- Top of page
Summary: Purpose: There is a notion that people with epilepsy have substantial and often unrecognized comorbidity of chronic conditions. However, most studies focus on selected patient groups; population-based studies are scarce. We compared the prevalence of chronic somatic conditions in people with epilepsy with that in the general population using Canadian, nationwide, population-based health data.
Method: We examined epilepsy-specific and general population health data obtained through two previously validated, independently performed, door-to-door Canadian health surveys, the National Population Health Survey (NPHS, N = 49,000) and the Community Health Survey (CHS, N = 130882), which represent 98% of the Canadian population. The prevalence of epilepsy and 19 other chronic conditions was ascertained through direct inquiry from respondents about physician-diagnosed illnesses. Weighted prevalence, prevalence ratios (PR), and 95% confidence intervals were obtained for the entire population and for males and females separately. Multivariate analyses assessed the strength of association of comorbid conditions with epilepsy as compared with the general population.
Results: People with epilepsy had a statistically significant higher prevalence of most chronic conditions than the general population. Conditions with particularly high prevalence in epilepsy (prevalence ratio ≥ 2.0) include stomach/intestinal ulcers (PR, CHS 2.5, NPHS 2.7), stroke (PR, CHS 3.9, NPHS 4.7), urinary incontinence (PR, CHS 3.2, NPHS 4.4), bowel disorders (PR, CHS 2.0, NPHS 3.3), migraine (PR, CHS 2.0, NPHS 2.6), Alzheimer's disease (PR, NPHS 4.3), and chronic fatigue (PR, CHS 4.1). There were no gender-specific differences in prevalence of chronic conditions among people with epilepsy.
Conclusions: People with epilepsy in the general population, not only those actively seeking medical care, have a high prevalence of chronic somatic comorbid conditions. The findings are consistent across two independent surveys, which show that people with epilepsy in the general population have a two- to five-fold risk of somatic comorbid conditions, as compared with people without epilepsy. This patient-centered comorbidity profile reflects health aspects that are important to people with epilepsy, and indicate the need for a more integrated approach to people with epilepsy. The impact of epilepsy relative to other comorbid conditions requires further analysis, as does the contribution of comorbidity to epilepsy intractability and to differential health care needs. Similarly, it remains to be determined whether the observed comorbidity patterns are specific to epilepsy or simply reflect a pattern that is common to chronic illnesses in general.
Originally coined by Feinstein, the term comorbidity is now used to refer to the greater than coincidental association of two conditions in the same individual (1). Epilepsy is comorbid with neurological and psychiatric conditions including stroke, migraine, and depression (2). Understanding the comorbidity of epilepsy is important from several perspectives. First, understanding comorbidity can improve the differential diagnosis of epilepsy because of the substantial symptomatic overlap with several of the comorbid conditions. For example, transient ischemic attacks (TIA), migraine, and epilepsy all cause transient alterations of consciousness, and the latter two also cause headache. Less well recognized is the problem of concomitant diagnosis (3). That is, the presence of epilepsy should increase, not reduce, the index of suspicion for epilepsy, depression, and anxiety disorders. Second, comorbidity influences treatment, creating both therapeutic limitations and opportunities (4) (e.g., topiramate's antiseizure, antimigranous, and mood stabilizing effect). Third, patients with multiple conditions require a different mix and often higher level of health care resources than those with simple conditions. Finally, the study of comorbidity may provide epidemiological clues to the fundamental mechanisms of epilepsy and associated conditions.
Typically, reports of comorbidity in epilepsy have focused on stroke, migraine, and psychiatric symptoms (4), and some reports have found a higher cooccurrence of psychiatric, but not somatic conditions (5).
