• Carbamazepine;
  • Gabapentin;
  • Cognition;
  • Mood;
  • Senior adults


  1. Top of page
  2. Abstract

Summary:  Purpose: This study compared the cognitive effects of carbamazepine (CBZ) and gabapentin (GBP) in healthy senior adults by using a randomized, double-blind crossover design.

Methods: Thirty-four senior adults were randomized to receive one of the two drugs followed by a 5-week treatment period. A 4-week washout phase preceded initiation of the second drug. Antiepileptic drugs (AEDs) were titrated to target doses of either CBZ (800 mg/day) or GBP (2,400 mg/day). Primary outcome measures were standardized neuropsychological tests of attention/vigilance, psychomotor speed, motor speed, verbal and visual memory, and the Profile of Mood State (POMS), yielding a total of 17 variables. Each subject received cognitive testing at predrug baseline, end of first drug phase, end of second drug phase, and 4 weeks after completion of the second drug phase.

Results: Fifteen senior adults (mean age, 66.5 years; range, 59–76 years) completed the study. Seniors completing the study did not differ significantly from noncompleting seniors in terms of demographic features or baseline cognitive performances. Fifteen of the 19 seniors not completing the study dropped out while receiving CBZ. Adverse events were frequently reported for both AEDs, although they were more common for CBZ. Mean serum levels for the completers were within midrange clinical doses (CBZ, 6.8 μg/ml; GBP, 7.1 μg/ml). Significant differences between CBZ and GBP were found for only one of 11 cognitive variables, with better attention/vigilance for GBP, although the effect was modest. Performances on the nondrug average were significantly better on 45% of cognitive variables compared with CBZ and 36% compared with GBP. The overall pattern of means favored GBP over CBZ on 15 of 17 (p < 0.001), nondrug over CBZ on 17 of 17 (p < 0.0000), and nondrug over GBP on eight of 17 (NS).

Conclusions: Mild cognitive effects were found for both AEDs compared with the nondrug average condition. The magnitude of difference between the two AEDs across the cognitive variables was modest. Self-reported mood was not significantly affected by either AED. However, overall tolerability and side-effect profile of CBZ were poorer than those of GBP in senior adults at doses and titration rates reported in this study.

Epidemiologic estimates indicate that the incidence of epilepsy in senior adults has increased over the past several decades, with annual estimates of new epilepsy cases in those older than 60 years approaching 45–50,000 in the United States (1). The prevalence rate of active epilepsy in persons older than 60 years now approaches 1% (2). Treatment for senior adults with epilepsy has relied on the incorporation of studies on younger patients that do not fully account for the complex physiologic and pharmacokinetic changes that occur with aging (3). This complexity makes an already difficult treatment selection even more so, when considering the range of medication side effects and interactions that are more problematic in the elderly.

The potential for adverse cognitive effects of antiepileptic drugs (AEDs) within the older population is considerable, given age-related increases in drug sensitivity, which could exacerbate already existing declining cognitive abilities. In addition, elderly patients are likely receiving multiple medications for non–epilepsy-related medical/neurologic conditions that could complicate already drug-sensitive biologic systems. Developing and using AEDs that minimize the risk to cognitive function in senior adults with epilepsy is an important issue that has not been sufficiently examined, especially in light of trends suggesting that the number of new cases of epilepsy in older adults is increasing because of declining mortality and general aging of the American population.

To date, the potential cognitive effect of AEDs in senior epilepsy populations has received limited attention. However, one study specifically examined the comparative cognitive effects of phenytoin (PHT) and valproate (VPA) monotherapy in groups of elderly-onset seizure patients (4). Patients with new-onset partial seizures or generalized tonic–clonic seizures were recruited from their community physicians into the study. They found only minor changes in cognitive test performance over the course of 1 year taking either AED compared with predrug baseline (4). Additionally, minimal differences were found between the two AEDs with regard to impact on cognitive function. However, they used a parallel-group design that tends to reduce the statistical power compared with a randomized crossover design. The two groups of seniors were, however, randomized to AED and were matched across several demographic variables.

To gain greater understanding of the potential anticonvulsant effects on cognition, healthy adult populations should be used because seizure effects, underlying cerebral injury, and concurrent medical factors can confound evaluating the cognitive effects of AEDs with epilepsy. Meador et al. (5) recently presented data demonstrating mild untoward effects of three common first-line AEDs in a group of young healthy adults with a randomized, double-blinded, incomplete-block crossover design. More than half of the neurocognitive variables exhibited negative AED effects when compared with nondrug baselines.

