Citation
Short-term memory in epileptic patients being treated with phenytoin

Material Information

Title:
Short-term memory in epileptic patients being treated with phenytoin
Creator:
McGuigan, Marianne L. D., 1935-
Copyright Date:
1984
Language:
English

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Marianne L. D. McGuigan. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
11886407 ( OCLC )
ACQ1593 ( LTUF )
30570135 ( ALEPH )

Downloads

This item has the following downloads:


Full Text
SHORT-TERM MEMORY IN EPILEPTIC PATIENTS
BEING TREATED WITH PHENYTOIN
By
MARIANNE L. D. MCGUIGAN

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

1984




ACKNOWLEDGMENTS

To the following people, who have contributed to my education and my life, I am deeply grateful:
Suzanne Johnson, your enthusiasm for scholarly research, magnanimity, and faith in a nonstandard prospective student fanned a spark and helped to make graduate study possible;
Roger Blashfield, encountering your sensitivity and devotion to
intellectual rigor, graced with a sense of humor and irony, made me know I was in the best of company;
Cynthia Belar, the discovery that my nobly professional and
intellectual mentor could as well be tender friend brought warmth and renewed hope;
Joe Wilder, getting to know you better, gentleman physician who permitted me to work with the patients you serve so well and value so highly, was a pleasure and a compliment;
Eileen Fennell, the generosity of spirit, the model of professional excellence, the wit, intelligence, and warm friendship you have shared with me have done me honor;
Finally, my thanks go to L. James Willmore on whose hypothesis of anticonvulsant effects on memory a portion of this study was based.




TABLE OF CONTENTS

PAGE

ACKNOWLEDGMENTS............ . .... .. .. .. .. ..

ABSTRACT CHAPTER

INTRODUCTION . . . . . . . . . .

Anatomy and Connections of the Temporal Investigation of Memory in Epilepsy.. The Issue of Antiepileptic Medications Rationale For This Study .......
METHOD . . . . . . . .
Subjects . . . . . . . .
Pilot Studies........ .. .. .. .. .
Materials and Apparatus... .. .. .. Procedure........ .. .. .. ....
Instructions to the Subjects ..... Statistical Analyses .........
RESULTS..... .... .. .. .. .. ..

Overview of Results... .. .. .. .. ..
Description of Groups and Relationships o Group and Categorical Blood Level Effects
on Test Scores ............
Group and Categorical Blood Level Effects
on Within-Session Performance. .. .. Overview of Mean Single and Combined
Test Performances by Subjects. ...
Relationships Between Memory Scores
and Categorical Phenytoin Levels.
Effects of Session Sequence.. .. ....
DISCUSSION . . . . . .
The Experimental Hypotheses.. .. .... Attention and Vigilance... .. .. .. ..
Summary and Conclusions... .. .. .. ..

FOUR

f Variables.

. 50
53
. 54 69 78

TWO

THREE

Lobe

. 33




APPENDICES
A SAMPLE VISUAL GEOMETRIC TEST STIMULI . . . . . 91
B VISUAL GEOMETRIC TEST DISTRACTION TASK . . . . 92
C VISUAL CONSONANTS TEST RESPONSE FORM . . . . . 93
D PATIENT INFORMATION FORM 95
E PATIENT INFORMED CONSENT TO BE A RESEARCH SUBJECT . 98
REFERENCES 100
BIOGRAPHICAL SKETCH 107




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
SHORT-TERM MEMORY IN EPILEPTIC PATIENTS BEING TREATED WITH PHENYTOIN
By
Marianne L.D. McGuigan
August 1984
Chairman: Eileen B. Fennell
Major Department: Clinical Psychology
Studies have confirmed the presence of long-term, but not shortterm, memory dysfunction in complex partial seizures. A controversy exists regarding the nature of these deficits. One view is that stimulus material, verbal or nonverbal, interacts with hemisphere of seizure focus, leading to verbal deficits in left-lateralized and nonverbal deficits in right-lateralized complex partial seizures. The other view holds modality is relevant, resulting in impaired memory for any material presented acoustically. Other studies have reported that antiepileptic drugs do/do not have deleterious effects on cognitive-intellectual performances in epileptics.
This study tested the hypotheses that (a) complex partial seizures are accompanied by greater short-term memory impairment than are




generalized seizures, and (b) phenytoin levels outside the therapeutic range are detrimental to memory performance. A minor hypothesis is that deficits are not the result of inattention.
Twenty-nine male epileptics having unilateral complex partial or
generalized seizures were tested three times with measures of short-term memory and attention at three blood levels of phenytoin.
Results of analyses did not support the hypothesis of greater memory deficit in complex partial seizures. All patients showed memory impairments. Left focus subjects scored lowest on verbal measures; right focus patients scored lowest on a figural measure. These findings were in the predicted direction, but not significant. Generalized seizure patients differed from complex partial seizure patients in number of commission errors made on the test of attention.
The hypothesis of greater impairment of memory at nontherapeutic
blood levels of phenytoin was not statistically confirmed. Two findings in accord with the hypothesis must be cautiously interpreted because of a nonsignificant MANOVA. These were (a) improvement in a verbal score as blood level became therapeutic, and (b) greater loss of information from short-term memory on a figural measure when blood level was toxic. When blood level was toxic, significantly more commission errors were made by all subjects.
The conclusions were that all patients had memory deficits, but were not inattentive. The modality hypothesis did not enhance understanding of any observations. Nonsignificant findings in the predicted directions leave issues open to further study.




CHAPTER ONE
INTRODUCTION
Cognitive-intellectual deficits accompanying epilepsy, described in the nineteenth century as leading in some patients to global mental deterioration, remain a topic of controversy regarding their specific nature and causation. One of these deficits, dysfunctional memory, is often associated with the presence of complex partial seizures (CPS, formerly called psychomotor), a type of paroxysmal cerebral dysrhythmia, arising in one or both temporal lobes. Greater attention has been given this seizure type since it came to be better differentiated clinically and electroencephalographically (Gibbs, Gibbs, & Fuster, 1948; Gibbs, Gibbs, & Lennox, 1938). Although the disruptive effects of other focal and generalized seizures have not been overlooked, the proximity of a periodically discharging temporal lobe focus to the limbic system and the hippocampal structures forming the anatomical substrate for memory processing has implications for that function (Hecaen & Albert, 1978; Walsh, 1978). Moreover, patients with seizures, their families, and their caretakers provide anecdotal support of dysfunction in epileptics. This dysfunction is characterized as failure to retain recently acquired information.
A number of factors have been shown to influence general and
specific deficits among epileptics. These have included, in addition to heredity, coexisting brain damage and seizure-related variables




such as seizure type, age at onset of seizures, frequency of seizures, duration of seizure history, laterality of a discrete focus, electroencephalographic patterns, and the effects of antiepileptic medications (Dikmen & Matthews, 1977; Dikmen, Matthews, & Harley, 1977; Dodrill, 1975, 1981; Guey, Charles, Coquery, Roger, & Soulayrol, 1967; Klove & Matthews, 1966; Lennox, 1960; MacLeod, Dekaban, & Hunt, 1978; Matthews & Klove, 1967).
Studies of heterogeneous groups of epileptics have often yielded uninterpretable or nonsignificant results which did not discriminate experimental subjects from nonepileptic controls on measures of memory (Loiseau, Strube, Broustet, Battellochi, Gomeni, & Morselli, 1980; Scott, Moffett, Mathews, & Ettlinger, 1967; Tomlinson, Stirling, Merrifield, & Reynolds, 1981). Separation of subjects into diagnostic groups to assess memory in CPS as compared to other seizure groups has become more common since the advent of surgical treatment of refractory CPS disorder, as well as the publication of data from the study of patients before and after temporal lobe resection.
Anatomy and Connections of the Temporal Lobe
The temporal lobe is the mass of cerebral tissue below the Sylvian fissure, extending posteriorly to approximately Brodmann's areas 40, 39, and 19 (see Figure 1). It is composed on its lateral surface of neocortex which forms its major gyri. On its medial unexposed surface it consists of phylogenetically older cortex, archicortex and paleocortex. These genetically older tissues form not only its medial gyri, but the subcortical hippocampus and amygdala as well. Together, these three




