Optimizing the diagnostic timing for drug-resistant epilepsy: balancing timeliness and stability with an antiseizure medication waiting period

Abstract

Background:

While drug-resistant epilepsy (DRE) is defined by the failure of two antiseizure medications (ASM), the optimal duration to confirm such resistance remains controversial. Clinical delays intended for treatment optimization may postpone life-changing nonpharmacological interventions and increase risks associated with uncontrolled seizures, highlighting the need for standardized diagnostic timing.

Objective:

We aimed to identify the optimal waiting period that balances the consistency, stability, and timeliness of DRE diagnosis.

Design:

Retrospective observational cohort study.

Methods:

This study evaluated a modified DRE definition featuring a structured “ASM waiting period” against the original criteria. Outpatients with epilepsy received initial ASM at a tertiary center in a self-paired diagnostic simulation design. A B-spline linear model verified the inflection point at which the DRE revision proportion plateaued. A trade-off score for DRE (TOS_DRE) was developed to identify the optimal waiting period by balancing diagnostic stability and delay.

Results:

This study included 1573 patients. We used 28-, 219-, and 364-days as waiting periods for the analyses. The diagnostic consistency with the original standard was high (κ ⩾ 0.879, p < 0.001). A significant increase in diagnostic delay was observed across waiting periods (p < 0.001), except for the 0- and 28-day strategies (p = 0.290). Kaplan–Meier analysis, truncated on day 540, revealed higher DRE maintenance possibility for the 219- (hazard ratio (HR) = 0.29, 95% confidence interval (CI) 0.10–0.87, p = 0.028), and 364-day criteria (HR = 0.33, 95% CI 0.11–0.98, p = 0.046) than those for the 0-day criterion. The TOS_DRE indicated a 219-day optimal waiting period for diagnostic accuracy (w₁ = 0.7). Subgroup analyses confirmed the robustness of the 219-day strategy in most scenarios.

Conclusion:

We propose a 219-day waiting period to operationalize an “adequate trial” of DRE. This timeframe may inform DRE diagnosis and evaluation of nonpharmacological therapy.

Plain language summary

Optimal diagnostic timing for drug-resistant epilepsy

For people with epilepsy whose seizures are not well controlled by medication, accurately determining if it is “drug-resistant epilepsy” is crucial. Making this judgment too early might lead to a misdiagnosis, while waiting too long could delay trying other treatments like surgery. Our research aimed to find the most appropriate “observation and waiting period.” We reviewed and analyzed the medical records of over 1,500 patients who started standard treatment at an epilepsy center. By testing different hypothetical waiting periods (such as 1 month, 7 months, or 1 year) and using a new scoring system to balance the reliability of the diagnosis against the required waiting time, we found that an observation period of approximately seven months (219 days) achieves the best balance. Diagnoses made after this period are very reliable and rarely change later. Although waiting a full year might be slightly more stable, it causes a much longer delay. Therefore, we suggest that if a patient’s seizures persist despite close monitoring and medication adjustments for more than seven months, doctors can confidently diagnose drug-resistant epilepsy. This clear timeframe can help standardize diagnosis and ensure patients are evaluated for other treatment options at an appropriate time.

Keywords

diagnostic delay diagnostic stability drug-resistant epilepsy TOS_DRE waiting period

Introduction

Epilepsy affects over 60 million people worldwide, approximately one-third of whom develop drug resistance, which remains a major challenge in clinical management.^1,2 According to the 2010 International League Against Epilepsy (ILAE) consensus, drug-resistant epilepsy (DRE) was defined as “failure to achieve a sustained seizure-free status despite adequate trials of two well-tolerated antiseizure medications (ASMs) used in combination or as monotherapies.”³ The “sustained seizure-free” status was defined as an event that met or exceeded three times the longest preintervention interictal interval or 12 months, whichever was longer. In addition, the dosage of an “adequate trial” for each ASM was estimated to be at least 50% of the defined daily dose (1/2 DDD).⁴ This definition is the minimum working criterion for creating a standard baseline to improve routine care and guide treatment decisions. Among patients with DRE, it serves as an early indicator for considering nonpharmacological interventions, such as epilepsy surgery, which could achieve high seizure-free rates.⁵

Although ASM dosages for an “adequate trial” are well defined, the necessary duration remains controversial. Clinicians usually introduce a prudential delay by waiting for an additional 1–12 months after the second ASM reaches the target dose before formally diagnosing the DRE. This hesitation is driven by the expectation that further drug optimization might improve outcomes,⁶ combined with concerns that assigning a DRE diagnosis would impose DRE-related psychological and financial burdens on patients. This waiting period allows the medications to reach steady-state concentrations and facilitates individualized ASM adjustments. However, this extended waiting period inevitably postpones the formal diagnosis of DRE and subsequent evaluation of potentially curative nonpharmacological therapies. Excessive delay might compromise the efficacy of interventions such as surgery⁷ and prolong exposure to risks associated with uncontrolled seizures, including mortality.^8,9

Therefore, it is crucial to determine the clinical necessity and, if essential, the optimal waiting period for DRE assessment. We aimed to comprehensively evaluate the effect of an additional ASM waiting period on the consistency, stability, and timeliness of DRE diagnoses according to the ILAE criteria. We conducted a retrospective analysis of outpatients receiving initial standardized ASM therapy to identify the optimal duration that balances these critical aspects and informs more standardized and effective clinical decision-making.