For the older patient with epilepsy, gabapentin (GBP) may offer an alternative to the more traditional AEDs that are known to have significant side-effect profiles in the elderly (6). GBP has demonstrated few sedative and cognitive effects in the young epilepsy population (7). One recent study found that senior adults patients with epilepsy taking an average GBP dose of 2,700 mg/day reported few adverse events (AEs), with three patients reporting fatigue, and two reporting unsteadiness (8). Unfortunately, no study to date has explored the potential cognitive effects of GBP in either senior adults with epilepsy or in healthy senior adults.

At present, carbamazepine (CBZ) is considered a standard drug of choice for seniors with epilepsy and ranks next to PHT in prescribed popularity (9). However, AEs do occur with CBZ, including cognitive effects. Recently, data from healthy young adults demonstrated mild negative effects of CBZ on cognitive function compared with baseline conditions when administered at common clinical dosage (10). Sparse effort has been directed to the study of the potential cognitive effects of CBZ in senior adults despite its widespread clinical use for chronic seizures and other conditions. Read et al. (11) recently found no significant changes in various cognitive task performances in senior adults who took an extra dose of their AED medication (either CBZ, VPA, or PHT). They concluded that cognitive impairment does not develop when modest dose increases are given in patients taking therapeutic monotherapy AED.

Our study compared the potential cognitive and behavioral effects of CBZ and GBP in a double-blind, randomized crossover design. The use of independent community-dwelling healthy senior adults negates the impact of the epileptic state and allows generalization to populations who may use these AEDs for non–seizure-related treatment.


  1. Top of page
  2. Abstract


The subjects were healthy senior paid volunteers who participated in the study after University IRB approved informed consent. Subjects were recruited from a university-sponsored senior focus newsletter and from a large citywide newspaper. All volunteers were between the ages of 60 and 80 years. Exclusion criteria consisted of history of any type of seizure, active renal or hepatic disease, the use of any concomitant medications known to affect GBP or CBZ, use of psychoactive medications, history of drug or alcohol abuse, history of clinical finding of a progressive encephalopathy including central nervous system (CNS) tumors. Subjects who reported a serious illness in the past 3 months also were excluded.

A total of 34 subjects were enrolled. Fifteen subjects completed the entire study, whereas 18 dropped out because of medication side effects (15 receiving CBZ and three receiving GBP). The number of subjects dropping out of the study while receiving CBZ compared with GBP was statistically significant (nonparametric Sign Test, p < 0.004). One patient dropped out after completion of phase I, during drug-washout phase, (received GBP during first drug phase) because of non–drug study-related medical condition that was identified during course of study. Of the 15 subjects discontinuing the study while taking CBZ, 12 dropped with CBZ as the first drug taken. Of the three subjects withdrawing from the study while taking GBP, two dropped with GBP as the first drug. Average number of days taking CBZ before study discontinuation was 9 days (range, 4–17 days), whereas for GBP, average number of days was 36. Two subjects completed the entire GBP phase but refused to complete phase II because of side effects experienced during the phase I period. One subject withdrew after 8 days taking GBP.

The 15 subjects completing the study were between the ages of 60 and 76 years (mean age, 66.5 years). Nine subjects were men, and six were women. All subjects had at least a high school education, and 73% of the sample had ≥1 year of college education. Ten subjects were retired, and five continued to work. Ten subjects were married, three were divorced, and two were widowed. Study completers and noncompleters were not statistically different across demographic characteristics or baseline cognitive test performance. Each of the 15 subjects completing the study reached GBP target dose, although only 12 of those same 15 subjects reached the CBZ target dose. Three subjects reached the end dose of 600 mg/day of CBZ. Mean CBZ serum concentration at the time of cognitive testing for the study completers was 6.8 μg/ml (range, 3.9–8.5 μg/ml), and mean GBP serum concentration for the same subjects was 7.1 μg/ml (range, 5.5–9.7 μg/ml). AE incidence rates are reported and summarized separately for study completers and for study dropouts in Table 1. Subjects reported several types of AEs under both drug conditions, although more so when receiving CBZ. Of the 23 AEs reported, incidence of AEs was equal in two and worse in 16 of remaining 21 AEs (sign test, p < 0.01) with CBZ compared with GBP. Fatigue/tiredness, unsteadiness, concentration/memory, dizziness, and weight gain were reported by more than one third of subjects when taking CBZ. Subjects receiving GBP reported similar, but less frequent complaints.

Table 1.  Incidence of adverse effects for carbamazapine and gabapentin in healthy senior completers and noncompleters
 Completers (n = 15)Noncompleters (n = 19)
CBZGBPCBZ (n = 15)GBP (n = 4)
  1. CBZ, carbamazepine; GBP, gabapentin; CNS, central nervous system.

