|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 THE NEUROPSYCHOLOGIC AL ASSESSMENT BATTER Y (NAB): A TEST OF CRITERION VALIDITY W ITHIN AN EPILEPSY PO PULATION By BRADLEY J. DANIELS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011
2 2011 Bradley J. Daniels
3 This dissertation is dedicated, in loving memory, to Donna Graham.
4 ACKNOWLEDGMENTS I would like to acknowledge the many people in my life who have supported, guided, and inspired me in various ways throughout my higher education. These people include my academic mentor, Russell M. Bauer; the neurologists and epileptologists at UF/Shands including Jean Cibula, Stephen Eisenschenk, and Ian Goldsmith; my research assistants/collaborators including Ashley Wilson, Paul Hemrick, and Isaac Lee; my parents, Lyle and Gloria Daniels, whose financial support kept my lights on; my family, Angel, Dan and Brooklynne Joy Farran, Gary Graham, Sr. and Gary Graham, Jr.; my labmates, Gila Reckess and Emily Green King; my extended family and friends, including the DiFalco and Rauch families, the McGregor Clan, Phil McCarty, Erica Murray, Julius Tobin, Danie l Aguel, Patrick Williams Pittard and Bryan Pittard, Andria and Matt Doty, Christian Popoli, and Adam and Amanda Kelley. Last, but certainly not least, I would like to thank my amazing wife, Tiffany Daniels, whose steadfast support and belief in me kept m e going.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 7 LIST OF FIGURES ................................ ................................ ................................ ......................... 8 ABSTRACT ................................ ................................ ................................ ................................ ..... 9 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................. 11 Neuropsychological Assessment Battery (NAB) Features ................................ ..................... 11 NAB Scores ................................ ................................ ................................ ............................ 12 Test Validation ................................ ................................ ................................ ........................ 14 Epilepsy ................................ ................................ ................................ ................................ .. 15 Non Epileptic Seizures (NES) ................................ ................................ ................................ 17 Prev ious Related Studies ................................ ................................ ................................ ........ 18 Hypotheses ................................ ................................ ................................ .............................. 20 Question One ................................ ................................ ................................ ................... 20 Question Two ................................ ................................ ................................ .................. 22 2 METHODS ................................ ................................ ................................ ............................. 27 Participants ................................ ................................ ................................ ............................. 27 Procedure ................................ ................................ ................................ ................................ 29 Statistical Analyses ................................ ................................ ................................ ................. 30 Classificati on with Neuropsychological Tests ................................ ................................ 30 Seizure Lateralization Index (SLI) ................................ ................................ .................. 30 Exploratory and Advanced Analyses ................................ ................................ .............. 31 3 RESULTS ................................ ................................ ................................ ............................... 34 Normality and Group Di fferences on the Neuropsychological Assessment Battery ............. 34 Group Differences by Demographics ................................ ................................ .............. 35 Group Differences by EMU Classification ................................ ................................ ..... 35 Group Differences by Lateralization Classification ................................ ........................ 36 Group Differences by Localization Classification ................................ .......................... 3 6 Correlations ................................ ................................ ................................ ............................. 37 Question One ................................ ................................ ................................ .......................... 37
6 Discriminant Function Analyses ................................ ................................ ..................... 37 Multiple Regression Analyses ................................ ................................ ......................... 39 Question Two ................................ ................................ ................................ .......................... 39 4 DISCUSSION ................................ ................................ ................................ ......................... 50 Results Summary ................................ ................................ ................................ .................... 50 Study Limitation s ................................ ................................ ................................ .................... 52 Possible Reasons for Study Outcome ................................ ................................ ..................... 54 Future Directions for Research ................................ ................................ ............................... 55 Conclusion ................................ ................................ ................................ .............................. 57 LIST OF REFERENCES ................................ ................................ ................................ ............... 58 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ......... 62
7 LIST OF TABLES Table page 1 1 Domains and s ubtests of the Neuropsychological Assessment Battery (NAB) ................ 24 1 2 Test c omposition of the UF Standard Neuropsychological Battery (SNB) ....................... 26 3 1 Variable name k ey ................................ ................................ ................................ ............. 43 3 2 Descri ptive statistics by seizure c lassificati on/epilepsy monitoring unit (EMU) o utcome ................................ ................................ ................................ .............................. 44 3 3 Descriptive statistics by lateralization c lassification ................................ ......................... 45 3 4 Percentage correct classification of seizure lateralization using NAB variables as p redictors ................................ ................................ ................................ ............................ 46 3 5 Descriptive statistics by localization classification (excluding participants with psychogenic non epileptic s eizures [PNES] ) ................................ ................................ ..... 47 3 6 Percentage correct classification of seizure localization (excluding participants with PNES) using NAB variables as p redic tors ................................ ................................ ......... 48
8 LIST OF FIGURES Figure page 3 1 Seiz ure Lateralization Index (SLI) distribution of participants experiencing electrographic s eizures ................................ ................................ ................................ ....... 49
9 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE NEUROPSYCHOLOGICAL ASSESSMENT BATTERY (NAB): A TEST OF CRITERION VALIDITY WITHIN AN EPILEPSY POPULATION By Bradley J. Daniels May 2011 Chair: Russell M. Bauer Major: Psychology This study attempted to assess the clinical utility of the Neuropsychological Assessment Battery (NAB) within an epilepsy population by analyzing how well NAB scores (as well as specific additional variables derived from NAB scores) pre dicted lateralization and localization of seizure focus in our sample. Forty five participants with documented or suspected epilepsy were tested while undergoing 24 hour video electroencephalographic ( V EEG ) monitoring during the ir stay as an inpatient in the epilepsy monitoring u nit (EMU) at UF/Shands Hospital in Gainesville, Florida over a 16 month period. Each participant was administered the NAB Screening Module as well as the five Core NAB Module s (Attention, Language, Memory, Spatial, and Executive Functions). Of the 45 participants tested, twenty were found to have ele ctrographic seizures (four left lateralized, eleven right lateralized, and five nonlateralized), twelve were found to have psychogenic non epileptic seizures (PNES), and thirteen sub jects had no clinical events during their stay and were therefore unable to be definitively diagnosed or classified. The results of our study provided some support for the clinical utility of the NAB within an epilepsy population. Specifically, our result s suggested that certain combinations of NAB scores
10 (as well as additional derived scores) were able to discriminate between individuals with temporal lobe epilepsy and individuals whose seizures do not originate in the temporal lobe at a rate greater than to successfully predict seizure laterality (from which cerebral hemisphere seizures originate ). though other sample specific constraints may have also played a role in this limitation. Future research with seizure focus when the results are a nalyzed by on e or more trained clinical n europsychologists, similar to the techniques used in current practice with this population using other neuropsychological batteries) is warranted to more definitively determine its clinical utility within an epilepsy population.
11 CHAPTER O NE INTRODUCTION of neuropsychological tests developed for the assessment of a wide array of cognitive skills and functions in adults aged 18 years to 97 years, with known or suspected disorders of the central comprehensive neuropsychological batteries such as the Halstead Reitan Neuropsychological Test Battery (Reitan & Wolfson, 1993) and the Luria Nebraska Neuropsychological Battery (Golden, Purisch, & Hammeke, 1985), including its significantly abbreviated administration time. It is estimated that all six Module s of the NAB (including the five C ore Modules and the Scre ening Module ) can be administered in less than four hours, nearly half the time it takes to administer the complete Halstead Reitan Battery. With an emphasis on increasing cost effectiveness of assessment services, the NAB offers an attractive alternative to established neuropsychological approaches by allowing for a reasonably comprehensive neuropsychological assessment that can be conducted in a relatively short amount of time. N europsychological Assessment Battery (N AB ) Features Major features of the NAB compared to other comprehensive neuropsychological batteries include its resilience to floor and ceiling effects, modifiable modes of use as both a flexible and a fixed battery, comprehensive coverage of all functional domains, use of a single standard ization group for the entire battery, availability of alternate forms and demographically corrected norms, ease of utility for both the administrator and patient, and a particular focus on the inclusion of ecologically valid tasks (Stern & White, 2003). A n example of efforts to provide more ecologically valid assessments is found in the digits backward task, a formal test of attention and working memory abilities that requires the subject to repeat back a series of orally presented
12 numbers in backwards ord er. Digits backwards is a commonly used task in neuropsychological assessment batteries (including the NAB), and although this task provides relevant information, it can do little by itself to predict impairment in functioning in a real world situation as the requirements of the task itself are not commonly found in everyday life. In contrast, in addition to assessing attention and working memory with a digits backward task, the NAB also includes a Driving Scenes Task, a test of attention in which the sub ject briefly views a drawing of a driving scene as shown from behind the wheel of a motor vehicle. The subject is then shown a different scene and required to point out the changes that occurred between the two scenes (Stern & White, 2003). A recent pub lication examined the ability of Driving Scenes subtest performance to classify groups of normal and mildly demented subjects into driving categories. All subjects (UFOV; Owsley et al., 1991) test and a comprehensive road course driving test. After performing a discriminant function analysis on the results, it was determined that the Driving Scenes score could be used to correctly classify subjects into their gl obal ratings of driving ability (rated as safe, marginal, or unsafe), approximately 66% of the time (Brown et al., 2005). This is merely one example ; in fact, t he NAB contains at least one Daily Living subtest in each of its five Module to be congruent with an analogous real (White & Stern, 2003, p.7). For a full list of tests included in the NAB, see Table 1 1. NAB Scores Administration of the NAB provides scores that can be compared to either demographically correct ed norms based on a national sample of 1,448 adults or to an alternative age based, U.S. census matched norm set based on a sample of 950 adults. Typical administration during clinical practice involves first administering the NAB Screening Module Upon successful adm inistration and scoring of the Screening Module, five S creening D omain
13 score s, as well as a Screening T otal score, would be available for interpretation. The NAB was designed such that the Screening Domain score s would inform the examiner wh ich subsequent full Domain Module s would be useful to administer. This design was structured to have dual screening capabilities, such that individuals performing significantly poorly or significantly well on a given Screening Domain score would not have the full Module from that Domain administered to them. The rationale for this approach is that administering the full Domain Module to these individuals would essentially be pointless as they would likely perform at floor during the full Module (much as t hey did on the briefer Screening Domain ) or would be equally Screening Domain Module was so high that they are clearly unimpaired in that particular Domain (Stern & White, 2003). In general, Domain score s o n the NAB Screening Module below a Standard Score value of 74 to 75 or above a Standard Score value of 114 to 126 will indicate that further testing in that Domain is not needed. According to Table 1.10 of the NAB Administration, Scoring, and Interpretati on Manual (Stern & White, 2003), the sensitivity of the NAB Screening Domain score s ranged from .95 to .96 while specificity ranged from .03 to .75. Specificity levels were lower for the Above Average cutoffs than the Moderately to Severely Impaired cuto ffs. The overall correct classification rate was .95 for the Moderately to Severely Impaired cutoffs and ranged from .67 to .79 for the Above Average cutoff. Although the NAB offers significant data on the reliability and stability of the battery, as w ell as comprehensive sets of both demographically corrected and age based norms (Stern & White, 2003), there have been few validity studies assessing the utility of the NAB during clinical assessments with special populations (e.g., patie nts with epilepsy, isease, dementia, etc) or with specific referral questions to which neuropsychological assessments are
14 typically applied. In practice, the NAB and other neuropsychological instruments are often used to make specialized decisions, such as whe ther or not to offer surgery, medical, or rehabilitative treatment, or to determine whether a treatment has been beneficial or detrimental to neurocognitive function. To date, the studies that have been conducted to assess the validity of the NAB have var iably supported its validity as a clinical instrument. However, these studies have focused primarily on assessing the validity of the NAB Screening Module alone (Grohman & Fals Stewart, 2004; Iverson, Williamson, Ropacki, & Reilly, 2007; Temple, Zgalijard ic, Abreu, Seale, Ostir, & Ottenbacher, 2009; Zgalijardic & Temple, 2010), the validity of one particular subtest or group of subtests of the NAB on predicting specific impairment (Brown et al., 2005; Cahn Weiner, Wittenberg, & McDonald, 2009; Gavett et al ., 2009; Yochim, Kane, & Mueller, 2009), or establishing base rates for low NAB scores in older adults (Brooks, Iverson, & White, 2007; Brooks, Iverson, & White, 2009). Thus far, limited data is available on the validity of the full NAB battery as a clini cally relevant neuropsychological tool. The purpose of our study was to conduct a validity study on the use of the NAB in neurocognitive evaluations of patients with documented or suspected epilepsy. Test Validation There are a number of formal ways in which new psychometric tests may be validated. Some of the most common types of test validity include construct validity, predictive validity, concurrent validity, and discriminant validity. According to Anastasi & Urbina (1997), construct validity is ge s ability to measure the underlying construct or ability it is attempting to measure. Demonstration of construct validity depends on showing that the test correlates highly with validated measure(s) of the same theoretical constr uct (convergent validity), and shows lower correlations with measures of different construct s (discriminant validity; Campbell & Fiske, 1959; Anastasi & Urbina, 1997). Criterion oriented validity refers to
15 extern al outcome measure (such as when the score on a neuropsychological test is used to predict the presence of a localized brain lesion). When the test is shown to predict some criterion available later in time, it is shown to demonstrate concurrent validity of the NAB in lateralizing and localizing seizure focus in a surgical epilepsy po pulation since neuropsychological test scores, neuroimaging findings, and electrographic localization data all result from the same interdisciplinary workup Epilepsy t p. 304). Approximately two million Americans have epilepsy, with nearly 125,000 new cases of epilepsy being diagnosed each year (Abel, 2005). The most common tr eatments for epilepsy involve the use of anti epileptic drugs (AEDs). Some other forms of treatment may also involve the patient following a ketogenic diet, or the implantation of a Vagus nerve stimulator (Mayo Clinic 2007 ). Unfortunately, as many as 35 % of patients with epilepsy have drug resistant, or seizures are not elimin ated with AEDs or other common treatments, resective surgery to remove neural tissue at the site of seizure onset may be considered. In 1992, Pilcher, Locharernkul, Primrose, Ojermann, and Ojermann estimated that as many as 1,000 resective seizure surgeries are conducted in the U.S. per year. With newer advancements in the field of neuro imaging to assist in determining seizure focus, it is likely that the number of resective seizure surgeries conducted in the U.S. has increased significantly since that time, with nearly 100 of these surgeries conducted at the Cleveland Clinic in 2007 alon e (Cleveland Clinic, 2007).
