Effectiveness of First Antiepileptic Drug


Address correspondence and reprint requests to Professor M. J. Brodie at Epilepsy Unit, Department of Medicine and Therapeutics, Western Infirmary, Glasgow G11 6NT, Scotland. E-mail: Martin.J.Brodie@clinmed.gla.ac.uk


Summary:  Purpose: To investigate the interaction among efficacy, tolerability, and overall effectiveness of the first antiepileptic drug (AED) in patients with newly diagnosed epilepsy.

Methods: The 470 patients were diagnosed, treated and followed up from January 1984 at a single center. Outcome was classified as seizure freedom for at least the last year or failure of initial treatment because of inadequate seizure control, adverse events, or for other reasons.

Results: Overall, 47% of patients became seizure-free with the first prescribed AED. A higher proportion (p = 0.025) of patients with symptomatic or cryptogenic epilepsy changed treatment because of intolerable side effects (17%), and a lower proportion (p = 0.007) became seizure-free (43.5%) compared with those with idiopathic epilepsy (8.5% and 58%, respectively). Most patients (83%) received carbamazepine (CBZ; n = 212), sodium valproate (VPA; n = 101), or lamotrigine (LTG; n = 78). The majority of seizure-free patients required only a moderate daily AED dose (93.1% with ≤800 mg CBZ, 91.3% with ≤1,500 mg VPA, 93.8% with ≤300 mg LTG), with commonest dose ranges being 400–600 mg for CBZ, 600–1,000 mg for VPA, and 125–200 mg for LTG. Most withdrawals due to poor tolerability also occurred at or below these dose levels (CBZ: 98%; VPA: 100%; LTG: 75%). Patients taking CBZ (27%) had a higher incidence of adverse events necessitating a change of treatment than did those treated with VPA (13%) or LTG (10%), resulting in fewer becoming seizure-free (CBZ vs. VPA, p = 0.02; CBZ vs. LTG, p = 0.002).

Conclusions: Nearly 50% of newly diagnosed patients became seizure-free on the first-ever AED, with >90% doing so at moderate or even modest dosing. Tolerability was as important as efficacy in determining overall effectiveness.

The “natural history” of newly diagnosed epilepsy in response to treatment has not been well documented. Long-term outcome studies (1) and randomized comparative trials (2–5) have suggested that <50% of patients become seizure-free with the first antiepileptic drug (AED). Failure due to lack of efficacy is associated with poor subsequent outcome (1). Effectiveness encompasses both efficacy and tolerability (6), but the importance of the latter is often inadequately assessed in randomized, particularly regulatory, studies that are not designed to address the everyday practical use of the drug (7). In patients with persistent seizures, it is unclear at what dose an AED should be deemed nonefficacious and when alternative treatment, such as with a second drug or a combination of two AEDs, should be considered. The usual recommendation of escalating the dose to near-toxic levels in patients with persisting seizures (8) assumes an appropriate dose–response relationship.

Although randomized controlled trials are essential to establish the efficacy of a new AED, whether these results can be extrapolated to clinical practice has been questioned (9). Observational studies may assist the translation of trial data into everyday use (10–12). We studied the interaction among efficacy, tolerability, and overall effectiveness in response to the first AED in patients with newly diagnosed epilepsy followed up at a single center.



The study included unselected patients in whom epilepsy was diagnosed and treatment initiated at the Epilepsy Unit in the Western Infirmary in Glasgow, Scotland between 1 January 1984 and 31 December 1997. Only those who had never received AED therapy before were included in the analysis. During the first clinic visit, a structured questionnaire was used to collect information from the patients and any witnesses to the seizure (13). Investigations, in particular surface electroencephalography and brain imaging, were carried out as clinically indicated. Epileptic seizures and syndromes were classified using information obtained from the history, physical examination, and investigations according to international guidelines (14,15). The classification for each patient represented the final decision at the time of analysis.

Treatment and follow-up

AED treatment for all patients was initiated at the epilepsy clinic, taking into consideration the types of seizures and epilepsy syndromes and other clinical characteristics (16,17). Most patients had two or more seizures before starting treatment. Twenty (4%) patients elected to start treatment after a single seizure. AEDs were introduced according to recommended dosing schedules (18,19) and adjusted during follow-up as clinical circumstances dictated, paying particular attention to efficacy and tolerability. Time was allowed at each dosage increment to establish whether the patient remained seizure-free. Doses were titrated to the limit of tolerability in patients still reporting seizures. Some patients took part in randomized, double-blind, flexible-dose, comparative monotherapy trials. Their outcomes were similar to those of the rest of the cohort, suggesting no bias in patient selection for these studies (1).

