Do patients with psychogenic nonepileptic seizures produce trustworthy findings on neuropsychological tests?
Address correspondence to Carl B. Dodrill, Ph.D., 4488 West Mercer Way, Mercer Island, WA 98040-3934, U.S.A. E-mail: firstname.lastname@example.org
Drane et al. (2006) has recently raised the possibility that patients with psychogenic nonepileptic seizures (PNES) may make poor effort in taking neuropsychological tests in comparison with patients with epilepsy (ES). Therefore, findings previously reported with PNES patients may be in error, especially with regard to tests of mental abilities. Using the same measure of effort used by Drane et al. (2006) but with more broadly selected patients, this paper attempts to replicate their findings with new samples of ES (n = 65) and PNES (n = 32) cases. However, their findings could not be replicated, and no differences in test taking effort could be demonstrated across the groups. The highly selected samples of Drane et al. (2006) appear to be responsible for their results, and neuropsychological findings with PNES patients appear to be as trustworthy as those with ES patients.
Approximately equivalent neuropsychological impairment in epilepsy (ES) and psychogenic nonepileptic seizures (PNES) has been reported for many years (Wilkus & Dodrill, 1984; Drake et al., 1993; Hermann, 1993). However, Drane et al. (2006) recently raised the possibility that PNES patients may often make an inadequate effort in taking the tests so that their findings may not be trustworthy. They used the Word Memory Test (WMT; Green et al., 1996) as a measure of effort in highly selected epileptic (ES) and PNES patient samples. In particular, they eliminated patients for all of the following reasons: (1) previous epilepsy surgery; (2) primary language other than English; (3) recent epileptic seizure; (4) inability to live independently; (5) obvious cognitive deficits; (6) incomplete neuropsychological testing; and (7) not seen by a full-time neuropsychologist. With these highly selected patient samples, only 8% of ES cases “failed” the WMT (showed inadequate effort) while 49% of the PNES cases failed. The neuropsychological test results of many PNES patients were not believed trustworthy and previously published findings with PNES patients were called into question.
The objective of the present study was to replicate the findings of Drane et al. (2006) using the same measure of effort but new and larger patient groups.
The 97 adults included in this study all received video-electroencephalographic (EEG) monitoring at the same facility where the patients studied by Drane et al. (2006) were monitored. They were selected to represent a much broader sample, however, and all adults (age 16 and over) without a history of epilepsy surgery were considered for inclusion who had received their EEG monitorings during a consecutive 42-month period. Every patient either had English as their language of origin or had been in the United States since childhood and all were proficient in English. Each adult (age 16 or over) had clear EEG findings of either ES only (n = 65) or PNES only (n = 32) with all other cases put aside. Each ES patient had at least one unequivocal seizure recorded during EEG monitoring but no PNES, and the reverse was true for PNES cases. None had been in the study by Drane et al. (2006). Descriptive information regarding these groups is found in Table 1. In this table, “Neurological history” reflects the report of the attending physician as to whether or not there was a specific neurological event in the patient's history (e.g., birth injury, definite head injury, infectious disorder, stroke, brain tumor, etc.).
