Address correspondence and reprint requests to Dr. D.L. Drane at Neuropsychology Laboratory, UW Regional Epilepsy Center, Harborview Medical Center, Box 359745, 325 Ninth Avenue, Seattle, WA 98104, U.S.A. E-mail: firstname.lastname@example.org
Summary: Purpose: Patients with psychogenic nonepileptic seizures (PNES) and those with epileptic seizures (ES) purportedly have roughly equal neurocognitive deficits. However, recent findings suggest that patients with somatoform disorders exhibit more variable effort on neurocognitive testing than do controls. We reexamined neurocognitive function in patients with ESs and PNES by using symptom validity testing to control for variability in effort.
Methods: Patients referred for video-EEG monitoring were administered the Word Memory Test (WMT), a measure of symptom validity, as part of neuropsychological evaluation. Patients classified with ictal video-EEG recordings as having ES (n = 41) or PNES (n = 43) were compared on neurocognitive and WMT performance and demographic, psychiatric, and medical variables.
Results: Striking rates of WMT failure were observed in the PNES (51.2%) group, but not in the ES (8.1%) group (p = <0.001) after controlling for false-positive errors. Although the PNES and ES groups reported equivalent neurologic histories, the PNES group exhibited less objective evidence of impairment as measured by valid neuropsychological testing, MRI of the brain, and video-EEG monitoring.
Conclusions: Many patients with PNES do not put forth maximal effort during neuropsychological assessment. When patients with PNES put forth valid effort, they demonstrate less objective evidence of neuropathologic injury or disease than do patients with ES. The cognitive impairment reported by this group appears to be more a function of motivational (although not necessarily intentional) factors than of verifiable neuropathology.
Although patients with psychogenic nonepileptic seizures (PNES) differ in significant ways from patients with epileptic seizures (ES), overall neurocognitive performance is not known to be one of these differences. Several studies have reported that patients with PNES have cognitive deficits on neuropsychological testing as severe as or worse than those of patients with ES (Wilkus and Dodrill, 1984; Wilkus and Dodrill, 1989; Drake et al., 1993; Hermann, 1993; Dodrill and Holmes, 2000), although this has not been a universal finding (Sackellares et al., 1985). Several studies (Wilkus and Dodrill, 1984; Wilkus and Dodrill, 1989; Dodrill and Holmes, 2000) reported that both groups perform outside of normal limits on approximately one half of the measures in a battery of neuropsychological tests. Another report (Drake et al., 1993) noted that four of 20 individuals with PNES performed in the mentally retarded range, whereas an additional 13 exhibited cognitive impairment on the Halstead-Reitan Battery (Reitan and Wolfson, 1993). Similarly, another study found no difference in neurocognitive performance between a small group of patients with PNES or ES (Hermann, 1993). It was proposed that the severity of neurocognitive deficits in patients with PNES is due to their other medical difficulties, as they frequently report more neurologic injury or disease than do patients with epilepsy (e.g., head trauma, CNS infection, possible birth traumas) (Wilkus and Dodrill, 1984; Wilkus and Dodrill, 1989; Drake, 1993; Dodrill and Holmes, 2000). Such histories are typically based on self-report rather than on objective data, however, and are rarely verified. This is potentially problematic, as patients with neurologically unexplainable symptoms such as PNES may be much less reliable historians (Schrag et al., 2004).
The rapidly developing neuropsychological literature on “symptom validity testing” (SVT) is relevant to these issues. SVT measures designed specifically to identify poor effort often appear difficult but are actually quite easy, so that patients with known impairments [e.g., moderate to severe traumatic brain injury (TBI), mental retardation, or even mild dementia] respond correctly >90% of the time (Sweet, 1999). Consequently, “failure” of an SVT implies the patient has performed significantly worse than patients with verified neurologic injuries or diseases with statistically and clinically significant effects on ability to learn and recall new information. Such poor performance raises a red flag as to the trustworthiness of neurocognitive testing results, especially when the patient in question has a vague or unverifiable neurologic history.