16 However, t here are many risks associated with resective brain surgery, including medical complications and neuropsychological morbidity In 1957, Scoville and Milner published the first account of the famous patient, H.M., who developed a severe permanent anterograde amnesia following a bilateral anterior temporal lobectomy in an attempt to eliminate his seizures. His case raised awareness of surgery induced neuropsychological morbidity, and today candidates for epilepsy surger y must undergo a wide variety of neuropsychological and neurological tests (e.g., electroencephalograph [ EEG ] magnetic resonance imaging [ MRI ] Wada) in order to ensure that losses to vital functions such as memory, motor function, vision, and language ar e minimized and that surgery will not place the patient at risk for developing functionally significant memory or language deficits (Valton & Mascott, 2004). Before epilepsy surgery is performed, a candidate is typically admitted to the hospital for noninv asive (Phase I) monitoring, during which patients undergo neurological testing, 24 hour video EEG monitoring with scalp electrodes, and structural neuroimaging with specialized MRI protocols (Abel, 2005). In addition, the patient also often undergoes comp rehensive neuropsychological assessment in order to further predict lateralization and localization of seizure focus. During this neuropsychological assessment, patients typically undergo a neuropsychological test battery that broadly evaluates functional domains including intellectual ability, language, executive functioning, material specific memory functioning, visuoperceptual/visuoconstructional skills, and motor f unctioning. During Phase I epilepsy monitoring, the patient is removed from their AEDs a nd may be subjected to sleep deprivation (a common epileptic trigger) in the hopes of eliciting seizure activity that can be monitored in order hospitalization can be quite stressful on the patient, the importance of quick and efficient
17 methods of neuropsychological evaluation is vital. The advent of briefer neurocognitive batteries (like the NAB) thus has potential for improving the cost effectiveness of clinical n europsychological evaluation in this setting. Non Epileptic Seizures (NES) Some patients seen within the context of clinical practice that are believed to have epilepsy are later found to have non epileptic seizures (NES). There are two types of non epile ptic seizures, physiologic and psychogenic (Benbadis, 2005). Physiologic non epileptic seizures are the result of a disruption of brain function due to physical causes other than true epilepsy, such as syncope, hypoglycemia, delirium tremens following alc ohol abuse or eclampsia as a complication of pregnancy (Benbadis, 2010). In contrast, psychogenic non epileptic seizures be seizure like in regards to observable behavior yet are psychological in nature (often conceptualized as sharing similarities with conversion disorders) and, unlike physiological non epileptic seizures, do not involve the characteristic electrical discharges associated with epilepsy. It is est imated that 20 to 30 % of patients seen in epilepsy clinics are eventually diagnosed with psychogenic non epileptic seizures (Martin, 2005). Approximately 75 % of patients diagnosed with PNES are female, and PNES patients often tend to have additional psychological problems and diagnoses, comorbidities including fibromyalgia or chronic pain, and/or previous histories of abuse (Benbadis, 2005). Although suspicion s of psychogenic non epileptic seizures may be raised following clinical interview in an epilepsy clinic, a definitive diagnosis of PNES is usually withheld until after a patient has undergone 24 hour video EEG monitoring. Whereas in genuine epilepsy this type of monitoring is useful in determining lateralization and localization of seizure focus, in PNES, it can be used to definitively rule out the presence of electrographic abnormalities during
18 observed behavioral events, and can provide evidence of beha vioral inconsistencies in presentation between such patients and individuals with genuine epilepsy. Previous Related Studies The current study is related to a previously published study from our lab conducted by David Moser (Moser et al., 2000). Moser and his colleagues sought to compare the utility of EEG, MRI, and neuropsychological data in predicting the lateralization and localization of seizure focus in presurgical epilepsy patients. For the purpose of their study, EEG data consisted of a seizure l at eralization i ndex ( SLI), which gave a measure of laterality along a continuum from 1 (purely left lateralized) to +1 (purely right lateralized). The SLI, and how it is calculated, will be discussed in more detail later as it will play an important role a s a criterion in the current study. MRI data was quantified as left right differences in hippocampal volume, and neuropsychological data was compiled from a number of tests administered to the patient as part a standard approach to presurgical neuropsycho logical evaluations at the University of Florida. In what follows, this ba ttery is referred to as the UF standard neuropsychological b attery, or SNB (See Table 1 2 for a full list of tests administered in the SNB). After the neuropsychological tests were administered, specific Domain score s (e.g., verbal memory, nonverbal memory, language, visuoconstruction, and motor) were calculated. For each functional domain (language, memory, etc.), demographically corrected scores on constituent tests were converte d to a common z score metric and averaged (by adding up the z scores and dividing by the total number of tests included in that Domain ) to derive overall Domain score s. These Domain score s, as well as the aforementioned EEG seizure lateralization index an d difference in hippocampal volume (DHF ) found from the MRI were entered, both separately and in various combinations, into leave one evaluate their ability to predict left vs. right temporal lobe seizure lateralization. Subjects were
19 44 right found that EEG data alone c orrectly lateralized patients 89 % of the time, whereas MRI data alone correctly lateralized patients 86 % of the time. Neuropsychological data alone was able to correctly lateralize patients 66 % of the time. Using all three in combination with each other improved the accuracy of lateralization prediction to 95 % From this data it becomes clear that the combination of EEG, MRI, and neuropsychological data serve as an excellent predictor of laterality in populations of individuals who ultimately undergo ant erior temporal lobectomy (ATL) It is known that presurgical neuropsychological evaluation is valuable in predicting post surgical outcome, provided that it correctly localizes the lesion. However, it could be argued that a full day neuropsychological exa mination, which costs $1,500, added relatively little to prediction and could be eliminated in a cost effective environment. Developing shorter, more cost effective approaches to presurgical neuropsychological assessment might be useful to tip that equati on in a more favorable direction that would support the use of neuropsychological assessment as an integral part of standard epilepsy care. The previous study has two notable limitation s The patients included in the study were all patients who eventually underwent surgery for their epilepsy. Therefore, the criterion variable which Moser and his colleagues used for group prediction was final data on which hemisphere was ultimately removed surgically. The first problem is that neuropsychological examinati on results contributed to the surgical decision and were thus not independent of patient classification. Also, i n order for an intensive surgical procedure such as an ATL to be conducted, doctors must be highly confident in the lateralization and localization of seizure onset in a single hemisphere. In the Moser et al. study, of the 26 patients who underwent left ATL
20 surgery, the mean SLI score was .83 (indicating modera tely strong left lateralization). Of the 18 patients who underwent right ATL surgery, the mean SLI score was .63. Thus, in both groups of patients the SLI scores tend to be at the farther ends of the spectrum, representing more classical cases of lateral ization. What is unclear is whether neuropsychological examination would have lateralized or in which all or some of the events were nonepileptic. The consecutive admission design of our study allowed a n evaluation of this possibility. Hypotheses Our study evaluated the validity and utility of the Neuropsychological Assessment Battery ( NAB ) in predicting the lateralization and localization of seizure f ocus in a sample of individuals with documented or suspected epilepsy by comparing the results of the NAB battery to numerical lateralization and localization data available from noninvasive inpatient video EEG. We attempted to address this by posing two primary research questions: Question One The first question related solely to lateralization of seizure focus. We examined how accurately NAB data predicted membership in one of three groups (left lateralized epilepsy, right lateralized epilepsy, and nonl ateralized epilepsy/PNES) using SLI data as the criterion measure. At first glance, it may seem confusing why individuals with genuine but nonlateralizing epilepsy (such as individuals with generalized seizures) would be included in the same group as indi viduals diagnosed with PNES. The reason for this decision was that, within the context of an epilepsy clinical practice, future treatment options for both of these patient populations would not include respective seizure surgery, whereas individuals diagn osed with intractable epilepsy and found to be highly lateralized to the left or right hemisphere with regards to seizure focus may in fact eventually undergo such surgery. In order to determine group membership, the
21 scores obtained on the SLI in our samp le were used. After the base rates of occurrence for each of the three categories within the current population sample were determined, a discriminant function analysis was conducted to examine how well specific NAB scores predicted group membership. Of p rimary interest were two construct score s that were derived from the Memory Module of the NAB, Verbal Memory and Nonverbal Memory The NAB M emory Module consists of four primary subtests, including List Learning, Story Learning, Shape Learning, and Daily Living Memory subtests The List L earning subtest and Story Learning subtests are entirely verbal memory tasks (See Table 1 1 for test descriptions), whereas the Shape Learning subtest is a nonverbal memory task. The Daily Living task was not included in the analysis because it contains both verbal and nonverbal components. In order to calculate these construct score s, methods consistent with those in the Moser et al. (2000) study were conducted. A Verbal Memory construct score was calculated by averagi ng z scores obtained on the List Learning List A Long Delayed Recall task (from the List Learning subtest) and the Story Learning Phrase Unit Delayed Recall task (from the Story Learning subset). A Nonverbal Memory construct score was based upon the obtain ed z score for their performance on the Shape Learning Delayed Recognition task of the Shape Learning subtest. As only one subtest was included in the Nonverbal Memory construct score no mathematical averaging was necessary. Once the Verbal Memory and No nverbal Memory construct score s were calculated, they, along with overall Domain score s for the remaining Module s (Attention, Language, Spatial, and Executive Functions), were entered into a discriminant function analysi s to evaluate the ability of each co nstruct score or Domain score to predict group membership. Hypothesis one : It was hypothesized that the Verbal Memory and Nonverbal Memory construct score s, along with the Language Domain score would be more useful at predicting
22 group membership than other Domain score s individually. In particular, it was hypothesized that the Verbal Memory construct score derived from the NAB would be able to correctly predict group membership at a rate greater than chance in the current sample. Hypothesis two : Th e Nonverbal Memory construct score derived from the NAB appears to have some significant limitations, especially when compared to the Nonverbal Memory construct score calculated in the Moser et al. (2000) study. Unlike the Moser study, whose Nonverbal Mem ory construct score was comprised of multiple spatial delayed recall tasks, the NAB Nonverbal Memory construct score has only one score, and, though clearly a test of spatial memory, is a recognition task rather than a spontaneous recall task. As recognit ion tasks are traditionally less difficult than recall tasks, and less specifically hippocampal dependent because of the contribution of familiarity to recognition performance, scores in this Domain may be a less sensitive predictor of laterality unless th e patient does extremely poorly. Therefore, it was predicted that the Nonverbal Memory construct score derived from the NAB would by itself be unable to correctly classify individuals in the current sample at a rate greater than chance. After the discrimi nant function analysis was conducted, a multiple regression was also conducted to examine the overall amount of variance accounted for by the NAB Module s in regards to laterality in order to give a more general indication of the utility of the NAB in makin g a laterality determination in this population. Question Two The second question related solely to localization of seizure focus, another important factor in determining candidacy for epilepsy surgery. For this question, our primary focus was on patterne d discrepancies in performance among different NAB Module s. Four performance discrepancy scores were calculated for each patient: Memory Language discrepancy Memory Executive Functions discrepancy Memory Attention discrepancy and Memory Spatial