CNS effects    
 Fatigue/Tired11 (73%)4 (33%)6 (40%)1 (33%)
 Unsteadiness7 (47%)6 (40%)8 (53%)1 (33%)
 Concentration/Memory7 (47%)2 (13%)3 (20%)0 (0%)
 Dizziness6 (40%)3 (20%)2 (13%)1 (33%)
 Nervousness3 (20%)1 (7%)0 (0%)0 (0%)
 Restlessness3 (20%)1 (7%)1 (7%)0 (0%)
 Insomnia3 (20%)1 (7%)0 (0%)0 (0%)
 Tremor2 (13%)0 (0%)2 (13%)0 (0%)
 Headaches2 (13%)2 (13%)1 (7%)0 (0%)
 Visual disturbance2 (13%)0 (0%)1 (7%)1 (33%)
 Depression1 (7%)0 (0%)1 (7%)0 (0%)
 Decreased libido1 (7%)1 (7%)1 (7%)0 (0%)
 Irritability1 (7%)2 (13%)1 (7%)0 (0%)
 Confusion1 (7%)0 (0%)0 (0%)0 (0%)
 Increased energy0 (0%)1 (7%)0 (0%)0 (0%)
Systemic effects    
 Weight gain6 (40%)2 (13%)2 (13%)1 (33%)
 Skin irritation2 (7%)2 (13%)3 (20%)0 (0%)
 Sore mouth/gums1 (7%)3 (20%)1 (7%)0 (0%)
 Constipation1 (7%)0 (0%)0 (0%)0 (0%)
 Diarrhea1 (7%)0 (0%)1 (7%)0 (0%)
 Increased appetite1 (7%)0 (0%)1 (7%)0 (0%)
 Dry mouth0 (0%)1 (7%)0 (0%)0 (0%)
 Difficulty urinating0 (0%)1 (7%)0 (0%)0 (0%)
Total incidence6233345

Before drug study baseline assessment, all subjects were administered a neuropsychological screening battery that consisted of the Immediate and Delayed Recall conditions of the Logical Memory subscale of the Wechsler Memory Scale–Revised (12), Boston Naming Test (13), Controlled Oral Word Association Test (14), and Animal Naming test (15). Subjects were excluded from the study if performance was greater than two standard deviations below normative values for three or more of the screening tests. Two individuals did not meet minimal inclusion criteria based on their screening test results and were not included in the study.

Neuropsychological tests

Subjects were administered several cognitive tests comprising measures of psychomotor speed, attention/concentration, memory, motor speed, and self-reported mood. This composition of tests has demonstrated sensitivity to AED effects in prior studies (5,10). Test descriptions have been detailed in previous work (5,16,17).

Choice Reaction Time (CRT) (5)

Total averaged (milliseconds) response time across 25 trials was calculated.

Lafayette Grooved Pegboard (15)

Total time (seconds) to place grooved pegs into slotted holes. Dominant-hand time was used.

Symbol Digit Modalities Test (SDMT) (18)

This is a complex graphomotor task that requires the subject to transcribe symbols to numbers as quickly as possible with a combination of direct visual identification and/or short-term memorization. Symbols with empty squares are presented, and the subject's task is to fill in the corresponding number as quickly as possible during a 90-s interval.

Selective Reminding Test (SRT) (19)

This is a verbal serial word list learning task consisting of six trials of 12 words with parallel forms scored for Continuous Long-Term Retrieval (CLTR). Subjects are told a list of words for immediate recall. On all subsequent stimulus presentations, subjects are told only those words they omitted on the previous trial. This procedure continues until the subject recalls all words or to the sixth trial.

MCG Complex Figures (5)

This is a task requiring subjects to copy a complex geometric design, similar to the Rey Complex Figure, and then recall that figure in immediate and delayed conditions. Parallel forms were used. Each figure has 36 scoreable units.

MCG Paragraph Memory (5)

This task consists of subjects being read a story and requested to recall the story from memory immediately and then again after a second reading of the story. Subjects are then requested to recall the story content after a 30-min delay. Parallel forms were used for different test days.

Visual Serial Addition Test (VSAT)

A visual analogue of the Paced Auditory Serial Addition Task (PASAT) (20). A total error across four trials was used.

Profile of Mood States (POMS) (21)

This is a checklist of adjectives describing mood states. The test is scored for six scales (Fatigue, Vigor, Depression–Dejection, Anger–Hostility, Tension–Anxiety, Confusion–Bewilderment). All scales but the Vigor scale are scored such that higher scores reflect greater mood disturbance; for the Vigor scale, higher scores reflect less mood disturbance.


