Mortality of Epilepsy in Developing Countries


  • The commission wishes to acknowledge the support of UCB Pharma for logistical support in the organization of this workshop.

Address correspondence and reprint requests to Dr. A. Carpio at School of Medicine and Research Institute, University of Cuenca, Ecuador. E-mail:


Summary:  During the last two decades, there has been a renewed interest in studying epidemiology of epilepsy in developing countries. While there are data on prevalence of epilepsy from many developing countries, there is very little information on the mortality of epilepsy in these same populations. This is because incidence studies of epilepsy are difficult to perform, death certificates are unreliable and often unavailable, and the cause of death is difficult to determine. We report on several studies of mortality in epilepsy in developing countries: Ecuador; the Parsi community of Bombay; a semiurban community in Vasai, India; Mali; Martinique; and Africa. Overall, these studies in general illustrate excess mortality among people with epilepsy when compared with the general population.

In developed countries, the overall mortality associated with epilepsy is two to three times that of the general population (1,2). There are few data from developing countries, because incidence studies are difficult, death certificates are not very reliable, autopsies are also not easy to obtain, and the cause of death is not usually known with certainty. The overall autopsy rate is probably less than 5% in public hospitals and extremely rare in private hospitals. There is also a relative absence of reliable coroner's reports, and, if available, they are incomplete.

An attempt was made to gather some information concerning mortality in developing countries for the symposium “Mortality in People with Epilepsy” held in Brussels in May 2000, organized by the Commission for the Burden of Epilepsy of the International League Against Epilepsy. This article presents a summary of that information, drawing on studies presented from Ecuador; the Parsi community of Bombay; a semiurban community in Vasai, India; Mali; Martinique; and Africa. For each study, we present the methods, results, and discussion. A separate summary discussion section at the end presents the collective findings.


Mortality in the general population in Ecuador (8/1,000 in 1997, adjusted to the U.S. 2000 population) is similar to the mortality in the general population in a developed country; as for example, the United States (8.7/1,000 in 1997, adjusted to the U.S. 2000 population). However, the causes of mortality are different. Infectious and parasitic diseases are the leading causes of mortality in Ecuador, and the risk of death from these diseases is virtually identical to the risk of death from cardiovascular disease in developed countries. There are no studies of mortality for epilepsy in Ecuador or other Latin American countries. This report is the first effort to provide information regarding this matter.



This was a prospective cohort study of patients with newly diagnosed epilepsy (recurrent unprovoked seizures) who were seen at five main hospitals in the three major cities of Ecuador from 01/1997 to 01/2000 (3). Eighty-two patients did not return for regular clinical evaluation. Of these, 49 patients (or their families) were interviewed by telephone, or a home visit was done to request information regarding seizure recurrence or mortality during the follow-up.


Type of seizure, or epilepsy, was classified in accordance with International League Against Epilepsy definitions and recommendations (4). We excluded cases with a first seizure more than 3 years prior to the index consultation or first medical contact for seizures more than 1 year prior to consultation. Patients with single unprovoked or acute symptomatic seizures were also excluded. Sudden unexpected/unexplained death in epilepsy (SUDEP) was defined in accordance with criteria developed by Annegers (5,6).


The standardized mortality ratio (SMR) was calculated by comparing the observed number of deaths in the epilepsy cohort with that expected in the Ecuadorian population, adjusting for age and time period. SUDEP rates were calculated using a patient-years denominator.


Characteristics of the cohort

Four hundred and twelve newly diagnosed epilepsy patients were enrolled. Thirty-three patients were lost to follow-up. Among the remaining 379 patients (1,137 patient-years of follow-up), 42% were adults and 58% children, 56% were males and 44% females. The mean age on entering the study was 18 years (±17). Forty-six percent of the patients had partial seizures, 48.7% had generalized seizures, and 5.3% were unclassifiable. Sixty-three percent of the patients had idiopathic/cryptogenic etiology, and 37% had remote symptomatic etiology.

All-cause mortality

During the follow-up period 7 patients died. The crude all-cause SMR for the study group (Table 1) was calculated as 6.3 (95% CI 2.0–10.0). The SMRs by age groups show that mortality is only significantly increased in children less than 19 years of age (Table 1). Seven deaths were reported, of which 1 patient had definite SUDEP (14.3%), 2 had probable SUDEP (28.6%), 1 had aspiration pneumonia (14.3%), 2 had seizure-related death (28.6%), and 1 had stroke (14.3%). The overall mortality rate in newly diagnosed epilepsy was 6.1 per 1,000 patient-years at risk. The overall SUDEP rate was 2.6 per 1,000 patient-years.