23 discre pancy It was hypothesized that these discrepancy scores would be useful in clinical study, extratemporal classification included all seizure onsets occurring o utside of the temporal lobe, including occipital, parietal, and frontal lobe foci, though the authors are aware that these seizure foci would likely present with very different patterns of neuropsychological performance. The aforementioned discrepancy sco res, along with the individual Module scores by themselves (not calculated as discrepancies between Module s) were entered into a discriminant function analyses to determine 1) which methods provided better model fit and 2) how well they predicted temporal vs. extratemporal group membership based on final clinical judgmen t of seizure focus made by the comprehensive epilepsy p rogram (CEP) team. This judgment took into account all available data, including video EEG, MRI and neuropsychological testing using a standard battery protocol. Hypothesis three : It was hypothesized that the Memory Spatial discrepancy score would provide the best prediction of group membership in regards to localization, as scores on the Memory, Attention, Language, and Executive Funct ions Module s are more likely to be positively correlated since all are in varying ways related to frontal lobe functioning; therefore, Memory Attention, Memory Executive Function, and Memory Language discrepancy scores were hypothesized to be less predicti ve than Memory Spatial discrepancy scores.
24 Table 1 1. Domain s and Subtests of the Neuropsychological Assessment Battery (NAB) Test Description Screening Orientation Questions about orientation to self, time, place, and situation Screening Digits Forward Repetition of orally presented digits Screening Digits Backward Repetition of orally presented digits Screening Numbers & Letters Two timed tasks (parts A & B) involving letter cancellation and letter cancellation plus serial addition, respective ly Screening Auditory Comprehension Three part test that requires the examinee to demonstrate comprehension of orally presented commands Screening Naming Visual confrontation naming task in which the examinee state the name of an object depicted in a photograph; semantic and phonemic cues are provided if necessary Screening Shape Learning Single trial visual learning task involving multiple choice immediate recognition recall of five visual stimuli, followed by a delayed recognition task Screening St ory Learning Verbal learning task involving immediate and delayed free recall of a two sentence story Screening Visual Discrimination Visual match to target paradigm, in which the examinee matches a target visual design from an array of four similar desig ns presented beneath the target Screening Design C onstruction Visuoconstruction assembly task using plastic manipulatives (tans) to copy two dimensional target designs (tangrams) Screening Mazes Three timed paper and pencil mazes of increasing difficulty Screening Word Generation Timed task in which the examinee creates three letter words from a group of six letters (two vowels, four consonants) that are presented visually Attention Orientation Questions about orientation to self, time, place, and situation Attention Digits Forward Repetition of orally presented digits Attention Digits Backward Repetition of orally presented digits Attention Dots Delayed recognition span paradigm, in which an array of dots is exposed for a brief period, followed by a blank inteference page, followed by a new array with one additional dot; examinee points to new dot Attention Numbers and Letters Four timed tasks (Parts A, B, C, and D) involving letter cancellation, letter counting, serial addition, and letter canc ellation plus serial addition, respectively Attention Driving Scenes Daily living task in which the examinee is first presented with a drawing of a driving scene as viewed from behind a steering wheel, and then shown another scene and asked to say and po int to everything that is new, different, or missing relative to the previous scene; this is continued for four additional scenes Language Oral Production Speech output task in which the examinee orally describes a picture of a family scene Language Auditory Comprehension Six part test that requires the examinee to demonstrate comprehension of orally presented instructions; tasks include performing one to four step commands, the concepts of before/after, above/below, and right/left, body part identification, and paper folding Table information obtained from Chapter 1 (pages 6 10) of the Neurop sychological Assessment Battery: Administration, Scoring and Interpretation Manual (Stern & White, 2003)
25 Table 1 1. Continued Test Description Language Naming Visual confrontation naming task in which the examinee states the name o f a pictured object; semantic and phonemic cues are provided if necessary Language Reading Comprehension Two part test that requires the examinee to demonstrate reading comprehension of single words and of sentences by pointing to multiple choice written words and sentences that match visual stimuli Language Writing Narrative writing task in which the examinee is shown the same drawing of a family scene used in the Oral Production test and asked to write about it; the writing sample is scored with regard to legibility, syntax, spelling, and conveyance Language Bill Payment Daily living task in which the examinee is given a utility bill statement, check ledger, c heck, and envelope, and asked to follow a series of eight commands requiring oral and written responses of increasing complexity Memory List Learning Verbal learning task involving three trials of a 12 word list, followed by an interference list, and then short delay free recall, long delay free recall, and long delay forced choice recognition tasks; the word list includes three embedded semantic categories with four words in each category Memory Shape Learning Visual learning task involving three learni ng trials and multiple choice immediate recognition of nine visual stimuli, followed by delayed recognition and forced choice recall Memory Story Learning Verbal learning task involving immediate and delayed free recall of a five sentence story; two learning trials are provided, and recall is scored for both verbatim and gist elements Memory Daily Living Memory Verbal learning task involving three trial learning with immediate recall, delayed recall, and delayed multiple choice recognition of informa tion encountered in daily living, including medication instructions, and a name, address, and phone number Spatial Visual Discrimination Visual match to target paradigm, in which the examinee matches a target visual design from an array of four similar de signs presented beneath the target Spatial Design C onstruction Visuoconstruction assembly task using plastic manipulatives (tans) to copy two dimensional target designs (tangrams) Spatial Figure Drawing Visuoconstruction drawing task involving a copy and immediate recall of geometric figure of moderate complexity; the production is scored for the presence, accuracy, and placement of the elements, as well as overall organizational skill Spatial Map Reading Daily living task in which the examinee answers questions (presented both orally and in writing) about a city map that has a compass rose and mileage legend Executive Functions Mazes Seven timed paper and pencil mazes of increasing difficulty Executive Functions Judgment Daily living test in which th e examinee answers 10 judgment questions pertaining to home safety, health, and medical issues Executive Functions C ategories Classification and categorization task in which the examinee generates two group categories based on photographs and verbal information about six people Executive Functions Word Generation Timed task in which the examinee creates three letter words from a group of eight letters (two vowels, six consonants) that are presented visually
26 Table 1 2. Test Composition of the UF Standard Neuropsychological Battery (SNB) Test Description Wechsler Abbreviated Scale of Intelligence (WASI; Psychological Corporation, 1999) Block Design, Matrix Reasoning, Vocabulary, an d Similarities subtests adapted from Wechsler Scales of Intelligence Rey Complex Figure Test and Recognition Trial (Myers & Myers, 2004) Figural memory; Measure of v isuoconstructive ability that incorporates immediate and delay recall. California Verbal L earning Test 2nd Edition (Delis et al., 2000) Verbal memory test that assesses learning strategy, immediate and delayed recall, recognition, interference, and errors. Wechsler Memory Scale R; Logical Memory I, II; Visual Reproduction, I, II; Wechsler, 1987) Measure of verbal and figure /nonverbal m emory, respectively Controlled Oral Word Association (COWA; S preen & Benton, 1977) Verbal fluency to alphabet letter (e.g., C,F,L). Category Fluency Animals (Tombaugh et al., 1999) Verbal fluency to a quarters. Boston Naming Test 2nd Edition (Goodglass & Kaplan, 2000) Confrontation naming using large ink drawings WASI Block Design subtest (Wechsler, 1997) Visuoconstruction measure requiring reproduction of 2 dimensional designs with blocks Wisconsin Card Sorting Test (Heaton, 1981) Participants sort the cards based on their own principle. Measure of mental flexibility and problem solving Trail Making Test (Reitan, 1958) Visuomotor speed, tracking and attention between randomly placed numbers and letters WAIS III Digit Span (Wechsler, 1997) R equires su stained attention and span Finger Tapping (Halstead, 1947; Reitan & W olfson, 1993) Speeded gross motor movement Grooved Pegboard Test (Klove, 1963) Speeded fine motor movement State Trait Anxiety Inventory (STAI; S peilberger, 1968) 40 item self evaluation questionnaire assessing state and trait anxiety Beck Depression Inventory II (Beck, Steer, & Brown, 1996) 21 item questionnaire assessing elements of depression
27 CHAPTER TWO METHODS Participants Forty five 1 individuals with documented or suspected epilepsy were recruited over a 1 6 month period from the comprehensive e p ilepsy p rogram (CEP) at Shands at the University of Florida in Gainesville, Florida. The CEP inpatient unit admits, on average, 2.7 new inpatients per week for noninvasive (Phase I) video encephalographic (V EEG) monitoring, and this admission rate was relatively consistent throughout the data collection phase of our study. All participants were identified, recruited and assessed during their Inpatien t Phase I hospital stay in the epilepsy monitoring u nit (EMU). Participants were included if they were 18 years o f age or older, had documented or suspected epilepsy (from history and/or prior neurodiagnostic tests), were capable of undergoing neuropsychological testing and were willing to participate. Participants were excluded if they suffered from severe developm ental disability or mental retardation resulting in IQ < 70 or an inability to participate, severe visual or auditory sensory deficits, history of Axis I psychiatric disturbance sufficiently significant to have resulted in hospitalization, or a current or past diagnosis of substance abuse or substance dependence 2 In addition, participants who experienced no electrographic events or definitive psychogenic non epileptic seizures ( PNES ) during phase I monitoring were excluded from discriminant function analy ses (DFA) and mul tiple r egression analyses as th e necessary calculation of the s ei zure lateralization i ndex (SLI) would be unable to be determined for these individuals. 1 This number was derived by estimating the minimum number of subjects per variable (5 to 7) required to provide adequate model stability during Discriminant Function Analyses. 2 Due to patient confiden tiality requirements, treating n eurologists performed i nitial pre screenings for recruitment. Precise numbers are unavailable for how many potential parti cipants were eliminated by the n eurologists prior to evaluating patient suitability for participation based on exclusionary criteria.