This study used a randomized, double-blind, crossover design. Every senior volunteer received each AED in monotherapy for a total of 6 weeks that included a 7- to 14-day titration period. Each treatment period was followed by a 1-week taper period and a 4-week washout period. Subjects were allowed to delay an increase for up to 3 days if side effects were present. A dose <800 mg/day of CBZ was accepted if patients could not titrate to higher doses. A study physician monitored each subject during AED treatment periods and made adjustments in AED dosages with the goal of reaching therapeutic doses of 2,400 mg/day of GBP and 800 mg/day of CBZ. Initial CBZ dose was 200 mg/day for the first 2 days, then increasing by 200 mg/day dosing on day 3, day 5, and then on day 7. Feasibility and pharmacokinetic studies of similar oral loading of CBZ in young adults with epilepsy has resulted in minimal AEs unless accompanied by history of static encephalopathy (22–24). Initial GBP dose was set at 800 mg/day till day 4, at which time dosing increased to 1,600 mg/day and finally to 2,400 mg/day by day 7.

Subjects underwent cognitive and behavioral testing at screening, at the end of each AED treatment, and at the end of the final drug washout. Cognitive testing was performed in a predetermined order, at the same time of day. Testing was completed ∼3 h after subject's morning dose. All study personnel in direct contact with subjects were blinded to AED treatment. A nonblinded M.D., pharmacist, or registered nurse (under supervision of a nonblinded M.D.) filled the drug trays.

Seniors were given medication in prepackaged sheets with matched capsules containing CBZ, 200-mg, or GBP, 400-mg doses. A constant number of capsules were given to subjects across the AED treatment phase for each day (active drug plus placebo). Subjects were examined at 2-week intervals, and blood was drawn for hematology, liver functions, and anticonvulsant blood levels (ABLs) ∼2 h after morning dose.

Statistical analyses

We performed three-level within-subjects repeated-measure analyses of variance (ANOVAs) comparing the two AED conditions and the average of the two nondrug conditions across each of the 17 variables. If significant effects were found, we then followed up with t test post hoc comparisons. We used Bonferroni correction (set at p < 0.017), given the large number of post hoc comparisons.


  1. Top of page
  2. Abstract

Cognitive test results

The within-subjects repeated-measures ANOVAs and post hoc comparisons are presented in Table 2. Of the 11 cognitive variables, six (55%) of 11 revealed statistically significant differences across the three conditions. Post hoc analyses revealed that the nondrug average condition was better than CBZ on five (45%) of 11 variables and better than GBP on four (36%) of 11 variables. Only one cognitive variable was performed more poorly with CBZ compared with GBP. The means and standard deviations are listed in Table 3. Examination of the difference scores between the three conditions revealed that performance for the VSAT was <0.5 SD between the two AEDs. However, larger magnitude differences on the order of 0.5 to 1 SD unit were found when examining the absolute differences comparing the nondrug average baseline score with the two AED conditions.

Table 2.  Statistical results with significant ANOVAs
Neuropsychologic variablesANOVAsSignificant t tests
  1. Significant analyses of variance (ANOVAs) and indicated follow-ups are depicted in bold. None of the follow-up post hoc tests on the POMS fatigue scale were significant with Bonferroni correction.

  2. GBP, gabapentin; CBZ, carbamazepine; ND, nondrug conditions (averaged baseline and washout period); CLTR, continuous long-term retrieval; POMS, Profile of Mood State.

 Visual Serial Addition TestF(2,13) = 5.52, p ≤ 0.01GBP  
Cognitive/motor speed    
 Choice Reaction TimeF(2,13) = 1.93, NS   
 Grooved Peg BoardF(2,13) = 5.04, p ≤ 0.01 NDND
 Complex Figure: copyF(2,13) = 2.00, NS   
 Complex Figure: immediateF(2,13) = 0.78, NS   
 Complex Figure: delayedF(2,13) = 0.91, NS   
 Selective Reminding TestF(2,13) = 2.6, NS   
 Paragraph: Immediate 1F(2,13) = 9.81, p ≤ 0.001 NDND
 Paragraph: Immediate 2F(2,13) = 9.10, p ≤ 0.001 NDND
 Paragraph: DelayF(2,13) = 12.73, p ≤ 0.001 NDND
Other cognitive measures    
 Symbol Digit Modalities TestF(2,13) = 3.61, p ≤ 0.04 ND 
Subjective mood measures    
 POMS: angerF(2,13) = 0.78, NS   
 POMS: confusionF(2,13) = 0.63, NS   
 POMS: depressionF(2,13) = 1.1, NS   
 POMS: fatigueF(2,13) = 4.6, p ≤ 0.02   
 POMS: vigorF(2,13) = 2.1, NS   
 POMS: anx-tensionF(2,13) = 2.2, NS   
Table 3.  Group mean and standard deviation values across cognitive and self-report measures at each drug condition for completers (n = 15)
 Averaged non-AED baselineCarbamaz- epineGabapentin
  • CRT, Choice Reaction Time; SDMT, Symbol Digits Modalities Test; VSAT, Visual Serial Addition Test; MCG Figures, Medical College of Georgia Complex Figures Immediate Recall Score; SRT–CLTR, Selective Reminding Test–Continuous Long-Term Retrieval Score; AED, antiepileptic drug.