ACF
LV
LT
AR
,''fi
AC O xFO- PCx c
Figure 1. Left cerebral hemisphere, lateral view.
Positions of subcortical amygdala (A) and hippocampus (H) are illustrated below Sylvian fissure (SF), with superior (STG) and inferior (ITG) lateral temporal gyri.
Bottom figures. AC = anterior coronal section at approximately dotted lin AC; PC = posterior coronal section at approximately dotted line PC.
Structure locations labeled. A = amygdala; C = caudate nucleus; DM = dorsomedial nucleus of the thalamus; H = hippocampus; ITG = inferior temporal gyrus; LT lateral thalamus; LV = lateral ventricle; P = putamen; STG = superior temporal gyrus.




distinguishable cortical tissues provide the temporal lobe with auditory, association, and limbic cortex.
The connections of the temporal lobe are similarly diversified. It
receives afferent projections from the sensory system and projects efferently to the parietal and frontal association regions, limbic system, and basal ganglia. The left and right lobes communicate via the corpus callosum (neocortex) and the anterior commissucre(archicortex) (Kolb & Whishaw, 1980).
Investigation of Memory in Epilepsy
The terminology used to describe memory processes varies across
disciplines and studies. Reviewing the studies below, which typically refer to a storage model of memory, I shall define several relevant terms (Russell, 1981).
The earliest stage of the memory process, that of input, is called sensory memory, which begins to decay in approximately 250 msec. In terms of its storage capacity, more information can be held than the subject is able to report. The visual form of sensory memory is termed "iconic" memory, the auditory form "echoic" memory. Immediate memory is a clinical term defined as that quantity of material reproducible directly after presentation. Short-term memory (STM) and long-term memory (LTM) refer to hypothetical processes mediating retention, each with its own storage capacity. In STM a limited number of items, generally seven plus or minus two, are maintained for about 20 to 30 seconds without rehearsal. Material in STM is subject to displacement by new input of information. Overlapping with STM from about 0.5 seconds after input, LTM is that process believed to consolidate, store, and




make information available for future retrieval. The storage capacity of LTM is presumed to be indefinitely large and relatively permanent, although there is some loss over time (Waugh & Norman, 1965). A final process, retrieval, refers to the locating of stored material in LTM by means of an STM probe, whereby the material is returned to STM for active use. Finally, recent and remote memory refer to the personal time scale of the patient or examinee.
Temporal Lobectomy Studies
Temporal lobe resection produces differential performance deficits related to side of operation and the extent of hippocampus removed (Kolb & Whishaw, 1980). The following studies are representative of those reporting effects of these procedures.
Milner (1962, 1967) reported that a left temporal lobe seizure focus interfered with verbal memory performance as tested with the Logical Memory subtest of the Wechsler Memory Scale (WMS). This deficit appeared to be material-specific rather than being related to the mode of presentation, appearing in the visual as well as the auditory presentational mode. Temporal lobectomy markedly worsened this verbal memory impairment, also seen for paired-associate words and recognition of numbers or letters. Considerable recovery took place during the extended postoperative period.
Auditory, but not visual, verbal deficits were found by Meyer (1959) and Meyer & Yates (1955) in left temporal lobectomy patients. The authors described their left temporal lobectomy subjects as doing less well than right temporal lobectomy subjects, if not significantly so,




on all measures employed. At follow-up testing 11-15 months later, this auditory verbal deficit was often still present.
Cherlow & Serafetinides (1976), comparing right-handed groups
of left and right temporal lobectomy patients, found left-lobectomized subjects significantly impaired in learning verbal material. These patients needed more repetitions to learn somewhat less material than did right lobectomy patients. Percentages of learned material recalled, however, did not differ between groups.
Deficits found in patients with right temporal lobe foci both preand postoperatively have been documented by Milner (1962, 1965, 1968), Kimura (1963), and Corkin (1965), among others.
Memory for visual material not easily verbally encoded was impaired in right temporal lobectomy patients on measures consisting of nonsense figures, random dot patterns, and photographs of human faces. Although preoperative comparisons with left lobectomy patients in performance on the Tonal Memory subtest of the Seashore Measures of Musical Talents were small, postoperative testing revealed highly significant differences representing losses by right-lobectomized patients (Kimura, 1963; Milner, 1962, 1968).
Right lobectomy subjects did poorly in tactual maze-learning,
relative to left lobectomy subjects, making significantly more errors and requiring more trials to criterion (Corkin, 1965). Milner (1965) had reported parallel findings in subjects with bilateral hippocampal damage on a visual maze-learning task.
Samuels, Butters, & Fedio (1972) attempted to differentiate perceptual from memory components of these deficits, as well as to evaluate




the material/modality specificity issue. They presented verbal and nonverbal stimuli in the visual, and verbal stimuli in the auditorymode to patients having unilateral temporal lobe excisions. Their method was that of Peterson & Peterson (1959) in which either immediate or delayed (3, 9, or 18 seconds) recall is tested after tachistoscopic presentation. When recall was delayed, backwards counting was introduced to prevent rehearsal or verbal labeling. They considered impairment on the immediate recall evidence for perceptual deficit, on the delayed recall alone as supportive of STM disturbance. Both right and left temporal lobectomy patient groups performed as well as normal controls on the visual tasks. This was consistent with earlier studies that had provided evidence for major participation of the parietal lobe in the mediation of visual retention (Butters & Barton, 1970; Butters, Barton, & Brody, 1970; Butters & Brody, 1968; Butters, Samuels, Goodglass, & Brody, 1970; Samuels, Butters, & Goodglass, 1971).
The authors suggested that the differences in their results from those of Milner, who used delays of 45 seconds or more, might be explained by the existence of a double dissociation in these processes relative to material/modality specificity and anatomical separation. Thus, in addition to the mediation of visual input by the parietal and auditory by the temporal areas, STM storage might be modality-specific and LTM storage material-specific.
In general, lobectomy studies found verbal deficits in left temporal lobectomy patients and nonverbal deficits in right temporal lobectomy patients with respect to their memory functioning. Inconsistencies with




these findings have yet to be resolved. There is, however, agreement that bilateral damage to both hippocampi leads to substantially greater impairment than does unilateral lesion or removal.
Artificial Stimulation Studies
Ojemann and his associates, performing brain stimulation in the context of surgically treating movement disorders, reported having observed effects on speech. They elicited anomia with stimulation of the left pulvinar of the thalamus and of the deep parietal paracallosal white matter of both hemispheres in right-handers. The right-sided findings were interpreted as being due to extension of the stimulating current to the speech area by way of afferent callosal fibers from both hemispheres. Later, using a Peterson & Peterson (1959) technique variant, they found differences dependent on stimulation time. Results showed significant interference with memory with stimulation on recall. There were no effects with stimulation during distraction or presentation plus recall. Slight nonsignificant improvement occurred with stimulation during presentation. No manipulation interefered with performance on recognition tasks (Ojemann & Fedio, 1968; Ojemann, Fedio, & Van Buren, 1968).
Ojemann (1978) also reported phenomena produced by temporal cortical stimulation in the course of mapping the area anatonomically during resection for intractable CPS. He concluded that the areas of the temporal lobe where stimulation affects language, interfering with naming of common objects, are adjacent to, but not identical with, the areas where it affects short-term verbal memory (STVM). He proposed that storage of