Table 1. Descriptive information regarding the subject groups
|Years of age (S.D.)||35.22 (12.96)||42.25 (11.59)||Student's t, p = 0.011|
|Age at seizure onset (S.D.)||16.34 (12.40)||35.00 (13.27)||Student's t, p < 0.001|
|Years of education (S.D.)||12.05 (3.02)||12.70 (2.37)||Student's t, p = 0.305|
|Gender (F/M)||34/31||19/13||Fisher's exact, p = 0.525|
|Handedness (R/L)||56/9||31/1||Fisher's exact, p = 0.158|
|Seizures/month—average (S.D.)||38.19 (72.61)||38.81 (40.10)||Student t, p = 0.966|
|Number of AEDs (S.D.)||1.98 (0.72)||1.52 (0.73)||Student t, p = 0.010|
|Neurological history, Y/N||27/35 (44%)||13/19 (41%)||Fisher's exact, p = 0.829|
|Family history of epilepsy, Y/N||14/43 (25%)||10/22 (31%)||Fisher's exact, p = 0.619|
|Focal neurological findings on examination, Y/N||16/49 (25%)||6/26 (19%)||Fisher's exact, p = 0.612|
|Psychiatric history, Y/N||20/42 (32%)||23/9 (72%)||Fisher's exact, p = 0.0004|
In order not to restrict the sample of patients due to a lack of testing facilities, testing time, or any other reason, only the WMT was required for study inclusion. Other tests were given when possible including the Wechsler Adult Intelligence Scale-III (WAIS-III) and tests from the Neuropsychological Battery for Epilepsy (Dodrill, 1978). No fixed rule was followed regarding a certain period of time that must have elapsed from a seizure until testing could be accomplished or resumed. However, no patient was tested when it was believed that the patient was suffering the effects of a recent epileptic seizure.
Each patient received the oral form of the WMT as described by Drane et al. (2006), and exactly the same criteria for passing and failing were used. This test calls for the learning of 20 logically related word pairs. Recognition memory is tested immediately and also after 30 min. Consistency of response from the immediate trial to the delayed trial is also computed. Passing the test requires a score of 82.5% correct or higher on each of the three scores (immediate recognition, delayed recognition, and consistency). One or more lower scores results in failing the test and is interpreted as reflecting insufficient test taking effort (Green et al., 1996).
With regard to the tests from the Neuropsychological Battery for Epilepsy (Dodrill, 1978), it is noted that the eight tests which had been given to at least 80% of the patients (and their criteria for performances inside vs. outside normal limits) were as follows: (1) Stroop Test, Part 1—simple reading (93 s or quicker vs. 94 s or more); (2) Wechsler Memory Scale-III Auditory Memory, Immediate (IQ equivalent score of 90 or higher vs. 89 or less); (3) Wechsler Memory Scale-III Visual Memory, Immediate (IQ equivalent score of 90 or higher vs. 89 or less); (4) Name Writing speed (0.85 letters/s or faster vs. 0.84 letters/s or less); (5) Finger Tapping Total—right + left (women: 92/10 s or greater vs. 91 or less; men: 101/10 s vs. 100 or less); (6) Trail Making Test Part B (81 s or less vs. 82 s or more); (7) Aphasia Screening Test—total language errors (2 or fewer vs. 3 or more); and (8) Aphasia Screening Test—distortion in drawings (none/questionable vs. mild/moderate/severe). In every case, these cutoff scores were set so that approximately 25% of persons with no histories of any neurological problems fell outside normal limits (Dodrill, 1978).
To provide general information about the ES and PNES groups, the performances of the groups on each test are given in Table 2. In most respects, the performances of the two groups were quite similar, a finding which no doubt reflected the comparability of patient groups on critical variables in Table 1 including the frequency of a positive neurological history other than epilepsy and the presence of focal findings on neurological examination. Where statistically significant differences were found between the groups in Table 2, they always favored the PNES group. Mild impairment overall was noted in both groups (50–70% of tests performed outside normal limits as specified by previously published norms; Dodrill, 1978).