SVT failure does not necessarily imply that impairment has been intentionally exaggerated. These tests do not speak directly to intention, but rather to the extent to which a given patient's neurocognitive performance provides a believable index of the functional integrity of the tested neurocognitive systems. For example, patients with very mild head injuries who failed the Word Memory Test (WMT) (Green et al., 1996), one of the most sensitive SVTs, produced far lower scores on a neurocognitive test battery than did patients with substantiated severe brain injuries (Green et al., 1999). As a large portion of the mild head injury patients in this study were involved in litigation, it appears possible that either conscious or unconscious factors were contributing to this increased rate of SVT failure. Of particular interest with regard to patients with PNES, approximately two thirds of a small sample of patients with somatization or conversion pathology demonstrated at least one noncredible performance on neurocognitive tests sensitive to effort (Boone and Lu, 1999).
Thus an alternative hypothesis for the apparent neurocognitive dysfunction of patients with PNES is that it results from inconsistent effort rather than true brain impairment. Two groups (Binder et al., 1998; Hill et al., 2003) found that patients with PNES are more likely than ESs patients to perform in an invalid fashion on neurocognitive testing, but neither study had sufficient numbers of patients to stratify neurocognitive performance by SVT performance. In the current study, we used a sensitive, well-validated SVT specifically to examine the presumed equivalence of neurocognitive impairment in the PNES and ES groups. We predicted that
1patients with PNES would fail SVT at a higher rate than those with ES;
2patients with PNES who pass SVT would significantly outperform both patients with ES and those with PNES who fail SVT on a well-validated neurocognitive battery sensitive to deficits seen with seizure disorders (Dodrill, 1978); and
3patients with PNES would report significantly more unverifiable neurologic diseases or injuries than did patients with epilepsy.
Our original sample was drawn from 166 patients referred for continuous video-EEG monitoring for evaluation of uncontrolled seizures at the University of Washington (UW) Regional Epilepsy Center. This represents inpatients admitted to the epilepsy monitoring unit over a 2-year period seen by the authors D.L.D. and E.S.S. who were able to participate in neuropsychological testing. Patients undergoing repeated monitoring after a previous surgery and those with English as their second language were excluded.
We classified patients on the basis of their ictal video-EEG recordings as experiencing (a) ES (n = 70): evidence of definite ictal EEG abnormalities; (b) PNES (n = 43): episodes of unresponsiveness or behavioral abnormality in the absence of epileptiform EEG changes; (c) indeterminate spells (IS: n = 44): no spells during monitoring or subjective feelings only, in the absence of EEG abnormality, unresponsiveness, or behavioral abnormality; (d) cooccurrence group (COG: n = 6): evidence of episodes fitting the criteria for both ES and PNES during the same or across multiple monitoring sessions; or (e) nonepileptic seizures of other origin (NESO: n = 3): this included patients with spells resulting from medical conditions other than epilepsy (e.g., syncopal episodes, sleep disturbance). EEG interpretation and diagnostic classification were made by two board-certified electroencephalographers (J.M. and M.H.) in accordance with recently published criteria used to determine base rates of PNES among patients with epilepsy (Martin et al., 2003).
Given the small size of the COG and NESO groups, we believed we would be unable to draw meaningful conclusions about them. Thus we removed them from further analyses. Likewise, we excluded the patients whose spells were of indeterminate origin. Finally, we excluded those patients (n = 29) who experienced electrographically confirmed seizure activity during any portion of the testing or within the 24-h period preceding the testing, as data suggest that postictal patients may perform below the level typical of their interictal functioning on neurocognitive tests (Rennick et al., 1969; Aarts et al., 1984; Aldenkamp and Arends, 2004a; Aldenkamp and Arends, 2004b). In addition, we recently presented data demonstrating that acute temporal lobe seizures can alter performance on even the WMT (Williamson et al., 2005). Demographic data for the two remaining diagnostic groups are provided in Table 1. The only significant difference in demographic variables was that more women than men were in the PNES group.
Table 1. Comparison of the diagnostic groups on demographic variables
This study was approved by the Institutional Review Board (application 04-3521-A-01) of the University of Washington Medical School, where all data were collected. Data from all subjects are stored in a secure database registry maintained by the UW Regional Epilepsy Center. Informed consent was obtained from each patient.