28 Of the 45 participants, gender composition was some what unbalanced, with 73% (n = 3 3) female par ticipants compared to 27% (n = 12) male participants. Participants ranged in age from 21 to 66 years ( M = 43.16, SD = 12.24). Education level ranged from 9 to 18 years ( M = 13.36, SD = 2.24). The sample was 78% Caucasian (n = 35), 16 % Afric an American (n = 7), 2 % Latino (n = 1), and 4% Native American (n = 2), a sample relatively in line with the Caucasian and African American demographics of the County and State in which the study took place, though the percentage of Latino participants was lower than expected. Approxima tely 93% of participants in the study (n = 42) identified themselves as being right handed, whereas approximately 7% of participants (n = 3) identified themselves as left handed. Sixty percent of participants in the study ( n = 27) were administered the Form 1 version of the N europsychological Assessment Battery (N AB ) while 40 % of participants (n = 18) were administered the Form 2 version. Of the 45 participants in the stud y, 20 participants (44.4% ) experienced clinical even ts during their inpatient stay that were in fact associated with genuine epileptic activity. Twelve participants (26% ) experienced clinical events during their inpatient stay which, upon review, involved no epileptiform activity and were of sufficient type and quality to be confidently diagnosed by the attending neurologist post discharge as psychogenic non epileptic seizures (PNES). Fina lly, 13 participants (29% ) who completed all testing for inclusion in our study, and who were subjected to multiple common triggers for seizure activity (e.g., sleep deprivation, exercise) nonetheless had no clinical events during their inpatient stay and as a result, their seizures were not definit ive ly diagnosed/localized. Previous demographic information data was available for comparison from an existing database of 362 patients participating in this program. Unlike our study, the database contained a relatively comparable number of men and women (Male n=173; Female n=179). Caucasians
29 ma de up approximately 88.4% of patients in the database (n= 320), with Afri can Americans (n=21, 5.8%), Hispanics (n=18, 5%), Asians (n=1, 0 0 .6% ) comprising the remainder of the sample. Mean level of education was approximately high school/ general equivalency diploma ( GED ) range (12 years), with a standard deviation of 2.4 years (minimum = one year education; maximum = 21 years education). Procedure Participants were given the full NAB during their admission to the inpatient epilepsy monitoring u nit (EMU) at Sh ands Hospital. All in dividuals participating in the comprehensive epilepsy p rogram (CEP) at the University of Florida undergo MRI and Phase I monitoring, and many also undergo neuropsy chological testing with the UF standard neuropsychological b attery (SNB) as part of their diagnostic workup if they are considered to be a potential surgery candidate. The magnetic resonance imaging ( MRI ) results are analyzed by a board certified neuroradiologist who provides diagnostic consultation to the CEP team. The elec troencephalographic ( EEG ) findings are analyzed by a team of board certified neurologists and e pileptologists to determine localization of onset and seizure type. The SNB is supervised by licensed or board certified neuropsychologists. Once all data are collected, the interdisciplinary CEP team meets to make a clinical judgment based on all available data in order to attempt to localize/lateralize seizure onset and select an appropriate mode of treatment. The average time of administration per participan t was approximately 4 hours. Each participant was tested while in his or her room on the EMU. To control for variations in NAB administration, only the principal investigator and three trained research assistants tested participants. Besides the Screeni ng Module which was always administered first (as is standard) the remaining five Module s were administered to each participant in random order to minimize the likelihood of practice or order effects.
30 Statistical Analyses Classification with Neuropsych ological Tests In surgical epilepsy centers, the results of the presurgical neuropsychological battery are used to predict lateralization and localization of seizures for purposes of surgical planning. Regarding lateralization, two neuropsychological pat terns appear to be most critical. First, discrepancies between verbal and nonverbal memory are commonly seen in unilateral temporal lobe cases, particularly those involving seizure onsets in the language dominant temporal lobe where verbal memory is selectively impaired Second, discrepancies between language and spatial functions are seen in unilateral cases, with language being most affected in left onset epilepsy and spatial ability being most affected with onsets in the ri ght hemisphere. Seizure Lateralization Index (SLI) To perform these analyses, we first classified patients as having left lateralized epilepsy, right lateralized epilepsy, or nonlateralized epilepsy/PNES and as having seizure foci in the ta (in the case of EEG, the seizure lateralization i ndex [SLI], adapted from Moser et al. (2000), was calculated in order to estimate eizures were lateralized). The i ndex, displayed in Equa tion 2 1: [(R L) 0.5N [R L/|R L|]/R+L+N (2 1) is calculated by determining the total number of ri ght lateralized (labeled in E quation 2 1 as R), left lateralized (labeled in the Equation 2 1 as L), and nonlateralized (labeled in E quation 2 1 as N) seizure s occurring while undergoing EEG monitoring The numerator of this equation represents the degree to which seizures are lateralized to the right (positive values) or left (negative values) by first calculating a lateralization index (R L 0.5N). The numbe r of nonlateralized seizures (N) is given a weight of 0.5 in the equation, as these seizures would have an equal chance of actually originating in either hemisphere, and are generally less definitive in
31 prediction than seizures with a clear hemispheric lat eralization. The (R L/|R L|) term in the numerator determines the arithmetic sign as positive or negative. The resulting index is then divided by the total number of electrographic eve nts (R+L+N), yielding a SLI that varies from 1 to +1 SLI values near ing 1.00 suggest seizure events lateralized entirely to the left hemisphere, whereas values nearing 1.00 suggest seizure events lateralized entirely to the right hemisphere. Values close to 0.00 suggest nonlateralized epilepsy (Moser, 2000). In this way a patient who has six left later alized seizures, no right l ateralized seizures, and no non lateralized seizures will obtain an SLI value of 1, a theoretically perfect prediction of left lateralization according to this equation. Conver sely, a patient wh o has 12 left lateralized seizure s, six right l ateralized seizures, and no non lateralized seizures, would obtain an SLI value of 0.5. Without appropriate weighting that considers the total number of seizures occurring, these two patients would have the s ame SLI value, even though they present with two very differ ent clinical presentations based on this data (Moser, 2000). Participants who underwent 24 hour video EEG monitoring and were later determined to have PNES were given an SLI value of 0.00 in or der to be included in quantitative analyses. The SLI was used as the primary criterion to address Question one As stated previously, the 13 participants who experienced no electrographic events or definitive PNES during phase I monitor ing were excluded from discriminant function and multiple r egression analyses as the necessary calculation of the seizure lateralization index (SLI) could not be calculated for these individuals. Exploratory and Advanced Analyses After the SLI was calculated and each derived scores specific to our stu dy including the Verbal Memory construct score, Nonverbal Memory construct score and four NAB Module discrepancy score s) were obtained along with their overall EMU findings regarding seizure localization, initial analyses were conducted to examine
32 the data including frequencies, descriptives, tests for normality, test for group differences, and correlations between variables. In order to test hypotheses relating to predict seizure lateralization separate discriminant function analyses were run using SLI as the outcome variable and (a) the Verbal Memory construct score as the sole predictor of lateralization, (b) the Nonverbal Memory construct score as the sole pr edictor of lateralization, and (c) the Verbal Memory and Nonverbal Memory construct score s together in combination with the four remaining full NAB Module score s [Attention, Language, Spatial, and Executive Functions] as predictor variables. The five Scre ening Module Domain score s in combination as predictor variables were also examined, though no a priori hypotheses were made with regard to their ability to predict lateralization. Two follow up multiple r egression analyses were then run using the SLI as the outcome variable In the first analysis, the Verbal Memory and Nonverbal Memory construct score s were used as predictor variables In the second, the five NAB Core Module score s (Attention, Language, Memory, Spatial, Executive Functions) served as pr edictor variables to examine the overall amount of variance accounted for by the NAB Module s in predicting seizure laterality of seizure focus, discriminant fun ction analyses were run for Question Two using all patients who displayed electrographic seizures ( excluding participants who were classified as having PNES ). These analyses were conducted with the rationale that, while reducing the overall sample size of would result With the PNES subjects removed, the goal was to use NAB data to predict membership in the temporal lobe (n = 14) vs. extratemporal (n = 6) groups. Separate analyses were run using localization s tatus ( temporal vs. e xtratemporal ) as the outcome variable and (a)
33 the Memory Attention discrepancy score as the sole predictor of localization, (b) the Memory Language discrepancy score as the sole predictor, (c) the Memory Spatial discrepancy score as the sole predictor, (d) the Memory Executive Functions discrepancy score as the sole predictor, (e) these four NAB derived discrepancy score s in combination as predictor variables and (f) the five Core NAB Module score s (Attention, Language, Memory, Spatial, Executive Functions) in combination as predictor variables. Since data were available, additional analyses were conducted using (a) the Verbal Memory construct score as the sole predictor, (b) the Nonverbal Memory c onstruct score as the sole predictor, and (c) the five Screening Module Domain score s in combination as predictor variables As above, no a priori hypotheses were made with regard to the ability of these variables to predict seizure localization to the te mporal lobe. Stepwise DFAs were also run when variables were used in combination and at least one variable was statistically significant. In all discriminant function analyses in our study, only cross validated grouped classification results were reviewe d in order to increase overall generalizability to epilepsy populations as a whole.
