Rash from Antiepileptic Drugs: Influence by Gender, Age, and Learning Disability


Address correspondence and reprint requests to Eylert Brodtkorb at Department of Neurology and Clinical Neurophysiology, St. Olav's University Hospital, 7006 Trondheim, Norway. E-mail: eylert.brodtkorb@ntnu.no


Summary: Purpose: Cutaneous adverse reactions from antiepileptic drugs (AEDs) are common, but have received little scientific attention from a clinical point of view. We wanted to study the incidence of skin reactions of current AEDs and to explore their relation to clinical parameters such as gender, age, and learning disability.

Methods: Consecutive patients with epilepsy were studied retrospectively. A detailed survey of medical records concerning all treatment with AEDs was performed.

Results: A total of 663 patients were included with altogether 2,567 exposures to 15 different AEDs. Skin reactions were found in 14% of the patients and in 5% of the exposures. Ninety-seven percent of the reactions occurred to either carbamazepine (CBZ, 11%), phenytoin (PHT, 8%), lamotrigine (LTG, 8%), oxcarbazepine (8%), or phenobarbital (2%). Skin reactions developed significantly more often in females than in males (19% vs. 8%), and significantly less often in patients with learning disability than in other patients (7% vs. 16%). These differences were significant for CBZ, PHT, and LTG when analyzed separately. Females displayed higher rash frequency during the reproductive years, while men experienced less frequent rash in the same phase of life.

Conclusions: Fertile females have a higher risk for skin reactions compared to males, probably due to hormonal factors. Patients with learning disability appeared to have a lower risk than other patients in this study. Hygiene factors may possibly be underlying.

Skin reactions belong to the idiosyncratic adverse drug reactions. Idiosyncratic toxicity is unpredictable from the known pharmacology of the drug and refers to a susceptibility unique to an individual. It shows no simple dose-dependency (Park et al., 2001). Skin rash is a common side effect of antiepileptic drugs (AEDs) and a major cause of treatment discontinuation. Such reactions range with increasing severity from the common and mild maculopapular rashes, to the hypersensitivity reactions with fever and internal organ involvement, and to Stevens–Johnson syndrome and toxic epidermal necrolysis.

The pathogenesis of skin reactions appears to be multifactorial and has in many cases been explained by the hapten hypothesis of drug hypersensitivity, which implies both metabolic and immunological mechanisms. An imbalance between metabolic bioactivation and detoxification of the drug may lead to an accumulation of reactive metabolites, which may bind irreversibly to endogenous proteins (Leeder, 1998). T-cell lymphocyte clones may react to these drug-modified proteins or to the parental drug itself and cause delayed immune responses in the skin, whereas immediate IgE responses are responsible for urticaria, angioedema, and anaphylaxis (Roujeau, 2006).

Among the traditional AEDs, the aromatic compounds phenytoin (PHT) and carbamazepine (CBZ) have been associated with relatively high incidences of cutaneous reactions in up to at least 10% of patients (Rapp et al., 1983; Brodie et al., 1995; Chadwick, 1999). Some of the newer drugs also frequently cause this problem, particularly lamotrigine (LTG) (Wong et al., 1999; Hirsch et al., 2006), oxcarbazepine (OCBZ) (Friis et al., 1993; Kalis and Huff, 2001), and felbamate (Galindo et al., 2002). Skin reactions may considerably limit the use of these AEDs. Serious reactions are rare, but may be life threatening (Mockenhaupt et al., 2005). Nevertheless, this problem has generally received little scientific attention from a clinical point of view. Many other factors than the drug itself may modify the susceptibility to this adverse effect and mechanisms may differ. Pharmacogenetic variations in drug biotransformation may be crucial (Hung et al., 2006), but gender, hormones, and age may also influence the predisposition to these reactions. Some clinical variables, like a high starting dose and a rapid dose escalation, have been identified as risk factors, particularly for LTG (Wong et al., 1999). Several studies have indicated that LTG-induced rashes are more frequent in children than in adults, at least in the case of potentially life-threatening reactions (Matsuo, 1999; Hirsch et al., 2006). In one study, females were at a higher risk of developing LTG-induced rash than males (Wong et al., 1999).