Table 1. All-cause mortality in patients with epilepsy in Ecuador

with epilepsy
(95% CI)
  1. CI, confidence interval.

  2. aCalculated by multiplying the number of person-years with epilepsy with the corresponding mortality rate of the Ecuadorian general population divided by 100.

  3. bObserved number of deaths divided by expected number of deaths.

0–192680.5647.7 (1.9–15.9)
20–591020.3127.1 (0.6–18.3)
>60  90.4113.2 (0.01–9.5)
Total3791.2876.3 (2.0–10.0)


This study suggests that overall mortality for epilepsy in Ecuador, a developing country, is higher than that reported in the general population of a developed country. A wide range of estimated SUDEP rates (<1.5 to >9 in 1,000 patient-years) has been published in recent cohort studies (2,7,8), depending on the methods and the population studied. The SUDEP rate in our study of incidence cases (2.9/1,000 patient-years) increased almost 10-fold when compared with the incidence of SUDEP in a study from Rochester Minnesota (0.3/1,000) (8) and as high as those reported in patients with severe epilepsy. See Tomson and colleagues (9).

It is possible that less severe new-onset cases of epilepsy were not ascertained at the hospitals. This may lead to an overestimation of the risk of death, if death is associated with severity of epilepsy. This bias may also effect the distribution of cause of death. Another problem is that patients lost to follow-up in this cohort may have a different mortality than patients we were able to follow. Other limitations of our study are that we cannot compare rates of sudden death by seizure type, seizure etiology, age, or sex, due to the small numbers of our cohort. As expected, only one patient had an autopsy performed.

On the other hand, there are some advantages of the cohort studied: diagnosis of epilepsy is well-documented, and definitions and analytic procedures are current.


Data regarding mortality in people with epilepsy in developing countries are scarce. India, a large developing country with an estimated 5 million people with active epilepsy, has no published studies of mortality in epilepsy. We describe and compare the results of two studies of mortality in different communities in India: one—the Parsis in Bombay; the other—a semiurban community in Vasai.

Mortality among Parsis with epilepsy


In 1985, we conducted a prevalence survey of Parsis with epilepsy living in housing colonies in Bombay (10). From January to March 2000, we carried out a combined postal and telephone survey of this group to identify survivors and also to ascertain the dates of death of those who had died. It is not easy to obtain the mortality rate for all Parsis in Bombay, because both numerator and denominator are difficult to obtain. For the numerator, we used death register data for the year 1992. A register of deaths is maintained for the majority of Parsis, who, after their demise are consigned to the Towers of Silence. The denominator was based on the 1981 Government of India population census data for Greater Bombay. However, this does not give age-specific rates (11). The age-specific percentage distribution was obtained from a 1982 sample survey and applied to 1981 data (12).

Cause of death was unknown. Consequently, we do not know if there were any cases consistent with SUDEP.


There were 109 people with epilepsy in 1985. Details for two of these people could not be found in 2000. Of the 107 remaining, 68 were males (63.6%) and 39 were females (36.4%), 96 were adults (89.7%), and 11 were children (10.3%); the median age was 45 years. Follow-up data were obtained on 104 out of 107 people with epilepsy. There were 34 deaths among those with epilepsy over the period March 1, 1985, to December 31, 1999: 24 males (70.6%) and 10 females (29.4%).

We calculated the age- and sex-specific mortality rates in the general population. Using these mortality rates and the observed deaths among those with epilepsy in our community study, we calculated SMRs. These were 0.73 (95% CI 0.52–1.2) for Parsi males and 0.81 (95% CI 0.29–1.1) for Parsi females, and 0.76 (95% CI 0.51–1.01) overall.

The Vasai study


Vasai Taluka in Thane District consists of a collection of villages and the town of Vasai located 50 km north of Bombay. Sixteen of these villages having a population of approximately 16,000 people were selected in 1986 for an epilepsy identification and treatment program. In 1989, a survey was carried out to identify individuals with epilepsy. All patients with epilepsy were followed for death through December 1999. EEG and CT scan information were unavailable in this cohort. Verbal autopsy was used to obtain some information on cause of death.