The oral version of the WMT requires learning a list of 20 word pairs (e.g., dog–cat, man–woman) presented in two consecutive trials. Patients are then asked to discriminate words originally presented from foils. Thirty minutes later, they are asked to discriminate words from the original list from a different set of foils. These word pairs are easily recalled by healthy adults and patient groups alike (with the exception of some patients with severe dementia or aphasia) and prove easy to discriminate from foils. After the delayed-recognition trial, patients are presented with the first word from each pair and asked to select the word with which it was initially presented from among eight choices. On completion, the examiner reads the target words again, and the patient is asked to recall the word that went with each of them. The patient is then asked to freely recall all of the original words in any order.
The cutoffs for the effort-sensitive components of this measure have been found to be simultaneously sensitive to effort and insensitive to genuine neurocognitive impairment (Green et al., 1999; Iverson et al., 1999). In addition, the sensitivity of these subtests persists even when the WMT is completed by sophisticated subjects who have been warned that the test is sensitive to effort (Green et al., 1996). Per recommendations in the SVT literature, the specific cutoffs that identify poor effort are not published here but are available in the test manual and from the authors (Sweet, 1999).
This battery of 16 neurocognitive tests effectively discriminates between healthy adults and patients with epilepsy. It includes the entire Halstead–Reitan Battery (Reitan and Wolfson, 1993), the Stroop Test (Dodrill, 1978), the Seashore Tonal Memory Test (Seashore, 1960), and portions of the 3rd edition of the Wechsler Memory Scale (Wechsler, 1997). It yields a Discrimination Index (the Dodrill Discrimination Index, or DDI) that reflects the percentage of tests on which a patient scores below normal limits. The DDI quantified the level of neurocognitive impairment in several of the primary articles, suggesting that patients with PNES have equivalent or greater neurocognitive dysfunction than do those with verified ES (Wilkus and Dodrill, 1984; Wilkus and Dodrill, 1989; Dodrill and Holmes, 2000).
All patients underwent inpatient video-EEG monitoring at the UW Regional Epilepsy Center to classify their episodes. Additionally, patients completed a comprehensive neuropsychological assessment, which included Dodrill's Neuropsychological Battery for Epilepsy (Dodrill, 1978) and the WMT.
Sensitivity and specificity are important issues to consider for any diagnostic method, with the relative importance of sensitivity and specificity dependent on the context and purpose of the test. A “false-positive” result is of particular concern in situations in which “failure” of a test is interpreted as not putting forth valid effort on neuropsychological testing. We classified patients as false positives if they failed the WMT but had clear evidence of aphasia, dementia, or severe cognitive dysfunction of a magnitude to render them unable to function independently. No patients from the PNES group met these criteria, as all were living independently at the time of their assessments. However, four patients from the ES group were not living independently and were deemed by the authors to have sufficient evidence of severe cognitive deficits. These patients exhibited profound memory deficits, significant confusion, and an inability to keep up with events from day to day, and were thus considered false positives and were excluded from all analyses examining group differences on WMT pass/fail rates. Of note, even if false positives were not excluded, the results of all primary analyses remained significant.
We performed all statistical analyses with SPSS 12.0 (SPSS, Inc., 2003). Chi-square analysis was used to test for significant group differences in failure of the WMT. Analysis of variance (ANOVA) was used to examine the impact of poor effort on actual neurocognitive test performance by comparing the DDI of the groups in light of WMT performance (i.e., pass/fail). When the DDI was analyzed as a dependent variable, it was treated as a proportion. Completion of at least half the tests of the DDI was required for calculating the DDI (Dodrill, personal communication), resulting in exclusion of seven patients (one ES and six PNES) from the analyses involving DDI (e.g., those addressing our second hypothesis). For those who did not complete the entire battery, we used an arcsine root transformation to avoid introducing nonuniform residuals that could result from comparing individuals whose DDI scores were based on differing numbers of tests. Finally, we conducted ANOVA (or Kruskal–Wallis test for nonparametric analyses) to determine if the groups reported differing numbers of unverifiable neurological diseases or injuries and to explore relevant medical and psychiatric variables.