Andermann and Andermann (6) reported a median epilepsy prevalence of 5.9 (range 1–17) in migraineurs, which greatly exceeds the population prevalence of 0.5%. The reported migraine prevalence in people with epilepsy ranges from 8 to 23% (7). Migraine risk was not related to age of epilepsy onset, but was higher in patients with partial and generalized seizures, and was highest in posttraumatic patients with epilepsy (relative risk 4.1) (6). On the other hand, stroke is a common cause of seizures in adults aged >50 (8). The frequency of reported seizures with stroke varies from 4.4 to 17% (9,10). Finally, the most common studied comorbid conditions in patients with epilepsy are psychiatric disorders. The rate of major depression ranges between 20 and 39%, anxiety from 19 to 66%, and psychosis from 6 to 9% (11), depending on the study population (12–14). However, few studies have systematically assessed the prevalence of other chronic conditions in epilepsy. Using International Classification of Diseases (ICD)-9 codes, a recent study from the United Kingdom (15) assessed the prevalence of various conditions in adults with and without epilepsy attending general practitioners' offices. A large number of somatic and psychiatric conditions were more prevalent in persons with epilepsy in this report.
Our aim was to assess the prevalence of physician-diagnosed and self-reported epilepsy and other chronic conditions in the Canadian general population. By focusing on the general population, and using different ascertainment methods, our analyses provide a different and complementary perspective of the somatic comorbidity of epilepsy.
- Top of page
We analyzed data from two ongoing omnibus health surveys of the Canadian population, the National Population Health Survey (NPHS) (cycle 3), and the Community Health Survey (CHS) (cycle 1.1). The design and conduct of both surveys have been extensively validated. The surveys are similar in most methodological aspects, and have been described extensively (16). Data from these surveys have been used to assess the national and regional prevalence of epilepsy in Canada (16). The content of both surveys is based on a systematic, nationwide consultation process with key users of health information (17). Using validated, targeted instruments, the surveys explore a large number of health-related areas and behaviors, as well as health state utilities, family function, social function, and sociodemographic variables. Field-testing assessed respondent reaction to sensitive items and respondent burden of questionnaires. Data collection occurred over a period of approximately 13 months for both interviews, as originally intended. Using strategies to minimize nonresponse, supervising survey administration, and checking validity of responses at the time of survey administration ensured a high rate of usable data. Interviewers with a wide range of language competencies were utilized to reflect the ethnographic composition of the Canadian society (16). Both surveys included all Canadian provinces, used cluster sampling, and provided weights to adjust for the probability of being selected. In both surveys, the response rate was over 85%, 50% of subjects were females, and certain small groups were excluded (Indian reserves, armed forces, and those in prisons and chronic institutions).
The NPHS (18) collected wide-ranging health-related and sociodemographic information from 49,060 respondents of all ages. We focus on cycle 3 (1998–1999), which contains the largest and most recent cross-sectional data. The CCHS (cycle 1.1) (17) collected information between September 2000 and November 2001 from 130,822 subjects aged 12 years or older, representing 98% of the Canadian population 12 years or older.
Validity of the CHS and NPHS
The design and conduct of both surveys have been extensively validated. In most aspects, both surveys are similar and share many characteristics (Table 1). In both surveys, content is based on a systematic, nationwide consultation process with key users of health information. For example, the CHS involved over 17 workshops and 225 participants. Specific survey components, tapping into areas such as health state utilities, family function, and social function, used validated, targeted instruments for the corresponding domains. Surveys were preceded by field tests to assess respondent reaction to sensitive items and respondent burden of questionnaires. Strategies to minimize nonresponse were implemented, including introductory letters and brochures delivered to each dwelling, flexibility in the timing of survey administration, and repeated attempts in cases of initial nonresponse. The survey administration required supervised, close checks on validity of responses at the time of administration, resulting on a high rate of usable data. When needed, interviewers with a wide range of language competencies were utilized, including English, French, Spanish, Portuguese, Italian, Chinese, Punjabi, and Inuktitut, among others. Data collection occurred over a period of approximately 13 months for both interviews, as originally intended.