short-term verbal memory is in the posterior language area cortex, and that anterior language cortex plays a part in retrieval from short-term memory.
Data from studies of electroconvulsive shock lend further support to temporal laterality effects on memory (Halliday, Davison, Browne, & Kreeger, 1968). Unilateral application of shock in treatment of depressed patients showed interference with nonverbal material retention with right-sided electrode placement. Left-sided electrode placement produced deficits in verbal learning and retention.
Overview of the Psychological Characteristics of Epileptics
Numerous studies of epileptics which have compared this patient population with normals or with other organically impaired groups have reported differences in epileptics' cognitive-intellectual performances (Dodrill, 1981). In addition to this considerable literature, another body of work evaluating epileptics' proneness to various forms of psychopathology has developed. Patients with CPS are seen as being at greatest risk, compared to other epileptics, for a variety of sequelae ranging from undesirable personality change to psychosis (Hermann & Whitman, 1984).
Studies of Medically TreatedEpileptics
Three types of studies of medically treated epileptics will be reviewed. The first of these has examined general cognitive and/or memory functioning in heterogeneous CPS samples, often including right, left, and bilateral CPS patients. A second set of studies, employing CPS patients grouped by side of focus, has provided some data to support




differential deficits between left-lateralized and right-lateralized CPS. Neither of the two types of studies described has included medication as a relevant factor. The third type of study, in which antiepileptic drugs have been considered or controlled, will be reviewed following a brief general discussion of the use of these agents in obtaining seizure control.
Heterogeneous group studies
Those investigations which have utilized mixed groups of CPS
subjects, often contrasted with a group of other focal or generalized seizure patients, have not found right CPS versus left CPS differences. A disparity in attentional capacity favoring the CPS group has been reported, as well as a variety of verbal and memory impairments seen on the Wechsler Adult Intelligence Scale (WAIS) and WMS favoring the subjects with generalized seizure types (Glowinski, 1973; Matthews & Klove, 1967; Mirsky, Primac, Marsan, Rosvold, & Stevens, 1960; Quadfasel & Pruyser, 1955).
Matthews and Klove (1967) used multiple groups including (a)
brain-damaged without seizures, (b) major motor, and (c) mixed seizures, each group being again divided into known and unknown etiology, to control for demonstrable organic pathology. They ranked their CPS groups with unknown and known etiology of seizure disorder second and third, respectively, after their nonneurological control group. Rankings were based on integrity of neuropsychological functioning, measured by the Halstead Impairment Index. The authors' earlier conclusions (Klove & Matthews, 1966) that epileptic seizures were associated with




impaired neuropsychological status regardless of etiology were amended following this investigation. Subjects having major motor or mixed (CPS and major motor) seizures continued to perform less well than controls. Their most impaired groups were (a) major motor of known etiology, and (b) brain-damaged without seizures, which did not differ from each other.
Homogeneous left and right group studies
A comparison of performance in 6 to 14 year old children with
either CPS or generalized absence (formerly called petit mal) seizures by Fedio & Mirsky (1969) revealed laterality differences. Their left CPS and right CPS subjects performed differently on the verbal and nonverbal components of a derived "supraspan" score. Verbal deficits were prominent in left CPS subjects; nonverbal deficits characterized performances of right CPS subjects. Children with absence seizures differed from both CPS groups in earning lowest scores on a measure of attention, the Continuous Performance Test.
Other investigators studying adult groups of divided left and right CPS epileptics have also reported laterality effects. Reynolds (1974) was able to classify an overall 79% of his right and left CPS subjects. A stepwise discriminant analysis of responses on verbal and nonverbal memory instruments provided the prediction equation he used.
In comparing CPS groups to lateralized groups with frontal foci, Ladavas, Umieta, and Provinciali (1979) referred to their findings as supporting long-term but not short-term deficits in CPS subjects, according to side of involvement. Among frontal subjects, those with




a left-sided focus were impaired in judging the recency of presentation of stimuli.
Mayeux, Brandt, Rosen, and Benson (1980) used a variety of verbal
and nonverbal measures to contrast groups of CPS and generalized seizure patients. They found that left-focus subjects differed from both other groups on a confrontation naming task, whereas right-focus subjects did not differ in any interpretable way. The authors proposed that leftlateralized CPS patients have a specific anomia as distinct from a memory deficit, and suggested that this impairment, with its accompanying word-finding difficulties, might explain the circumstantiality of speech so often attributed to left CPS patients.
Delaney, Rosen, Mattson, and Novelly (1980) compared right and left CPS patients with frontal focal epileptics and with normal controls. Right CPS subjects performed worst on one visual nonverbal measure; left CPS subjects earned the lowest derived verbal score. On other measures of verbal and nonverbal memory both CPS groups were alike, but differed from contrast groups, with left CPS subjects showing verbal and right CPS subjects nonverbal deficits. The authors reported low correlations between anticonvulsant serum levels and performances on psychometric measures, but a consistent correlation between longer duration of seizure history and declining memory scores.
The Issue of Antiepileptic Medications
Consideration of anticonvulsant blood level by Delaney et al. (1980) reflected a concern many have had for the participation of this variable in epileptics' psychological




functioning. Studies of epilepsy which do not control these agents must at least acknowledge the omnipresence of anticonvulsants in patient treatment. Until rather recently, only enumerating of drugs, quantifying of dose, and seizure-frequency monitoring have been available as methods.
Evaluation of the role antiepileptic medications have in psychological performances has become possible in the last 10 to 12 years as laboratory methods of quantitative analysis have been developed and made widely accessible (Wilder & Bruni, 1981). An invaluable tool for the neurological clinician, ready measurement of anticonvulsant blood level (ABL) has been a methodological aid to the researcher as well. Drug Treatment in Epilepsy
Antiepileptic drugs may be used alone or in combination, the choice of agent being in part dictated by the clinical seizure type (Millichap, 1972; Wilder & Bruni, 1981). Coatsworth and Penry (1972), writing on the general effectiveness of antiepileptic drugs, estimated that 70 to 80 percent of chronic epileptics obtain some seizure reduction with their use. Astute management is essential after careful selection of an appropriate drug. Pharmacokinetic variation, i.e., differences in serum concentration produced by a given dose, is common and may be seen when the same quantity of a given drug is administered to different patients (Richens, 1976). As with most effective drugs, there is a potential for undesired side effects which may be manifested in several body systems. These unlooked-for outcomes extend as far as toxicity, which is sometimes expressed psychologically in alterations in mental functioning ranging from mild dulling, through memory and learning




problems, to relatively rare toxic psychosis (Glaser, 1972; Wilder & Bruni, 1981; Wilder & Ramsay, 1974). Pharmacological effects are believed difficult to separate from those of seizure activity itself, particularly when several medications are simultaneously in use (Glaser, 1972, 1974, 1980).
The pharmacological properties of the antiepileptic drugs continue to be clinically studied, descriptions of their efficacy and characteristics forming an extensive literature. One area of interest has been that of determining the range of serum concentration for each agent within which seizure control is enhanced without production of symptoms of intoxication. Less is known of the psychological effects of these medications, but it is evident that both extremes of the dosage continuum require monitoring to secure an optimal clinical treatment effect. Studies of Cognitive-Intellectual Effects of Antiepileptic Drugs
Guey et al. (1967) studied the effects of ethosuximide on 25 referred children with severe petit mal seizures, 15 of whom were mentally retarded. They found intellectual deterioration manifested by memory disorders, speech problems, and emotional disturbances in their subjects. Anticonvulsants already prescribed were continued, the test drug being added to the regimen. The authors, testing with the Wechsler Intelligence Scale for Children (WISC), and the Benton Visual Motor and Retention tests an average of seven months later, reported declining scores on all these measures. Serum levels of drugs were not measured.