After commencement of treatment, patients were reviewed at the epilepsy clinic every 4–6 weeks for the first 6 months and at least every 4-monthly thereafter. A dedicated telephone line was set for the patients and their primary care physicians to contact the Epilepsy Unit should problems arise between scheduled clinic appointments. Compliance was monitored at the clinic with the aid of on-site measurement of serum AED concentrations (20). Patients who persistently did not comply with the treatment regimen were excluded from the study at the time of analysis. Each patient's demographic data, clinical history, investigation results, and progress on treatment were recorded in a database (13).


Response to the first AED was classified as seizure freedom; failure of treatment due to inadequate seizure control despite being able to tolerate the medication (lack of efficacy) or due to intolerable adverse events (side effects or idiosyncratic reactions); or change of treatment for reasons unrelated to efficacy or tolerability, such as concern about potential side effects, planning a pregnancy, etc. Patients were considered to be seizure-free if they had no auras or seizures of any type for a minimum of 1 year while receiving the same AED dose.

The major aim of the project was to identify clinical factors that were predictive of response to AEDs. The χ2 test and Fisher's exact test were used for comparisons of categorical data, and the Mann–Whitney test for nonparametric continuous data.

Multiple nominal logistic regression was used to examine the potential interaction between factors such as the type of epilepsy and the AED prescribed, and to assess the difference between individual drugs in terms of treatment response. All statistical tests were two-tailed. Calculations were made using Minitab for Windows (Release 11.21) software.


Patient demographics

Four hundred and seventy (51% male patients) previously untreated patients with newly diagnosed epilepsy were included in the analysis. The mean follow-up period was 5.6 years (standard deviation, 3.4 years); 91% of patients attended the clinic for ≥3 years. The mean age at referral was 35.2 years (standard deviation, 19.4 years), and the mean age at onset of epilepsy was 32.8 years (standard deviation, 19.6 years). The great majority of patients (79.4%) were in the adult age group, with 9.8% between 9 and 15 years, and the remaining 10.8% between ages 66 and 93 years. One hundred nineteen (25.3%) patients were classified as having idiopathic epilepsy, 138 (29.4%) as having symptomatic epilepsy, and 213 (45.3%) as having cryptogenic epilepsy.

Epilepsy type and drug response

Overall, 47% patients became seizure-free on the first AED. A higher proportion of patients with symptomatic or cryptogenic epilepsy came off their first drug because of intolerable side effects (17% compared with 8.5% in patients with idiopathic epilepsy; p = 0.025; relative risk, 2.00; 95% confidence interval, 1.06 to 3.78), and a lower proportion became seizure-free (43.5% compared with 58% in patients with idiopathic epilepsy; p = 0.007; relative risk, 0.75; 95% confidence interval, 0.62 to 0.91). There was no significant difference in other outcome parameters between the two patient groups (Fig. 1). Seizure-free rates between the symptomatic (43%) and cryptogenic (44%) patient groups were almost identical.

Figure 1.

Response of previously untreated patients with newly diagnosed epilepsy to the first antiepileptic drug. *p < 0.01, **p < 0.05.

The most frequently prescribed AEDs were carbamazepine (CBZ, n = 212), sodium valproate (VPA, n = 101), and lamotrigine (LTG, n = 78). The idiopathic group received lower doses of CBZ (n = 29; 469 mg/400 mg) compared with the symptomatic/cryptogenic patients (n = 183, 568 mg/600 mg; p = 0.05). Patients with idiopathic and symptomatic/cryptogenic epilepsy took similar mean/median daily doses of VPA (n = 56; 1,040 mg/1,000 mg and n = 45, 1,202 mg/1,000mg, respectively) and LTG (idiopathic: n = 20, 207.5 mg/200 mg; symptomatic/cryptogenic: n = 58, 234.9 mg/200 mg).

No significant differences in dosage were apparent between the two patient groups within each response category for all three AEDs.

Efficacy, tolerability, and dosage

Patients who became seizure-free (CBZ, p < 0.0001; VPA, p < 0.01; LTG, p < 0.001), and those who had to change treatment because of intolerable side effects (CBZ, p < 0.0001; VPA, p < 0.01) took lower AED doses than did those with inadequate seizure control (Table 1).

Table 1.  Median (interquartile range) final doses (mg/day) of first antiepileptic drug in different response groups
(n = 212)
Sodium valproate
(n = 101)
(n = 78)
  • a

     p < 0.0001 compared with patients taking carbamazepine with inadequate seizure control.

  • b

     p < 0.01 compared with patients taking sodium valproate with inadequate seizure control.

  • c

     p < 0.001 compared with patients taking lamotrigine with inadequate seizure control.

