Epilepsy is a brain disorder caused by different etiologies, and its prognosis varies. Most patients with epilepsy have satisfactory seizure control, but 30% to 40% of patients with epilepsy treated with appropriate Anti-Epileptic Drugs (AEDs) still do not achieve seizure freedom after a certain period of time, i.e. Refractory Epilepsy (RE). Frequent seizures, long-term use of AEDs, and social discrimination not only bring physical damage to RE patients, but also cause serious social-behavioral, psychological, and psychiatric disorders, which affect patients’ study, employment, marriage, and childbirth, and significantly reduce their quality of life. Therefore, studies to explore the etiology, early prediction and intervention of RE have become a hot topic of academic interest. However, there is no strict definition of RE at home and abroad, and the criteria used in various studies are not uniform, which affects the reliability and comparability of the results of different studies. How long epilepsy patients should be observed, how many AEDs they should take, and how often seizures or how long remission during the observation period can define RE are all inconclusive, and are reviewed below. 1. Overview RE has different titles in foreign literature, such as “Chronic Epilepsy”, “Refractory Epilepsy”, and “Intractable Epilepsy”. Intractable Epilepsy”, “Drug resistant Epilepsy”, “Drug non-responsive Epilepsy”, etc., which reflect the different characteristics of RE from different perspectives, namely However, due to the lack of specific connotation, no uniform definition of RE has been made so far. RE has been defined in both broad and narrow senses. Broadly defined RE refers to epilepsy and epilepsy syndromes that cannot be terminated with current AEDs, with reasonable medication during the effective treatment period, or that have been clinically proven to be refractory. Broad RE has high sensitivity and can identify most RE patients, and can be used to diagnose and screen RE early and take appropriate measures to improve the success rate of treatment and avoid unnecessary waiting, so it is widely recognized by scholars in various countries and has become an important reference indicator for academic and management governance. However, it also has the drawback of false positives, i.e., including those patients with poor medication compliance, mild or atypical symptoms, thus overestimating the incidence of RE. With the continuous application of various therapeutic and invasive interventions, especially the application of surgery in the early stage of the disease has become an important means of treatment for some RE, it would be blind to classify all those with poor treatment results as RE, which makes some non-RE patients undergo unnecessary tests, treatments and even surgeries. Therefore, it is particularly important to regulate RE with a strict definition. In recent years, the definitions of RE used in various studies in foreign literature are inconsistent. The definition proposed by Wu Xun and Shen Dinglie is mostly used in the domestic literature, which summarizes RE as frequent seizures, at least 4 times per month, regular treatment with appropriate AEDs and drug blood levels in the effective range, at least 2 years of observation, still uncontrollable and affecting daily life, and no progressive CNS disease or intracranial occupying lesions. The definition of RE differs in the number of AEDs used, the duration of observation, the frequency of seizures or how long remission reaches seizure-free, and the incidence of RE varies from 6.7% to 74% due to differences in study population, RE inclusion criteria and follow-up time, making some patients who meet the criteria for RE in some studies fall back into the non-RE category in other studies.Sillanpaa et al. studied 144 Finnish patients with childhood epilepsy patients with onset of seizures in Finland were followed for an average of 37 years from diagnosis and found that during the follow-up period, only 16% of patients remained seizure-free after initial treatment, 19% were resistant to medication without remission since the beginning, and the remaining 65% alternated between seizures and remission, suggesting that RE is not a specific state but is dynamic, and that for most epilepsy patients, they do not always belong to the refractory group or non-refractory group. 2. Core elements of the definition of refractory epilepsy: The above studies have shown that a strict definition of RE requires the following elements: number of treatment-failed AEDs, seizure frequency or duration of remission, and duration of treatment observation, etc. However, there are differences in the current definition, and the characteristics of RE are described below from different perspectives. (1) Number of treatment-failed AEDs There are more than 20 clinically available AEDs internationally. Before determining whether epilepsy is drug-refractory, seizures should be selected rationally according to seizure type, and AEDs need to be used until the patient’s maximum tolerated dose and effective blood levels are achieved. How many drugs should be tried to determine RE Theoretically, patients with epilepsy should try each AEDs and perform different rational dosing to prove their drug effectiveness, but this consumes a lot of time, money, and is unethical and difficult to achieve in practice. In a prospective, multicenter study of childhood epilepsy, Carpay et al. found that of 416 children treated with AEDs, 40% were treated with at least 2 AEDs, regardless of pre- and post-monotherapy or combination, and the proportion achieving at least 1 year seizure-free after treatment with 2, 3, or 4 AEDs was 51%, 29%, and 10%, respectively In a study of early predictors of RE in children, Kwan et al. used achieving at least 1 year seizure-free during or after discontinuation of AEDs as a criterion for seizure control and showed that 47%, 13%, and 1% achieved seizure control after monotherapy with the first, second, and third AEDs, respectively, and only 3% of patients achieved seizure control with the combination of two AEDs. Aso et al. enrolled 169 patients with cryptogenic or symptomatic epilepsy and followed them for more than 5 years, starting with one AED, and found that 11%, 11%, and 3% achieved at least 2 years of seizure freedom after failure of treatment with two, three, and four AEDs, respectively. The above results suggest that if good control is not achieved with monotherapy, the likelihood of further application of 2, 3 or even more AEDs for complete control of seizures decreases significantly. Therefore, clinically, when a patient does not achieve seizure control despite regular treatment with 2 well-tolerated AEDs in consecutive trials, RE can be basically established. (2) Criteria for seizure control There is a lack of uniformity regarding the criteria for seizure control. Some studies have used a certain frequency of seizures during the observation period as a diagnostic criterion for RE, while some studies have used the absence of seizures up to a certain period during the observation period as an