases) and 132 hospitalisations with an associated diagnosis of SJS/TEN (main diagnosis in 89 cases). This corresponds to a 6-year incidence of 13.2 hospitalisations with EM and 12.2 hospitalisations with SJS/TEN per million inhabitants (Table 1). In 2014, we observed a 1-year incidence of 2.1 EM hospitalisations and 3.8 SJS/TEN hospitalisations per million inhabitants, while the 1-year incidences in 2009 were of 3.2 EM hospitalisations and 0.7 SJS/TEN hospitalisations per million inhabitants (Additional file 2: Table S2), suggesting a learning effect as SJS/TEN codes were firstly used in 2009.

Table 1 Demographic and clinical characteristics of hospitalised patients with an associated diagnosis of erythema multiforme (EM) or Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN); Mainland Portugal, 2009–2014 (n = 254 hospitalisations)

SJS (n = 73) accounted for 55% of all SJS/TEN hospitalisations, while SJS–TEN overlap (n = 18) and TEN (n = 41) accounted for 14% and 31%, respectively (Table 1). Patient readmissions accounted for 3% (n = 7) of all hospitalisations (2% for EM vs. 4% for SJS/TEN). All readmissions occurred within 1 year of the first hospitalisation and most cases (all EM readmissions and 33% of SJS/TEN readmissions) were due to exposure to the same drug class. A majority of hospitalisations occurred in females, both for EM (65%) and SJS/TEN (55%). The median age in each case was 63 years. Six percent (n = 7) of EM episodes and 8% (n = 10) of SJS/TEN episodes (8 with SJS and 2 with TEN diagnosis) occurred in children, and 43% and 70% occurred, respectively, in girls. No significant differences were observed for gender or age distribution between the distinct SJS/TEN entities (SJS, SJS–TEN overlap, and TEN).

HIV co-infection was more common in hospitalisations with associated diagnosis of SJS/TEN (9%) than with EM (2%) (P = 0.009). No significant differences were observed for the frequency of any of the other comorbidities studied.

The median length of hospital stay was 10 days for EM versus 15 days for SJS/TEN (P = 0.007). Within the SJS/TEN group, TEN episodes were associated with the shortest median length of stay (14 days); however, considering non-fatal cases only, TEN was associated with a median length of stay of 22 days.

A fatal outcome was reported for 7% of hospitalisations with associated diagnosis of EM and for 24% of SJS/TEN episodes (P < 0.001). In the SJS/TEN group, TEN had the highest proportion of fatal cases (44%), followed by SJS (16%), and SJS–TEN overlap (11%) (P = 0.002). No fatal cases were registered among paediatric patients. Variables significantly associated with in-hospital mortality in hospitalisations with a diagnosis of SJS/TEN after multivariable analysis included advanced age (OR 1.1 per year; 95% CI 1.0–1.1; P = 0.002), a diagnosis of liver disease (OR 7.9; 95% CI 1.2–50.7; P = 0.031), and TEN (OR 6.5; 95% CI 2.3–18.8; P = 0.001) (Table 2). For EM, the variables significantly associated with in-hospital mortality were advanced age (OR 1.1 per year; 95% CI 1.0–1.2; P = 0.005) and male gender (OR 23.9; 95% CI 3.4–168.6; P = 0.001). The multivariable models for SJS/TEN and EM hospitalisations had an AUC-ROC of 0.843 and 0.912, respectively. The Hosmer–Lemeshow goodness-of-fit test did not evidence lack of fit in the multivariable model for SJS/TEN (P = 0.942) or EM (P = 0.995). The models did not show evidence of multicollinearity.

Table 2 Results from binomial logistic regression with in-hospital mortality as the dependent variable

Drug classes most frequently associated with adverse reactions in hospitalised patients with SJS/TEN were antibiotics (26%), uric acid metabolism drugs (20%), and anticonvulsants (17%) (Table 3). The same drug classes were identified in cases of EM: 30% for antibiotics, 17% for uric acid metabolism drugs, and 11% for anticonvulsants. In paediatric patients, antibiotics were responsible for the greatest proportion of adverse reactions in EM (57%) and SJS/TEN (30%) hospitalisations. The two registered cases of TEN in children were associated with antiviral and psychotropic drugs.