Booker (1972) reported the results of an attempt to relate seizure control achieved to serum concentration of anticonvulsants. He found no significant differences in the serum levels of controlled and uncontrolled patients. Seizure control status was seen to be associated with type of clinical seizure, measured IQ, Halstead-Reitan Impairment Index, and the nature of background electroencephalogram (EEG). Prospects for effective control were better in those persons having major motor seizures, less evidence of encephalopathy, and normal (8-12 Hz) EEG background activity.
The WAIS, WMS, and Beck Depression Inventory were given to a general sample of 118 male and female epileptics by Reynolds and Travers (1974). Serum anticonvulsant concentrations were monitored as well. Those patients who had clear signs of drug toxicity, mental retardation from birth, or evidence of gross cerebral lesions were excluded. The investigators looked at such dependent variables as psychomotor slowing, intellectual deterioration, psychiatric diagnosis, and personality change. These were evaluated from records, prior psychiatric reports, relatives' responses to inquiry, a brief mental state examination, and the psychometric measures. The authors found that those patients with psychomotor slowing, intellectual deterioration, psychiatric diagnosis, or personality change had, as a group, higher serum concentrations of phenytoin and phenobarbitone than the group showing no evidence of change. The mean serum values of those persons showing mental changes fell within a recommended therapeutic range, although there was a wide scatter both below and above that range.




Dodrill (1975), evaluating the neuropsychological status of 70 diagnosed major motor and psychomotor seizure patients receiving phenytoin, found differences between groups designated High and Low on the basis of blood level. The High group did less well on tests of gross and fine motor performance. Measures of higher mental functioning, as measured by the Halstead-Reitan Battery, showed no significant intergroup differences.
As part of a larger study of cognitive performance in epileptic patients receiving phenobarbital at two therapeutic dosage levels, MacLeod et al. (1978) reported short-term but not long-term memory retrieval problems at the higher dose. Their tasks consisted of (a) sequential presentations of single digits followed by a probe which the subject responded to as in/not in the set, and (b) presentation of pairs of letters, e.g., AA, Aa, AB, Ab, to be identified as same/ different in terms of the letter name. The experimenters proposed these as measures of speed of access to STM and LTM, respectively, using reaction time (assuming accuracy to be perfect) as the dependent variable. Observing increases in reaction time to the digit task on the higher dosage, MacLeod et al. concluded that increased phenobarbital concentration reduced efficiency of STM scanning. No report was made of efforts to separate attentional mechanisms from those of memory.
Matthews and Harley (1975) administered 15 measures of intellectual and adaptive abilities, sensory discrimination, and motor proficiency to 63 subjects currently being treated with phenytoin, phenobarbital, and primidone. Serum concentrations of the drugs were seen to be in the




toxic range of one or two of the agents for 35 subjects. Comparing these toxic patients with the 28 whose serum levels were nontoxic, the investigators found significantly poorer performance for the former group on WAIS Coding, the Seashore Rhythm, Knox Cube, Maze Coordination, and Static Steadiness tests. They concluded that these findings reflected impairment of attention and concentration, as well as motor coordination and steadiness in those individuals with serum concentrations on the toxic range. They noted, however, that the groups had not been equated for seizure control status, which obscured the source of the deficits found.
Danish researchers Sommerbeck, Thielgaard, Rasmussen, Lohren, Gram, and Wulff(1977) tested a subset of the patients participating in a triple blind study of valproate sodium to assess reported psychotropic effects of the drug. They found differences only of a negative kind in psychomotor tempo and visuospatial analytic and synthetic functioning. Other anticonvulsants being used by these subjects, who had therapyresistant seizures, were held constant. These agents averaged 3.5 per patient. Interpretation of the authors' results is difficult in view of a 33 percent sample attrition, failure of most subjects to complete most assessment measures, and the possibility that the introduction of valproate sodium produced an additive effect rather than one reflecting its unique characteristics.
Dekaban and Lehman (1975) adjusted the anticonvulsant doses of
15 epileptic patients upward twice by 30 to 50 percent of their initial doses. Before each dose change, tests of vigilance, reaction time, and verbal memory were given. The authors reported that "the greatest




number of patients performed best on the lowest dose of medication" (Dekaban & Lehman, 1975, p. 319) on the vigilance and reaction time measures. Verbal recall scores showed similar but nonsignificant trends. Serum levels of anticonvulsants, various combinatiorsof phenytoin, phenobarbital, and primidone, were monitored but not reported.
Thompson and Trimble (1982) studied 15 epileptics receiving
mutiple drugs, either reducing or withdrawing their medications and adding or introducing carbamazepine. Two other groups of 20 and 10 patients had reductions only or no change, respectively. Measures of cognitive and motor functioning improved from the initial to the third session six months later for subjects in the medication change and reduction groups, but no differences were observed in the control group.
Gannaway and Mawer (1981), also seeking to reduce polypharmacy (the use of several drugs of the same or different classes), transferred 18 patients, poorly controlled on their anticonvulsants, to phenytoin therapy alone. After patients achieved blood levels of 15 micrograms/milliliter of serum the physicians increased doses by 25 mg increments until seizures were completely suppressed or clinical intoxication appeared. Tests of speech, motor coordination, verbal memory, and vigilance were given as blood levels ascended. Only the test of vigilance proved sensitive to clinical intoxication states.
Not all studies of anticonvulsant effects on psychological
functioning are cautionary. Dodrill and Troupin (1977) reported results of a double-blind, crossed comparison study of the anticonvulsants carbamazepine and phenytoin. All seizure patients spent four months




on each drug. As opposed to an unmedicated state, the presence of therapeutic blood levels of anticonvulsants coincided with improvements in test scores. During phenytoin dosage periods, patients showed significant reduction of errors on tests of receptive aphasia, the Stroop test (a test of perceptual interference which also measures word-reading fluency), and a brief self-administered mental ability test. While taking carbamazepine, error reduction was seen in a test of constructional dyspraxia. Greater improvements were seen in patients with initially lower average IQ, longer duration of seizures, and more emotional disturbance on the MMPI. The authors saw these effects as being related to attention maintenance and to the fostering of performance of higher-level complex skills requiring manipulation.
Instances of detrimental psychological effects stemming from antiepileptic drug treatment appear to receive more investigative attention. These effects tend to be associated with high or toxic serum concentrations, although idiosyncratic responses in individuals to amounts within recommended bounds are not unusual. It is also evident that therapeutic dosage which reduces seizure activity can provide some concomitant improvement in neuropsychological performance (Dodrill & Troupin, 1977; Thompson, Huppert, & Trimble, 1980, 1981; Thompson & Trimble, 1980, 1981; Trimble, 1981; Trimble & Reynolds, 1976; Trimble & Thompson, 1981).
Rationale For This Study
The preceding review lends support to the existence of memory deficits in complex partial seizure patients. These impairments are more prominent in them than are other kinds of cognitive deficits and,




in LTM performance, often reflective of the hemisphere of involvement. The point in memory processing at which these difficulties arise is obscure, although current evidence supports the view that it occurs beyond sensory memory. The material-specific effects on LTM according to side of focus have not been shown to be due to problems of retrieval or as taking place at earlier stages during the registration process.
Early studies using heterogeneous CPS groups reported general memory deficits in patients in this seizure group, but described no impairments related to side of involvement. The view that side of focus might well be an issue has stimulated investigators to look at left and right CPS groups separately. However, these studies have presented some problems in interpretation of their results because of inappropriate contrast groups, lack of adequate control of relevant variables (e.g., anticonvulsant medication), limitation of measures to those assessing LTM, and unclear reporting of or incorrectly used statistical analyses. The material-to-be-remembered versus modality-of-presentation controversy continues. In part, this is due to the difficulty inherent in developing genuinely nonverbal measures and to the omission of any nonverbal auditory measure.
Studies taking into consideration or controlling antiepileptic drugs have often used patient samples taking multiple medications, which obscures the source of effects seen. These studies have more usually employed LTM measures or, in using STM measures, have not evaluated STM with inclusion of rehearsal prevention.
This study was designed to examine STM functioning in CPS patients
with respect to material-to-be-remembered versus modality-of-presentation.