exclusion criterion for RE. Since seizure types and their seizure frequencies vary greatly, it is clearly inappropriate to use a uniform seizure frequency criterion to define RE. Clinically, RE can also be diagnosed for different seizure types, provided that the diagnosis is correct and the medication is used reasonably, after a certain period of illness without reaching seizure-free. (3) Duration of therapeutic observation Defining RE requires determining an appropriate duration of therapeutic observation. If the observation time is too long, the recurrent and persistent attacks will further aggravate the brain damage and lead to serious psychosocial and cognitive impairment, and also increase the risk of accidental death; however, if the observation time is too short, patients who can eventually achieve good control with medication may be misdiagnosed as RE. Dlugos, in identifying RE patients with indications for surgery, included persistent attacks after more than 2 years of observation in the criteria for RE. Go et al. proposed that the duration of treatment observation should be defined as 1.5 to 2 years when diagnosing RE in children. in a retrospective study of predictors of RE in children, Oskoui et al. limited the period of treatment observation to more than 1 year, during which time RE was diagnosed in patients taking at least 3 different AEDs (before and after or in combination with single medication) but still having at least 1 seizure per month on average. An absolute seizure-free period of at least 1 year is currently considered to be associated with improved quality of life for patients, and in many countries, patients with 1 seizure per year are restricted from driving. Therefore, a minimum observation period of 1 year for patients with newly diagnosed epilepsy is clinically relevant. However, for special populations such as newborns and infants, a 1-year observation period is clearly inappropriate, and some epilepsy syndromes, once diagnosed, should be treated aggressively rather than choosing to wait too long before judging whether they are RE. (4) Quality of life requirements Quality of life is a comprehensive measure of the individual’s or group’s perceived state of well-adjusted physical, psychological, and social aspects of life. With the shift in the biopsychosocial medicine paradigm and the increasing demand for quality of life and satisfaction with treatment from epileptic patients, defining RE from a sociological perspective has become an inevitable trend in the development of epileptology. Although sometimes patients’ seizure frequency and severity do not meet the current criteria for RE, or patients’ lives are not negligibly affected despite significant treatment effects, in such cases, they cannot be hastily excluded from the scope of RE. In a study of quality of life and its influencing factors in adult patients with epilepsy, Zhu Dantong et al. found that the quality of life was significantly lower in the epilepsy group compared with the control group, as evidenced by fear of seizures, dissatisfaction with daily life, poor mood, low energy, decreased cognitive function, more concerns about long-term use of AEDs, and social and work limitations, and the factors that had a greater impact on patients’ quality of life were the high number of medications and frequent seizures. Wang Xuefeng et al. investigated the quality of life of RE patients and showed that 23% had not completed 9 years of compulsory education, 62% of young people of marriageable age had unstable or irregular marriages, 61% were financially dependent on family or social support, 93% had psychosocial disorders, and 79% experienced side effects of AEDs.Sillanpaa et al [26] also found that the rate of marriage and childbirth in patients with epilepsy compared to healthy controls significantly lower, and in terms of employment rate and socioeconomic level, the post-withdrawal remission group was similar to the healthy control group, while the continuous medication group was lower, with statistically significant differences. The above study suggests that the recurrent seizures and the long-term use of AEDs make the quality of life of patients seriously affected. Because individual patients have different quality of life requirements, e.g., one to two seizures per year may not have a significant impact on young children, but may have a greater impact on adolescents who are at a critical stage of developing independence and social skills, it is difficult to assess quality of life uniformly and reflect it in the definition of RE. In response to the current lack of uniformity in the definition of RE, in 2009 the International League Against Epilepsy (ILAE) developed a unified approach to the definition of RE in terms of treatment strategies to better guide clinical work and facilitate research progress. The scheme was divided into two levels: firstly, the treatment effects of AEDs were classified, and the core definition of RE was proposed based on the former classification. In the first level, treatment outcomes are classified into 3 categories: (1) “seizure remission” means that the duration of seizure-free time during the treatment observation period is at least 3 times the longest interval between seizures before treatment or 1 year (depending on which is longer between the two); (2) “treatment failure (2) “Treatment failure” means that the above criteria for seizure remission were not met after reasonable treatment, including a change in seizure type within 1 year; (3) “Unknown outcome” means that the duration of seizure-free episodes was 3 times the longest pretreatment interval but shorter than 1 year. In the second level, based on the classification of treatment effectiveness in the first level, the responsiveness of epilepsy to AEDs treatment was classified as “drug effectiveness”, “drug resistance”, and “unknown drug responsiveness” by combining the number of AEDs used. The response of epilepsy to AEDs was classified as “drug effectiveness”, “drug resistance”, and “unknown drug responsiveness”. This resulted in a precise definition of “drug-resistant epilepsy” (RE), which is defined as a seizure-free period of less than 3 times the maximum pre-treatment seizure interval or 1 year (depending on which is longer), depending on the type of seizure, after the rational selection and correct use of at least 2 well-tolerated AEDs, either before or after a single drug or in combination. which is longer between the two). A clear definition is the theoretical basis for all studies and plays an important role in further research on the etiology and pathogenesis of RE, predicting early risk factors, standardizing diagnosis and developing and validating new drugs. The definition of RE will continue to be improved as various studies are carried out and deepened. At present, ILAE has provided a relatively complete definition of RE, but this definition cannot be used as the gold standard, and more prospective, large sample, multicenter, and long-term follow-up studies should be conducted to provide new and strong evidence to further improve the definition of RE.