Seizure-free600 (400–600)a1,000 (800–1,000)b187.5 (150–200)c
Inadequate seizure control800 (600–800)1,500 (1,000–2,000)300 (200–400)
Intolerable side effects400 (275–600)a800 (500–1,200)b300 (150–375)
Idiosyncratic reactions200 (200)100 (50–150)
Changed for other reasons600 (400–800)1,250 (1,000–1,875)175 (150–200)
All patients600 (400–600)1,000 (825–1,500)200 (150–300)

Figures 2–4 show the range of doses of the three AEDs in seizure-free patients, the frequency of these doses, and the cumulative percentages of patients seizure free up to and including these dose levels. Among patients who became seizure-free with CBZ (n = 88), the daily dose ranged from 200 to 1,600 mg, with 93.1% taking ≤800 mg; the most common dosage range was 400–600 mg/day (Fig. 2). The daily dose of VPA among seizure-free patients (n = 58) ranged from 200 to 3,000 mg, with 91.3% receiving ≤1,500 mg; the most common range was 600–1,000 mg/day (Fig. 3). Forty-eight patients treated with LTG became seizure-free on daily doses ranging from 50 to 600 mg, with 93.8% requiring ≤300 mg. The commonest range in these patients was 125–200 mg/day (Fig. 4). Most withdrawals due to adverse events also occurred at ≤800 mg CBZ (n = 57, 98.2%), 1,500 mg VPA (n = 13, 100%), or 300 mg LTG (n = 8, 75%) per day.

Figure 2.

Daily doses of carbamazepine taken by seizure-free patients. The numbers in the boxes are percentages remaining seizure-free within that dosage range. The line represents the cumulative percentages of patients seizure-free up to and including each dose range.

Figure 3.

Daily doses of sodium valproate taken by seizure-free patients. The numbers in the boxes are percentages remaining seizure free within that dosage range. The line represents the cumulative percentages of patients seizure-free up to and including each dose range.

Figure 4.

Daily doses of lamotrigine taken by seizure-free patients. The numbers in the boxes are percentages remaining seizure- free within that dosage range. The line represents the cumulative percentages of patients seizure-free up to and including each dose range.

Comparison of antiepileptic drugs

There was no significant difference in the proportion of patients with inadequate seizure control who were able to tolerate each of the three major AEDs, suggesting similar efficacy (Table 2). However, more patients receiving CBZ changed treatment because of adverse events compared with those given VPA or LTG and, consequently, fewer became seizure-free (CBZ vs. VPA: p = 0.02; relative risk, 0.43; 95% confidence interval, 0.20 to 0.89; CBZ vs. LTG: p = 0.002; relative risk, 0.27; 95% confidence interval, 0.12 to 0.62). There was no difference in outcome between patients treated with VPA or LTG.

Table 2.  Response to the first antiepileptic drug in newly diagnosed epilepsy
 CarbamazepineSodium valproateLamotrigine
  • a

     Carbamazepine versus sodium valproate, p = 0.02; carbamazepine versus lamotrigine, p = 0.002.

Inadequate seizure control53(25)26(26)20(26)
Changed because of adverse eventsa57(27)13(13)8(10)
Withdrew for other reasons14(6.5)4(4)2(2.5)

Although the proportions of patients receiving CBZ, VPA, and LTG were different among the idiopathic and symptomatic/cryptogenic groups, nominal logistic regression analysis did not reveal significant interaction in terms of treatment outcome between epilepsy type and AED prescribed. Nevertheless, separate subanalyses were undertaken for the two patient populations. A lower proportion of patients with idiopathic epilepsy given CBZ (n = 29) became seizure free than those treated with VPA (n = 56) or LTG (n = 20) [CBZ vs. VPA: 41 vs. 64%, p = 0.04; CBZ vs. LTG: 41 vs. 80%, p = 0.007], again partly because they had higher withdrawal rates due to adverse events (CBZ vs. VPA: 28 vs. 11%, p = 0.05; CBZ vs. LTG: 28% vs. none, p = 0.01). Among patients with symptomatic/cryptogenic epilepsy, a higher proportion receiving CBZ (n = 183) changed treatment because of adverse events than those prescribed LTG (n = 58) [CBZ vs. LTG: 27 vs. 14%, p = 0.04], although the difference in seizure-free rates did not reach statistical significance (CBZ vs. LTG: 41.5 vs. 55%, p = 0.07). Among patients with symptomatic/cryptogenic epilepsy, there was no significant difference in seizure-free rates between those treated with CBZ and VPA (n = 45; 41.5 vs. 49%).

The most common adverse event necessitating drug discontinuation was rash (Table 3), which developed more often with CBZ than with VPA (p = 0.001). The differences in rash rate, however, between CBZ and LTG and between VPA and LTG were not statistically significant. Patients not tolerating CBZ often complained of more than one symptom. For VPA and LTG, the most frequent side effect leading to drug discontinuation was headache. Nonneurologic side effects were uncommon with both these AEDs (Table 3).