  • a

     Higher mean values indicate better performance for all variables except those denoted.

 VSATa55.0 (37.2)61.4 (38.1)47.8 (36.6)
Cognitive/motor speed   
 CRTa865.5 (149)918.9 (224)871.4 (195)
 Grooved Pegboarda75.8 (7.9)83.6 (10.8)82.0 (10.8)
 MCG Figure: Copy35.3 (.8)34.3 (2.4)34.3 (1.8)
 MCG Figure: imm23.4 (6.1)22.8 (6.8)24.2 (4.7)
 MCG Figure: delay22.5 (6.8)21.7 (6.1)23.2 (5.0)
 SRT–CLTR25.7 (9.5)20.6 (9.2)23.5 (11.2)
 Paragraphs: imm I19.0 (8.3)14.2 (6.4)12.7 (7.6)
 Paragraphs: imm II32.9 (11.9)25.4 (11.6)26.8 (12.3)
 Paragraphs: delay30.5 (11.8)21.5 (10.6)23.7 (12.4)
Other cognitive measures   
 SDMT38.0 (7.1)33.6 (8.2)36.7 (11.2)
Profile of Mood States   
 Angera4.6 (5.2)4.7 (5.3)3.7 (3.9)
 Confusiona7.5 (4.8)8.4 (5.1)7.4 (4.5)
 Depressiona6.3 (8.5)7.1 (7.2)5.2 (5.5)
 Fatiguea6.0 (5.7)9.8 (7.3)5.9 (5.6)
 Tension–Anxietya12.8 (4.7)13.7 (6.0)11.1 (4.2)
 Vigor18.3 (5.8)15.7 (8.1)19.3 (5.7)

Subjective self-report

Profile of mood state

Within-subject repeated-measures ANOVAs revealed only one statistically significant effect for the Fatigue subscale (p < 0.02). However, none of the post hoc comparisons reached the a priori level of significance.

Nonparametric analyses

Because of the large subject-dropout rate and small sample size, a concern for type II errors is raised. Therefore, nonparametric sign test analyses were performed to examine the consistency of the findings. We compared the means across the variables to determine whether the number favoring one condition over another was significant (see Table 4). Overall, 15 of 17 means favored GBP over CBZ (sign test, p < 0.001), whereas all 17 means favored the nondrug average over CBZ (p < 0.0000). In contrast, the nondrug average was favored over GBP on only eight of 17 means (p = NS). Of note, all six of the POMS means favored GBP over CBZ (p < 0.016), and all six means favored the nondrug average over CBZ (p < 0.016; see Table 4). In contrast, the means favored GBP over the nondrug average on all six means (p < 0.016).

Table 4.  Statistical results of nonparametric analyses of the cognitive tests and mood ratings
 CBZ favoredGBP favoredNondrug favoredp Value
  1. In the CBZ versus GBP comparisons, one cognitive measure was equal between the two groups.

  2. CBZ, carbamazepine; GBP, gabapentin.

CBZ vs. GBP    
 Cognitive tests1/119/110.01
 Mood ratings0/66/60.016
Nondrug vs. CBZ    
 Cognitive tests0/1111/110.0005
 Mood ratings0/66/60.016
Nondrug vs. GBP    
 Cognitive tests3/118/11NS
 Mood ratings6/60/60.016


  1. Top of page
  2. Abstract

Despite the prevalence of epilepsy in the elderly, little is known about the potential cognitive and behavioral effects of commonly used AEDs in this population. This study enlisted a group of healthy senior adults exposed to GBP and CBZ. This study used a healthy senior population to eliminate the confound of seizures on cognitive performance and to allow extrapolation of results to other conditions treated with these AEDs. The central aim of this study was to examine differential cognitive effects of GBP and CBZ. Previous attention to potential neuropsychological effects of AEDs in seniors has been limited and has compared only more traditional AEDs (4,11). This study demonstrated that for the 11 cognitive variables, only a single measure of attention was statistically different between the two AEDs. Although mean variable comparison (sign test) found that performance with GBP was better than that with CBZ on nine of 11 variables, the magnitude of difference between the AEDs was modest (typically <0.5 SD).