Materials and methods

Patient recruitment

This retrospective observational study was conducted at the Tertiary Epilepsy Diagnosis and Treatment Centre of First Hospital of Jilin University. Outpatients who visited the center between August 2010 and May 2025 were screened for eligibility. Patients diagnosed with epilepsy who subsequently commenced ASM treatment were enrolled in this study. The exclusion criteria were maintenance of any ASM at the initial clinic visit, follow-up duration of <7 days (to exclude records without meaningful longitudinal data), and seizure type not being perceived or recorded. A flowchart of the process is shown in Figure S1 in the Supplemental Material. No patient underwent epilepsy surgery, neurostimulation, or ketogenic diet therapy during the follow-up period.

Demographic and clinical data, including age, sex, and disease duration, were extracted from outpatient records during the first visit. Additionally, we collected data on seizure frequency, seizure type, types and dosages of all administered ASMs, associated side effects at each visit, and the final diagnosis established at the last visit.

Definitions and outcomes

The primary outcome of this study was meeting the selected diagnostic criteria for DRE or achieving seizure-free status after treatment. The “sustained seizure-free” condition, according to the ILAE criteria, was termed the Expected Remission Duration in this study and defined as the longer of three times the longest interictal interval or 1 year. Different diagnostic criteria were applied as follows:

1. Original ILAE criteria (0-day criteria): Patients were classified as having DRE if they experienced any seizure after failing two tolerated and adequate ASM regimens before completing the expected remission duration.³ In this framework, the Observation Window, which was the period of seizure monitoring used to determine the outcome, began immediately upon titration of the second ASM regimen to 1/2 DDD.

2. Modified DRE criteria: This definition incorporates an additional ASM Waiting Period after two adequate ASM trials when assessing DRE. The observation window began after the waiting period ended. Patients were diagnosed with DRE only if seizures occurred during this observation window; seizures occurring exclusively during the waiting period were not considered DRE.

The seizure-free diagnosis retained the original criterion, which depended solely on the seizure-free duration exceeding the Expected Remission Duration, regardless of ASM type or dose.

Of note, this was a diagnostic simulation study conducted on a single, identical cohort of patients. Different diagnostic criteria were applied to the same clinical dataset to evaluate how the ASM waiting period affects the subsequent stability of DRE diagnosis. Thus, all comparisons are within-subject rather than between different patient groups.

Longitudinal monitoring was conducted after diagnosis to verify the stability of the outcome. Patients who maintained a seizure-free status following a DRE diagnosis were classified as exhibiting DRE remission, whereas those who experienced seizures after seizure freedom were classified as having relapse. The total diagnostic delay was defined as the time from the first ASM to the time the diagnostic criteria were met. A schematic representation of these outcome definitions and their temporal sequences is shown in Figure 1.

Figure 1.

Conceptual schematic of epilepsy diagnostic evolution with the ASM waiting period.

Statistical analysis

Continuous data are presented as mean ± standard deviation or median (interquartile range, IQR), whereas categorical variables are summarized using frequency counts and percentages. Kruskal–Wallis or Mann–Whitney U tests were used for comparisons involving skewed distributions. Inter-rater agreement was quantified using Cohen’s kappa coefficient. Linear models with B-spline basis functions were used to characterize the nonlinear relationship between waiting periods and diagnosis-specific proportions. The DRE proportion was calculated by dividing the number of patients diagnosed using each modified criterion by the number diagnosed using the original ILAE DRE standard. The same method was used to calculate the proportion of patients reaching seizure freedom. The revised DRE proportion indicates the percentage of patients who initially met the DRE criteria according to the original criterion and were identified as seizure-free under the modified criteria. The inflection point was identified by detecting substantial reductions in the first derivative of each fitted curve and was defined as the threshold at which the absolute slope was <50% of its maximum value. A 15% buffer zone was applied to both temporal boundaries to exclude edge effects. The stability of the diagnosis was analyzed using Kaplan–Meier curves and the marginal Cox proportional hazards models with robust standard errors.