The aim of this study was to compare the incidence of drug rashes within the current armamentarium of AEDs consisting of both traditional and newer drugs, and to investigate whether clinical factors such as age, gender, epilepsy syndrome, and learning disability influence AED-related skin reactions.


We systematically reviewed the medical records of consecutive adult patients with epilepsy for documentation of cutaneous adverse reactions from AEDs. The study was carried out in three specialist outpatient clinics in Middle Norway served from neurologists from Trondheim University Hospital. A cutaneous adverse reaction was defined as a diffuse rash, which had no other obvious reason than a drug effect and resulted in contacting a physician. Dates of onset and termination of any AED treatment were recorded and the reasons for termination were assessed. Circumscribed eruptions were not included. As initial symptoms of hypersensitivity reactions most frequently occur up to 8 weeks after starting the drug (Knowles et al., 1999), treatments lasting less than 3 months and stopped for any other reason than a rash were not included as an exposure. Date of occurrence of rash and its clinical description, when available, were documented. Telephone interviews were performed if written documentation was considered insufficient. Age, gender, epilepsy syndrome, and presence and degree of learning disability were also recorded.

Groups were compared using Fisher's exact test for categorical data, linear-by-linear test for trend for ordinal data, and the 2-sample t-test for continuous data. Ninety-five percent confidence intervals (CI) for odds ratios (ORs) and corresponding p-values were computed using exact conditional methods. Some analyses were done separately for each gender because logistic regression analyses (not included here) showed interactions between gender and several other variables. The ORs for learning disability were also calculated stratifying on age and gender groups. The mid-p approach was applied for tests and confidence intervals for categorical data, to avoid conservatism in small datasets (Hirji, 2006). Two-sided p-values <0.05 were considered significant. Analyses were carried out in SPSS 13 (Chicago, IL, U.S.A.) and StatXact 7 (Cytel Statistical Software, Cambridge, MA, U.S.A.).


A total of 663 patients were included in the study, of which 320 (48%) were males and 343 (52%) were females. Partial epilepsy was diagnosed in 433 patients (65%), 154 (23%) had generalized epilepsy, and 74 (11%) had unclassified epilepsy. Learning disability was present in 151 patients (23%) and was classified according to the ICD-10 guide for mental retardation as mild in 56 (8%) and as severe in 95 (14%).

The 663 patients had altogether 2,567 exposures to 15 different AEDs. A skin reaction to at least one AED was found in 93 patients (14%) and in 118 exposures (5%). The treatment was discontinued in all patients, except in two. As much as 97% of the reactions occurred either to CBZ, PHT, LTG, OCBZ, or phenobarbital. The frequency of rash from each AED is shown in Table 1. A clinical description was available for 111 events (Table 2). Most patients had common morbilliform rashes; in 13 febrile reactions were reported, two with multiorgan affection including elevated liver enzymes. However, internal organ affection and eosinophilia was not routinely searched for. In five subjects Stevens–Johnson syndrome was reported; toxic epidermal necrolysis did not occur. Most rashes (86%) occurred less than three months after beginning medication. The median time from exposure to development of symptoms was two weeks (interquartile range 2–5.25 weeks). Clinical and demographic patient characteristics as well as comedications are listed in Table 3 for the four AEDs with the highest frequencies of rash. Epilepsy syndrome classification had no impact on rash frequency (results not shown).

Table 1. Frequency of skin reactions in 2,567 exposures to antiepileptic drugs
DrugNumber exposuresFrequency of rash (%)
Carbamazepine498 54 (11)
Phenytoin22919 (8)
Lamotrigine35929 (8)
Oxcarbazepine114 9 (8)
Phenobarbital211 4 (2)
Valproate391 1 (0)
Levetiracetam155 1 (1)
Vigabatrin144 0
Topiramate141 0
Clonazepam 68 1(1)
Clobazam 76 0
Gabapentin 73 0
Primidone 59 0
Etosuksimide 38 0
Felbamate 11 0
Total2,567  118(5) 
Table 2. Clinical characteristics and associated features of skin reactions from antiepileptic drugs as reported in medical records
CharacteristicNumber of events
Skin reaction
 Urticaria 2
 Erythema nodosum 1
 Stevens–Johnson Syndrome 5
Associated features
 Multiorgan affection 2
Table 3. Clinical and demographic characteristics of patients exposed to antiepileptic drugs with high risks for skin reactions
 GenderMean age (±SD)Learning disabilityCotreatment
  1. Some patients are not included in the calculation of mean age due to missing data.