Question 1: Do patients with PNES fail the WMT at a higher rate than patients with ES?
The ES and PNES groups clearly failed the WMT at different rates, chi-square (1, n = 79) = 16.32, p < 0.001. With odds ratios (ORs), patients who failed the WMT were >11 times more likely to be diagnosed with PNESs than with ESs (OR = 11.33; p < 0.001). Only three (8.1%) of 37 patients with epilepsy scored below cutoffs on one or more of the WMT tests most sensitive to effort, whereas 22 (51.2%) of 43 patients with PNES did so. Three patients with PNES actually scored below chance on the WMT, whereas none of the patients with epilepsy did so.
Question 2: Do patients with PNES who pass WMT outperform patients with PNES who fail WMT and both ES groups on neurocognitive testing?
Table 2 depicts the overall cognitive performance of the groups, as indexed by the DDI. Consistent with expectations, the ES and the PNES groups did not differ significantly, with overall scores in the abnormal range on approximately half of the DDI neurocognitive measures. This changes dramatically, however, if one views performance in light of the WMT results. The small number of patients with epilepsy who failed the WMT precluded their inclusion into an ANOVA. However, comparing the other three groups with each other (i.e., patients with epilepsy who passed the WMT and PNES patients who passed or failed) reveals significant group differences in the DDI, F(2, 68) = 15.99; p < 0.001. Planned Bonferroni comparisons revealed that all three groups differed from each other in the expected directions: the PNES-Pass group demonstrated less impairment on the DDI than did the PNES-Fail group (p < 0.001) and the ES-Pass group (p < 0.03), whereas the PNES-Fail group demonstrated more impairment than the other groups (p = 0.001 and p < 0.001 for the ES-Pass and PNES-Pass groups, respectively).
Table 2. Performance on the Dodrill Discrimination Index (DDI) by diagnostic group and WMT performance
ES (n = 37)
PNES (n = 37)
One patient from the ES group and six patients from the PNES group did not complete enough of the neurocognitive battery to produce a reliable DDI score. Four patients were excluded from the ES group, as they were considered to represent false positives on the WMT because of their inability to live independently.
DDI, Dodrill Discrimination Index (the percentage of 16 test scores falling into the abnormal range); WMT, Word Memory Test; ESs, epileptic seizures; PNES, psychogenic nonepileptic seizures.
(n = 34)
(n = 3)
(n = 19)
(n = 18)
Although ANOVA statistics certainly provide helpful information, they do not always translate to clinical application as readily as do ORs and graphic depiction of data. These perspectives demonstrate the striking extent to which the PNES cognitive test results vary once effort is taken into account. As Fig. 1 depicts, all but one patient with PNESs who failed the WMT scored in the impaired range on ≥50% of the tests on the DDI, whereas all but three patients with PNES who passed the WMT performed in the abnormal range on fewer than half of the measures.
In predicting who will perform abnormally on ≥50% of the DDI measures, knowing a patient's diagnosis is not helpful; 49% of the ES group and 57% of the PNES group performed in such a manner. Likewise, knowing how a patient with epilepsy performed on the WMT does not significantly improve prediction (OR = 2.25, NS). In marked contrast, knowing that a patient with PNES failed the WMT adds a great deal of confidence, as our patients with PNES who scored in the invalid range on the WMT were nearly 64 times more likely to score abnormally on ≥50% of the DDI measures (OR = 63.75; p < 0.001).
Figure 2 depicts the performance of our current sample relative to previously defined expectations for patients with epilepsy and for normal controls at our center. As can be seen, our current sample's performance deviates markedly from expectations once one considers the results of SVT. The average performance of our sample, without stratification, follows the expected pattern of ∼50% of tests impaired, regardless of event etiology. If stratified by WMT results, however, the PNES group diverges significantly from expectations, whereas the ES group does not. The 95% confidence interval (CI) of the PNES group that fails the WMT falls entirely above the 95% CI for patients with epilepsy, indicating more severe impairment than expected from patients with verified epilepsy. In contrast, the 95% CI of the PNES group that passes the WMT falls entirely below the epilepsy CI, indicating less severe impairment than expected from patients with verified epilepsy. This CI actually overlaps the CI of normal controls.