The trade-off score for DRE (TOS_DRE) was calculated for each candidate’s waiting period to balance the diagnostic accuracy against delay. For each waiting period strategy, DRE remission rate (DRemR) estimates and their 95% confidence intervals (95% CI) were normalized to a 0–1 scale using min-max normalization based on the observed range across all strategies. The diagnostic delay, represented by the median time, was normalized to a cost score (0–1) using a robust min-max normalization approach, with range boundaries estimated from the IQR across all strategies. These normalized scores were combined into a composite score using the following formula:

TOS_DRE = w₁ × DRemR_Norm + w₂ × Delay_Norm,with weights set to prioritize accuracy (w₁ = 0.7, w₂ = 0.3)

The optimal waiting period was selected based on the lowest TOS_DRE point estimate and the smallest upper and lower confidence bounds. This ensures a robust performance under expected, worst-case, and best-case scenarios. We visualized the trade-offs across weight combinations and identified decision boundaries at which optimal strategies based on point estimates changed. Ambiguous regions were delineated by identifying inconsistencies between the point estimates and upper (pessimistic) or lower (optimistic) bounds. Following a predefined rule, subgroup analyses using forest plots indicated the optimal strategy at w₁ = 0.7.

Textual data records were extracted using custom Python scripts (v3.11; scripts are in‑house developed, no manufacturer). The resulting dataset was cleaned and organized using IBM SPSS Statistics (v24.0, IBM Corp., Armonk, NY). Statistical analyses and graphical representations were performed using R (v4.5.1; R Foundation for Statistical Computing, Vienna, Austria) and SPSS. Statistical significance was set at p < 0.05.

Results

This study included 1573 patients with epilepsy. The demographic and clinical data are shown in Figure 2. Males constituted 55% of the study population, and the mean age was 35.5 ± 17.2 years. The median disease duration was 730 days (IQR 2838 days). The median follow-up duration was 285 (IQR 654) days, and 42.7% (n = 672) of the patients completed 1-year follow-up. The most prevalent seizure type was focal (73.8%), and the most common identified etiology of epilepsy was structural lesion (17.4%). The baseline seizure frequency was 1.0 (IQR 1.0) per month. Detailed information on the primary outcomes according to the original criteria is presented in Table S1.

Figure 2.

Demographic and clinical information.

Primary outcomes across waiting periods

A total of 366 (23.3%) patients achieved seizure freedom, whereas 203 (12.9%) had DRE according to the original ILAE criteria (0-day waiting period criterion). Diagnostic outcomes varied based on ASM waiting periods. The proportions of DRE and seizure-free cases (referred to as the original ILAE criteria) are shown in Figure 3(a). As the waiting period increased, up to 8.4% of patients with DRE diagnosed using the original ILAE criteria were revised to exhibit a seizure-free status (Figure 3(b)). An inflection point was observed on day 219, after which the number of patients leveled off. The detailed diagnostic distributions are listed in Table 1. The diagnostic results for each waiting period demonstrated good consistency with the original standards (κ ⩾ 0.879, p < 0.001). Based on pharmacokinetics and the inflection point indicating a plateau, we selected 28-, 219-, and 364-days (the longest) as candidate waiting periods for further analysis. Among patients who reached 1/2 DDD with the second ASM (n = 385), 60.8% (n = 234) achieved ⩾50% reduction in seizure frequency compared with baseline.

Figure 3.

Diagnosis proportion trends with inflection points. (A) DRE and seizure‑free proportions at each waiting period based on original ILAE criteria. (B) Proportion of DRE patients revised to seizure‑free with extended waiting periods; plateau after day 219.

Table 1.

Comparison of diagnostic outcomes across waiting periods, n (%).

Waiting period	0-day			Total^b	κ	p Value
Waiting period	DRE^a	Seizure-free^a	Undetermined^a	Total^b	κ	p Value
7 days					1.000	<0.001
DRE	203 (100.0)	0 (0.0)	0 (0.0)	203 (12.9)
Seizure-free	0 (0.0)	366 (100.0)	0 (0.0)	366 (23.3)
Undetermined	0 (0.0)	0 (0.0)	1004 (100.0)	1004 (63.8)
28 days					0.996	<0.001
DRE	200 (98.5)	0 (0.0)	0 (0.0)	200 (12.7)
Seizure-free	1 (0.5)	366 (100.0)	0 (0.0)	367 (23.3)
Undetermined	2 (1.0)	0 (0.0)	1004 (100.0)	1006 (64.0)
91 days					0.963	<0.001
DRE	173 (85.2)	0 (0.0)	0 (0.0)	173 (11.0)
Seizure-free	6 (3.0)	366 (100.0)	0 (0.0)	372 (23.6)
Undetermined	24 (11.8)	0 (0.0)	1004 (100.0)	1028 (65.4)
182 days					0.925	<0.001
DRE	143 (70.4)	0 (0.0)	0 (0.0)	143 (9.1)
Seizure-free	13 (6.4)	366 (100.0)	0 (0.0)	379 (24.1)
Undetermined	47 (23.2)	0 (0.0)	1004 (100.0)	1051 (66.8)
219 days					0.913	<0.001
DRE	134 (66.0)	0 (0.0)	0 (0.0)	134 (8.5)
Seizure-free	15 (7.4)	366 (100.0)	0 (0.0)	381 (24.2)
Undetermined	54 (26.6)	0 (0.0)	1004 (100.0)	1058 (67.3)
364 days					0.879	<0.001
DRE	109 (53.7)	0 (0.0)	0 (0.0)	109 (6.9)
Seizure-free	17 (8.4)	366 (100.0)	0 (0.0)	383 (24.3)
Undetermined	77 (37.9)	0 (0.0)	1004 (100.0)	1081 (68.7)
Total^b	203 (12.9)	366 (23.3)	1004 (63.8)	1573 (100.0)

Proportion of 0-day waiting periods.