Table 2. Performance of epileptic (n = 65) and psychogenic nonepileptic (n = 32) groups on all neuropsychological tests
|Word Memory Test, % immediate recognition||65||90.92|| 7.52||32||88.67||13.66||0.390|
|Word Memory Test, % delayed recognition||65||92.27|| 7.28||32||88.98||15.41||0.260|
|Word Memory Test, % consistency||65||87.73|| 9.34||32||86.56||12.06||0.632|
|WAIS-III Verbal IQ||63||88.22||12.24||30||92.60||15.89||0.189|
|WAIS-III Performance IQ||63||88.10||14.48||30||95.10||16.53||0.052|
|WAIS-III Full Scale IQ||63||87.30||13.00||30||93.27||16.31||0.086|
|Stroop Test (simple reading)||53||115.66 ||43.04||26||99.85||47.47||0.158|
|WMS-III Auditory Memory, Immediate||62||91.32||13.37||28||97.39||14.12||0.061|
|WMS-III Visual Memory, Immediate||62||85.84||16.05||29||96.24||15.04||0.004|
|Name Writing Total, (right + left)||60|| .80|| .34||30|| .77|| .28||0.734|
|Finger Tapping Total (right + left)||61||89.55||15.69||26||89.27||14.37||0.935|
|Trail Making Test, Part B||56||97.55||56.17||31||112.94 ||81.70||0.356|
|Aphasia Screening Test, total language errors||60|| 6.05|| 7.21||28|| 5.00|| 4.16||0.391|
|Aphasia Screening Test, distortion in drawings||60|| 1.20|| .95||28|| 1.18|| 1.16||0.932|
|% of neuropsychological tests outside normal limits||65||59.26||24.19||32||53.14||29.78||0.318|
To replicate the findings of Drane et al. (2006), two sets of analyses were conducted. First, frequencies of passing and failing the WMT by PNES and ES patients were compared using Fisher's Exact statistic. Second, decreased overall neuropsychological performance in association with WMT failure was evaluated with two-way ANOVA (WMT pass vs. fail; ES vs. PNES). In order to provide a larger framework for the interpretation of the findings on the WMT, exactly the same analysis was run for persons who had performed inside vs. outside normal limits (“pass” vs. “fail”) on each of the tests from the Neuropsychological Battery for Epilepsy (Dodrill, 1978) that had been administered to at least 80% of all patients.
The first analysis evaluated the strength of the relationship between failure on the WMT and presence of PNES rather than ES. The findings of Drane et al. (2006) are presented initially in Table 3 to provide a basis for comparison, and using the Fisher's exact statistic, a highly significant finding was observed with their data (p = 0.0002). However, when the same computations were made on the basis of data from the current study, there were no such differences between the groups (p = 0.806).
Table 3. Numbers of ES and PNES patients passing and failing the Word Memory Test
|Drane et al. (2006)||34 (92%)|| 3 (8%)||19 (51%)||18 (49%)|| 0.0002|
|Current study||49 (75%)|| 16 (25%)||23 (72%)|| 9 (28%)||0.806|
The second series of analyses attempted to confirm the finding of decreased neuropsychological performances generally when failure on the WMT occurred. The first lines of Table 4 show that indeed when the WMT was failed, a larger percentage of neuropsychological tests fell outside normal limits than when the test was passed (p = 0.010). However, this was observed with both ES and PNES patients. Also, when the same analysis was undertaken with the other neuropsychological tests, they showed findings that were almost identical with those of the WMT. When a person scored lower on any one of these tests (except Name Writing speed), they were also likely to score lower on the battery as a whole. In addition, except for Finger Tapping, neuropsychological performances were not significantly different for ES and PNES groups on any of the tests. Finally, a simultaneous consideration of patient group (ES vs. PNES) and test performance (passing vs. failing) never resulted in the identification of a significant interaction effect.