Patients having generalized major motor seizures served as a contrast group. Antiepileptic medication, limited as much as possible to the drug phenytoin, was controlled. A variety of tests of attention and memory were compared at subtherapeutic, therapeutic, and toxic levels of blood phenytoin to assess the effects of seizure medication on performance in these three groups. This design would allow the investigation of the extent to which serum concentration of phenytoin and its therapeutic range, arrived at by criteria of optimum seizure reduction without clinical toxicity, is meaningful to STM functioning.
The following hypotheses were tested. First, groups will perform differentially with right and left CPS groups showing greater STM impairment than the generalized major motor seizure group. Of the CPS subjects, those with a predominantly left-lateralized temporal focus will perform less well than subjects with a predominantly rightlateralized temporal focus on verbal memory measures. Subjects with a right-lateralized temporal focus will show more impairment in memory for less readily verbalized "nonverbal" material than will subjects with a left-lateralized focus. Second, subjects in the sample will show better STM performance at phenytoin serum concentrations within the therapeutic range than at subtherapeutic or toxic blood levels. Such effects on performance are likely to be particularly detrimental at toxic levels, resulting in the enhancement of the material-specific vulnerabilities according to side of involvement.




CHAPTER TWO
METHOD
Subjects
The subjects in the study were 37 male and 2 female epileptic patients attending the Seizure Clinic of the Gainesville Veterans Administration Medical Center or of the Shands Teaching Hospital of the University of Florida, Gainesville. They were classified into groups of predominantly right-lateralized complex partial seizures (R-CPS), predominantly left-lateralized complex partial seizures (L-CPSJ, and generalized major motor seizures (GEN) on the basis of clinical appearance of the ictus and electroencephalographic recording (EEG). All subjects had complete neurological examinations, computerized axial tomography (CT) scans of the head, and periodic laboratory studies of hepatic and other physiological functioning, in addition to serum anticonvulsant level determinations as needed. They were seen during the investigation either as outpatients or as inpatients if their medical status warranted hospitalization.
Minimum inclusion criteria for the study were presence of a
documented unilateral-focus complex partial or generalized major motor seizure disorder, current or intended treatment with phenytoin (diphenylhydantoin, DPH, Dilantin), age between 18 and 62 years, inclusive, and a minimum Full Scale IQ of 85. Mental retardation, gross brain damage, or other major neurological disease were exclusion criteria.




Ten subjects, two women and eight men, did not complete the study. Two women subjects did not return for the final testing session. Treatment requirements of three men led to use of an alternate anticonvulsant medication; two men did not achieve serum concentrations of phenytoin within the therapeutic range over repeated testings. Of the three remaining subjects, one withdrew, one moved out of the state, and one was disqualified because of prominent clinical symptoms of narcolepsy.
The final sample numbered 29 male subjects of whom 12 had a predominantly right-lateralized temporal seizure focus, 10 a predominantly left-lateralized temporal focus, and 7 evidence of generalized epileptogenic activity on EEG. Sample means for these subjects were age, 45.6 years (SD = 10.2); education, 12.7 years (SD = 3.3); age at onset of seizures, 31.9 years (SD = 14.3); daily frequency of seizures, 0.27 (SD = 0.66); and duration of seizure disorder, 13.7 years (SD = 11.4). Mean IQ scores for the sample were Full Scale IQ, 107.2 (SD = 10.2); Verbal IQ, 108.3 (SD = 12.6); and Performance IQ, 104.7 (SD = 10.5). Within the sample racial balance was 90% white, 10% black; handedness, 93% right-handed, 7% left-handed; etiology of seizures, 59% unknown, 41% known; and CT scans, 59% normal, 41% abnormal. Twenty-one of the subjects (72%) were taking phenytoin as their sole anticonvulsant. The other eight subjects (28%) were taking one additional anticonvulsant drug as follows: methsuximide 3, valproic acid 3, carbamazepine 1, and gamma vinyl GABA 1. Anticonvulsants other than phenytoin were monitored to maintain therapeutic range serum concentrations. With the exception of gamma vinyl GABA, an experimental agent for which a range




of efficacy was not yet established, all medications were at therapeutic or lower concentrations during subject testing.
Subjects received no payment in return for their participation.
They were given feedback on their anticonvulsant serum concentrations, and questions concerning the testing were answered. Subjects were offered the opportunity to discuss their perceived memory problems with the examiner at the end of their participation.
PilJot Studies
Three measures of short-term memory, which variously combined
modality of presentation and material to be remembered, had displayed differences in degree of difficulty in an earlier study with one visual test producing higher scores (Samuels et a]., 1972). The use of a verbal distraction task across the three tests appeared to interfere negligibly with the instrument composed of geometric figures presented visually. Seven staff members and graduate students of the University of Florida were tested individually on separate occasions and found to exhibit greater score variability when a visual task employing unfamiliar figures was introduced as a distractor. This task, rather than the former verbal one, was adopted for use with the measure described.
A second pilot study was done to examine the properties of these
three measures of memory processes to look for the presence of learning effects, specifically, improved scores upon readministration. Six University of Florida freshmen students, three men and three women, were individually given the measures of visual memory for geometric figures, visual memory for consonant trigrams, and auditory memory for consonant




trigrams on three separate occasions. Four of the subjects were given the three tests within five days. The remaining two subjects were tested three times over 10 and 20 days.
Pair-difference t tests were used to compare mean differences in scores between first and second, then first and third testing sessions. All t tests were nonsignificant at p < .05. Because mean scores increased with each testing session on the visual consonant trigrams measure, three forms of this instrument using different sets of trigrams were prepared for use in repeated testing.
Materials and Apparatus
The following test instruments were employed:
The Wechsler Adult Intelligence Scale (WAIS). A short form was
given which includes all of the subtests of this measure. Two subtests are left intact and every other or every third item of the remaining eight are used (Satz & Mogel, 1962). A validation study using this abbreviated version for retesting of schizophrenic patients revealed correlations of r = .98 with Verbal IQ (VIQ), r = .93 with Performance IQ (PIQ), and r = .97 with Full Scale IQ (FSIQ) with the full WAIS (Mogel & Satz, 1963).
The Visual Vigilance Test (VVT). This is a measure of attention in which a series of pairs of simple visual geometric figures is presented in random order. A Terak Model 8532-1 Monitor computer is employed in presenting stimuli. The examinee's response is a bar press to the target stimulus, two vertically displayed circles on the left or right of the screen. Presentation time is 1.0 second; interstimulus