Proportion of total population.

DRE, drug-resistant epilepsy.

The interval from the first ASM administration to the fulfillment of the diagnostic criteria was analyzed to evaluate the effect of the ASM waiting period on the overall diagnostic delay. The 364-day waiting period criterion experienced the longest delay, which was significantly longer than those of the 0-day (Z = −7.394, p < 0.001), 28-day (Z = −7.073, p < 0.001), and 219-day (Z = −3.658, p < 0.001). In addition, the 219-day criterion had a significantly longer delay than the 0-day (Z = −5.740, p < 0.001) and 28-day (Z = −5.235, p < 0.001) criteria. The 0- and 28-day criteria showed no significant differences (Z = −1.058, p = 0.290). In contrast, no significant difference was observed among the ASM waiting period criteria in terms of the delay in achieving seizure freedom (p = 0.883; Table 2).

Table 2.

Diagnostic delay and probability of diagnosis maintenance across waiting periods.

	Waiting period				Test
	0-day	28-day	219-day	364-day	χ²	p Value
Diagnostic delay, days (median (IQR))
DRE	296 (606)	322 (598)	507 (594)	658 (579)	86.775	<0.001
Seizure-free	434 (498)	434 (498)	449 (510)	450 (511)	0.660	0.883
Probability truncated at 540 days, % (95% CI)
Persisting DRE	88.6	89.2	96.7	96.2
	(83.1–94.4)	(83.9–95.0)	(93.1–100.0)	(92.0–100.0)
DRE remission	11.4	10.8	3.3	3.8
	(5.6–16.9)	(5.0–16.1)	(0.0–6.9)	(0.0–8.0)
Sustained seizure-free	81.9	81.9	82.0	81.7
	(77.6–86.5)	(77.7–86.5)	(77.7–86.4)	(77.4–86.1)
Seizure relapse	18.1	18.1	18.0	18.3
	(13.5–22.4)	(13.5–22.3)	(13.6–22.3)	(13.9–22.6)

CI, confidence interval; DRE, drug-resistant epilepsy; IQR, interquartile range.

Postdiagnosis analysis of stability

Follow-up after the primary diagnosis revealed probable revisions. Approximately 16.7% (n = 34, total DRE = 203), 16.5% (n = 33, total DRE = 200), 13.4% (n = 18, total DRE = 134), and 14.7% (n = 16, total DRE = 109) of patients with DRE achieved DRE remission at the end of the study by the 0-, 28-, 219-, and 364-day criteria, respectively. The Kaplan–Meier curve was used to analyze DRE maintenance (Figure 4). The number-at-risk table showed substantial censoring beyond 540 days, suggesting considerable uncertainty in the long-term survival estimates. To account for the within-subject correlation among the different waiting period criteria, marginal Cox proportional hazards models with robust standard errors were employed. In the model truncated at day 540, the overall Wald test showed a p-value of 0.130. However, planned pairwise comparisons revealed that the 219-day (hazard ratio (HR) = 0.29, 95% CI 0.10–0.87, p = 0.028) and 364-day criteria (HR = 0.33, 95% CI 0.11–0.98, p = 0.046) significantly improved diagnostic stability compared to the 0-day original standard. In contrast, the 28-day criteria showed no significant improvement in stability over the 0-day criteria (HR = 0.95, p = 0.501). Regarding the diagnostic revision of seizure-free patients, approximately 27.0% (n = 99, total-seizure-free = 366), 27.0% (n = 99, total-seizure-free = 367), 27.3% (n = 104, total-seizure-free = 381), and 27.4% (n = 105, total-seizure-free = 383) of patients experienced seizure relapse until the study ended by the 0-, 28-, 219-, and 364-day criteria, respectively. The Kaplan–Meier curve is presented in Figure S2 in the Supplemental Material. The pattern of substantial censoring beyond 730 days warrants caution when interpreting the long-term survival estimates. No significant differences were observed among the four criteria (p = 0.451).

Figure 4.

Kaplan–Meier curves of DRE remission. Kaplan–Meier analysis showing the probability of diagnostic maintenance of DRE by criteria during follow-up (main) and at 540 days (insert).