34 CHAPTER T HREE RESULTS Data were analyzed using the Statistical Package for the Social Sciences (SPSS) 18.0 software package. A minimum significance level of p < .05 was used throughout the analyses. Normality and Group Differences on the Neuropsychological Assessment Battery For ease of table reading, a variable name k ey was created and is displayed in Table 3 1. Descriptive statistics, including mean sc ores, standard de viations, and one w ay analysis of variance ( ANOVA ) results by group based on epilepsy monitoring unit (E MU ) findings (electrographic seizures, psychogenic non epileptic seizures [ PNES ] n o events) on all key N europsychological Assessment Battery (N AB ) variables for our study are included in Table 3 2. Seizure lateralization i ndex (SLI) distribution in participants experiencing electrographic seizures are included in Figure 3 1 (note: both subjects shown in Figure 3 1 with an SLI value of 0.33 had two seizures which localized to the right temporal lobe and one seizure which localized to the left temporal lobe. As both of these subjects were ultimately diagnosed with bitemporal lobe epilepsy and not considered to be potential candidates for seizure sur gery, they were subsequently classified in our nonlateralized group during laterality based analyses). As can be seen from Table 3 2, NAB scores obtained from the sample in the present study were positively skewed, meaning the mean z score obtained for ea ch NAB specific variable in the study was lower than the average z standardization sample Because of this, normality was examined using the Kolmogorov Smirnov statistic (Corder & Foreman, 2009), which examines goodness of fi t between the observed distribution and the normal distribution. With the exception of the Nonverbal Memory construct score (p < .01), all variables fit normality (p > .05). Because of this, subsequent analyses regarding group differences and the Nonverb al Memory construct score were conducted using i ndependent
35 s amples Mann Whitney U tests (none of these tests showed significant group differences on any variables in our study with respect to the Nonverbal Memory construct score ). For all other variab les, normality was assumed and one w ay ANOVAs with follow up independent samples t tests were run. Group Differences by Demographics No significant group differences (p > .05) were found on any NAB test Domain variables with regard to gender or h andedness. An unexpected group diffe rence was found with regard to f orm, such that the 27 participants in the study who were tested with Form 1 of the NAB had a lower Language Module Mean Domain score (z = .86, s.d. = 1.34) than did the 18 participants who were test ed with Form 2 of the NAB (z = .01, s.d. = 1.44) with a t (43) = 2.07, p = .045. This is an unusual finding and may potentially be explained by error variance given our small sample as there is no significantly plausible reason why this group differ ence would have occurred (both f orms were given in alternating fashion to participants, and all trained test administrators utilized both forms). Psychometric data included in Chapter 5 of the NAB Psychometric and Technical Manual (White & Stern, 2003) showed that tests of equivalent forms reliability between Form 1 and Form 2 of the NAB Language Module using generalizability theory produced a generalizability (G) coefficient of .62. According to Cicchetti & Sparrow (1981), generalizability coefficients of .60 or higher are regarded as demonstrating very good reliability. Group Differences by EMU Classification When comparing the individuals in our study who had electrographic seizures (n = 20) to those who were classified as having psychogenic NES (n = 12) v s. those who experienced no clinical events during one w ay ANOVAs, no statistically significant group differences in any NAB variable were found.
36 Group Differences by Lateralization Classification When comparing groups based on late ralization classificatio n (left lateral ized seizures [n = 4] vs. right lateralized seizures [n = 11] vs. nonlateralized seizures/PNES [n = 17]), a statistically significant group difference ( F [2, 29] = 3.73, p < .05) was found on the Memory Sp atial discrepancy score during one w ay ANOVAs. Subsequent t tests revealed statistically significant group differences between the left lateralized and right lateralized groups (t [13 ] = 3.24, p < .01) as well as the left lateralized and nonlateralized groups (t  = 2.22, p < .05), whi le tests of group differences between the right late r alized and nonlateralized groups were nonsignificant (t  = .396, p > .05). Individuals in the left lateralized group had a mean z score on this NAB derived variable over one standard deviation (Mean z = 1.76 s.d. = 0.93 ) lower than individuals in the right lateralized (Mean z = 0.45 s.d. = 0.60 ) or nonlateralized (Mean z = 0.58 s.d. = 0.96 ) groups. Further explication of this group difference is discussed i n the Correlations section Group Dif ferences by Localization Classification With regards to individuals in the study who were classified as having temporal lobe differences were found on the Scr eening Spatial Domain score (t  = 2.25, p < .05) Screening Executive Functions Domain score (t  = 2.47, p < .05), Attention Module score (t  = 2.31, p < .05) Memory Attention discrepancy score (t  = 2.38, p < .05) and Verbal Memory c onstruct score s (t  = 4.30, p < .001) Individuals with temporal lobe epilepsy performed approximately one standard deviation or more below individuals having seizures of extratemporal focus on all five of these variables. T test values and associat ed significance levels for each of these variables are also displayed in Table 3 5.
37 Correlations Bivariate correlations were conducted to further explore relationships between variables. Positive correlations between education and all NAB derived scores (excludi ng the discrepancy s cores) were found. Nearly all of these scores were in fact statistically significant at the p <.05 level, with the exceptions of the NAB Total Score, Attention Module score, and Language Module score. All NAB derived scores (a gain, excluding the discrepancy scores) were positively correlated with each other (p < .001). Notably, we observed a positive correlation of SLI score with Memory Spatial discrepancy score r = .376, p < .05 ), suggesting that as the Memory Spat ial discrepancy score increased (suggesting more intact memory than spatial ability), the SLI value also increased (towards a value of 1.0, which represents a strongly right lateralized seizure focus). This correlation suggests that individuals with a hig her Memory Spatial discrepancy score were associ ated with more right lateralized seizures, a correlation that provides some support for the utility of the NAB in predicting seizure lateralization. Question One Discriminant Function Analyses Descriptive st atistics (including means, standard deviations and one w ay ANOVAs) for groups by lateralization classification (left lateralized seizures, right lateralized seizures, and nonlateralizing seizures/PNES) are included in Table 3 3. The classification rates fo r each predictor variable in our hypotheses, including percentage correct classification and information regarding base rates of occurrence for each lateralization subset, are included in Table 3 4. As can be seen in Table 3 4, no variables or combinations of variables in our study were able to correctly classify individuals by lateralization at a rate greater than chance given the base rates of occurrence in our study Therefore, the hypothesis that the Verbal Memory construct
38 score derived from the NAB w ould be able to correctly predict lateralization at a rate greater than chance was not supported. Our second Hypothesis (that the Nonverbal Memory construct score derived from the NAB would be unable to predict lateralization at a rate greater than chance ) was not disconfirmed Stepwise DFAs were not conducted for these lateralization based analyses since no predictor variables reached statistical significance. Even the Memory Spatial discrepancy score the only variable in our study which was significan tly correlated with SLI r = .376, p = .034) was unable to classify individuals based on lateralization at a rate greater than chance. The following is a description of each set of predictors used in these discriminant analyses, the resul ting Wilks values, and levels of statistical significance. Entering the Verbal Memory construct score as the sole predictor of lateralization yielded the following results: Wilks = 0.949 (p = .47), n = 32; the Nonverbal Memory construct score as the s ole predictor: Wilks = .985 (p = .80), n = 32; The two Memory derived construct score s in combination with the four remaining NAB Module score s (Attention, Language, Spatial, Executive Functions) as predictors: Wilks = .542 (p = .18), n = 32; The five Screening Module Domain score s in combination as predictors: Wilks = .791 (p = .79), n = 32. Similar discriminant function a nalyses were run for Question One excluding participants who were classified as having NES, ca using the groups to be: left lateralized n = 4 right lateralized n = 11, non lateralized n = 5. None of these analyses yielded statistically significant results, and none were able to achieve greater than chance correct classification into laterality gro ups based on participant NAB scores
39 Multiple Regression Analyses Multiple regressions were conducted to examine the overall amount of variance accounted for by the NAB Module s in regards to laterality in order to give a more general indication of the util ity of the NAB with this particular patient population. A multiple regression using the Verbal Memory and Nonverbal Memory construct score s as predictor variables was nonsignif icant (F = .570, p = .572, a djusted R 2 = .029). A second multiple regressi on using the five Core NAB Module s (Attention, Language, Memory, Spatial, Executive Functions) as predictor variables, while appearing to be a better overall model, was also nonsignifi cant (F = 1.846, p = .139, a djusted R 2 = .120). The small n availab le for inclusion (n = 32) likely reduced the overall stability of these regression analyses. Question Two Descriptive statistics (including means, standard deviations, and independent samples t tests) by localization classification excluding participants with PNES are included in Table 3 5. The classification rates for each predictor variable in our hypotheses, including percentage correct classification and information regarding base rates of occurrence for each localization subset when PNES patients wer e excluded from the analysis are included in Table 3 6. Unlike lateralization prediction, some variables in our study were able to predict seizure localization at a rate greater than chance. Five variables predicted localization at statistically signifi cant levels, including the Screening Spatial Domain score r = .469, p = .037), Screening Executive Functions Domain score r = .504, p = .024), Attention Module score r = .479, p = .033), and Memory Attention discrepancy sco re r = .489, p = .029). Lastly, the Verbal Memory construct score which was derived primarily to evaluate lateralization in our analyses (and which did a poor job at doing so), turned out to be r = .712, p < .001), such that
40 higher (better) Verbal Memory c onstruct score s were associated with membership in the extratemporal seizure focus group. In DFA a nalyses, the Memory Attention discrepancy score by itself was able to correctly classify 92.9% of temporal lobe cases, but was able to classify cases with e xtratemporal seizure focus with only 50% accuracy. Even with PNES subjects excluded, the other three discrepancy scores (Memory Language discrepancy Memory Spatial discrepancy and Memory Executive Functions discrepancy ) were unable to adequately classif y individuals based on group membership. Using all four discrepancy score variables in combination in a s tepwise DFA provided no greater classification accuracy than using the Memory Attention discrepancy score alone. Using the 5 core NAB Module s (Attent ion, Language, Memory, Spatial, Executive Functions) in combination in a s tepwise DFA, temporal seizure focus was predicted at a rate slightly greater than chance (85.7%) However, this combination of variables was able to predict extratemporal seizure fo cus with only 50% accuracy. Similar findings were found when using the 5 Screening Module s Domain score s (Screening Attention, Screening Language, Screening Memory, Screening Spatial, Screening Executive Func tions) in combination, as this s tepwise DFA was able to correctly classify 92.9% (13/14) of the cases with temporal lobe seizures, but only 50% (3/6) of the cases whose seizures were of extratemporal focus. The Verbal Memory construct score while originally intended to examine lateralization rather than localization of seizure focus, appeared to exhibit the best predictive ability of any variable in our entire study. As the sole predictor of seizure localization in individuals with known seizures, the Verbal Memory construct score correctly classifi ed 90% of cases correctly grouping 92.9% (13/14) of the participants with temporal lobe seizure focus and 83.3% (5/6) of the participants with extratemporal seizures
41 Despite our predictions that the NAB derived discrepancy scores (Memory Attention discrep ancy Memory Language discrepancy Memory Spatial discrepancy and Memory Executive Functions discrepancy ) would be particularly useful in predicting localization, only the Memory Attention discrepancy score was able to predict localization at a rate great er than chance. As such, our third Hypothesis (that the Memory Spatial discrepancy score would provide the best prediction of group membership in regards to localization) was not supported. The following is a description of each set of predictors used i n these discriminant analyses, the resulting Wilks values, and levels of statistical significance. Entering the Memory Attention discrepancy score as the sole predictor of localization when PNES participants were excluded yielded the following results: Wilks = .761 (p = .029), n = 20; the Memory Language discrepancy score as the sole predictor: Wilks = .997 (p = .80), n = 20; the Memory Spatial discrepancy score as the sole predictor: Wilks = .999 (p = .90), n = 20; the Memory Executive Functions discrepancy score as the sole predictor: Wilks = .955 (p = .37), n = 20. When the four NAB derived discrepancy scores (Memory Attention discrepancy Memory Language discrepancy Memory Spatial discrepancy and Memory Executive Functions discrepancy ) were used in combination as predictors, the following results were obtained: Wilks = .652 (p = .14), n = 20; the four NAB derived discrepancy scores (Memory Attention discrepancy Memory Language discrepancy Memory Spatial discrepancy and Memory Execu tive Functions discrepancy ) in combination as predictors in a s tepwise DFA: Wilks = .761 (p = .029), n = 20. The Verbal Memory construct score as the sole predictor of localization yielded the following: Wilks = 0.492 (p < .001), n = 20, while the No nverbal Memory construct score as the sole predictor produced a Wilks = .836 (p = .076), n = 20.