interval is 1.5 seconds. Target stimuli comprise 10% of 900 presentations. Scoring includes omission and commission errors, separately and summed, in addition to correct responses. The task is analogous to the Continuous Performance Test described by Rosvold, Mirsky, Sarason, Bransome, and Beck (1956) which has been used to evaluate the effect of phenothiazine medication on schizophrenics, showing score improvements in treated versus untreated groups (Kornetsky, 1972).
The following three instruments are adaptations of measures used in earlier investigations of short-term memory (Butters et al., 1970; Peterson & Peterson, 1959; Samuels et al., 1972; Witmer, 1935).
The Visual Geometric Test (VGT). This is a measure of short-term visual memory. An unfamiliar geometric line drawing on a 4-inch by 6-inch card (see Appendix A for examples) is manually presented briefly, followed by either immediate or delayed (3, 9, or 18 seconds) selection from nine individually presented patterns. In the delay condition a distraction task, consisting of copying unfamiliar graphics,is interposed between stimulus presentation and response (see Appendix B). Percentages of trials correct in the immediate and in each delay condition are the scores.
The Visual Consonants Test (VCT). This is a measure of short-term visual memory. A consonant trigram of less than 21% association value is manually presented briefly, in horizontal array, on a 4-inch by 6-inch card followed by either immediate or delayed (3, 9, or 18 seconds) recognition and selection from 10 consonants (Appendix C). In the delay condition a distraction task, counting backward from a three-digit number presented by the examiner, is interposed. Percentages of letters




correct in the immediate and in each delay condition are the scores. Three forms are available for use in repeated testing.
The Auditory Consonants Test (ACT). This is a measure of short-term auditory memory. A consonant trigram of less than 21% association value is verbally presented at the rate of 1 second per letter, followed by either immediate or delayed (3, 9, or 18 seconds) verbal report of letters recalled. The distraction task and scoring are the same as those used in the VCT.
The Tonal Memory subtest (TMT) of the Seashore Measures of Musical Talents. This instrument is composed of 30 pairs of tone sequences. The examinee is required to identify the number of the single tone changed in the second playing of each pair (Seashore, Lewis, & Saetviet, 1960). Tone sequences are 3, 4, and 5 in length for 10 trials of each. The subtest is recorded and was presented on an Ampex cassette tape using a Panasonic RQ-212DS recorder/player. Scores are based on percentage of correct responses per 10 trials and over 30 trials. This measure has proven sensitive to memory problems in right temporal lobectomy as compared with left temporal lobectomy patients, but not to preoperative differences in the small groups examined (Milner, 1962).
A Patient Information Form. This is an instrument developed for
this study and designed to record demographic and seizure-related medical information (see Appendix D). Used in an interview context, this instrument provides an ongoing record of the subject's anticonvulsant dose, serum concentration, and dates of testing sessions, as well as relevant medical record data. Seizure frequency estimations recorded here were based on reported seizure activity during the preceding 90 days.




Procedure
Patients were invited to take part in this investigation if one of three conditions was present:
1.) seizure dyscontrol believed due to subtherapeutic anticonvulsant serum concentration whether by noncompliance, altered dose requirements, or new onset of seizure disorder;
2.) clinical symptoms of phenytoin toxicity; or
3.) expressed interest on the part of a patient in perceived memory problems with or without a documented treatment history of difficulty in maintaining serum concentrations of anticonvulsants within the therapeutic range.
Thus, subjects began participation at any categorical blood level of phenytoin: subtherapeutic (LOW), 0 to < 10, therapeutic (NORMAL), 10 to < 20, or supratherapeutic (TOXIC), > 20 micrograms per milliliter of serum concentration. A blood sample for measurement of anticonvulsant serum level was drawn within hours of each study testing session. Testing was done blind to serum concentration of phenytoin or other medication because of an approximately 48-hour delay in receiving laboratory reports on routine requests for this procedure.
After reading and signing the consent form (see Appendix E),
subjects were interviewed to obtain a history of their seizure disorder, then given the set of measures described. The abbreviated WAIS was given during a session when the subject was able to undertake the task, i.e., not clinically drug-intoxicated or otherwise appearing disadvantaged. The neurologist then adjusted the dose of phenytoin if present symptoms and/or history warranted it, or awaited the laboratory results to evaluate the amount and direction of change needed.




Subsequent sessions were scheduled for a time when the anticonvulsant level was expected to have changed in response to the new dose. A minimum of three testing sessions took place until at least one was concurrent with a phenytoin blood level within the therapeutic range and one concurrent with a level outside the range. The memory measures were presented in differing orders on each occasion of testing. The VVT was given first or last because of the necessity for use of stationary equipment located at a distance from the usual testing site. No data collected within 24 hours of a seizure were used in analyses; the data were discarded and the session rescheduled.
Instructions to the Subjects
Visual Vigilance Test. Subjects were instructed by means of programmed instructions, reinforced verbally by the examiner, to watch the screen, pressing the space bar only at the appearance of two circles, vertically displayed, on the left or right side. After a brief practice session, the 900 stimuli of the measure were presented. Average length of testing was 40 minutes.
Visual Geometric Test. In the immediate response condition
subjects were shown a line drawing of a figure, then asked to identify it by stopping the examiner during a sequential review of the nine stimuli. The delay condition featured presentation of the target stimulus, followed by the examiner' s saying, "Copy." The subject then copied unfamiliar graphics for either 3, 9, or 18 seconds until hearing, "Stop," and again selecting the most recently presented drawing from those reviewed. The number of trials was 24, 6 immediate and 18 delayed




with the intervening distraction task. Immediate responses were grouped at the beginning as a perceptual check; delayed responses were arranged randomly. A practice period employing six nontest stimuli with both immediate and delayed response conditions preceded the actual test.
Visual Consonants Test. Subjects were given instructions similar to those above, except for the response format and distraction task. The response requested was that of selecting the letters in the displayed trigram from an array of 20 different letters and circling them. Correct order of letters was not a requirement. The distraction task consisted of counting backward from a three-digit number supplied by the examiner immediately after withdrawing the stimulus. The number of trials was 16, 4 immediate and 12 with intervening distraction. A practice session using nontest trigram stimuli preceded the test.
Auditory Consonants Test. Subjects were instructed to listen to the three letters spoken by the examiner. In the immediate response condition they were to repeat the letters heard. The delay condition imposed the backward-counting task before a verbal response was given. In all other respects this measure was identical to the VCT.
Tonal Memory Test. Subjects were given the standard instructions and practice experience described in the manual. The examiner told subjects they would hear several tones, a pause, then the same tones repeated except for one which would be changed. Subjects were told their task was to decide which note was different in the second playing and report it by number. After listening to several randomly selected pairs from the measure, subjects began the actual test. Response consisted of circling the number of the changed tone after hearing a pair of sequences.