Table 1. Characteristics of the population surveys
|Prevalence of epilepsy||5.2 (95% CI 4.9, 5.4)||5.6 (95% CI 5.1, 6.0)|
|Weighting to adjust for survey design||Yes||Yes|
|Survey administration and data collection method||Computer-assisted personal interview||Computer-assisted personal interview|
|Ages included (median)||All groups (30 years old)||12 years and older (40 years old)|
|Groups excluded||Indian reserves, Canadian Forces bases, and some remote areas in Quebec and Ontario||Indian Reserves or Crown lands, clientele of institutions, full-time members of the Canadian Armed Forces and residents of some remote regions|
|Years of the survey||1998–1999||2001|
In the CHS, an interviewer asked subjects directly whether they had any of 20 specific chronic conditions, including epilepsy. In the NPHS, the interviewer asked whether they had any of 19 specific conditions diagnosed by a physician. Subjects responding “yes” were considered as cases. The CHS studied the following chronic conditions: epilepsy, stomach/intestinal ulcers, stroke, urinary incontinence, bowel disorders/Crohn's disease/colitis, cataracts, thyroid conditions, cancer, glaucoma, chronic fatigue, allergies, asthma, arthritis/rheumatism, back problems, high blood pressure, migraine headaches, chronic bronchitis/emphysema, diabetes, heart disease, and fibromyalgia. The NPHS studied the same conditions, excluding chronic fatigue and fibromyalgia, and adding Alzheimer's disease (17,18).
In both surveys to obtain the prevalence of chronic conditions, the numerator was the sum of weights of persons with the chronic conditions divided by the sum of weights of persons at risk. We obtained gender-specific prevalence ratios and 95% CI of the prevalence of chronic conditions in persons with epilepsy and in the general population. To assess the magnitude of the prevalence ratios, we used a logistic regression model in which the dependent variable was the presence or absence of epilepsy and the independent variables were the chronic conditions associated with epilepsy with a prevalence ratio ≥1.5. We performed this analysis with data from the CHS to have an adequate sample size.
- Top of page
We compared the prevalence of chronic somatic disorders in persons with epilepsy and in the general population using two large national health surveys in Canada. Our results confirm the notion that common chronic somatic disorders in the general population are even more common in men and women with epilepsy. As have others, we found that persons with epilepsy have a two- to five-fold risk of stroke, migraine, Alzheimer's disease, upper gastrointestinal (GI) tract disorders (ulcers), urinary incontinence, and chronic fatigue.
Hesdorffer et al. (19) and other investigators have identified a potential causal association between epilepsy and hypertension, which may be explained by hypertensive cerebrovascular epileptogenic lesions. In our study, the prevalence of hypertension was the same in patients with epilepsy and in the general population. This may be due to the nature of the surveys, which did not include formal measurement of blood pressure.
The high prevalence of stomach and intestinal problems in people with epilepsy is of note and may have at least two explanations. First, many persons with epilepsy have ictal autonomic symptoms, including GI upset (20). Second, drugs used for comorbid conditions may produce gastritis and even hemorrhage (antiplatelet drugs, antiinflammatories). Gastrointestinal complaints are common in patients that are taking antiepileptic drugs (AEDs), especially older drugs (21). Interestingly, the UK study found a higher prevalence of upper gastrointestinal bleeding in patients with epilepsy (15).
The higher frequency of urinary incontinence in people with epilepsy is likely related to ictal incontinence, as well as that related to neurological comorbidity, such as stroke, and Alzheimer's disease.
Another interesting observation is the higher frequency of chronic fatigue in patients with epilepsy. Possible explanations include an overlap between chronic fatigue and psychiatric disorders, which are more prevalent in patients with epilepsy than the general population (11–15). Fatigue is also a major side effect of AEDs, especially in polytherapy (22). Finally, seizures can result in substantial postictal fatigue. Further studies should be done in this field.
The higher frequency of heart disease in people with epilepsy is consistent with similar findings in the UK study (15). This aspect is not well studied in epilepsy. For example, it may play a role in cerebrovascular lesions, which can result in seizures and epilepsy. It may also be associated to the excess mortality seen in medically refractory epilepsy. Individuals with epilepsy have a mortality rate 2–5 times that of the general population, and bradyarrhythmias, sinus tachycardia, and a prolonged QT interval have been suggested as contributors (23–26). Although it is tempting to attribute some role to sudden unexpected death in epilepsy (SUDEP), victims of this condition typically do not have known cardiac disease.
Our analysis focused on somatic rather than psychiatric comorbidity. This is because psychiatric comorbidity is extensive and complex in people with epilepsy, and is being investigated separately, along with alcohol and substance abuse.