A secondary aim of this study was to examine AED effects compared with a nondrug condition. Untoward cognitive effects were found for both GBP and CBZ compared with the average nondrug condition. Both AEDs produced statistically worse performance on tests of verbal memory and motor speed compared with the nondrug average condition. Seniors experienced more difficulty recalling prose passages while receiving either AED compared with nondrug baseline, even with the practice of a second learning trial. This effect persisted over the delayed-recall condition. These drug effects ranged in magnitude from 0.5 to 1 SD unit difference from nondrug average baseline performances. Overall, AEDs negatively affected performances ranging from 9 to 30% worse compared with the nondrug average condition. No statistical differences were noted favoring either AED over the nondrug average condition across any of the cognitive tasks. These effect sizes appeared similar in magnitude to test–retest effects reported in a study of cognitive effects of either VPA or PHT monotherapy in seniors with epilepsy (4). Additionally, when the overall pattern of AED versus average nondrug condition was examined, all 17 variables (cognitive tests and mood ratings) favored the average nondrug condition over CBZ, whereas eight of the 17 variable means favored the nondrug average condition over GBP. However, it should be noted that this study did not have a placebo condition, so subjects were always aware when they were receiving an AED.

These results have some consistency with recent comparative AED studies using healthy young adults (5,10). Healthy senior adults, like young adults, do experience drug effects that affect some areas of cognitive functioning. AEDs such as phenobarbital (PB) and, to a lesser degree, PHT, VPA, and CBZ, produce modest cognitive effects in younger adults. Likewise, standardized cognitive tasks measuring attention/vigilance, memory, and cognitive/motor speed appear sensitive to the effects of therapeutic range dosage of these older AEDs in senior adults.

Previous research has identified positive mood effects and subjective ratings of improved well-being in seizure patients after unblinded treatment with GBP (25,26). Other studies using low-dose GBP for augmentative treatment for bipolar and schizoaffective disorders have found significant mood-stabilizing effects (27,28). However, the present study found no statistically significant differences between the two AEDs across any of the six POMS subscales. However, when examined for overall distribution of the mean drug differences, the POMS subscales consistently favored GBP over both CBZ and the nondrug average condition across all subscales. The statistical probability of finding such consistency is quite low, although like that of the cognitive variable comparisons, the magnitude of effect was modest. However, the results suggest that GBP may have a mild mood-enhancing effect consistent with findings from previous research. The mild positive mood finding for GBP was somewhat surprising, in that our senior adults were experiencing normal mood to begin with, whereas more dramatic mood effects from GBP may be more likely in patient groups who are more likely to be depressed. The robustness of these findings awaits larger sample sizes and use in elderly seizure patients.

In this study, subjects receiving either AED frequently experienced AEs. Seniors reported a high incidence of sedative effects across both drugs that tended to occur early in the titration phase. Senior adults are known to have increased susceptibility to the effects of AEDs for both pharmacokinetic and pharmacodynamic reasons (29). Side effects were the critical factor to the study dropout rate. Potentially dangerous symptoms of unsteadiness and dizziness were commonly reported, especially when receiving CBZ. Interestingly, although AEs were reported with GBP, these did not lead to study discontinuation as compared with CBZ. Arguably, slower titration schedules and lower end-state dose levels for this study may have reduced the incidence of side effects. In this study, titration times and end-dose points were both designed to reflect recent experimental findings (22,23). The titration rate used in our study may not be typical of the clinical situation for management of seizures in the elderly and therefore may have contributed to the high study dropout. However, it should be pointed out that AED tolerability, in a recent multicenter study of CBZ and lamotrigine (LTG) efficacy in elderly epilepsy patients with slower titration rates, was similar to that in our study (30). In that study, the titration rate for CBZ was very slow, occurring over a 6-week period. The final median dose was 400 mg/day (range, 200–800 mg/day) with median ABL of 6.9 mg/l. However, even with the slower titration in that study, the dropout rate for CBZ at 70 days was 40% and ∼55% at the end of 168 days. In comparison, the 70-day dropout rate for the present study was 56%.