42 The five Core NAB Module score s (Attention, Language, Memory, Spatial, Executive Functions) used in combination as predictors of localization yielded the f ollowing : Wilks = .672 (p = .29), n = 20; the five Core NAB Module score s (Attention, Language, Memory, Spatial, Executive Functions) used in combination as predictors of localization in a stepwise analysis: Wilks = .771 (p = .033), n = 20. When the f ive NAB Screening Module Domain score s (Screening Attention, Screening Language, Screening Memory, Screening Spatial, and Screening Executive Functions) were entered in combination as predictors of localization, the following results were obtained: Wilks = .448 (p = .029), n = 20; the five NAB Screening Module Domain score s (Screening Attention, Screening Language, Screening Memory, Screening Spatial, and Screening Executive Functions) in combination as predictors of localization in a stepwise analysis: W ilks = .746 (p = .024), n = 20.
43 Table 3 1. Variable Name Key Variable Variable c oding Screening NAB Total Score S NAB Screening Attention Domain score S ATT Screening Language Domain score S LAN Screening Memory Domain score S MEM Screening Spatial Domain score S SPT Screening Executive Functions Domain score S EXE Full NAB Total s core NAB Attention Module score ATT Language Module score LAN Memory Module score MEM Spatial Module score SPT Executive Functions Module score EXE Verbal Memory construct score VMEM Nonverbal Memory construct score NVMEM Memory Attention discrepancy score MEMATTDISC Memory Language discrepancy score MEMLANDISC Memory Spatial discrepancy score MEMSPTDISC Memory Executive Functions discrepancy score MEMEXEDISC
44 Table 3 2. Descriptive Statistics by Seizure Classification/E pilepsy Monitoring Unit (E MU ) Outcome Grand m ean (SD) (N=45) Electrographic seizures m ean (SD) (N=20) PNES m ean (SD) (N=12) No events m ean (SD) (N=13) One w ay ANOVA* Age 43.16 (12.24) 40.95 (12.97) 47.17 (9.13) 43.85 (13.51) F (2,42) = 0.97 Education 13.36 (2.24) 13.15 (2.18) 13.50 (2.71) 13.54 (1.98) F (2, 42) = 0.15 S NAB 1.21 (1.30) 1.53 (1.16) 1.25 (1.33) 0.70 (1.43) F (2, 42) = 1.63 S ATT 1.11 (1.33) 1.18 (1.28) 1.18 (1.39) 0.94 (1.45) F (2, 42) = 0.14 S LAN 1.60 (1.23) 1.94 (1.10) 1.52 (1.45) 1.16 (1.14) F (2, 42) = 1.70 S MEM 1.56 (1.27) 1.76 (1.20) 1.48 (1.47) 1.34 (1.24) F (2, 42) = 0.45 S SPT 0.70 (1.28) 0.93 (1.39) 0.60 (1.11) 0.44 (1.29) F (2, 42) = 0.64 S EXE 0.99 (1.24) 0.88 (1.28) 1.13 (1.28) 1.03 (1.21) F (2, 42) = 0.16 NAB 0.78 (1.06) 0.65 (1.02) 1.12 (1.13) 0.68 (1.10) F (2, 42) = 0.80 ATT 1.05 (1.26) 1.02 (1.24) 1.08 (1.18) 1.06 (1.45) F (2, 42) = 0.01 LAN 0.52 (1.43) 0.71 (1.02) 0.60 (1.43) 0.13 (1.93) F (2, 42) = 0.66 MEM 0.23 (1.59) 0.14 (1.47) 0.77 (1.52) 0.12 (1.83) F (2, 42) = 1.03 SPT 0.80 (1.28) 1.18 (1.20) 0.68 (1.27) 0.33 (1.34) F (2, 42) = 1.87 EXE 0.91 (1.35) 1.20 (1.30) 0.54 (1.35) 0.80 (1.43) F (2, 42) = 0.97 VMEM 0.34 (0.94) 0.30 (1.03) 0.55 (0.89) 0.22 (0.88) F (2, 42) = 0.41 NVMEM 0.02 (1.13) 0.20 (1.12) 0.13 (1.32) 0.11 (1.02) F (2, 42) = 0.44 MEMATTDISC 0.52 (0.93) 0.74 (0.76) 0.40 (1.01) 0.28 (1.09) F (2, 42) = 1.11 MEMLANDISC 0.05 (0.80) 0.14 (0.85) 0.05 (0.80) 0.03 (0.80) F (2, 42) = 0.21 MEMSPTDISC 0.83 (1.08) 0.90 (0.93) 0.31 (0.83) 1.18 (1.36) F (2, 42) = 2.24 MEMEXEDISC 0.14 (0.90) 0.18 (0.74) 0.54 (0.72) 0.26 (1.13) F (2, 42)= 2.84 F rati os reaching significance at p < .05 are annotated by a *; p < .01, **; p <.001, ***
45 Table 3 3. Descriptive Statistics by Lateralization Classification All p articipants m ean (SD) (N=32) Left lateralized s eizures m ean (SD) (N=4) Right lateralized s eizures m ean (SD) (N=11) Nonlater alizin g s eizures/PNES m ean (SD) (N=17) One w ay ANOVA* Age 43.28 (11.92) 46.75 (14.18) 42.55 (12.85) 42.94 (11.44) F (2, 29) = 0.18 Education 13.28 (2.36) 14.25 (2.06) 12.91 (2.39) 13.29 (2.47) F (2, 29) = 0.46 S NAB 1.42 (1.21) 2.23 (0.25) 1.21 (1.26) 1.37 (1.28) F (2, 29) = 1.10 S ATT 1.18 (1.30) 1.98 (0.50) 0.98 (1.28) 1.13 (1.42) F (2, 29) = 0.90 S LAN 1.79 (1.24) 3.00 (0.51) 1.60 (1.08) 1.60 (1.34) F (2, 29) = 2.38 S MEM 1.65 (1.30) 2.77 (0.45) 1.38 (1.27) 1.57 (1.35) F (2, 29) = 1.85 S SPT 0.80 (1.28) 1.54 (0.94) 0.66 (1.46) 0.73 (1.24) F (2, 29) = 0.74 S EXE 0.98 (1.26) 1.47 (0.86) 0.74 (1.35) 1.01 (1.30) F (2, 29) = 0.49 NAB 0.82 (1.07) 0.43 (0.64) 0.32 (0.79) 1.24 (1.16) F (2, 29) = 3.21 ATT 1.04 (1.20) 1.84 (0.56) 0.70 (1.16) 1.08 (1.28) F (2, 29) = 1.37 LAN 0.67 (1.17) 0.98 (0.26) 0.85 (0.92) 0.48 (1.43) F (2, 29) = 0.48 MEM 0.38 (1.50) 0.08 (0.95) 0.30 (1.00) 0.50 (1.87) F (2, 29) = 0.15 SPT 1.00 (1.23) 1.58 (0.76) 0.83 (1.40) 0.97 (1.22) F (2, 29) = 0.54 EXE 0.95 (1.34) 2.23 (0.36) 0.87 (1.32) 0.71 (1.38) F (2, 29) = 2.30 VMEM 0.39 (0.98) 0.90 (0.66) 0.18 (0.93) 0.41 (1.06) F (2, 29) = 0.79 NVMEM 0.08 (1.19) 0.10 (0.73) 0.27 (1.23) 0.01 (1.28) F (2, 29) = 0.23 MEMATTDISC 0.61 (0.86) 0.93 (0.88) 0.68 (0.64) 0.49 (1.00) F (2, 29) = 0.46 MEMLANDISC 0.07 (0.83) 0.37 (1.02) 0.04 (0.90) 0.06 (0.77) F (2, 29) = 0.35 MEMSPTDISC 0.68 (0.93) 1.76 (0.93) 0.45 (0.60) 0.58 (0.96) F (2, 29) = 3.73* MEMEXEDISC 0.09 (0.80) 0.40 (0.47) 0.17 (0.76) 0.37 (0.81) F (2, 29) = 2.55 F rati os reaching significance at p < .05 are annotated by a *; p < .01, **; p <.001, ***
46 Table 3 4. Percentage Correct Classification of Seizure Lateralization Using NAB Variables as Predictors Predictor v ariables Left lateralized s eizures (n=4; 12.5% base rate of occurrence) Right lateralized s eizures (n = 11; 34.4% base rate of occurrence) Nonlateralized s eizures/PNES (n = 17; 53.1% base rate of occurrence) VMEM 0.00% (0/ 4 ) 0.00% (0/11) 88.2% (15/17) NVMEM 0.00% (0/4) 0.00% (0/11) 100% (17/17) VMEM, NVMEM, ATT, LAN, SPT, EXE 50.0% (2/4) 18.2% (2/11) 70.6% (12/17) S ATT, S LAN, S MEM, S SPT, S EXE 0.00% (0/4) 0.00% (0/11) 58.8% (10/17)
47 Table 3 5. Descriptive Statistics by Localization Classification ( Excluding Participants with P sychogenic Non epileptic Seizures [P NES ] ) All p articipants m ean (SD) (N=20) Temporal lobe seizure localization m ean (SD) ( N=14) seizure l ocali zation m ean (SD) (N=6) Independent s amples t tests (N = 20, df =18) Age 40.95 (12.97) 43.86 (12.86) 34.17 (11.41) 1.60 Education 13.15 (2.18) 12.86 (2.41) 13.83 (1.47) 0.91 S NAB 1.53 (1.16) 1.77 (1.13) 0.95 (1.08) 1.52 S ATT 1.19 (1.28) 1.54 (1.26) 0.37 (0.98) 2.02 S LAN 1.95 (1.11) 2.13 (1.22) 1.52 (0.71) 1.13 S MEM 1.76 (1.20) 1.90 (1.31) 1.43 (0.92) 0.79 S SPT 0.93 (1.39) 1.35 (1.36) 0.03 (0.94) 2.25* S EXE 0.88 (1.28) 1.30 (1.30) 0.08 (0.45) 2.47* NAB 0.64 (1.02) 0.70 (0.96) 0.52 (1.22) 0.35 ATT 1.02 (1.24) 1.40 (1.20) 0.14 (0.87) 2.31* LAN 0.71 (1.02) 0.99 (0.94) 0.05 (0.97) 2.03 MEM 0.14 (1.47) 0.50 (1.33) 0.71 (1.52) 1.80 SPT 1.18 (1.20) 1.24 (1.29) 1.06 (1.05) 0.30 EXE 1.20 (1.31) 1.48 (1.38) 0.55 (0.93) 1.50 VMEM 0.30 (1.03) 0.77 (0.75) 0.80 (0.73) 4.30*** NVMEM 0.20 (1.12) 0.09 (1.07) 0.87 (1.00) 1.88 MEMATTDISC 0.74 (0.76) 0.51 (0.77) 1.30 (0.37) 2.38* MEMLANDISC 0.14 (0.85) 0.10 (0.94) 0.21 (0.68) 0.25 MEMSPTDISC 0.90 (0.93) 0.89 (0.88) 0.94 (1.11) 0.12 MEMEXEDISC 0.19 (0.74) 0.09 (0.78) 0.42 (0.65) .921 Independent samples t tests described in the rightmost column describe differences between temporal and extratemporal groups. T tes ts reaching significance at p < .05 are annotated by a *; p < .01, **; p <.001, ***
48 Table 3 6. Percentage Correct Classification of Seizure Localization (Excluding Participants with PNES) Usin g NAB Variables as Predictors Predictor v ariables Temporal lobe seizure l ocalization (n=14; 70% base rate of occurrence) l ocalization (n=6; 30% base rate of occurrence) MEMATTDISC 92.9% (13/14) 50.0% (3/6) MEMLANDISC 100% (14/14) 0.00% (0/6) MEMSPTDISC 100% (14/14) 0.00% (0/6) MEMEXEDISC 92.9% (13/14) 0.00% (0/6) MEMATTDISC, MEMLANDISC, MEMSPTDISC, MEMEXEDISC 85.7% (12/14) 33.3% (2/6) MEMATTDISC, MEMLA NDISC, MEMSPTDISC, MEMEXEDISC (s tepwise) 92.9% (13/14) 50.0% (3/6) VMEM 92.9% (13/14) 83.3% (5/6) NVMEM 92.9% (13/14) 16.7% (1/6) ATT, LAN, MEM, SPT, EXE 78.6% (11/14) 16.7% (1/6) ATT, LAN, MEM, SPT, EXE (s tepwise) 85.7% (12/14) 50.0% (3/6) S ATT, S LAN, S MEM, S SPT, S EXE 85.7% (12/14) 66.7% (4/6) S AT T, S LAN, S MEM, S SPT, S EXE (s tepwise) 92.9% (13/14) 50.0% (3/6)
49 Figure 3 1. Seizure Lateralization Index (SLI) Distribution of Participants Experiencing Electrographic Seizures
50 CHAPTER FOUR DISCUSSION Results Summary This study attempted to assess the clinical utility of the Neuropsychological Assessment Battery (NAB) in an epilepsy population by analyzing how well NAB scores (as well as specific additional variables derived from NAB scores) predicted lateralization and localization of seizure focus in our sample. Over a 16 month data collection period, 45 participants wi th documented or suspected epilepsy were tested while undergoing 24 hour video electroencephalographic (V EEG ) monitoring during their inpatient stay o n the epilepsy monitoring u nit (EMU) at UF/Shands Hospital in Gainesville, Florida Each participant was administered the NAB Screening Module as well as the five Core NAB Module s (Attention, Language, Memory, Spatial, and Executive Functions). With the exception of the Screening Module which was designed to be administered first, the remaining five NAB Mo dule s were administe red in random order to minimize pr actice or order effects. Both f orms of the NAB were also used with alternating participants to attempt to increase generalizability. After testing was completed and NAB scores were obtained, additiona l NAB derived scores were calculated for additional use in our analyses. Specifically, various subtest scores within the NAB Memory Module s were used to calculate Verbal Memory and Nonverbal Memory construct score s based on which inherent form of Memory a bility these particular subtests assessed. These scores were primarily our sample. It was hypothesized that the derived Verbal Memory construct score would be able to predict lateralization of seizure focus in our sample at a rate greater than chance during Module contains only one measure