  2. CBZ, carbamazepine; PHT, phenytoin: LTG, lamotrigine; OCBZ, oxcarbazepine; VPA, valproate; F, females; M, males; SD, standard deviation.

  3. ap < 0.001, bp < 0.01, cp < 0.05

CBZ no rash21822624 ± 16a120324 5773 24
 n = 427
CBZ rash 35 1934 ± 18a  450 610  3
(14%)b(8%)bn = 53(3%)a(13%)a (10%)(13%) (6%)
PHT no rash10210818 ± 15c7513526 39 17
 n = 187 
PHT rash 15  427 ± 16c  2 17 3 01 5
(13%)b(4%)bn = 18(3%)c(11%)c(10%) (10%)(23%)c
LTG no rash18714332 ± 1510222810526 42 111
 n = 321 
LTG rash 22  732 ± 16 326 9 0 1 11
(11%)c(5%)cn = 27(3%)c(10%)c(8%) (2%)(9%)
OCBZ no rash 505532 ± 142283 6 812   48
 n = 100 
OCBZ rash  7  237 ± 15 1 8001   4
(12%)(4%)n = 8(4%)(9%) (8%) (8%)

Among the 93 patients experiencing a rash to any AED, 27 (8%) were males and 66 (19%) females. Females were overall more than twice as likely to develop a rash as were males (OR = 2.6, CI 1.6–4.2, p < 0.001). For the five drugs analyzed separately the results were: CBZ OR = 1.9, CI 1.1–3.5, p = 0.026; PHT OR = 3.9, CI 1.3–14.2, p = 0.011; LTG OR = 2.4, CI 1.0–6.2, p = 0.040; OCBZ OR = 3.8, CI 0.80–28, p = 0.13 (Table 3). The co-medication did not appear to influence the occurrence of rash, other than an increased rate for PHT in patients on VPA pretreatment.

The risk of AED-related rash increased with age. The mean age of patients experiencing CBZ and PHT rash was higher (34 and 27 years, respectively) than that of nonrash patients (24 and 18 years, respectively). When subdividing the population into three age groups (<16 years, 16–49 years, >50 years), both PHT and CBZ displayed higher incidences of rash in the two highest age groups (Table 4). For LTG, there was a nonsignificant tendency to the opposite, with more rashes in children (15%) than in adults (7%). In children and individuals above 50 years there were no gender difference in overall rash rates, but between the age of 16 and 50 this difference was very pronounced, 5% in males and 22% in females (OR = 4.8, CI 2.4–10, p < 0.001).

Table 4. Skin reactions in relation to age (% of exposures)
DrugsAge groupP
<16 years16–49 years>50 years
Both genders
 Carbamazepine10/172 (6%) 33/270 (12%)10/53 (19%)0.012
 Phenytoin 4/98 (4%)11/100 (11%) 3/11 (27%)0.022
 Lamotrigine  6/40 (15%)18/272 (7%) 3/45 (7%)0.177
 Oxcarbazepine  0/7 (0%) 6/89 (7%) 2/16 (13%)0.547
 All drugs19/143 (13%)55/388 (14%)14/66 (21%)0.287
 Carbamazepine 3/86 (3%)27/143 (19%) 4/27 (15%)0.001
 Phenytoin 2/49 (4%) 10/46 (22%)  2/9 (22%)0.016
 Lamotrigine  3/21 (14%)15/153 (10%)2/29 (7%)0.696
 Oxcarbazepine  0/2 (0%)  5/44 (11%)1/10 (10%) 0.833
 All drugs 10/75 (13%)45/205 (22%) 8/37 (22%)0.259
 Carbamazepine 7/86 (8%)6/127 (5%) 6/26 (23%)0.011
 Phenytoin 2/49 (4%) 1/54 (2%)  1/2 (50%)0.033
 Lamotrigine  3/19 (16%)3/119 (3%)1/16 (6%)0.026
 Oxcarbazepine  0/5 (0%) 1/45 (2%)  1/6 (17%)0.269
 All drugs  9/68 (13%)10/183 (5%)  6/29 (21%)0.008