A major limitation of this study was the small sample size of completers, only 15 subjects. Despite the double-blinded crossover design that has subjects serving as their own controls, the sample size may have limited the power necessary to minimize a type II error. However, with a double-blinded crossover design, subjects served as their own controls, attenuating potential sample-selection bias that may be found in parallel-group designs. The impact of practice effect is acknowledged, but scores for the non-AED conditions were averaged to lessen any impact from repeated testing exposures. Such repeated measurement of nondrug average performance has been used successfully in previous AED cognitive studies (5). Despite the small sample, statistical differences did emerge when comparing the nondrug average condition with the AED conditions. However, with the exception of one cognitive variable, no drug–drug comparison was statistically different. When we applied nonparametric pattern comparisons to the data, CBZ resulted in a consistently less favorable pattern of cognitive performance and self-report mood compared with the non-AED condition (95% of the comparisons) and with GBP (88% of the comparisons). However, again emphasis must be placed on the relatively small magnitude of between-AED differences. Although these differences could be clinically significant in some patients, it should be emphasized that potential cognitive effects are only one of many factors in determining AED selection for epilepsy in the individual patient.

In terms of potential clinical relevance of the present findings, much is known about the impact of aging on drug metabolism and increased risk of drug AEs (31,32). The increased susceptibility to mental impairments related to age-associated diseases (i.e., stroke and dementia) makes the potential for adverse events related to AEDs even more important (29). For example, rehabilitative efforts in the dementia or stroke patient may be further complicated by exacerbation of cognitive difficulties from AED treatment. In addition, AEDs are achieving a wider use for non–epilepsy-related conditions such as neuropathic pain (33). The present study found that the magnitude of cognitive and behavioral effects for the two AEDs was modest for most tasks (i.e., mean performance change, ∼0.25 to 0.5 SD from baseline), but more significant for the verbal memory and motor speed tasks (0.5–1 SD from baseline). Such findings have potential clinical relevance because CBZ is one of the most prescribed AEDs in the United States and is the first- or second-choice prescribed AED for seizures in the elderly (9). CBZ is also commonly prescribed for the treatment of non–epilepsy-related psychiatric conditions, particularly bipolar affective disorder (34). Both 800 mg/day of CBZ and 2,400 mg/day of GBP are typical end-dose points in several controlled trial studies on AED seizure efficacy (35,36). However, future studies will be needed to clarify issues related to the appropriate titration rate of AEDs in senior adults to minimize AEs, as well as whether long-term cognitive habituation is affected by longer-term use of AEDs.

Acknowledgment: This study was supported by a grant from Parke-Davis. The results were presented in part at the 1999 annual meeting of the American Epilepsy Society in Orlando, FL.