51 assessing nonverbal memory abilities (NAB Shape Learning) and this measure is a recognition measure, which are in general less difficult than free recall measures (such as the List Learning and Story Learning subtests which derived the Verbal Memory construct score ), it was also hypothesized that the Nonverbal Memor y construct score by itself would be unable to predict lateralization during discriminant function analyses. With regard to prediction of localization of seizure focus, it was hypothesized that NAB derived Module discrepancy scores (Memory Attention discr epancy score Memory Language discrepancy score Memory Spatial discrepancy score and Memory Executive Functions discrepancy score ) would be able to predict localization of seizure focus at a rate greater than chance within our sample, with the Memory Spa tial discrepancy score predicted to have the best ability to classify subjects as having seizures of temporal vs. extratemporal focus. Of the 45 participants tested, twenty were found to have ele ctrographic seizures (four left lateralized, eleven right lateralized, and five nonlateralized), twelve were found to have psychogenic non epileptic seizures (PNES), and thirteen subjects had no clinical events during their stay and were therefore unable to be definitively diagnosed or classified. Despite data b eing positively skewed for our sample, tests of normality using Kolmogorov Smirnov statistics for the variables in our study were nonsignificant (p > .05), with the exception of the Nonverbal Memory construct score (p < .01). As such, tests for group diff erences for this particular variable were assessed using nonparametric tests and all were nonsignificant). No group differences were found on key demographic variables including gender or handedness. A group difference was found on the Language Module sc ore by which NAB Form was administered; however, this difference is likely due to error given our small sample size. No group differences were found on NAB scores with respect to EMU classification (electrographic seizures vs. PNES
52 vs. No events). With r egard to late ralization classification (left lateralized seizures vs. right lateralized seizures vs. nonlateralized seizures/PNES), a statistically significant group difference was found on the Memory Spatial discrepancy score which lent some credence to our initial hypotheses regarding the potential utility of this score. With regard to localization classification (temporal lobe seizures vs. seizures originating outside the temporal lobe), statistically significant group differences were found on five se parate NAB variables (Screening Spatial Domain score Screening Executive Functions Domain score Attention Module score Memory Attention discrepancy score and the Verbal Memory construct score ). Discriminant function analyses (DFAs) showed an inabilit y for the NAB scores and additional NAB derived scores obtained in our study to predict lateralization at a rate greater than chance. Even the Memory Spatial discrepancy score the only variable in our study where group differences based on laterality wer e found, was unable to adequately predict group membership. In contrast, a number of variables in our study (including the Memory Attention discrepancy score NAB Core Module s in combination, NAB Screening Module s in combination, and Verbal Memory discrep ancy score ) predicted localization of seizure focus at a rate greater than chance. The Verbal Memory construct score in particular was able to correctly classify individuals by localization with 90% accuracy, providing the strongest evidence in our study Study Limitations The results of our study, along with the study itself, have several limitations that warrant acknowledging. The small sample size for our study likely reduced the robustness and power of some of our analyses (e.g., multiple regressions) to detect group differences ( though the current scores obtained would not have been significant at the p < .05 level even if our sample included
53 over 1,000 participants ) There were a number of unforeseen logistical challenges that made data collection with this specialized population more challenging than was originally anticipated, including enhanced privacy protections (which, while vital, presented as additional barriers to access), availa bility of trained research assistants, etc. In addition, data were collected on 45 participants but only 20 (4 4 % ) actually experienced electrographic seizures, while 12 (2 7 % ) experienced PNES and 13 (2 9% ) had no clinical events While these outc ome rates were confirmed by UF n eurologists to be fairly consistent with average EMU outcomes, the return rate of participants experiencing electrographic seizures (4 4 % ) compared to participants experiencing other outcomes (5 6 % ) was substantially lower than anticipated at study outset. Another potential limi tation involves the high female to male (33:12) ratio of participants in our study. Previous data available for comparison from an existing database of 362 patients participating in this program contained a more comparable number of men and women (Male n=173; Female n=179). It is unknown why our rate of female to male participants differed so greatly. Research has shown that there are notable differences between men and women with epilepsy, including men being significantly more like to experience generalized seizures than women (Janszky, Shulz, Janszky, & Ebner, 2004) while women are more likely to experience isolated auras and have lateralized electroencephalographic ( EEG ) seizure patterns more often tha n men (McHugh & Delanty, 2008). In addition, Frings et al. (2006) found that, during memory processing, hippocampal activation is significantly more left lateralized in healthy women and more right lateralized in healthy men (with women subsequently ratin g their strategies during memory testing as being more verbal than men in the same study). Limited information is currently available in the literature however, with regard to gender differences on neurocognitive test performance specifically within epil epsy populations and therefore the
54 extent to which this gender discrepancy may have impacted the results of the current study is unknown. As our study was a single site study sampling participants from a limited geographical region of the Southeastern United States, it is unknown to what extent our results may generalize to epilepsy populations located in other regions of the United States or other countries. Possible Reasons for Study Outcome In addition to the small sample size mentioned previously, there was a notable size discrepancy based on lateralization outcome, such that only four participants were in the left lateralized group, while eleven participants were in the right lateralized group and seventeen participants were in the nonlateralized/ PNES group (five participants with nonlateralized electrographic seizures and twelve participants diagnosed with PNES). These differences in group size in conjunction with the initial small sample size and nonsignificant differences between group mean sc ores on NAB variables, caused many of our laterality specific DFAs to perform poorly, even w hen follow up analyses were run excluding participants with PNES W hile it appears that the NAB genuinely possesses weaker sensitivity to right temporal dysfunctio n due to reduced sampling of nonverbal as compared to verbal memory abilities (e.g., Shape Learning is purely recognition based, thus reducing sensitivity of the Nonverbal Memory construct score ) grossly unequal group sizes in the lateralization groups also likely reduced the temporal dysfunction, it is worth noting that the Shape Learning test in the NAB Memory Module while being a purely recognition meas u re, is not the only nonverbal m emory test in the NAB. The Figure Drawing test included in the Spatial Module of the NAB bears much similarity to the Rey Complex Figure Test (Myers & Myers, 2004). The test assesses visuoperceptual/visuoconstructional abil ities by
55 requiring the individual being tested to reproduce a complex geometric figure with as much fidelity as possible. The Rey Complex Figure Test includes a copy, immediate recall, and delayed r ecall condition that occurs after 30 minutes (as well an optional recognition task) which allows the ability to assess nonverbal memory abilities while simultaneously assessing the individuals visuoperceptual/visuoconstructional skills. During initial test development and standardization, the creators of the N AB decided to include a similar task (the Figure Drawing Test) within the Spatial Module of the NAB. However only a copy and immediate r ecall condition was included in the published NAB, and therefore norms are available only for these two conditions. H ad the author s also chosen to include a d elayed r ecall condition for this measure during standardization and in the published NAB, this score could have potentially been used as an optional additional test to be transferred to the Memory Module in the even t both Module s were given. It is believed that, had the authors chosen to do this, the overall sensitivity of NAB Memory Module to right temporal impairment would likely have increased. localization of seizure focus in our study however, should not be downplayed in light of its limitations in predicting laterality. The specific prediction of seizure localization is an important component of neuropsychological testing within epilepsy pop ulations as localization of seizure focus more so than lateralization is more likely to be ambiguous from EEG data alone. Future Directions for Research While our findings provide some limited support for the clinical utility of the NAB within an epilepsy population, there is not enough information yet available to definitively determine its utility with this population. The current study evaluated concurrent validity using quantitative NAB indicators and did not involve a clinical judgment component ( e.g. how NAB data would be used and interpreted within a multidisciplinary epilepsy management team). Future research
56 and localization of seizure focus when b eing clinica lly interpreted by a qualified n europsychologist. I n routine clinical practice, a n europsychologist typically administers a batte ry of measures (such as the UF s tandar d neuropsychological b attery mentioned in Chapter One) and uses the scores from these measures to make predictions regarding lateralization and localization of seizure focus. Though the quantitative scores and subsequent derivations of such scores are heavily factored into the se predictions, additional qualitative factors concerning particular measures) are also potentially contributory but were not included in the present study. Just as t he NAB has focused on including ecologically valid Daily Living subtests within each Domain Module it could be argued that the most ecologically valid method of determining the clinical utility of the NAB within an epilepsy population would involv e provid ing a group of trained n europsychologists with all available NAB data (including scores and raw data) from a sample of patients with known electrographic seizures of varying lateralizations (left vs. right vs. nonlateralizing) and localizations (temporal l obe vs. extratemporal) and assessing their rates of correct classification (and rates of interrater reliability) based on available EEG and magnetic resonance imaging ( MRI ) data. If these individuals demonstrated sufficient reliability and predictive abil ity to classify seizure lateralization and localization using NAB data, this would provide substantial support for the clinical utility of the NAB with this population. Additional studies could further compare the predictive ability of the NAB Screening M odule alone compared with the five Core NAB Module s in a s imilar manner (using a trained n measures of neurocognitive assessment with this population.