Statistical Analyses
The independent variables were type of diagnosed seizure disorder (R-CPS, L-CPS, GEN) and phenytoin blood level, either absolute (ABL) or categorical (LOW, NORMAL, TOXIC). Dependent variables were the absolute scores obtained on the measure of attention (VVT) and its omission, commission, and total errors (OE, CE, TE, respectively); percentage correct responses on the tests of short-term memory, both immediate and delayed (VGT-IM and -STM, VCT-IM and -STM, ACT-IM and -STM); and, for the TMT, both total (TMT) and partial (TMT3, TMT4, TMT5) percentages correct. Scores obtained at increasing delay intervals within the testing session on each short-term memory test were converted to regression lines representing loss of information from short-term memory over time. Each session thus provided 3 lines for a total of 87 in the sample. The slopes of these lines (SLOPEVGT, SLOPEVCT, SLOPEACT) became dependent measures in multivariate analyses for group and blood level effects.
Univariate one-way analyses of variance (ANOVA) were employed to test for heterogeneity of seizure diagnostic groups. Relationships between subject variables and between dependent measures were examined using Pearson product-moment correlations.
The research hypotheses of differential performance of memory tasks by seizure groups and of impairment of this performance at phenytoin blood levels outside the therapeutic range were tested with two-way repeated measures analysis of variance and two-way repeated measures multivariate analysis of variance (MANOVA) using the GLM procedure of the Statistical Analysis System (SAS) for unbalanced data (Goodnight, 1979; Shaffer, 1979). Dependent measures in these procedures were




absolute scores, numbers of errors, and percentage of correct responses as appropriate to the instruments administered. The contributions to group effects of each of five uncontrolled memory- and seizure-related variables were evaluated post hoc by means of GLM analyses of covariance.
Tests of the above hypotheses as they relate to loss of information from short-term memory over time were performed on within-session scores converted to regression slopes. Because few subjects (n = 4) experienced all three possible phenytoin blood level categories, group effects were first tested by applying the MAANOVA procedure at each categorical blood level. Repeated measures MANOVA was used to test both seizure group and blood level effects upon movement from LOW to NORMAL levels and from NORMAL to TOXIC levels separately.
Tests for the presence of a linear versus quadratic relationship between phenytoin blood level and short-term memory performance were done using the CONTRAST statement within the GLM procedure.
With the exception of hand-calculated paired t tests comparing mean scores of short-term memory tests across sessions in order of occurrence to test for the presence of a sequence effect, all statistical analyses were performed at the Northeast Regional Data Center, University of Florida, Gainesville.




CHAPTER THREE
RESULTS
Overview of Results
The hypothesis that epileptic subjects with focal complex partial seizures have greater short-term memory impairment than subjects with generalized major motor seizures was not supported at a significant level. Although short-term memory test data tended in the predicted direction, and to be consistent with earlier investigators' reports of material-specific deficits according to hemisphere of involvement, test behavior related to attentional functioning was a better predictor of seizure type in this sample.
Similarly, the hypothesis that phenytoin serum concentrations outside the accepted therapeutic range of 10 to 20 micrograms per milliliter of serum exert detrimental effects on short-term memory was not confirmed. Once again, attention-related performance was more responsive than measures of memory functioning to alterations in anticonvulsant level.
Subjects in the study demonstrated short-term memory impairments
across seizure types and anticonvulsant serum levels relative to control and normative groups. Material-related rather than modality-related information loss problems were seen in subjects with focal seizures.




Description of Groups and Relationships of Variables
The three subject groups, R-CPS, L-CPS, and GEN, did not differ in age, education, age at seizure onset, duration of seizure history, frequency of seizures, FSIQ, or VIQ. They were also alike with regard to racial composition, handedness, etiology of seizure disorder, evaluation of CT scan, and anticonvulsant medications being taken. They were, however, dissimilar in PIQ, F(2, 26) = 3.49, p < .05, the two CPS groups being equal and scoring higher than the GEN group. Group and sample characteristics are presented in Table 1.
Correlations Between Subject Variables
Somewhat differing strengths of correlations between subject variables among groups can be seen in Table 2. The R-CPS and GEN groups showed less association of PIQ with FSIQ than did the L-CPS group. Both CPS groups exhibited positive association between age and age of seizure onset, a reflection of those groups' older members' later development of seizure disorder. In the R-CPS group the relationship of age to etiology, r= .70, p < .01, revealed an association between earlier-developed seizures and known etiology, as well as later-developed seizures and unknown etiology in these subjects. This group also showed a negative correlation between PIQ and CT scan, r = .75, p < .01, indicating that lower PIQ scores were associated with abnormal CT scans of the head, higher PIQ scores with normal scans.
Correlations Between Subject Variables and Dependent Measures
Correlations between subject descriptor variables and test scores presented in Table 3 show relationships between FSIQ and VIQ and




Table 1
Characteristics of Subjects

Group Variable Number Age in Years Race White Race Black Right-handed Left-handed Education in
Years
FSIQ
VIQ PIQ
Etiology
Known Etiology
Unknown

Right CPS n M SD

47.8

14.2 110.4 111.5 107.8

9.9

2.8 9.8 13.7 8.5

Left CPS Generalized
n M SD n M SD
10 7

46.7 10.1

11.4 105.7 104.1 106.9

4.0 9.6 9.4 11.5

Total Sample N M SD

29 45.6

40.3 10.5

11.9 103.7 108.9 96.3*

2.5 11.2 14.6 9.0

12.7 107.2 108.3
104.7

10.2

3.3 10.2
12.6 10.5

4 6




CT Scan
Normal CT Scan
Abnormal
Receiving DPH
Only
Age at Onset Duration
Seizures
Seizure b
Frequency

a Seizure history in years. 90 days.
*p< 0.05.

btennumber of seizures

daily, based on reported occurrence during prior

14.4 9.5

34.2 13.6
0.4

34.2 12.6 0.3

14.8 10.3 0.3

24.9 15.4 0.1

12.8
16.4 0.1

31.9 13.7 0.3

14.3 11.4 0.7




Table 2
Correlations Between Subject Variables

Variable
n = 12
Age in Years Education in Years
FSIQ
VIQ PIQ
Age at Onseta Duration in Years

Frequency Etiologyb CT Scanc
n = 10
Age in Years Education in Years
FSIQ
VIQ

Age

Educ FSIQ

-.13
1.0

.25
.60*
1.0

VIO

.17 .57 .93**
1.0

PIQ Onset
Right CPS .30 .76**
.28 -.30 .55 .15
.19 -.06 1.0 .55

Dur Freq Etio CT Scan

-.11
.32 .03 .27
-.52
-. 73**

1.0

1.0

.53 1.0

.51 .44
1.0

Left CPS
.70* .20 .72*
.63* .12 -.35
.89*** .89*** .72* 1.0 .58 .73*

-.06
-.31
-.54
-.37

.39 .18
-.03
-.06
.06
-.28
.05 1.0

.21
-.17
-.27
-.30

.70**
-.20
-.16
.08
.08 .62*
-.20
-.34

-.21
-.44
-.26
.04
-.75**
-.32
.26
-.32
-.17
1.0

-.04
.13
-.40
-.17




PIQ
Age at Onseta Frequency Etiologyb CT Scanc

n=7

Age in Years Education in Years
FSIQ
VIQ PIQ
Age at Onseta Duration in Years

Frequency Etiologyb CT Scan

1.0

-.73
1.0

.20 .38 1.0

-.005
.59
.95**
1.0

Generalized
.30 .008
-.20 -.03
.74 .11
.49 .009
1.0 .29

.63
-.51
.04
-.01
.13
-. 77*
1.0

-.44
.07
-.16
-.11
-.08
-.08
-.22

.39
-.03
.66
.51 .76 .04 .22 .33

1.0

.02
.32 .38
.44
-. 005 .42
- .32
-.65
-.35

aAge at Onset formula: Age Duration. positive correlation reflects association with unknown etiology; negative with known etiology. cNegative correlation refelcts association with normal scan; positive with abnormal scan.
* p < 0.05. ** p < 0.01 *** p < 0.001

.55
1.0

-.58
-. 73*

-.17
-.13
.38
1.0

-.56
-.03
-.002
-.22
1.0

.30
.10
-.04
-.15
-.41




Table 3
Correlations of Subject Variables With Test Scores
All Groups Combined
Variable Age Onset Duration Frequency FSIQ VIQ PIQ
Visual Vigilance
Omission Errors -.10 .02 -.12 .34*** -.08 -.07 -.04
Commission Errors -.16 .05 -.21 .10 -.22* -.11 -.30**
Total Errors -.18 .05 -.22* -.04 -.23 -.12 -.31**
Visual Geometric
Immediate .08 -.09 .19 .03 .26** .27** .15
Delayed (STM) -.03 -.03 .01 -.04 .21* .17 .18
Visual Consonants
Immediate .24* .06 .14 .08 .13 .16 .03
Delayed (STM) .26** -.18 .46*** .09 .23* .32** -.004
Auditory Consonants
Immediate .09 -.05 .15 .04 .003 -.01 .02
Delayed (STM) .11 -.18 .33** .09 .26** .35*** .001




Tonal Memory
3 Tones 4 Tones 5 Tones
Note. N = 29
* < 0.05.