Question 3: Do patients with PNES report significantly more unverifiable neurologic diseases or injuries than do patients with epilepsy?
Table 3 depicts group differences on a number of medical and psychiatric variables. Analyses revealed that patients with PNES were much more likely to report histories of fibromyalgia (p < 0.01) and chronic pain (p < 0.001) than were patients with verified ES. Overall self-reported neurologic history did not differ significantly between the groups.
Table 3. Comparison of diagnostic groups on medical and psychiatric variables
ESs (n = 41)
PNESs (n = 43)
Independent samples t tests were used to compare groups on age at spell onset, frequency of spells, and current number of AEDs, whereas chi-square analyses were used with all other variables.
ESs, Epileptic seizures; PNESs, psychogenic nonepileptic seizures; SD, standard deviation; AEDs, antiepileptic drugs; MRI, magnetic resonance imaging.
Age at spell onset (in yr) (SD)
Frequency of spells (per mo) (SD)
Number of current AEDs (SD)
Duration of spells (yr since onset) (SD)
Focal neurologic exam
Baseline EEG findings
History of neurologic insult (self-report)
Neurologic history (consensus)
History of closed head injury (self-report)
History of sexual abuse
History of physical abuse
History of emotional abuse
History of fibromyalgia
History of chronic pain
The PNES group reported a later onset of seizures (p < 0.01) and more frequently reported a history of physical abuse (p < 0.01). Patients with ES, in contrast, were taking more antiepileptic medications (p < 0.01) and were more likely to have abnormal findings on MRI (p < 0.001) and on baseline EEG (p < 0.001).
Our data are consistent with our primary hypotheses. Specifically:
1Significantly more patients with PNES than those with ES failed SVT, suggesting that statements about neurocognitive performance of patients with PNES are confounded by invalid data,
2Patients with PNES who performed within normal limits on the WMT significantly outperformed patients with ES on a well-validated neurocognitive battery previously shown to be insensitive to cognitive differences between the two groups, and
3Patients with PNES reported histories of fibromyalgia and chronic pain disorder more frequently than did those with ES, and equivalent rates of neurologic disease and injury, despite less objective evidence of actual brain dysfunction.
These results suggest that ∼50% of patients with PNES produce neurocognitive test results that are more a function of inadequate or variable effort than of brain impairment. Patients with PNES who passed the WMT produced neurocognitive scores more similar to those expected from healthy adult controls than to those expected from patients with confirmed epilepsy. In contrast, patients with PNES who failed performed significantly worse than patients with confirmed epilepsy. Furthermore, the patients with epilepsy exhibited overwhelmingly greater evidence of genuine brain impairment on a variety of objective diagnostic tests (e.g., video-EEG monitoring, MRI) than did the patients with PNES, despite equivalence in self-reported history of neurologic insults.
Direct comparison of PNES and ES groups who failed the WMT in terms of observable functional status reveals significant differences in their capacity to function independently. Whereas only seven patients with ES in our overall sample exhibited atypical WMT scores, we believe that four of these individuals represented false positives (i.e., all were profoundly impaired and were not functioning in an independent fashion). For example, one patient who had been diagnosed with a mitochondrial disorder had dementia and was living in a state institution. He could not recall daily events with any degree of accuracy and could provide little information during clinical interview beyond his name and date of birth. Although this patient's scores on the WMT fell below the cutoffs for valid performance, his performance actually exceeded that of 10 of the 21 patients with PNES who produced abnormal scores on the WMT. In contrast to this patient's profound impairment in daily life, all 10 of these patients with PNES were living independently and managing their own finances.