Ours and Gaitatzis's UK study (15) are the only large scale analyses of unselected populations exploring the comorbidity of epilepsy, and allowing for the estimation of meaningful prevalence ratios. The methodology, strengths and weaknesses of each study differ. Ours was performed in the general population, whereas the UK study focused on general practitioners' registries, which could have selected sicker patients and resulted in higher prevalence ratios, as the authors acknowledge. On the other hand, the UK study included a broader range and more specific comorbid conditions (e.g., brain tumors, GI hemorrhage, etc.), and it used physician diagnoses on medical records to capture index cases, classified using ICD codes. Yet, the accuracy and validity of ICD codes for epilepsy and other chronic conditions remain to be determined. Interestingly, although the methods of ascertainment differ, both studies share the main observations and therefore validate each other.
Jalava et al. (5) pointed out that the frequency of psychiatric but not somatic conditions was higher in patients with epilepsy compared with matched controls, Gaitatzis et al.'s and our findings refute this observation with a high level of certainty in the general population (e.g., narrow confidence intervals around the prevalence ratios). Possible explanations for Jalava et al.'s findings include incomplete data, small sample size (220 patients), and patient selection (5). This finding remains to be confirmed in other general population studies.
Forsgren et al. (27) evaluated the prevalence of psychiatric and somatic comorbidity in epilepsy in Sweden. Up to 50% of persons with epilepsy had somatic conditions. Unfortunately, the absence of controls precluded comparisons with the general population.
We do not have a clear explanation for the higher prevalence in people with epilepsy of cataracts, arthritis, bronchitis, back pain, or allergies. Prevalence studies such as ours can shed light into disease associations, but rarely into causality. In general, it is possible that this is an age-related phenomenon, as the prevalence of bronchitis, cataracts, arthritis, and back pain is higher in older age groups, and the prevalence of epilepsy in these surveys was also somewhat higher in older adults. The possibility of common pathophysiological mechanisms remains to be investigated.
Our analysis has several strengths. First, it is truly a population-based analysis, which is likely to capture both healthy and ill individuals. This circumvents the problems acknowledged by Gaitatzis et al. in their general practice-based analysis, which is likely to capture a sicker population (15). Second, direct report of chronic conditions, without relying on ICD codes, reduced the chance of invalid coding. There is little evidence supporting the accuracy of ICD coding for epilepsy and for many other conditions. For example, ICD accuracy is known to be highly variable for psychiatric conditions, especially if codes for specific syndromes are used. Therefore, the magnitude of uncertainty introduced by using ICD codes is unknown and could be substantial. Third, the remarkable agreement between two different surveys also strengthens our analysis. The majority of conditions were significantly more prevalent in epilepsy in both surveys (10 of the 17 conditions that were ascertained in both surveys). Two conditions were more common only in the CHS, three were more common only in the NPHS, and two were not more prevalent in either survey (Table 2). The same level of agreement between the surveys was seen for gender-specific comorbidity (Table 3 and 4), which further supports the validity of the results.
Some weaknesses of our study also require comment. Cases were ascertained through door-to-door personal interviews, during which participants or proxies reported medical conditions diagnosed by physicians. This method of ascertainment creates opportunity for under- and overreporting, but it is the only practical approach to explore population-based, nationwide morbidities. We have previously discussed issues of validity with this method, and conclude that analyses of prevalence of epilepsy using these health surveys yield results that are compellingly similar to those of classic epidemiological studies (28), and are also consistent across several health surveys explored (16). Although the survey validity for all reported conditions does not have an external gold standard, we maintain that the nature of the questions in the CHS and NPHS, the broad categories of illnesses explored, and the similarity of results with that of other studies, support a valid reflection of active prevalence of the chronic conditions explored here.
Our analysis of comorbidity by gender (Table 3 and 4) showed the same general pattern of comorbidity as that for the entire group. Both genders had the same associated chronic conditions including stroke, heart disease, stomach/intestinal ulcers, chronic fatigue, and chronic bronchitis/emphysema. The UK study reported similar findings (15). Thus, no strong differences in comorbidity emerged between males and females. It is important to point out that subgroup analyses (e.g., by age and gender) are limited by sample size in some categories in the NPHS. For example, the number of respondents with epilepsy aged 50 years and older is relatively small (Table 4). This is inadequate to reflect and compare the prevalence of conditions that occur mostly in this age group, such as glaucoma and Alzheimer's disease.