  1. Top of page
  2. Abstract
  • 1
    Hauser WA. Epidemiology of seizures in the elderly. In: RowanAJ, RamsayRE, eds. Seizures and epilepsy in the elderly. Newton, MA: Butterworth-Heinemann, 1997: 720.
  • 2
    Hauser WA, Annegers JK, Kurland LT. The incidence of epilepsy in Rochester, Minnesota, 1935-1984. Epilepsia 1993;34: 4539.
  • 3
    Willmore LJ. Management of epilepsy in the elderly. Neurol Clin 1996;11: 787804.
  • 4
    Craig I & Tallis R. Impact of valproate and phenytoin on cognitive function in elderly patients: results of a single-blind randomized comparative study. Epilepsia 1994;35: 38190.
  • 5
    Meador KJ, Loring DW, Moore EE, et al. Comparative cognitive effects of phenobarbital, phenytoin, and valproate in healthy adults. Neurology 1995;45: 14949.
  • 6
    Reynolds EJ & Trimble MR. Adverse neuropsychiatric effects of antiepileptic drugs. Drugs 1985;29: 57081.
  • 7
    Leach JP, Girvan J, Paul A, et al. Gabapentin and cognition: a double-blind, dose-ranging, placebo-controlled study in refractory epilepsy. J Neurol Neurosurg Psychiatry 1997;62: 3726.
  • 8
    Morris GL. Use of gabapentin in patients over 50 years of age [Abstract]. J Epilepsy 1996;9: 144.DOI: 10.1016/0896-6974(96)85620-9
  • 9
    Faught E. Epidemiology and drug treatment of epilepsy in elderly people. Drugs Aging 1999;15: 25569.
  • 10
    Meador KJ, Loring DW, Ray PG, et al. Differential cognitive effects of carbamazepine and gabapentin. Epilepsia 1999;40: 127985.
  • 11
    Read CL, Stephen LJ, Stolarek IH, et al. Cognitive effects of anticonvulsant monotherapy in elderly patients: a placebo-controlled study. Seizure 1998;7: 15962.
  • 12
    Wechsler D. Wechsler Memory Scale–Revised. New York: Psychological Corporation, 1987.
  • 13
    Kaplan E, Goodglass H, Weintraub S. The Boston Naming Test. 2nd ed. Philadelphia: Lea & Febiger, 1983.
  • 14
    Benton AL & Hamsher DdeS. Multilingual aphasia examination. Iowa City, IA: AJA Associates, 1983.
  • 15
    Lezak M. Neuropsychological assessment. 2nd ed. New York: Oxford University Press, 1995.
  • 16
    Martin R, Kuzniecky R, Ho S, et al. Cognitive effects of topiramate, gabapentin, and lamotrigine in healthy young adults. Neurology 1999;52: 3217.
  • 17
    Meador KJ, Loring DW, Allen ME, et al. Comparative cognitive effects of carbamazepine and phenytoin in healthy adults. Neurology 1991;41: 153740.
  • 18
    Smith A. Symbol digit modalities test manual. Los Angeles: Western Psychological Services, 1973.
  • 19
    Buschke H & Fuld PA. Evaluation storage, retention, and retrieval in disordered memory and learning. Neurology 1974;24: 101925.
  • 20
    Gronwall D & Wrightson P. Delayed recovery of intellectual function after minor head injury. Lancet 1974;2: 20710.
  • 21
    McNair DM, Lorr, M, Droppleman LF. Manual: profile of mood states-revised. San Diego, CA: Educational and Industrial Testing Service, 1992.
  • 22
    Cohen H, Howland MA, Luciano DJ, et al. Feasibility and pharmacokinetics of carbamazepine oral loading doses. Am J Health Syst Pharm 1998;55: 113440.
  • 23
    Kanner AM, Bourgeois BF, Hasegawa H, et al. Rapid switchover to carbamazepine using pharmacokinetic parameters. Epilepsia 1998;39: 194200.
  • 24
    Pullianen V & Jokelainen M. Effects of phenytoin and carbamazepine on cognitive functioning. Acta Neurol Scand 1994;89: 816.
  • 25
    Harden CL, Lazar LM, Pick LH, et al. A beneficial effect on mood in partial epilepsy patients treated with gabapentin. Epilepsia 1999;40: 112934.
  • 26
    Handforth A & Treiman DM. Efficacy and tolerance of long-term, high-dose gabapentin: additional observations. Epilepsia 1994;35: 10327.
  • 27
    Bennett J, Goldman WT, Suppes T. Gabapentin in the treatment of bipolar and schizoaffective disorders. J Clin Psychopharmacol 1997;17: 1412.DOI: 10.1006/geno.1993.1295
  • 28
    Schaffer CB & Schaffer LC. Gabapentin for the treatment of bipolar disorder [Letter]. Am J Psychiatry 1997;154: 291.
  • 29
    Meador KJ. Cognitive effects of epilepsy and of antiepileptic medications. In: WyllieE, ed. The treatment of epilepsy: principles and practice. 2nd ed. Baltimore: Williams & Wilkins, 1996: 112130.
  • 30
    Brodie MJ, Overstall PW, Giorgi L. Multicentre, double-blind, randomized comparison between lamotrigine and carbamazepine in elderly patients with newly diagnosed epilepsy. Epilepsy Res 1999;37: 817.DOI: 10.1016/s0920-1211(99)00039-x
  • 31
    United States General Accounting Office. Prescription drugs and the elderly. GAO/HEHS-95-152, 1995.
  • 32
    Cameron H & Macphee GJA. Anticonvulsant therapy in the elderly: a need for placebo controlled trials. Epilepsy Res 1995;21: 14957.DOI: 10.1016/0920-1211(95)00016-4
  • 33
    Mattson RH. New antiepileptic drugs: from discovery to practical use. Epilepsia 1996;(suppl 6);37.
  • 34
    Elphick, M. Clinical issues in the use of carbamazepine in psychiatry: a review. Psychol Med 1989;19: 591604.
  • 35
    Beydoun A, Fischer J, Labor DR, et al. Gabapentin monotherapy II: a 26-week, double-blind, dose-controlled, multicenter study of conversion from polytherapy in outpatients with refractory complex partial or secondarily generalized seizures. Neurology 1997;49: 74652.
  • 36
    Radulovic LL, Wilder BJ, Leppik IE, et al. Lack of interaction of gabapentin with carbamazepine or valproate. Epilepsia 1994;35: 15561.