Table 3 Drug classes deemed responsible for the cutaneous adverse reactions occurred in the context of hospitalisations with an associated diagnosis of erythema multiforme (EM) or Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN); Mainland Portugal, 2009–2014 (n = 254 hospitalisations)

Eighty-eight percent of episodes with reported adverse reactions to antivirals were in HIV+ patients. In hospitalisations with associated diagnosis of EM, chronic kidney disease was more common in episodes with adverse reactions attributed to uric acid metabolism drugs (P = 0.006). In hospitalisations with SJS/TEN, however, chronic kidney disease was significantly associated with adverse reactions to antibiotics (P = 0.023). Liver disease was not significantly associated with adverse reactions to any of the drug classes analysed.

The drug classes associated with a higher number of SJS/TEN episodes per million packages sold were antiviral drugs (8.7 episodes), followed by anti-neoplastic/immunosuppressive drugs (5.6), uric acid metabolism drugs (5.0), and anticonvulsants (1.2). The corresponding classes for EM were anti-neoplastic/immunosuppressive drugs (3.9 episodes), followed by uric acid metabolism drugs (2.4) and antiviral drugs (1.5)(Table 4).

Table 4 Number of hospitalisations with an associated diagnosis of erythema multiforme (EM) and Stevens–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN) per million sold packages of drug classes involved in adverse reactions; Mainland Portugal, 2010–2014 (n = 215 hospitalisations)


We used an administrative database to assess SJS/TEN and EM hospitalisations, and found that hospitalisations in patients with SJS/TEN were associated with significantly longer hospital stays and higher in-hospital mortality than hospitalisations in patients with associated diagnosis of EM, highlighting the need to accurately distinguish between these two clinical conditions. Drug classes responsible for the greatest proportion of adverse reactions in patients with SJS/TEN were antibiotics, uric acid metabolism drugs, and anticonvulsants. In-hospital mortality in SJS/TEN cases was significantly associated with liver disease, advanced age, and TEN diagnosis.

Most hospitalisations with associated diagnosis of SJS/TEN and EM occurred in females and older patients, a finding not totally consistent with several other studies, which found EM to be more frequent among males and younger patients [11, 29]. One the one hand, it is possible to hypothesise that this study predominantly assessed severe cases of EM, as most cases of this condition do not require hospitalisation (i.e. SJS/TEN episodes regardless of their severity were only compared with the most severe EM cases). However, it is also possible to infer that a substantial number of cases classified with a diagnosis of “EM” might actually consist of misclassified cases. In fact, the diagnosis of “drug-related EM” appears to be over-attributed [30]. A recent review found that, from 36 articles published from 2010 to 2016 and describing putative cases of drug-related EM, only 6 described cases compatible with probable/definite EM [30].

In our database analysis, HIV co-infection was significantly more common among SJS/TEN hospitalisations. Previous studies have found that HIV+ patients have a higher risk of developing SJS and TEN [22]. In fact, HIV+ patients are more likely to use some of the drugs most frequently implicated in SJS/TEN, such as antiretrovirals, sulfamethoxazole/trimethoprim, and antituberculosis drugs, and they often use them at higher doses [31, 32]. Secondly, these patients have a decreased number of skin CD4+ regulatory T cells and appear to have altered drug metabolism [31, 32]. HIV might also contribute to the pathogenesis and local cytotoxic mechanisms of SJS/TEN, as suggested by the presence of HIV antigens in the skin lesions of patients with these reactions [33].

The frequency of fatal cases in our series is consistent with other studies [34, 35]. Advanced age was identified as a risk factor for in-hospital mortality among SJS/TEN episodes; in fact, advanced age is an independent risk factor in the SCORTEN severity scale [36, 37]. Liver disease was also found to be associated with increased in-hospital mortality in our analysis, and while hepatic involvement in SCARs is associated with high mortality [38], chronic viral hepatitis has also been hypothesised by some authors to be a risk factor for TEN [23]. Impaired drug metabolism secondary to chronic liver disease could also enhance the risk of a fatal outcome.