Table 4. Overall performance (DDI—percentages of tests outside normal limits) on neuropsychological tests as a function of passing or failing performances on individual test measures
|Word Memory Test||55.00||72.30||48.82||64.19||.250||.010||.876|
|Stroop Test (simple reading)||40.50||62.93||39.26||53.48||.318||.001||.442|
|WMS-III Auditory Memory, Immediate||51.44||69.35||44.91||62.70||.295||.005||.992|
|WMS-III Visual Memory, Immediate||48.27||63.55||42.40||65.74||.749||.001||.484|
|Name Writing Total (right + left)||55.74||55.51||36.36||53.31||.079||.162||.256|
|Finger Tapping Total (right + left)||42.06||62.26||27.52||50.59||.040||.001||.806|
|Trail Making Test, Part B||46.16||63.82||38.19||69.38||.816||.001||.192|
|Aphasia Screening Test, total language errors||46.86||63.28||35.68||59.50||.197||.001||.522|
|Aphasia Screening Test, distortion in drawings||47.82||72.66||47.86||67.95||.656||.001||.650|
The essential findings of the Drane et al. (2006) study could not be replicated. PNES patients do not fail the WMT more often than do ES patients. Consequently, there was no evidence for diminished test taking effort by PNES patients, and their results on neuropsychological tests appear to be as trustworthy as those of patients with ES. People who do poorly on any of the tests in this study do poorly on the battery as a whole, not just on the WMT. It would have been a mistake to look only at the WMT and to conclude that since scores on the WMT are related to test performance overall, the WMT must be “driving” general neuropsychological performance. By looking at the results with the tests other than the WMT, it is clear that the other tests are relating to the battery as a whole in a way similar to that of the WMT. Thus, if the WMT is said to be a test of “effort,” the possibility should be considered that all of the tests examined here may be tests of that construct as well. Clearly, all tests require a degree of “effort.” However, the point is that the WMT, attempting to use a paradigm of simple test materials to detect diminished effort, does not produce results which are clearly different from other tests in any respect studied here. Thus, a unique role for the WMT in measuring test-taking motivation could not be confirmed in this study.
As the present study and the Drane et al. (2006) investigation were performed in the same neuropsychology laboratory of the same epilepsy center, many factors that might account for the differences in results can be ruled out. The conspicuous difference in design of the two studies is the difference in sample selection. While very few patients were eliminated for any reason in the present study, in the paper by Drane et al. (2006), 47% of the epilepsy sample was further eliminated for a host of reasons. An examination of these reasons reveals that such eliminations almost certainly had the effect of dropping the more severe ES patients. These are the patients most likely to do poorly on all tests including the WMT, and the results are not likely to be representative of ES patients as a whole. The dropping of patients with recent seizures was especially problematic since none of the studies cited actually reported cognitive effects of seizures even a few hours after their occurrence (Rennick et al., 1969; Aarts et al., 1984; Aldenkamp & Arends, 2004a, b). A poster presented at a meeting and available only in abstract form (Williamson et al., 2005), using a small unselected sample of patients, also did not present clear evidence for lingering cognitive effects. Furthermore, a very recent study (Dodrill & Ojemann, 2007), not available to Drane et al. (2006) when they wrote their paper, was specifically directed to discovering lingering cognitive effects after seizures but none could be found. Thus, many patients dropped by Drane et al. (2006) may have been excluded unnecessarily.
Drane et al. (2006) studied only those PNES cases who had been assigned to full-time neuropsychologists. However, following a longstanding practice in this laboratory established by the author of the present report, those patients were likely the more severe cases who more often required interpretations and family conferences on short notice. These cases included the blatantly hysteroid patients with obvious and prominent psychiatric difficulties which even from the time of admission would obviously require detailed clinical care and follow-up. When the less severe cases seen by the part-time neuropsychologists were also included as was true in the present paper, the WMT failure rate dropped from 49% to 28%.
The failure rate of 25% on the WMT for the ES group may seem high. Indeed, typically using several methods to determine symptom exaggeration, neuropsychologists commonly believe that less than 10% of nonlitigating medical patients evidence malingering (Mittenberg et al., 2002). However, this statistic is based upon the subjective impressions and recollections of neuropsychologists, neither of which are data based. However, in one research study in which the effort test scores of unselected epilepsy cases were actually examined, 28% of the sample had questionably valid or definitely invalid scores (Loring et al., 2005). This agrees closely with the finding of the present study, and it appears that in unselected epilepsy patient samples, the failure rate may be much higher than is commonly believed.
In conclusion, the majority of the findings of Drane et al. (2006) could not be verified in the present investigation. The reason for the lack of verifiability of their findings almost certainly reflects their restrictions in sample selection, and their results must therefore be interpreted with care.
The author has read the Journal's position on issues involved in ethical publication and affirms that this report is consistent with those guidelines. There are no conflicts of interest to report.