An epileptic seizure is characterized by the clinical manifestation of a discrete, abnormal discharge of neurons in a part of the brain or throughout the brain. Epileptic seizures must be distinguished from nonepileptic seizures and from other conditions that can produce clinical manifestations that are highly similar to those caused by epileptic seizures (6).

An acute provoked seizure is one that occurs in the context of an acute brain insult or systemic disorder, such as, but not limited to, stroke, head trauma, a toxic or metabolic insult, or an intracranial infection (7).

A seizure that occurs in the absence of an acute provoking event is considered unprovoked (7).

Epilepsy is generally diagnosed after the occurrence of at least two unprovoked seizures on separate days, at least 24 hours apart. An individual with a single unprovoked seizure or with two or more unprovoked seizures within a 24-hour period is not considered to have epilepsy per se (7), although it is still entirely reasonable to identify the underlying form of the epileptic disorder at the time of the initial seizure (8).

Etiology is considered symptomatic if there is an antecedent condition that is causally related to an increased risk of developing epilepsy. Symptomatic factors include, but are not limited to, history of stroke, brain malformation, clear neurodevelopmental abnormality such as cerebral palsy, history of bacterial meningitis or viral encephalitis, a variety of genetic conditions, and tumors. Epilepsy in the context of a progressive condition (e.g., neurodegenerative disease or an aggressive tumor) is often considered a subgroup within symptomatic epilepsy. In contrast, epilepsy is considered nonsymptomatic in the absence of such antecedent factors. The nonsymptomatic epilepsies are further divided into idiopathic (i.e., those in which other clinical evidence including electroencephalograms [EEGs] indicates that the individual has a form of idiopathic epilepsy) and cryptogenic (in which the underlying cause remains unclear). Recently, the distinction between these two designations has become blurred by the realization that there may be some commonalities between syndromes traditionally considered idiopathic and others considered cryptogenic (9,10).

Neonatal seizures (i.e., those occurring in an infant <28 days old) are usually differentiated from epilepsy for a variety of reasons; however, several specific forms of epilepsy have been reported in this age group (11). Febrile seizures are a well-described and recognized seizure disorder that, for historical reasons, has been distinguished (both clinically and in research) from epilepsy. This has had important clinical implications for the treatment of children with such seizures. From a scientific standpoint, however, this may obscure significant aspects of the genetics and underlying neurophysiologic mechanisms of seizure-related disorders.

Epilepsy syndromes are presented in greater detail in subsequent chapters of this book; however, they are important to modern epidemiologic endeavors and thus merit mention in this context. As with cancer, epilepsy is not a single disease. In theory, epilepsy syndromes represent forms of epilepsy that have different causes, different manifestations, different implications for short- and long-term management and treatment, and different outcomes. The classification of epileptic syndromes was first proposed more than 30 years ago. The current version, which was published in 1989, is undergoing major restructuring and revision (12, 13, 14, 15, 16, 17).

Seizures and epilepsy present a complex situation because the diagnosis is not based on a single source or type of information. Rather, epilepsy is a clinical diagnosis supported to a greater or lesser extent by a wide range of data obtained from several sources: the medical history, the history (both from the patient and witnesses) of the events believed to be seizures, a neurologic examination, a reliable EEG, and sometimes neuroimaging (18). To have a valid diagnosis, one must also be able to rule out many other conditions that mimic seizures. These disorders include, but are not limited to, movement disorders, parasomnias, attention deficit hyperactivity disorder (ADHD), pseudo- or nonepileptic seizures, transient ischemic attacks, and syncope. In addition to the simple diagnosis of epilepsy, adequate information is needed to identify as accurately as possible the specific syndrome and its underlying cause. In turn, this depends largely on clinical features and on the EEG, and often on neuroimaging.

Screening questionnaires have been used in epidemiologic studies for the purpose of obtaining a rough estimate of the frequency of seizure disorders (19, 20, 21). Their diagnostic
specificity, however, can be quite poor (i.e., they identify individuals as having epilepsy who, in fact, do not have the disorder). Individuals identified via such an approach are then evaluated more thoroughly by physicians. Unfortunately, the training of these physicians or other health care professionals and the access to such basic diagnostic tools as EEGs are often less than ideal.

The importance of professional training has been highlighted in several studies. The British National General Practice Study of Epilepsy (NGPSE) included all individuals referred by general practitioners. Of these patients, only 71% were confirmed by an expert as having epilepsy (22). The Dutch study of epilepsy in childhood reported that greater than 25% of children referred for epilepsy consultation by general practitioners and pediatricians did not have the disorder (23). By contrast, the epidemiologic study of incident epilepsy in the Gironde region of France (EPIGIR), which identified patients through hospital neurology services and EEG laboratories, excluded only about 10% of patients (24). In the Connecticut study, only about 2% of referrals from child neurologists were not confirmed (by the research team) as having epilepsy (25). Finally, the Coordination Active du Réseau Observatoire Longitudinal de L’Epilepsie (CAROLE) study, which also relied on referral after diagnosis by a network of neurologists, did not exclude any patients because of an unconfirmed diagnosis (8). In all these studies, epileptologists reviewed the medical data; however, the level of expertise of the individual(s) who collected the initial information and how that information was interpreted and recorded varied.

Health care systems vary enormously in terms of (a) sophistication of care, (b) access to that care, and (c) use of available care. These differences affect how the same study might be conducted in different settings. To obtain adequate evaluations and diagnostic accuracy, either the study itself must provide all of the evaluations, or the evaluations must depend on the health care system. The first approach can be prohibitively expensive, and the feasibility of the second approach depends on where the study is conducted. In health care systems that rely heavily on specialist care, new-onset seizures are often evaluated by a specialist (26, 27, 28, 29). In settings that rely on general practitioners, on the other hand, only the more difficult cases tend to be seen by a specialist with the necessary expertise to render an accurate diagnosis (30). In many developing nations, there is no system in place for epilepsy care and for most other medical needs (31, 32, 33, 34, 35).

Epidemiologic studies often focus on a disorder as it occurs in a defined population, with the goal of determining the incidence or prevalence of the disorder and of its subtypes. This requires that a complete ascertainment method be used or that the sampling fraction be known. Even when complete ascertainment or systematic partial sampling is not feasible, a sample that is representative of the patients in the general population from which it was drawn allows one to make inferences regarding the relative significance of various causes and the frequency of particular types of epilepsy. There are several biases to avoid.