The drug classes responsible for adverse reactions most frequently associated with SJS/TEN included antibiotics, antivirals, anticonvulsants, and uric acid metabolism drugs. Although we were not able to identify the specific culprit drugs within these groups, the latter include drugs that have frequently been previously described to be associated with SCARs, such as allopurinol, and lamotrigine [14, 15]. After adjusting for the number of drug packages sold, we also found a high rate of hospitalisations associated with adverse reactions to antineoplastic and immunosuppressive drugs, supporting some underlying immune deregulation associated with malignancies or autoimmune diseases [14]. While many cases of EM and SCARs have been reported following the use of several of these drugs, the underlying immunological mechanisms are still only partially identified [20, 39, 40]. Some antineoplastic drugs (e.g., EGFR tyrosine kinase inhibitors) interfere with keratinocyte proliferation, differentiation, and migration, and, thus, might facilitate the development of more severe SCARs. Similarly, radiotherapy could also increase the risk of EM and SCARs, possibly by inhibiting hepatic enzymes responsible for drug metabolism [41].

The use of an administrative database to study rare conditions such as SJS/TEN might have some advantages comparing to traditional case–control and cohort studies. Although these are the ideally preferred studies, case–control and cohort studies are usually resource-consuming and difficult to conduct, particularly on a nationwide and frequent basis [16, 42,43,44]. Additionally, administrative database studies may yield results consistent with these registry-based studies; for instance, the results described in our study concerning SJS/TEN demographic characteristics and mortality are similar with those of a recent study conducted in Italy [14]. On the other hand, in the lack of other registries, administrative database studies might feasibly complement pharmacovigilance studies and help to detect regional differences. The comprehensiveness of administrative databases regarding this condition is another important advantage—due to its severity, SJS/TEN is a condition requiring hospitalisation; additionally, the nationwide scope of the database allows for an overcome of possible biases related to the assessment of participants of a single region [16].

Nevertheless, although coding is standardised and frequently audited in Portugal, it should be noted that these databases might be incomplete or inaccurate [42]. A systematic review found that only 53–60% of ICD-9-CM code 695.1 reports consisted of validated cases of EM, SJS and TEN [45], while Davis et al. [18] found that, among inpatients, ICD-9-CM codes 695.13–695.15 correctly identified 50% of patients, and up to 57–92% when only patients hospitalised for three or more days were considered. While, in our study, we did not select patients according to their length of stay, only three patients had been hospitalised for less than 3 days (two of them died when hospitalised). Additionally, we identified drug hypersensitivity cases by using a combination of both ICD-9-CM diagnostic codes and E codes. According to Saff et al. [46], this combination identifies drug allergy patients more accurately than the use of a single code, but it underestimates the true incidence of drug allergic reactions. This algorithm lacks, however, validation regarding episodes with associated diagnosis of drug-related EM. In fact, not only there are several drug-related conditions which do not have a specific ICD-9-CM code (e.g. DRESS and AGEP), but also some heterogeneous drug-induced skin-eruptions may present as EM-like and, therefore, might be misclassified as EM [47, 48]. While these conditions do not have a specific ICD-10 code either [49], they are planned to have a specific ICD-11 code [50, 51]—in fact, with the development and adoption of ICD-11, the accuracy of administrative databases in the assessment of SCARs may improve, as a greater diversity of diagnosis procedures, drugs (and not only drug classes) and clinical entities have been ascribed specific codes—EJ00-EJ18 codes concern “adverse cutaneous reactions to medication” and include, among others, specific codes for DRESS, AGEP and fixed drug eruption). Additionally, it is paramount to ensure the validity of the hospital discharge codes for EM [18], as well as to educate clinicians on the differential diagnoses of drug-related skin disorders [30].

Another major limitation concerns the impossibility of identifying the specific drugs associated with each episode. Thus, it is only possible to speculate about the identity of the culprit drugs. For instance, it is highly probable that most hypersensitivity reactions to uric acid metabolism drugs were due to allopurinol, while most cutaneous reactions to antivirals were probably associated with antiretroviral drugs, since these reactions mostly occurred in HIV+ patients. We also lack information on the specific clinical presentation of each episode, criteria used by the physicians to deem a specific drug class responsible for the corresponding reactions, co-occurrence of herpes reactivation and patient ethnicity or birthplace. Another possible limitation concerns the indirect method used to identify hospital readmissions (based on inpatients’ gender, birth date and residence). Although this method does not identify distinct patients with complete certainty, we confirmed that episodes identified as readmissions and the respective “first admissions” had a similar set of associated diagnoses and, for the cases occurred in the same hospital, an equal hospital-specific inpatient identifier number.