We identified 51 people with epilepsy (estimated prevalence of epilepsy = 3.2 per 1,000 population). All cases identified in the survey had generalized tonic–clonic seizures and idiopathic or cryptogenic etiology. No EEGs or CT scan was available. There were 30 males (58.8%) and 21 females (61.2%), with a median age of 23 years. By the end of the follow-up period in December 1999, 10 patients with epilepsy had died (case fatality = 19.6%): 8 were males and 2 were females. All 10 deaths occurred during the first 5 years of follow-up. Thirty-eight people with epilepsy were known to be alive and the status of 3 was unknown.

In 1989, the median age of the people who died subsequently was 30 years and 6 months. The causes of death were not known, but an informal “verbal autopsy” of relatives and friends revealed that no one had had a seizure or status epilepticus just prior to death. There were two possible sudden and unexpected deaths, one in a 15-year-old girl and one in a 30-year-old man. Seven of the deaths occurred in people who were termed alcoholic by the community. SMRs were calculated for the period December 1989 to December 1994 and December 1989 to December 1999, using rural Maharashtrian census death rates for 1988. The SMR for the 5-year period was 7.81 and the SMR for the 10-year period was 3.9. This suggests that those who are going to die, do so early.


These results suggest that epilepsy confers a protective effect for mortality in the Parsi community of Bombay. This community is a small, urbanized, literate community with a higher socioeconomic status than the whole Indian population. One possible explanation is that Parsis with epilepsy may take better care of their health than Parsis without epilepsy. Another explanation could be that in the Parsi study the cases were mainly prevalent and therefore biased the results in favor of survivors whose epilepsy was less active than that of the patients in Vasai. The advantages of this study are that the diagnosis of epilepsy is definite, that an entire community has been studied, and follow-up is virtually complete. Comparison of the Vasai and Parsi studies suggests that the poorer, rural population with epilepsy has a much higher mortality rate than the urban one. Factors possibly responsible for this are being explored elsewhere. Our study points to the importance of separately evaluating different sections of a population, even in a large developing country.


Mali is a large sub-Saharan country (surface 1,240,142 km2) located in west-central Africa, with a sparse population (total 6,308,320 or 5.1 inhabitants per km2). The illiteracy rate is high and the predominant economic source is agriculture (85%). The government of Mali is a republic and the predominant religion is Muslim (65%).

Health-care administration in patients with epilepsy

In Mali, traditional medicine is the prevailing modality of health-care administration. The Traditional Medicine Regional Center (“Centre de Médecine Traditionnelle,” CMTR) of the Veme Region is the only service for the care of epilepsy (13). In the CMTR, the only technical equipment for the diagnosis of epilepsy is an EEG, and there is only one physician on the staff (PPM). Phenobarbital is the only available antiepileptic drug, and even so the lack of efficient social services prevents its availability everywhere in the country.

In Mali, as in other developing countries, successful health-care programs are based on the different levels of knowledge of a given disease (biological, clinical, sociocultural, symbolic, etc.). Epilepsy is no exception. Popular beliefs regarding epilepsy (“tibu sugo” or “kirikiri” in the local language) must be also understood. Case ascertainment is possible only with the contribution of the so-called “key-informants” (village chiefs, traditional healers, etc.).


Health-care program for epilepsy in Bandiagara

Between 1988 and 1992, a health-care program for epilepsy was started in the city of Bandiagara (population 7,158). The program was part of a joint venture between the Italian Ministry of Foreign Affairs and National Research Council and the Mali Ministry of Public Health and Social Affairs.

The aims of the program were to inform the local population about the availability and utility of a health-care program for epilepsy and to disseminate knowledge about the modes of utilization of the local health-care facilities for the treatment of epilepsy. The program for epilepsy was part of a more extensive research program designed to promote professional involvement and development in a country where traditional medicine was unable to provide effective care.

The health-care program for epilepsy included a study of the prevalence and main characteristics of the disease, which was conducted in two population samples: an urban sample (the city of Bandiagara) and a rural sample of equal size (28 villages within the Bandiagara administrative district, which included 48 villages).