Similar findings of inconsistency between neurocognitive performance and daily functioning have been reported for previous samples of PNES patients (Brown et al., 1991; Bortz et al., 1995). The latter study found that a subset of PNES patients performed normally on a variety of neurocognitive tests, whereas those who exhibited impairment often performed inconsistently across tasks. In addition, some PNES patients were said to have appeared significantly more impaired on testing than would have been expected from their behavioral presentation and overall level of success in daily-life functioning.
The patients with PNES in our study exhibited far less objective evidence of brain impairment than did patients with epilepsy, despite reporting equivalent rates of closed head injury and neurologic disease. For example, MRI abnormalities were observed in 65% of the epilepsy patients but in only 26% of the patients with PNES who received imaging. Recent neuroimaging studies that included healthy adult control groups have indicated that as many as 20% of such individuals will demonstrate MRI abnormalities (Katzman et al., 1999). Given that only those patients with PNES with a strong index of suspicion for possible neuropathology were referred for MRI, our results for the PNES group may to a large extent reflect normal variations in brain integrity. This is also suggested by an examination of the clinical findings, as many of the MRI abnormalities in the PNES group reflected incidental pituitary adenomas, mild periventricular white matter hyperintensities (presumed to be related to small-vessel disease), and other findings of dubious clinical importance. In contrast, the MRI abnormalities in the epilepsy sample included temporal lobe disease (e.g., mesial temporal sclerosis and general temporal lobe atrophy), developmental abnormalities (e.g., cortical dysplasia), tumors, and vascular malformations. In addition, ES patients exhibited significantly more baseline EEG abnormalities than did patients with PNES.
Like many prior studies (Wilkus and Dodrill, 1984; Wilkus and Dodrill, 1989; Dodrill and Holmes, 2000; Reuber et al., 2002), our results demonstrate that patients with PNES are more likely to be women and to have later ages of spell onset than do patients with verified epilepsy. Both groups experience an elevated rate of psychiatric disturbance, although this is slightly more marked for the PNES group. Our findings also raise important questions regarding the presumed psychiatric and neurologic underpinnings of this disorder.
Our PNES group also demonstrated a tendency to endorse conditions that often have an unclear medical explanation and may be influenced by psychiatric factors (e.g., chronic pain disorder, fibromyalgia). This is consistent with other emerging research, as one recent study demonstrated that 75% of patients seen in a university-based epilepsy clinic with a history of either fibromyalgia or chronic pain were diagnosed with PNES after comprehensive evaluation including video-EEG monitoring (Benbadis, 2005). The rate of endorsement of either of these syndromes among epilepsy patients was virtually nonexistent, whereas significant proportions of patients with PNES claimed to have these syndromes. We believe that this reflects a general tendency of those whose spells appear to be driven by psychological mechanisms to overendorse clinical symptoms and syndromes.
For all of the stated reasons, we believe our data strongly suggest that large numbers of patients with PNES produce invalid scores on neuropsychological assessment, whereas most patients with verified epilepsy produce valid data. The reasons for producing invalid data are less easy to discern and are likely multifactorial in nature. Possible causes for invalid effort include a host of factors that could interfere with ability or inclination to engage in the assessment process. Some researchers (Binder et al., 1998) have suggested that a very small subset of patients with PNES are likely producing suboptimal effort because of conscious intent (i.e., malingering), whereas the remainder who perform poorly on SVT lacked the psychological resources necessary to persist through a challenging neuropsychological assessment. Our data likely include patients who would fall into both of these categories. Some have argued that below-chance performance on any test should raise suspicions of malingering, the rationale for this claim being that such persons must recognize the right answer to choose the wrong one consistently. It is argued that they are actually putting forth effort to perform poorly. Three of the 21 patients in the PNES group who produced atypical WMT scores exhibited below-chance performance.