57 Conclus ion In summary, our study provided some support for the clinical utility of the NAB within an epilepsy population. Specifically, our results suggested that certain combinations of NAB scores (as well as additional derived scores) were able to disseminate between individuals with temporal lobe epilepsy and individuals whose seizures originated outside the temporal lobe Our seizure laterality Future research with th lateralization and localization of seizure focus when the results are analyzed by one or more trained clinical n europsychologists, similar to the techniques used in current practice with this popul ation using other neuropsychological batteries) is warranted to more definitively determine its clinical utility within an epilepsy population.
58 LIST OF REFERENCES Abel, B. (2005). Intractable epilepsy: Retrieved Ma y 12, 2007 from http://healthlink.mcw.edu/article/1031002476.html Anastasi, A., & Urbina, S. (1997). Psychological Testing (7 th ed.) Newark, NJ : Prentice Hall/Simon & Schuster. Beck, A.T., Steer, R.A., & Brown, G.K. (1996). Beck Depression Inventory II. San Antonio, TX: Psychological Corporation. gnosis of psychogenic seizures. Epilepsy Behavior, 6, 264 265. Benbadis, S.R. (2010 September 21 ). Psychogenic nonepileptic seizures Retrieved July 16, 2010 from http://emedicine.me dscape.com/article/1184694 overview older adults: Base rates of low memory scores in neuropsychological assessment. Journal of the International Neuropsych ological Society, 13 490 500. Brooks, B.L., Iverson, G.L., & White, T. (2009). Advanced interpretation of the Neuropsychological Assessment Battery with older adults: Base rate analyses, discrepancy scores, and interpreting change. Journal of the Inter national Neuropsychological Society, 24 647 657. Brown, L.B., Stern, R.A., Cahn Wiener, D.A., Rogers, B., Messer, M.A., Lannon, M.C. Maxwell, C., Souza, T., White, T., & Ott, B.R. (2005). Driving scenes test of the N europsychological A ssessment B attery (NAB) and on road driving performance in aging and very mild dementia. Archives of Clinical Neuropsychology 20 209 215. Cahn Weiner, D.A., Wittenberg, D., & McDonald, C. (2009). Everyday cognition in temporal lobe and frontal lobe epilepsy. Epileptic Disorders, 11 222 227. Campbell, D.T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait multimethod matrix. Psychological Bulletin, 56 81 105. Cicchetti, D.V., & Sparrow, S.S. (1981). Developing criteria for es tablishing interrater reliability of specific items: Applications to assessment of adaptive behavior. American Journal of Mental Deficiency, 86 (2), 127 137. Cleveland Clinic (2007). Adult epilepsy Retrieved May 12, 2007 from http://cms.clevelandclinic.org/neuroscience/body.cfm?id=124 Corder, G.W., & Foreman, D.I. (2009). Nonparametric Statistics for Non Statisticians: A Step by Step Approach Hoboken, N J : John Wiley & Sons
59 Delis, D.C., Kramer, J.H., Kaplan, E., & Ober, B.A. (2000). C alifornia Verbal Learning Test Second Edition (CVLT II). San Antonio, TX: Psychological Corporation. Frings, L., Wagner, K., Unterrainer, J., Spreer, J., Halsband, U., & Schulze Bonhage, A. (2006). Gender related differences in lateralization of hippocampal activation and cognitive strategy. Neuroreport, 17 (4), 417 421. Gavett, B.E., Poon, S.J., Ozonoff, A., Jefferson, A.L., Nair, A.K., Green, R.C., & Stern, R.A. (2009 amnestic mild cognitive impairment. Journal of the International Neuropsychological Society, 15, 121 129. Golden, C.J., Purisch, A.D., & Hammeke, T.A. (1985). Luria Nebraska Neuropsychological Battery: Forms I and II. Los Angeles CA : Western Psychological Services. Goodglass, H. & Kaplan, E. (2000). Boston Naming Test Philadelphia PA : Lippincott, Williams & Wilkins. Grohman, K., & Fals Stewart, W. (2004). The detecti on of cognitive impairment among substance abusing patients: The accuracy of the Neuropsychological Assessment Battery S cree ning Module Experimental Clinical Psychopharmacology 12 (3), 200 207. Halstead, W.C. (1947). Brain and I ntelligence Chicago IL : University of Chicago Press. Heaton, R.K. (1981). Wisconsin Card Sorting Test Odessa, FL: Psychological Assessment Resources Inc Iverson, G.L., Williamson, D.J., Ropacki, M., & Reilly, K.J. (2007). Frequency of abnormal scores on the Neuropsychological Assessment Battery Screening Module (S NAB) in a mixed neurological sample. Applied Neuropsychology, 14 178 182. Janszky, J., Schulz, R., Janszky, I., & Ebner, A. (2004). Medial temporal lobe epilepsy: Gender differences. Journal of Neurology, Neurosurgery, and Psychiatry, 75 (5), 773 775. Klove, H. (1963). Clinical neuropsy chology. In F.M. Forster (Ed.) The medical clinics of North America New York NY : Saunders. Lowe, N.K., & Ryan Wenger, N.M. (1992). Beyond Campbell and Fiske: Assessment of convergent and discriminant validity. Research in Nursing and Health, 15, 67 75. Martin, J. (200 8, October 30 ). The truth about psychogenic nonepileptic seizures Retrieved July 16, 2 010 from http://www.epilepsy.com/articles/ar_1112967056 Mayo Clinic. (2007). Epilepsy treatment options Retrieved May 10, 2007 from http://www.mayo clinic.org/epilepsy/ treatment.html
60 McHugh, J.C., & Delanty, N. (2008). Epidem io logy and classification of epilepsy: Gender comparisons. International Review of Neurobiology, 8 3 11 26. Moser, D.J., Bauer, R.M., Gilmore, R.L., Dede, D.E., Fennell, E.B., Algina, J.J. Jakus, R., Roper, S.N., Zawacki, T.N., & Cohen, R.A. (2000). Electroencephalographic, volumetric, and neuropsychological indicators of seizure focus lateralization in temporal lobe epilepsy. Archives of Neurology, 57 (5) 707 712. Myers, J.E. & Myers, K.R. (2004). Rey Complex Figure Test and Recognition Trial Odessa FL: Psychological Assessment Resources Inc Owsley, C., Ball, K., Sloane, M. E., Roenker, D. L, & Bruni, J. R. (1991). Visual/cognitive correlates of vehicle accidents in older drivers. Psychology and Aging, 6 403 415. Pilcher, W.H., Locharernkul, C., Primrose, D., Ojemann, L.M., & Ojemann, G .A. (1992). Update in epilepsy III : Surgical therapy of intractable epilepsy. New York State Journal of Medicine, 92 92 96. Reitan, R.M. (1958). Validity of the Trail Making Test as an indicator of organic brain damage. Perceptual and Motor Skills 8, 27 1 276. Reitan, R.M., & Wolfson, D. (1993). The Halstead Reitan Neuropsychological Test Battery. Tucson AZ : Neuropsychological Press. Schmitt, N., & Stults, D.M. (1986). Methodology reviews: Analysis of multitrait multimethod matrices. Applied Psycho logical Measurement, 10, 1 22. Scoville, W.B., & Milner, B. (1957). Loss of recent memory after bilateral hippocampal lesions Journal of Neurology, Neurosurgery, and Psychiatry, 20 11 21. Snyder, P.J. (1998). Epilepsy. In P.J. Snyde r & P.D. Nussbaum (Eds.), Clini cal n europsychology: A pocket handbook for a ssessment Washington, DC: American Psychological Association. Spielberger, C.D. (1968). State Trait Anxiety Inventory Odessa, FL: Psychological Assessment Resources Inc Spreen, O. & Benton, A. (1977). Neurosensory Center Comprehensive Examination for Aphasia Victoria BC : University of Victoria Neuropsychology Laboratory. Stern, R.A., & White, T. (2003). Neuropsychological Assessment Battery : Administration, Scoring, and Interpretation M anual Lutz, Fla: Psychological Assessment Resources Inc Temple, R.O., Zgaljardic, D.J., Abreu, B.C., Seale, G.S., Ostir, G.V., & Ottenbacher, K.J. (2009). Ecological validity of the Neuropsychological Assessment Battery Screening Module in post acute brain injury rehabilitation. Brain injury, 23 45 50.
61 Tombaugh, T.N., Kozak, J., & Rees, L. (1999). Normative data stratified by age and education for two measures of verbal fluency: FAS and animal naming. Archives of Clinical Neuropsych ology, 14, 167 177. Valton, L., & Mascott, C.R. (2004). What is the current role of a Wada test in the pre surgical work up of pharmacologically intractable epilepsy in adults? Revue Neurologique, 160 (1), 164 169. Wechsler, D. (1997). WAIS III/WMS III Technical Manual. San Antonio, TX: The Psychological Corporation. Wechsler, D. (1987). Wechsler Memory Scale Revised Manual. San Antonio, TX: The Psychological Corporation. White, T., & Stern, R.A. (2003). Neurop sychological Assessment Battery: Demo graphically Corrected Norms M anual Lutz, F L : Psychological Assessment Resources Inc. White, T., & Stern, R.A. (2003). Neuropsychological Assessment Battery : Psychometric and Technical Manual. Lutz, FL : Psychological Assessment Resources Inc White T., & Stern, R.A. (2003). Neuropsychological Assessment Battery : U.S. Census M atched N orms M anual. Lutz, F L : Psychological Assessment Resources Inc Yochim, B.P., Kane, K.D., & Mueller, A.E. (2009). Naming test of the Neuropsychological Assessment Battery: Convergent and discriminant validity. Archives of Clinical Neuropsychology, 24 575 583. Zgaljardic, D.J., & Temple, R.O. (2010). Reliability and validity of the Neuropsyc hological Assessment Battery Screening Module (NAB SM) in a sample of patients with moderate to severe acquired brain injury. Applied Neuropsychology, 17 27 36.
62 BIOGRAPHICAL SKETCH Bradley J. Daniels graduated summa cum laude with a Bachelor of Arts de gree in p sychology from the University of Central Florida in 2003. He soon relocated to Gainesville after being accepted into the clinical p sychology doctoral program in the Department of Clinical and Health Psychology at the University of Florida. He re ceived his Master of Science degree in 2005 and continued on with his doctoral p ursuits. His graduate work focused primarily on clinical neuro psychology and his research focused primarily on the neuropsychology of epilepsy. His clinical interests include clinical n europsycholog y rehabilitation psychology, and forensic psychology He rece ntly completed an APA approved n europsychology internship offered Birmingham, Ala bama, where he was awarded the C.J. Rosecrans psychology internship a ward, awarded annually to the intern who has demonstrated outstanding promise in the field of clinical p sychology upon completion of the internship. He also greatly enjoys teaching an d has taught p sychology as an adjunct assistant p rofessor in the Department of Social and Behavioral Sciences at Santa Fe College in Gainesville, Florida since 2006. He has published multiple anthology essays relating psychological principles to popular c ulture, a tool he regularly utilizes to enhance learning in the classroom. Currently, he is completing a 2 y ear postdoctoral fellowship in rehabilitation p Florida. In addition, Mr. Daniels has b een happily married to his wife Tiffany since 2008. He resides in Lakewood Ranch Florida with his wife and their rambunctious dog, Jack Daniels.