Epidemiological survey on epilepsy

The methods of investigation used for case ascertainment included three subsequent steps: (1) An informative meeting with any village chief and his counselors to identify the “key informants” (political committee members, family chiefs, traditional healers); (2) Individual contacts with the “key informants” to ascertain eligible patients and disseminate knowledge on the availability of effective and free medications; (3) Home visits to any eligible case to confirm the diagnosis of epilepsy (which was made by PPM in the presence of generalized tonic–clonic seizures). During the period February 14 to March 5, 2000, efforts were made to trace these patients to assess mortality and causes of death.


A total of 36 patients with epilepsy were identified in the two communities, giving a prevalence rate of 3.3 per 1,000 in the urban sample and 1.7 per 1,000 in the rural sample.

Two additional cases were identified after the completion of the prevalence survey. Thirty-one cases were traced (82%). Of these, 18 were alive (6 on conventional treatment and 8 on traditional treatment), and 13 had died (41.9%). Death was unrelated to epilepsy in 8 patients, 3 patients died after status epilepticus, and 2 patients after single seizures. Case-fatality ratio was 5:31 or 16%.


This study has several limitations. First of all, in the prevalence study the diagnosis of epilepsy was based on the presence of generalized tonic–clonic seizures, which represent a limiting factor and a source of under ascertainment of cases. Second, although in this sample the case fatality is high, the lack of data on the overall mortality in the entire study population prevents any reasonable inference on the mortality in patients with epilepsy compared to other clinical conditions and to the expected mortality.


Martinique is a French department, situated in the Caribbean Sea. The way of life in Martinique could be considered as between that of an industrialized country and a developing country.


Between May 1, 1994, and April 30, 1995, we collected all suspected cases of provoked and unprovoked epileptic seizures admitted to the hospitals or referred to the private neurologists or pediatricians on the island of Martinique. The diagnosis probability was based on clinical data from the patient charts and the EEG data. The syndromic and etiologic classifications were obtained on the basis of past medical history, neurological examination, and findings on CT scan.

The cohort was followed for 1 year to determine the risk of recurrence and the mortality rate, using only death certificates.


Three hundred and nine cases have been collected. The rate of initial diagnosis of provoked and unprovoked seizures standardized to the U.S. population was 77.7/100,000. We observed a bimodal distribution of the cases with age (86 in 0–10 years age group and 203 patients aged more than 60 years). Sixty-three cases were classified as provoked seizures (incidence: 16.4/100,000). Alcohol consumption, cerebrovascular diseases, and cranial trauma were the most frequent causes (respectively, 30.1%, 20.6%, and 18.7%). Two hundred and forty-six cases were classified as unprovoked seizures (incidence: 64.1/100,000) with the following distribution: seizures in relation with a stable condition—74 cases (incidence: 19.3/100,000), seizures in relation with an evolving condition—17 cases (incidence: 4.5/100,000), and seizures of unknown etiology—155 cases (incidence: 40.4/100,000).

During the 1-year follow-up, we identified 22 deaths in our cohort. The mortality rate was 5.73/100,000 and the SMR was 4.25. Unfortunately, we have no data concerning the cause of death, nor the exact relationship of epilepsy with the death.


The global incidence rate of newly referred persons with an initial diagnosis of epileptic seizures in Martinique Island is clearly higher than that observed in the department of Ironed in France and in the canton of Geneva. Nevertheless, we think that this is a minimal rate, probably due to an under ascertainment of situation-related seizures. A realistic estimate of the crude incidence rate would be between 87 and 102 per 100,000 population. This rate is between that observed in the industrialized countries and in developing countries. The major risk factors are represented by alcohol consumption, followed by cerebrovascular disease, cranial trauma, and infectious diseases.


It is extremely difficult to analyze the epilepsy death rate in the general population of a developing country because the number of deaths for epilepsy in a specific year is almost impossible to ascertain. Prospective large population-based mortality studies of epilepsy are also very difficult to carry out, due to the migration phenomenon.

The studies reported in this section are a first attempt to study the mortality of epilepsy in developing countries. It seems that mortality of epilepsy in these countries is generally higher than the mortality reported in developed countries. The SMR ranges broadly from 0.76 in a selected population of India to 5.7 in Martinique and 6.3 in a prospective cohort study in Ecuador. We cannot generalize these mortality numbers to the general population because they were obtained from very selected populations.

Prospective studies are needed in developing countries to follow a large new-onset cohort with epilepsy for a defined length of time to determine the mortality. These studies will be expensive and difficult to perform, but they are necessary if we are to understand the force of mortality in epilepsy in developing countries.