In this administrative database-analysis, SJS and TEN were associated with higher in-hospital mortality and longer hospital stays than other drug-related mucocutaneous conditions. In hospitalisations with a diagnosis of drug-related SJS or TEN, an increased risk of in-hospital mortality was associated with advanced age, with a TEN diagnosis, and liver disease. Our findings provide an epidemiological characterisation of SJS/TEN hospitalisations, as well as an identification of factors significantly associated with higher in-hospital mortality. This epidemiological knowledge might prove useful for performing an earlier diagnosis of SJS/TEN, allowing for an earlier start of the most adequate therapy; additionally, identifying factors associated with higher fatality will be essential for defining the most appropriate measures to prevent fatal outcomes. These results suggest that administrative databases are useful in the assessment of SJS/TEN drug-related hospitalisations in a nationwide basis, allowing for epidemiological studies to be conducted in a frequent- and low-resource-consuming basis. While this may be particularly advantageous for obtaining knowledge on SJS/TEN and for healthcare planning, further studies on this methodological approach are needed.



acute generalised exanthematous pustulosis


area under the receiver operating characteristics curve


confidence interval


drug reactions with eosinophilia and systemic symptoms


erythema multiforme


odds ratio


severe cutaneous adverse reactions


Stevens–Johnson syndrome


toxic epidermal necrolysis


  1. 1.

    Wheatley LM, Plaut M, Schwaninger JM, Banerji A, Castells M, Finkelman FD, et al. Report from the National Institute of Allergy and Infectious Diseases workshop on drug allergy. J Allergy Clin Immunol. 2015;136(262–71):e2.

    Google Scholar 

  2. 2.

    Pavlos R, Mallal S, Ostrov D, Pompeu Y, Phillips E. Fever, rash, and systemic symptoms: understanding the role of virus and HLA in severe cutaneous drug allergy. J Allergy Clin Immunol Pract. 2014;2:21–33.

    Article  PubMed  PubMed Central  Google Scholar 

  3. 3.

    Swanson L, Colven RM. Approach to the patient with a suspected cutaneous adverse drug reaction. Med Clin N Am. 2015;99:1337–48.

    Article  PubMed  Google Scholar 

  4. 4.

    Bouvresse S, Valeyrie-Allanore L, Ortonne N, Konstantinou MP, Kardaun SH, Bagot M, et al. Toxic epidermal necrolysis, DRESS, AGEP: do overlap cases exist? Orphanet J Rare Dis. 2012;7:72.

    Article  PubMed  PubMed Central  Google Scholar 

  5. 5.

    Leclair MA, Maynard B, St-Pierre C. Acute generalized exanthematous pustulosis with severe organ dysfunction. CMAJ. 2009;181:393–6.

    Article  PubMed  PubMed Central  Google Scholar 

  6. 6.

    Lin YF, Yang CH, Sindy H, Lin JY, Rosaline Hui CY, Tsai YC, et al. Severe cutaneous adverse reactions related to systemic antibiotics. Clin Infect Dis. 2014;58:1377–85.

    CAS  Article  PubMed  Google Scholar 

  7. 7.

    Borchers AT, Lee JL, Naguwa SM, Cheema GS, Gershwin ME. Stevens–Johnson syndrome and toxic epidermal necrolysis. Autoimmun Rev. 2008;7:598–605.

    CAS  Article  PubMed  Google Scholar 

  8. 8.

    Harr T, French LE. Toxic epidermal necrolysis and Stevens–Johnson syndrome. Orphanet J Rare Dis. 2010;5:39.

    Article  PubMed  PubMed Central  Google Scholar 

  9. 9.

    Bachot N, Roujeau JC. Differential diagnosis of severe cutaneous drug eruptions. Am J Clin Dermatol. 2003;4:561–72.

    Article  PubMed  Google Scholar 

  10. 10.

    Roujeau JC. Stevens–Johnson syndrome and toxic epidermal necrolysis are severity variants of the same disease which differs from erythema multiforme. J Dermatol. 1997;24:726–9.

    CAS  Article  PubMed  Google Scholar 

  11. 11.

    Auquier-Dunant A, Mockenhaupt M, Naldi L, Correia O, Schroder W, Roujeau JC, et al. Correlations between clinical patterns and causes of erythema multiforme majus, Stevens–Johnson syndrome, and toxic epidermal necrolysis: results of an international prospective study. Arch Dermatol. 2002;138:1019–24.