Table 3.  Adverse events among patients not tolerating the first antiepileptic drug
Adverse eventsCarbamazepine
(n = 212)
Sodium valproate
(n = 101)
(n = 78)
  • a

     Some patients reported more than one adverse event.

  • b

     Carbamazepine versus sodium valproate, p = 0.02; versus lamotrigine, p = 0.002.

Cognitive disturbance1(<1)1(1)0(0)
Change in mood1(<1)0(0)2(3)
Visual disturbance1(<1)1(1)0(0)
Total adverse eventsa78 15 9 
Patients withdrawnb57(27)13(13)8(10)


This study aimed to assess the response to treatment in a cohort of patients with newly diagnosed epilepsy followed up at a single center over a 15-year period. Nearly 50% became seizure-free with the first prescribed AED across a wide range of doses. Patients with symptomatic/cryptogenic epilepsy were more likely to change treatment because of side effects compared with those with idiopathic epilepsy. This reduction in the ability of patients with localization-related epilepsy to tolerate AED therapy contributed substantially to their lower seizure-free rate compared with patients with idiopathic epilepsy.

Tolerability also appeared to be a contributing factor in determining overall effectiveness of individual AEDs. VPA and LTG were better tolerated than CBZ. Comparisons between drugs in observational studies inevitably suffer from potential confounding factors, including bias in selection of AEDs for individual patients. It is possible, for instance, that CBZ was prescribed for a few patients with absence or myoclonic seizures, which it may exacerbate (18) and which might not have been apparent at initial presentation. This could have contributed to its lower seizure-control rate compared with VPA and LTG, which have a wider spectrum of activity (18,19). Similar findings, however, have been reported in randomized trials comparing CBZ with VPA and LTG, also involving previously untreated patients (3,4,21–23). CBZ was reported to be more effective than VPA in controlling partial seizures in the Veterans Administration study (2). This trial, however, had different patient and treatment characteristics, including enrollment of previously treated and mainly older men and use of high VPA doses (24). A randomized study in newly diagnosed epilepsy is required to confirm the differences reported here in overall seizure-free rates among these three AEDs.

Because AED doses are usually progressively increased in patients with uncontrolled seizures, it is not surprising that they received a higher average dose compared with the seizure-free patients. Those intolerant of AED treatment took a lower average dose than the responding patients, as the majority of withdrawals due to adverse events occurred at relatively low dosage for all three AEDs. The reasons for such marked differences in outcome in individual patients (i.e., seizure freedom or intolerability at low AED dosage) is poorly understood. This may reflect the diversity of underlying neuropathologies (25,26) and possible pharmacogenetic factors influencing drug response (27). These data also support the suggestion of two distinct populations of patients with newly diagnosed epilepsy [i.e., those responding to monotherapy and those requiring treatment with more than one AED (1)].

Examination of response to AEDs at different dose levels in newly diagnosed epilepsy has not been explored in randomized studies with flexible dosing schedules (3,4,21–23). A similar pattern emerged for CBZ, VPA, and LTG, the three most commonly prescribed AEDs in our cohort. Successful dosing in seizure-free patients appeared to be normally distributed, with the majority receiving middle-of-the-range amounts. More than 90% of seizure-free patients were taking ≤800 mg CBZ, 1,500 mg VPA, or 300 mg LTG per day. However, the commonest dosage ranges were more modest (i.e., 400–600 mg for CBZ, 600–1,000 mg for VPA, and 125–200 mg for LTG).

These observations support the widely held view that many patients with newly diagnosed epilepsy respond well to moderate doses of AED therapy (18). This perhaps explains why randomized monotherapy trials using fixed doses of AEDs in newly diagnosed epilepsy failed to demonstrate positive dose–response relationships (28,29). These data also suggest that in patients unresponsive to moderate AED doses, the chance of achieving seizure freedom with higher amounts is likely to be small. Alternative or combination therapy should, perhaps, be introduced earlier in the treatment paradigm than is currently recommended (8). Undoubtedly, other factors need to be considered before making this decision for the individual patient, including the extent of the response and the presence or absence of side effects. More prospective studies are required to determine what factors influence the likelihood of success with AED monotherapy, and at what stage combination therapy should be introduced.

In conclusion, ∼50% of patients with newly diagnosed epilepsy became seizure-free with the first prescribed AED. More than 90% did so at modest or moderate doses of CBZ (≤800 mg/day), VPA (≤1,500 mg/day), or LTG (≤300 mg/day). As has been shown in a number of randomized comparative monotherapy trials (4,21,30), tolerability was as important as efficacy in determining overall effectiveness. A better understanding of the natural history of treated epilepsy would allow more accurate assessment of the factors influencing prognosis and help formulate a strategic approach to management in patients in whom monotherapy with the first-choice AED fails.