Other factors that could have a role in suboptimal effort might include severe pain, physical fatigue, and emotional distress (e.g., depression, anxiety, posttraumatic stress disorder). Although some members of the PNES group had such complaints, so did many patients with verified epilepsy (based on self-reported ratings of pain, mood, and fatigue). In addition, some evidence is emerging that despite elevated rates of SVT failure in many of these groups, SVT performance does not always covary with identified problems. For example, studies examining patients diagnosed with fibromyalgia, rheumatoid arthritis, and other chronic pain syndromes suggest that SVT performance is independent of pain intensity at the time of testing (Gervais et al., 2001, 2004). We believe that future research should endeavor to explore the reason for elevated rates of invalid test performance in the PNES sample, comparing this group with patients with verified epilepsy on a variety of mood and personality variables, medication factors, litigation status, financial incentive, and ratings of fatigue and pain. Because of the rather heterogeneous nature of PNES etiologies and variations between patient populations seen at different epilepsy centers, our results also must be replicated in other settings.
As conversion disorders are put forth as one of the most common explanations for PNES events, one is forced to consider the possibility that such tendencies affect cognitive processes as well as physical ones. Many case studies suggest that patients can experience psychogenic cognitive problems (Kopelman, 1987; Campodonico and Rediess, 1996). In addition, at least one study has suggested that patients with somatization/conversion personality styles may have noncredible cognitive complaints in addition to implausible physical findings (Boone and Lu, 2003). Whether such cognitive failures reflect conscious or nonconscious processes is a matter of debate. A thorough comparison of patients with PNES who pass SVT measures versus those who do not may shed additional light on these issues.
These data suggest that many existing studies examining neurocognitive functioning in these populations are distorted by invalid results. Consistent with some preliminary reports (Loring et al., 2005), a small percentage of our patients with epilepsy demonstrate suboptimal effort to perform accurately on neurocognitive measures. Therefore it is likely that invalid data are present in investigations of neurocognitive deficits in various epilepsy syndromes and studies exploring the outcome of surgical intervention. Even a few negatively skewed results due to the impact of variable effort could significantly alter group data used in these studies. We believe that it is essential to use SVT with these populations if one plans to use neuropsychological test performance as a marker for brain impairment. The rate of suboptimal effort in our sample (8.1%) is remarkably similar to the rate of symptom exaggeration in a general medical population (8%) in an independent study using different symptom-validity measures (Mittenberg et al., 2002).
Further exploration of our results may assist in psychometric differentiation of patients with PNES from patients with verified epilepsy. It appears possible that SVT data provided by the WMT may be useful for predicting group membership for some patients (i.e., epilepsy vs. PNES), as such a discrepancy exists between the failure rates of these two groups. Nevertheless, because nearly half of the PNES group appeared to put forth valid effort on neurocognitive tests, performance on this measure should be combined with demographic information and results of formal personality measures if it is to be used for diagnostic purposes. The primary use of this test is to ensure that one is working with valid neurocognitive data, whether at the individual or at the group level.
These results have clinical relevance for planning treatment interventions with patients with PNES. Although we do not know the extent to which SVT failure may predict treatment response, those patients who are “screened out” of some treatment protocols because of low neurocognitive test scores may be failing to receive treatment from which they may actually benefit. Our results suggest that many patients with PNES exhibit normal neurocognitive functioning and have the intellectual capacity to benefit from standard psychotherapeutic interventions. For those who seem to exaggerate their neurocognitive problems, some facet of their therapy may focus on having them recognize their intact areas of ability. Some patients with PNES could have actual neurocognitive deficits. However, it appears that a tendency exists in this population to exaggerate the deficits in the same fashion that they produce behavioral manifestations similar to a seizure episode, despite no electrophysiologic correlate present on EEG.
Overall, we strongly recommend that all epilepsy centers adopt the use of modern SVT measures to improve validity of their neuropsychological results for most patient populations. Although false-positive errors will occur at times across diagnostic groups, it appears that this is infrequent, and that methods exist for more definitively determining if a problem with effort and task engagement exists. We encourage future research to examine possible false positives and to ascertain better whether any environmental or internal factors contribute to their occurrence. Given that group-level neuropsychological data have been used to assist in surgical planning and to determine postoperative cognitive decline, the accuracy of these data is critical.
Acknowledgment: Parts of this article were presented at the annual meeting of the American Epilepsy Society in December 2003 (Boston, MA) and in December 2004 (New Orleans). This research was supported in part by the Drueding Foundation.