Among 151 patients with learning disability, 11 (7%) developed rash to any AED, compared to 82 out of 512 patients (16%) without learning disability (OR = 0.41, CI 0.20–0.78, p = 0.004). Patients with severe learning disability exhibited less skin reactions (5 out of 95, 5%) than patients with mild learning disability (6 out of 56, 11%, test for trend p = 0.004). The difference in rash frequency between individuals with learning disability and other patients was significant for CBZ, PHT, and LTG when each drug was analyzed separately (Table 3). There was no significant difference in gender distribution between these groups.

Stratifying on the six age and gender groups for each of the drugs, the common ORs for learning disability were: CBZ OR = 0.30, CI 0.087–0.80, p = 0.014; PHT OR = 0.35, CI 0.50–1.5, p = 0.16; OCBZ OR = 0.72, CI 0.028–5.8, p = 0.8; LTG OR = 0.27, CI 0.062–0.85, p = 0.018. Hence, adjusting for age and gender group, we observed a lower frequency of rash in patients with learning disability for all these four drugs, statistically significant for CBZ and LTG.

Rashes from more than one AED occurred in 18 patients (in 3% of all patients and in 19% of those with rash); 14 (78%) were females; only one had learning disability. Four patients experienced reactions from more than two drugs.


As in previous studies, AEDs with aromatic ring structures had the highest incidences of cutaneous adverse reactions (Mattson et al., 1985; Hyson and Sadler, 1997; Matsuo, 1999). The risk was most pronounced for CBZ (Table 1). The somewhat higher frequency of OCBZ-related rash in the present study (8%), compared to the findings derived from controlled clinical studies (3%) (Schmidt and Sachdeo, 2000), may be due to the fact that OCBZ often had been tried in patients with a previous CBZ rash. We found several factors to be associated with the development of rash. In general, increasing age and female gender appeared to increase the risk of rash, whereas learning disability appeared to be protective.


It is already recognized that drug-induced skin reactions in general are more frequent in women (Tran et al., 1998). In the present study the difference was striking (19% vs. 8%, p < 0.001), and present across all AEDs commonly associated with rash. This difference has previously been reported for LTG (Wong et al., 1999), but has to our knowledge not been demonstrated for other AEDs. A rash is an immunologically mediated process, which is influenced by steroid hormones. Female and male sex hormones influence T-cell populations, the production of specific antibodies and proinflammatory mediators (Grossman, 1984). In general, female sex steroids enhance immune responses in both physiological and pathological states, whereas androgens dampen inflammatory responses even more than endogenous glucocorticoids do (Talal, 1987; Da Silva, 1999; Osman, 2003). In our study, there were no gender differences in rash rates in children and in patients above the age of 50, whereas females in their reproductive period had a strongly increased risk compared to males in the same age group (22% vs. 5%, p < 0.001). A similar shift to female predominance during puberty is reported for asthma, atopic conditions, and hay fever (Shamssain and Shamsian, 1999; Osman, 2003). In a recent study, no significant gender difference was demonstrated for LTG rashes (Hirsch et al., 2006), a finding that may be explained by a relatively higher number of children in that study population.


In general, it has been assumed that elderly patients experience skin reactions from drug therapy at higher rates (Sullivan and Shear, 2002). Accordingly, the total amount of CBZ, PHT, and OCBZ-rashes increased with age, a finding that may seem to correlate with falling testosterone levels in elderly men (Table 4). Decreasing liver volume, blood flow, and metabolism, as well as other biological factors may contribute to this effect (Greenwood, 2000). In a randomized, controlled multicenter trial the rash rate for CBZ was surprisingly high, 25%, causing withdrawal in 19% of patients with new onset epilepsy above 65 years, (Brodie et al., 1999). Females had a significantly higher frequency of CBZ- and PHT-rash during fertile age (Table 4). Our findings indicate that the variability of skin reactions with age to a great extent is hormonal dependent and gender specific. However, we found no significant association between LTG-rash and age, but a tendency toward higher rash frequency in young patients. Some previous studies (Mackay et al., 1997; Hirsch et al., 2006), but not all (Wong et al., 1999), have reported an increased risk in children, particularly for severe skin reactions (Guberman et al., 1999). In the above mentioned multicenter trial in elderly patients, rash occurred in 9% from LTG, causing withdrawal only in 3%.