    Article  PubMed  Google Scholar 

  12. 12.

    Marzano AV, Frezzolini A, Caproni M, Parodi A, Fanoni D, Quaglino P, et al. Immunohistochemical expression of apoptotic markers in drug-inducederythema multiforme, Stevens–Johnson syndrome and toxic epidermal necrolysis. Int J Immunopathol Pharmacol. 2007;20:557–66.

    CAS  Article  PubMed  Google Scholar 

  13. 13.

    Watanabe R, Watanabe H, Sotozono C, Kokaze A, Iijima M. Critical factors differentiating erythema multiforme majus from Stevens–Johnson syndrome (SJS)/toxic epidermal necrolysis (TEN). Eur J Dermatol. 2011;21:889–94.

    PubMed  Google Scholar 

  14. 14.

    Diphoorn J, Cazzaniga S, Gamba C, Schroeder J, Citterio A, Rivolta AL, et al. Incidence, causative factors and mortality rates of Stevens–Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) in northern Italy: data from the REACT registry. Pharmacoepidemiol Drug Saf. 2016;25:196–203.

    CAS  Article  PubMed  Google Scholar 

  15. 15.

    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;333:1600–7.

    CAS  Article  PubMed  Google Scholar 

  16. 16.

    Chan EW, Liu KQ, Chui CS, Sing CW, Wong LY, Wong IC. Adverse drug reactions—examples of detection of rare events using databases. Br J Clin Pharmacol. 2015;80:855–61.

    Article  PubMed  PubMed Central  Google Scholar 

  17. 17.

    Mateus C. Portugal. Results of 25 years of experience with DRGs. In: Busse R, World Health Organization, Regional Office for Europe, EuroDRG (Project), , editors. Diagnosis-related groups in Europe: moving towards transparency, efficiency and quality in hospitals. Berkshire: McGraw-Hill Open University Press; 2011.

  18. 18.

    Davis RL, Gallagher MA, Asgari MM, Eide MJ, Margolis DJ, Macy E, et al. Identification of Stevens–Johnson syndrome and toxic epidermal necrolysis in electronic health record databases. Pharmacoepidemiol Drug Saf. 2015;24:684–92.

    Article  PubMed  Google Scholar 

  19. 19.

    Instituto Nacional de Estatística (Portuguese Institute of Statistics). Accessed 26 Mar 2016.

  20. 20.

    Sousa-Pinto B, Correia C, Gomes L, Gil-Mata S, Araújo L, Correia O, et al. HLA and delayed drug-induced hypersensitivity. Int Arch Allergy Immunol. 2016;170:163–79.

    CAS  Article  PubMed  Google Scholar 

  21. 21.

    Ramasamy SN, Korb-Wells CS, Kannangara DR, Smith MW, Wang N, Roberts DM, et al. Allopurinol hypersensitivity: a systematic review of all published cases, 1950–2012. Drug Saf. 2013;36:953–80.

    CAS  Article  PubMed  Google Scholar 

  22. 22.

    Rzany B, Mockenhaupt M, Stocker U, Hamouda O, Schopf E. Incidence of Stevens–Johnson syndrome and toxic epidermal necrolysis in patients with the acquired immunodeficiency syndrome in Germany. Arch Dermatol. 1993;129:1059.

    CAS  Article  PubMed  Google Scholar 

  23. 23.

    Gao W, Tiwary A, McGann A, Rajmane R. Chronic infection with hepatitis C as a risk factor for toxic epidermal necrolysis in a cirrhotic patient with streptococcus pneumoniae bacteremia. Chest. 2011;140(4_MeetingAbstracts):124A.

    Article  Google Scholar 

  24. 24.

    INFARMED—Autoridade Nacional do Medicamento e Produtos de Saúde I.P. Estatística do Medicamento 2010 (Medicines Statistic 2010). Accessed 26 Mar 2016.

  25. 25.

    INFARMED—Autoridade Nacional do Medicamento e Produtos de Saúde I.P. Estatística do Medicamento 2011 (Medicines Statistic 2011). Accessed 26 Mar 2016.

  26. 26.