Composite score evaluation of ASM waiting periods for DRE

Subsequent analyses were restricted to DRE modifications. To determine the optimal waiting period, we analyzed TOS_DRE to assess its ability to balance the trade-off between minimizing the diagnostic revision rate (DRemR) and shortening the diagnostic delay across the four candidate waiting period strategies. DRemRs truncated on day 540 were applied to the TOS_DRE. With an increased emphasis on diagnostic accuracy (w₁), the TOS_DRE of the 0- and 28-day criteria increased, whereas that of the 219- and 364-day criteria decreased (Figure 5). When only point estimates were considered, the 219-day ASM waiting strategy was optimal when w₁ exceeded 0.32 (prioritizing diagnostic accuracy). The 0-day strategy was optimal when w₁ was <0.32 (prioritizing diagnosis timeliness). When the point estimates and confidence bounds indicated different optimal strategies, a decision-ambiguity region emerged (w₁ between 0.28 and 0.40), preventing the identification of a single optimal strategy. Nevertheless, the limited range suggests that the comparative outcomes are reasonably stable, with no effect on the overall preference for waiting strategies. The 219-day strategy emerged as the optimal waiting period for prioritizing diagnostic stability (w₁ = 0.7).

Figure 5.

Trade-off analysis of diagnostic waiting period strategies. Analysis of the trade-off between the DRemR and diagnostic delay across the four strategies. (Upper) TOS_DRE with CIs showing a major decision boundary at w₁ = 0.32 (black, dashed line). (Lower) Strategy preference map derived directly from point estimates. Ambiguous regions are highlighted: the white dashed box (0.28–0.32) indicates inconsistency between point estimates and upper confidence limits, whereas the gray dashed box (0.32–0.40) indicates inconsistency between point estimates and lower confidence limits.

Subgroup analysis

To verify the applicability of the waiting period strategies across subgroups, we analyzed stratified TOS_DRE outcomes by applying the four strategies within each subgroup, prioritizing the diagnostic stability (w₁ = 0.7). For all subgroups, the 219-day waiting period appeared to be the optimal strategy when comparing the point estimates and the upper bound, which represented the worst-case scenario (Figure 6). In contrast, when comparing the lower bound (best-case scenario), the 219-day strategy was robust in female patients, adult patients, patients with focal seizures, unknown etiology, and ⩾50% reduction in seizure frequency compared to baseline. The 28-day waiting period was more suitable for male patients, whereas the 0-day waiting period was preferred for the other four subgroups.

Figure 6.

Composite TOS_DRE forest plot by waiting period strategy and patient subgroup (w₁ = 0.7).

Discussion

In this single-center retrospective study, we evaluated the need to implement an ASM waiting period in clinical practice and sought to determine the optimal waiting period that balances diagnostic stability and timeliness. We found that implementing an additional waiting period resulted in up to 8.4% of patients who initially met the original DRE criteria being revised to seizure-free. The DRE revision rate increased over an extended waiting period and plateaued at a reflex point on day 219. While the waiting period inherently delayed the formal DRE diagnosis, it tended to enhance the diagnostic certainty. When diagnostic accuracy (w₁ = 0.7) was prioritized, a 219-day waiting period was identified as optimal for confirming the DRE. In stratified subgroup analyses, the 219-day strategy remained robust across the expected and worst-case scenarios for all subgroups. In contrast, best-case scenario findings were subgroup-dependent, favoring the 219-day strategy for specific subgroups (females, adults, focal seizure, unknown etiology, and ⩾50% reduction in seizure frequency); however, shorter waiting periods (0 or 28 days) were supported for others.

The original 2010 definition of DRE established a minimum threshold for alerting clinicians to the potential for drug resistance. The emphasis on timeliness aims to prevent unnecessary prolongation of ineffective pharmacotherapy, preserving a valuable window for referral to nonpharmacological treatments such as epilepsy surgery or neurostimulation.³ Surgery offers a superior probability of achieving seizure freedom and subsequent improvement in quality of life, particularly for patients with temporal lobe epilepsy, compared with the effects of continued ASM treatment.^5,10,11 From a health economics perspective, although epilepsy surgery and its presurgery evaluation require a notable initial outlay, they are long-term, cost-effective strategies for surgically eligible patients compared with ASMs alone.^12,13 Therefore, prompt evaluation for surgery at comprehensive epilepsy centers is recommended when the DRE criteria are met.¹⁴

However, when a patient exhibits a substantial reduction in seizure frequency following a second ASM at 1/2 DDD, clinicians usually hesitate to diagnose DRE, even when formal criteria are met definitively. This hesitation to diagnose reflects an informally adopted ASM waiting period that is both physiologically and clinically grounded in the literature. From a pharmacokinetic perspective, the stabilization of plasma drug concentrations after reaching the target oral dose is not spontaneous and requires time. Certain ASMs with high metabolic stability, such as phenobarbital and perampanel, exhibit prolonged titration periods.^15,16 Furthermore, the addition of a third or more adequate ASMs could increase the likelihood of ultimate seizure freedom.^6,17 Moreover, temporary fluctuations in seizure frequency could be influenced by bidirectional comorbidities, such as sleep or mood disorders,^18
–20 and extended observation could help distinguish these fluctuations from true drug resistance.