Learning disability

Overall, patients with learning disability had a lower rash frequency than others. This phenomenon was also present across all prevailing drugs (Table 3) and was supported by analysis stratified on age and gender groups. We do not believe that this finding is due to communication difficulties or a diminished ability to perceive or identify a rash. These patients are usually well looked after and their access to habilitation clinics and specialist health services is generally good in the study areas. Their medical records were as complete as for other patients and the reasons for AED terminations could be equally well assessed. The majority of patients with cognitive deficits had severe learning disability. In this group, the frequency of rash was particularly low. These people need assistance in all aspects of daily living and are surrounded by caregivers at all times. The threshold for contacting a physician is generally low. It is unlikely that maculopapulous exanthemas should develop unnoticed in patients starting on a new drug. Thus, the reason for the low tendency to skin reactions is obscure, but it may possibly be related to the so-called “hygiene hypothesis” (Strachan, 1989; Vercelli, 2006). This is a much-debated topic in immunology and preventive medicine, which may explain the increased prevalence of allergic disease in the Western society. Less childhood infections and improved personal cleanliness may be a two-edged sword. Individuals with learning disability may possibly be more exposed to unhygienic contacts and various environmental contaminants and hence be less liable to allergies.


VPA comedication is an established risk factor for LTG-rash due to its inhibiting effect on uridine diphosphate glucuronyltransferase, but was not identified as a risk factor in our study, probably due to careful LTG dosing in accordance with current guidelines. Titration procedures could not be included in this retrospective survey. To our knowledge, no earlier study has reported VPA co-medication as a risk factor for PHT-induced rash (Table 3). VPA increases the unbound concentration of PHT, by displacing PHT from protein-binding sites and may cause accumulation of reactive PHT metabolites through its enzyme inhibiting effect (Cloyd, 1991; Maggs et al., 2000; Anderson, 2002). However, the number of patients with PHT who received VPA pretreatment was small (Table 3).

Reliability of the study

The best estimates of rash rates are obtained from prospective clinical trials, which permit careful follow-up and detailed descriptions of the development of all emerging adverse events. Our study has obvious weaknesses due to its retrospective character and the small number of patients in some of the subgroups. Nevertheless, it allows comparisons among currently used drugs. Other diffuse skin symptoms, due to infections or other immunological reactions (Knowles et al., 1999) may have occurred in some patients. We consider the problem of over-diagnosing as small since rashes were very rare with known low-risk drugs. Medical records may have been incomplete for some patients, but we believe that the Norwegian health care system provides a suitable background for this kind of investigation. The study areas have relatively stable and predominantly rural populations. It was carried out in neurology clinics served by the same specialists representing the only neurological services available in these areas, except for one single private specialist practice. In Norway, epilepsy is traditionally treated and followed up by hospital specialists and detailed medical charts were available for most patients back to seizure onset, or could be collected from other hospitals when needed. The reliability of the findings is supported by similar rash frequencies in randomized controlled trials and other surveys, even for the older drugs (Hyson and Sadler, 1997).


In this study, the incidence of skin reactions was somewhat higher for CBZ than for PHT, LTG, and OCBZ. VPA, levetiracetam, vigabatrin, and topiramate are rarely associated with skin rashes. Various endogenous and environmental factors may influence the propensity to these reactions. Females seem to be at higher risk than males, particularly in fertile age. Hormonal effects are likely. Surprisingly, individuals with learning disability appeared to have a low risk. We hypothesize that behavioral factors related to hygiene and environmental contaminants may reduce the tendency to allergic reactions in these patients.