    INFARMED—Autoridade Nacional do Medicamento e Produtos de Saúde I.P. Estatística do Medicamento 2012 (Medicines Statistic 2012). Accessed 26 Mar 2016.

  27. 27.

    INFARMED—Autoridade Nacional do Medicamento e Produtos de Saúde I.P. Estatística do Medicamento 2013 (Medicines Statistic 2013). Accessed 26 Mar 2016.

  28. 28.

    INFARMED—Autoridade Nacional do Medicamento e Produtos de Saúde I.P. Estatística do Medicamento e Produtos de Saúde 2014 (Medicine and Healthcare Products Statistics 2014). Accessed 26 Mar 2016.

  29. 29.

    Samim F, Auluck A, Zed C, Williams PM. Erythema multiforme: a review of epidemiology, pathogenesis, clinical features, and treatment. Dent Clin N Am. 2013;57:583–96.

    Article  PubMed  Google Scholar 

  30. 30.

    Roujeau JC. Re-evaluation of ‘drug-induced’ erythema multiforme in medical literature. Br J Dermatol. 2016;175:650–1.

    Article  PubMed  Google Scholar 

  31. 31.

    Rotunda A, Hirsch RJ, Scheinfeld N, Weinberg JM. Severe cutaneous reactions associated with the use of human immunodeficiency virus medications. Acta Derm Venereol. 2003;83:1–9.

    Article  PubMed  Google Scholar 

  32. 32.

    Yang C, Mosam A, Mankahla A, Dlova N, Saavedra A. HIV infection predisposes skin to toxic epidermal necrolysis via depletion of skin-directed CD4(+) T cells. J Am Acad Dermatol. 2014;70:1096–102.

    CAS  Article  PubMed  Google Scholar 

  33. 33.

    Correia O, Delgado L, Santos C, Miranda AM. HIV-1 in blister fluid of a patient with toxic epidermal necrolysis and AIDS. Lancet. 1994;344:1432–3.

    CAS  Article  PubMed  Google Scholar 

  34. 34.

    Schneck J, Fagot JP, Sekula P, Sassolas B, Roujeau JC, Mockenhaupt M. Effects of treatments on the mortality of Stevens–Johnson syndrome and toxic epidermal necrolysis: a retrospective study on patients included in the prospective EuroSCAR Study. J Am Acad Dermatol. 2008;58:33–40.

    Article  PubMed  Google Scholar 

  35. 35.

    Finkelstein Y, Macdonald EM, Li P, Hutson JR, Juurlink DN. Recurrence and mortality following severe cutaneous adverse reactions. JAMA. 2014;311:2231–2.

    CAS  Article  PubMed  Google Scholar 

  36. 36.

    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;32:237–45.

    Article  PubMed  Google Scholar 

  37. 37.

    Heng YK, Lim YL. Cutaneous adverse drug reactions in the elderly. Curr Opin Allergy Clin Immunol. 2015;15:300–7.

    CAS  Article  PubMed  Google Scholar 

  38. 38.

    Devarbhavi H, Raj S, Aradya VH, Rangegowda VT, Veeranna GP, Singh R, et al. Drug-induced liver injury Associated with Stevens-Johnson syndrome/toxic epidermal necrolysis: patient characteristics, causes and outcome in 36 Cases. Hepatology. 2016;63:993–9.

    CAS  Article  PubMed  Google Scholar 

  39. 39.

    Payne AS, James WD, Weiss RB. Dermatologic toxicity of chemotherapeutic agents. Semin Oncol. 2006;33:86–97.

    CAS  Article  PubMed  Google Scholar 

  40. 40.

    Sanborn RE, Sauer DA. Cutaneous reactions to chemotherapy: commonly seen, less described, little understood. Dermatol Clin. 2008;26:103–19.

    CAS  Article  PubMed  Google Scholar 

  41. 41.

    Vern-Gross TZ, Kowal-Vern A. Erythema multiforme, Stevens Johnson syndrome, and toxic epidermal necrolysis syndrome in patients undergoing radiation therapy: a literature review. Am J Clin Oncol. 2014;37:506–13.

    Article  PubMed  Google Scholar 

  42. 42.

    Miguel A, Bernardo M, Freitas A, Lopes F, Azevedo L, Pereira AC. Detection of adverse drug reactions using hospital databases—a nationwide study in Portugal. Pharmacoepidemiol Drug Saf. 2013;22:907–13.