In our study, a 364-day waiting period differentiated 8.4% of patients who achieved sustained seizure freedom from those who met the original DRE criteria. Hence, these patients were protected from the psychological burden of being labeled as having DRE. Correspondingly, the probability of DRE remission decreased from 11.4% under the original criteria, a rate consistent with the report of early remission,²¹ to 3.8% under a waiting period of 364 days. This reduction enhanced the reliability of DRE diagnosis, strengthening clinicians’ confidence in recommending surgical evaluations. Such prudence was warranted not only because of concerns regarding potential psychological distress to the patient but also because of the substantial cost of comprehensive presurgical evaluation (approximately £7000–£50,000).²² For patients whose seizures can be controlled with medication optimization in a relatively short time, this costly and likely invasive workup is not cost-effective or justifiable. Therefore, applying an acceptable ASM waiting period is crucial for a more reliable DRE diagnosis in the future.

The analysis of the trade-off score (TOS_DRE) revealed that a 219-day strategy was the optimal waiting period for a stable, time-saving DRE diagnosis when diagnostic stability was prioritized. This strategy delayed the confirmation of DRE by approximately 7 months; however, it prevented DRE diagnosis in 7.4% of patients who achieved seizure freedom. Moreover, when we assigned equal significance to diagnostic stability and timeliness (w₁ = 0.5), the 219-day strategy remained optimal. The subgroup analysis validated this conclusion across expected TOS_DRE and worst-case scenarios, indicating that the 219-day waiting period was optimal even under the most pessimistic assumptions. However, considering the best-case scenario, the 219-day strategy was more robust in female patients, possibly because it enabled a prolonged observation period that could capture the hormonally driven cyclicity of seizures in women. Oestrogen and progesterone can influence seizure susceptibility.^23,24 In addition, female patients may have a higher risk of experiencing mood disorders or sleep disturbances owing to the menstrual cycle, which can induce temporary and frequent seizures.^25
–27 Prolonged observation can help reveal the true and relatively stable seizure condition. Without these disturbances, a 28-day waiting period may be more beneficial for male patients by avoiding unnecessary delays in a longer strategy. In the pediatric population, a prolonged ASM waiting period should be approached with caution. Pediatric epilepsy syndromes are highly complex, as the age of onset is critically linked to specific electro-clinical syndromes and long-term neurodevelopmental outcomes.²⁸ Moreover, early initiation of nonpharmacological interventions, such as the ketogenic diet and vagus nerve stimulation, is essential to prevent irreversible cognitive decline in these syndromes with poor drug response. Therefore, a strategy favoring no waiting period may be more appropriate for children. Regarding the etiology of epilepsy, we found that the 219-day waiting strategy was more clinically relevant for patients with epilepsy of unknown cause. This might be explained by evidence suggesting that an unknown etiology is usually associated with a more favorable response to ASMs,²⁹ allowing continued drug optimization for potential, long-term remission. In contrast, for patients with a known etiology (primarily structural, with magnetic resonance imaging-visible lesions), the observed lower DRemR reduced the need for a waiting period before initiating treatment. This finding supports the clinical principle that these patients are optimal candidates for early surgery, offering superior outcomes.^7,30,31 Additionally, we found that patients with focal seizures required a 219-day waiting period, possibly because the optimal ASM was less syndrome-specific, requiring longer observation to establish treatment failure.³² Conversely, drug responses in nonfocal epilepsy are heterogeneous and polarized according to the syndrome and cause.^33,34 Consequently, the optimal waiting period for these patients should be individualized. Furthermore, we found that more responsive cases preferred the 219-day strategy, whereas less responsive cases preferred diagnosis without a waiting period. This finding is consistent with clinical practice, in which clinicians tend to delay DRE diagnosis in patients with substantially reduced seizure frequency.

Limitations

This study has some limitations. First, as this was a single-center investigation, treatment adjustments could have been influenced by individual clinicians’ prescription preferences. However, all clinicians involved were specialized epileptologists at a tertiary epilepsy center, ensuring the standardization of management. Second, the retrospective design introduced potential variability in the timing of medication adjustments and DRE confirmation owing to the inherent inconsistency of follow-up intervals. Third, more than half of the patients were undetermined owing to insufficient follow-up. Prospective multicenter studies are required to validate and generalize these findings. Finally, the proportion of pediatric patients in our cohort was relatively small. Given that certain early-onset epilepsy syndromes require urgent nonpharmacological intervention to prevent developmental regression, a 219-day waiting period might be inappropriately long for this population. Future studies with larger pediatric cohorts are needed to define age-specific optimal waiting periods.

Conclusion

We introduced a waiting period to operationalize the temporal dimension of an “adequate trial,” providing a concrete, measurable reference point for clinical practice. Based on the preliminary ILAE criteria for DRE, we suggest a waiting period of 7 months (219 days) with close monitoring of seizures and adjustments to treatment. If seizures persist beyond this interval, a definitive diagnosis of DRE can be established, warranting consideration of nonpharmacological therapies. Clinical decisions should be tailored to individual patient factors, including age, sex, etiology, seizure type, and drug response.

Supplemental Material

sj-docx-1-tan-10.1177_17562864261464160 – Supplemental material for Optimizing the diagnostic timing for drug-resistant epilepsy: balancing timeliness and stability with an antiseizure medication waiting period

Supplemental material, sj-docx-1-tan-10.1177_17562864261464160 for Optimizing the diagnostic timing for drug-resistant epilepsy: balancing timeliness and stability with an antiseizure medication waiting period by Nan Li, Yanyan Chen, Guangjian Li, Jing Li and Weihong Lin in Therapeutic Advances in Neurological Disorders

Supplemental Material

sj-docx-2-tan-10.1177_17562864261464160 – Supplemental material for Optimizing the diagnostic timing for drug-resistant epilepsy: balancing timeliness and stability with an antiseizure medication waiting period

Supplemental material, sj-docx-2-tan-10.1177_17562864261464160 for Optimizing the diagnostic timing for drug-resistant epilepsy: balancing timeliness and stability with an antiseizure medication waiting period by Nan Li, Yanyan Chen, Guangjian Li, Jing Li and Weihong Lin in Therapeutic Advances in Neurological Disorders

Supplemental Material

sj-tif-1-tan-10.1177_17562864261464160 – Supplemental material for Optimizing the diagnostic timing for drug-resistant epilepsy: balancing timeliness and stability with an antiseizure medication waiting period

Supplemental material, sj-tif-1-tan-10.1177_17562864261464160 for Optimizing the diagnostic timing for drug-resistant epilepsy: balancing timeliness and stability with an antiseizure medication waiting period by Nan Li, Yanyan Chen, Guangjian Li, Jing Li and Weihong Lin in Therapeutic Advances in Neurological Disorders

Footnotes

Acknowledgements

The authors would like to express their sincere gratitude to all the colleagues who contributed to the long-term collection and documentation of outpatient data. Their diligent efforts were essential for this study. We extend our special thanks to J. Li, D. Yang, C. Chu, Z. Wang, Z. Jing, X. Ma, R. Zhong, X. Zhang, M. Li, Y. Lu, Q. Zhao, H. Zhang, Y. Han, X. Guo, X. Zhang, K. Yan, Y. Liu, L. Sun, J. Ge, X. Wang, Y. Zhou, W. Gao, R. Zheng, and K. Lang. We appreciate their meticulous work and unwavering commitment.

Declarations

ORCID iDs

Nan Li

Jing Li

Weihong Lin

Supplemental material

Supplemental material for this article is available online.

Other disclosure statements

We confirm that we have read the journal’s position on ethical publication issues and affirm that this report is consistent with those guidelines.

References

Lerche

Drug-resistant epilepsy—time to target mechanisms. Nat Rev Neurol 2020; 16: 595–596.

Kalilani

Sun

Pelgrims

, et al. The epidemiology of drug-resistant epilepsy: a systematic review and meta-analysis. Epilepsia 2018; 59: 2179–2193.

Kwan

Arzimanoglou

Berg

, et al. Definition of drug resistant epilepsy: consensus proposal by the ad hoc task force of the ILAE commission on therapeutic strategies. Epilepsia 2010; 51: 1069–1077.

Brodie

Barry

Bamagous

, et al. Effect of dosage failed of first antiepileptic drug on subsequent outcome. Epilepsia 2013; 54: 194–198.

Sivaraju

Hirsch

Gaspard

, et al. Factors predicting outcome after intracranial EEG evaluation in patients with medically refractory epilepsy. Neurology 2022; 99: e1–e10.

Chen

Brodie

Liew

, et al. Treatment outcomes in patients with newly diagnosed epilepsy treated with established and new antiepileptic drugs. JAMA Neurol 2018; 75: 279–286.

Janszky

Schulz

, et al. Temporal lobe epilepsy with hippocampal sclerosis: predictors for long-term surgical outcome. Brain 2005; 128: 395–404.

Zhang

Hall

Lam

SK.

Comparison of long-term survival with continued medical therapy, vagus nerve stimulation, and cranial epilepsy surgery in paediatric patients with drug-resistant epilepsy in the USA: an observational cohort study. Lancet Child Adolesc Health 2023; 7: 455–462.

Klein

Krauss

Steinhoff

, et al. Failure to use new breakthrough treatments for epilepsy. Epilepsia 2023; 64: 1458–1465.

10.

Wang

Chen

, et al. Resection of the temporal piriform cortex in mesial temporal lobe epilepsy: association with seizure freedom in a retrospective cohort study. Epilepsia 2025; 66(11): 4211–4222.