    Article  PubMed  Google Scholar 

  43. 43.

    Iezzoni LI. Assessing quality using administrative data. Ann Intern Med. 1997;127:666–74.

    CAS  Article  PubMed  Google Scholar 

  44. 44.

    Hsu DY, Brieva J, Silverberg NB, Paller AS, Silverberg JI. Pediatric Stevens–Johnson syndrome and toxic epidermal necrolysis in the United States. J Am Acad Dermatol. 2017;76(811–7):e4.

    Google Scholar 

  45. 45.

    Schneider G, Kachroo S, Jones N, Crean S, Rotella P, Avetisyan R, et al. A systematic review of validated methods for identifying erythema multiforme major/minor/not otherwise specified, Stevens–Johnson Syndrome, or toxic epidermal necrolysis using administrative and claims data. Pharmacoepidemiol Drug Saf. 2012;21(Suppl 1):236–9.

    Article  PubMed  Google Scholar 

  46. 46.

    Saff RR, Camargo CA, Jr, Clark S, Rudders SA, Long AA, Banerji A. Utility of ICD-9-CM codes for identification of allergic drug reactions. J Allergy Clin Immunol Pract. 2016;4:114–9.

    Article  PubMed  Google Scholar 

  47. 47.

    Nomura H, Takahashi H, Suzuki S, Kurihara Y, Chubachi S, Kawada I, et al. Unexpected recalcitrant course of drug-induced erythema multiforme-like eruption and interstitial pneumonia sequentially occurring after nivolumab therapy. J Dermatol. 2017;44:818–21.

    CAS  Article  PubMed  Google Scholar 

  48. 48.

    Laurinaviciene R, Sandholdt LH, Bygum A. Drug-induced cutaneous lupus erythematosus: 88 new cases. Eur J Dermatol. 2017;27:28–33.

    PubMed  Google Scholar 

  49. 49.

    World Health Organization. ICD-10 Version 2016. Accessed 29 Aug 2016.

  50. 50.

    Tanno LK, Calderon MA, Li J, Casale T, Demoly P, Joint Allergy Academies. Updating allergy and/or hypersensitivity diagnostic procedures in the WHO ICD-11 revision. J Allergy Clin Immunol Pract. 2016;4:650–7.

    Article  PubMed  Google Scholar 

  51. 51.

    World Health Organization. ICD-11 Beta Draft (Last uptadte 2017 November 25). Acessed 26 Nov 2017.

Download references

Authors’ contributions

BSP participated in study design, data analysis and manuscript writing; LA participated in study design and critical revision of the manuscript; AF participated in study design and data analysis; OC participated in study design and critical revision of the manuscript; LD participated in study design, manuscript writing and critical revision of the manuscript. All authors read and approved the final manuscript.


The authors wish to thank the Portuguese Ministry of Health for providing access to the hospitalisation data managed by the Portuguese Central Health System Administration (Administração Central do Sistema de Saúde). The authors also wish to thank Cláudia Correia, Lídia Gomes and Sara Gil-Mata for helping retrieving the analysed data. The authors wish to thank the project “NORTE-01-0145-FEDER-000016” (NanoSTIMA), financed by the North Portugal Regional Operational Programme (NORTE 2020), under the PORTUGAL 2020 Partnership Agreement, and through the European Regional Development Fund (ERDF).

Competing interests

The authors declare that they have no competing interests.

Availability of data and materials

The data that support the findings of this study are available from Administração Central do Sistema de Saúde (Portuguese Central Administration of the Health System) but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Administração Central do Sistema de Saúde (Portuguese Central Administration of the Health System).

Consent for publication

Not applicable.

Ethics approval and consent to participate

Not applicable. Data had been previously been anonymised.


Not applicable.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Author information



Corresponding author

Correspondence to Bernardo Sousa-Pinto.

Additional files

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Sousa-Pinto, B., Araújo, L., Freitas, A. et al. Stevens–Johnson syndrome/toxic epidermal necrolysis and erythema multiforme drug-related hospitalisations in a national administrative database. Clin Transl Allergy 8, 2 (2018).

Download citation


  • Epidemiology
  • Erythema multiforme
  • Drug allergy
  • Stevens–Johnson syndrome
  • Toxic epidermal necrolysis