11.

Bala

Olejnik

Kułak

, et al. Determinants of the quality of life in patients with drug-resistant temporal lobe epilepsy: a comparison of the results before and after surgery. Brain Sci 2024; 14: 241.

12.

Ngan Kee

Foster

Marquina

, et al. Systematic review of cost-effectiveness analysis for surgical and neurostimulation treatments for drug-resistant epilepsy in adults. Neurology 2023; 100: e1866–e1877.

13.

Sheikh

Kattan

Steinmetz

, et al. Cost-effectiveness of surgery for drug-resistant temporal lobe epilepsy in the US. Neurology 2020; 95: e1404–e1416.

14.

Wiebe

Jetté

Epilepsy surgery utilization. Curr Opin Neurol 2012; 25: 187–193.

15.

Jing

Shiba

Morita

, et al. A single- and multiple-dose pharmacokinetic study of oral Perampanel in healthy Chinese subjects. Clin Drug Investig 2023; 43: 155–165.

16.

Lappin

Shishikura

Jochemsen

, et al. Comparative pharmacokinetics between a microdose and therapeutic dose for clarithromycin, sumatriptan, propafenone, paracetamol (acetaminophen), and phenobarbital in human volunteers. Eur J Pharm Sci 2011; 43: 141–150.

17.

Barnard

Chen

Holmes

, et al. Treatment response to antiseizure medications in people with newly diagnosed focal epilepsy. JAMA Neurol 2025; 82(10): 1022–1030.

18.

Roliz

Kothare

The interaction between sleep and epilepsy. Curr Neurol Neurosci Rep 2022; 22: 551–563.

19.

Gruenbaum

Sandhu

MRS

Bertasi

RAO

, et al. Absence seizures and their relationship to depression and anxiety: evidence for bidirectionality. Epilepsia 2021; 62: 1041–1056.

20.

Krishnan

Depression and anxiety in the epilepsies: from bench to bedside. Curr Neurol Neurosci Rep 2020; 20: 41.

21.

Choi

Hayat

Zhang

, et al. Drug-resistant epilepsy in adults: outcome trajectories after failure of two medications. Epilepsia 2016; 57: 1152–1160.

22.

Khoo

Martin

de Tisi

, et al. Cost of pre-surgical evaluation for epilepsy surgery: a single-center experience. Epilepsy Res 2022; 182: 106910.

23.

Carvalho

Colonna

Curia

, et al. Sex steroid hormones and epilepsy: effects of hormonal replacement therapy on seizure frequency of postmenopausal women with epilepsy—a systematic review. Eur J Neurol 2023; 30: 2884–2898.

24.

Shiono

Sun

Batabyal

, et al. Limbic progesterone receptor activity enhances neuronal excitability and seizures. Epilepsia 2021; 62: 1946–1959.

25.

Wei

Wakim

Martinez

, et al. Differential effects of ovarian steroids in women with and without premenstrual dysphoric disorder: a replication and extension of findings. Am J Psychiatry 2025; 182: 922–934.

26.

Shechter

Lespérance

Ng Ying Kin

NMK

, et al. Nocturnal polysomnographic sleep across the menstrual cycle in premenstrual dysphoric disorder. Sleep Med 2012; 13: 1071–1078.

27.

Meth

EMS

Nôga

Dubol

, et al. The impact of pharmacotherapy for premenstrual dysphoric disorder on sleep. Sleep Med Rev 2025; 80: 102069.

28.

Menon

Helen Cross

Childhood epilepsy. Lancet 2025; 406: 636–649.

29.

Atalar

Vanlı-Yavuz

Yılmaz

, et al. Long-term follow-up of a large cohort with focal epilepsy of unknown cause: deciphering their clinical and prognostic characteristics. J Neurol 2020; 267: 838–847.

30.

Jehi

Braun

Does etiology really matter for epilepsy surgery outcome?

Brain Pathol 2021; 31: e12965.

31.

Gill

Deleo

Bernhardt

, et al. MRI-negative epilepsy: a systematic review and meta-analysis. Epilepsia 2025; 66(12): 4939–4949.

32.

Marson

Burnside

Appleton

, et al. The SANAD II study of the effectiveness and cost-effectiveness of levetiracetam, zonisamide, or lamotrigine for newly diagnosed focal epilepsy: an open-label, non-inferiority, multicentre, phase 4, randomised controlled trial. Lancet 2021; 397: 1363–1374.

33.

Hirsch

French

Scheffer

, et al. ILAE definition of the idiopathic generalized epilepsy syndromes: position statement by the ILAE task force on nosology and definitions. Epilepsia 2022; 63: 1475–1499.

34.

Hamdan

Myers

Cossette

, et al. High rate of recurrent de novo mutations in developmental and epileptic encephalopathies. Am J Human Genet 2017; 101: 664–685.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.03 MB

0.41 MB

1.32 MB