-. 003 .07
.15

** p < 0.01.

-.03
.01
.10

*** p < 0.001.

.04 .05 .004

.13
.004 .02

.19
.17 .22*

.14 .20




performance on the short-term memory aspect of the two tasks employing verbal materials, the VCT and ACT. Although FSIQ was associated with better performance on both immediate and short-term memory aspects of the VGT, which is composed of line drawings of geometric stimuli, VIQ was related only to the immediate memory portion of this instrument. The verbal measures were also positively correlated with longer seizure duration, the visually presented VCT with age as well. These and other relationships seen in the full sample are better traced to sources in the groups by examining Table 4 where age, age of onset, and duration can be seen to be differentially related to test scores within each group. For example, better scores on short-term memory components of verbal measures by R-CPS subjects were associated with earlier onset and longer duration. Conversely, however, the scores of GEN subjects on short-term memory components of the geometric test were associated with later onset and shorter duration. The L-CPS group showed relationships between better VCT scores and VIQ (r = .60, p < .001), and between better VGT scores and PIQ (r = .35, p < .05), suggesting their use of abilities other than verbal skills in VGT performance. Correlations Between Frequency and.Duration of Seizure Disorder and Dependent Measures
Several correlations between the subject variables of duration,
frequency, and IQ with errors made on the VVT were found to be differentially distributed across groups. As Table 4 shows, the negative relationship of duration to total errors became nonsiqnificant in any single group. Seizure frequency remained associated with omission




Table 4
Correlations Between Subject and Test Score Variables by Group
Variable Age Onset Duration Frequency FSIQ VIQ PIQ
Right CPS
Visual Vigilance
Omission Errors -.16 -.09 -.03 .86*** .01 -.01 .08
Commission Errors .13 .06 .05 -.10 -.02 .06 -.18
Total Errors .09 .03 .04 .10 -.02 .05 -.15
Visual Geometric
Immediate -.07 -.20 .24 .11 .08 .14 -.09
Delayed (STM) -.27 -.24 -.09 -.05 .25 .18 .22
Visual Consonants
Immediate .27 .18 .004 .11 .39* .39* .18
Delayed (STM) -.14 -.49** .61*** .17 .07 .33* -.56***
Auditory Consonants
Immediate .00 .00 .00 .00 .00 .00 .00
Delayed (STM) -.31 -.56*** .52*** .20 .06 .22 -.37*
Tonal Memory
3 Tones -.42** -.34* .07 .23 -.07 .02 -.24
4 Tones -.47** -.33* .02 -.05 -.10 -.04 -.20
5 Tones -.38* -.25 -.01 .01 .05 .13 -.18
Left CPS
Visual Vigilance
Omission Errors -.05 -.02 -.02 -.28 -.08 .02 -.17
Commission Errors -.09 .16 -.31 -.16 .09 .09 .06
Total Errors -.09 .12 -.27 -.24 -.05 .08 -.01
Visual Geometric
Immediate .20 -.08 .30 -.11 .24 .23 .23
Delayed (STM) .34 .02 .30 .06 .36* .31 .35*




Visual Consonants
Immediate .32 .05 .24 .17 .05 .06 .04
Delayed (STM) .52** .11 .35* .02 .52** .60*** .35
Auditory Consonants
Immediate -.23 -.31 .23 .15 -.25 -.27 -.17
Delayed (STM) .41* .06 .31 -.04 .41* .48** .27
Tonal Memory
3 Tones .31 .49** -.40* .07 .21 .09 .28
4 Tones .56*** .64*** -.37* .16 .25 .19 .25
5 Tones .56 .70 -.46** .13 .46** .33 .48**
Generalized
Visual Vigilance
Omission Errors -.39 .21 -.41 .74*** -.16 -.13 -.14
Commission Errors -.23 .25 -.34 -.23 .42 -.32 -.47*
Total Errors -.26 .27 -.37 -.19 -.43* -.33 -.48*
Visual Geometric
Immediate .16 .21 -.06 .31 .67*** .56** .69***
Delayed (STM) -.23 .43* -.49* -.20 .09 .20 -.13
Visual Consonants
Immediate .30 .09 .12 -.36 -.13 -.23 .11
Delayed (STM) .45* -.30 .52* -.38 -.06 -.05 -.03
Auditory Consonants
Immediate .27 -.002 .17 -.38 .07 .10 -.03
Delayed .39 .09 .18 -.61** .30 .30 .18
Tonal Memory
3 Tone .30 .27 .40 -.11 .65** .52* .67***
4 Tones .18 -.54** .53** .18 .52* .46* .43*
5 Tones .34 -.43* .55** -.31 .17 .03 .36

* p < 0.05. ** p < 0.01.

*** p < 0.001.




errors for the R-CPS and GEN groups (for both, p < .001). However, a negative relationship of FSIQ and PIQ to total errors and of PIQ to commission errors continued only in the GEN group.
Because duration and frequency of seizures have often emerged as relevant variables in studies of epileptics' neuropsychological functioning, these relationships were inspected by dividing the groups on these variables to further examine their effect on error rates. Subjects were separated into those having fewer and more than 10 years seizure history, and those reporting fewer and more than four seizures a month (B. J. Wilder, personal communication, February 17, 1984). Errors on the VVT by group and duration are presented in Table 5. Subjects in the L-CPS and GEN groups showed decreased VVT commission errors with more than 10 years duration, whereas the R-CPS group showed an increase in this type of error. Omission errors also differed by groups according to duration. The R-CPS group showed little change, the L-CPS group showed an increase with longer duration as compared with the GEN group whose mean error rate decreased with longer seizure history.
Table 6 presents the VVT error rates by group and frequency of seizures. As shown, greater numbers of omission errors were made by R-CPS and GEN subjects, but L-CPS subjects having more than four seizures a month had fewer errors. All groups made fewer commission errors at the higher seizure frequency.




Table 5
Mean Visual Vigilance Test Errors in Subjects Having Seizures Fewer/More Than Ten Years

Group
n
Error Type Subgroup n Fewer Than 10 Years Subgroup n More Than 10 Years

Right CPS
12
Omi t Commi t

1.22

1.81

1.78

7.59

Left CPS
10
omit Commit

2.67

4.17

10.67

1. 11

General ized
7
Omit Commit

2.89

49.2

0.25

2.67




Table 6
Mean Visual Vigilance Test Errors in Subjects Having Fewer/More Than Four Seizures a Month

Group
n
Error Type Subgroup n Fewer than 4/month Subgroup n More than 4/month

Right CPS
12
Omit Commit

0.87

2.95

10.53

2.50

Left CPS
10
Omit Commit

6.56

2.54

12.17

1.58

Generalized
7
Omit Commit

0.35

6.25

28.42

12.25




Correlations Between Phenytoin Dose, Blood Levels, and Dependent Measures
Mean phenytoin doses ordered and serum concentrations measured at each categorical level in the three subject groups are presented in Table 7. There was a significant association between ABL and CE in the GEN seizure group (r = .52, p < .05). Although dose of phenytoin and ABL were related at the lowest serum concentration measures in each subject (r = .43, p < .02), subsequent measurements of phenytoin ABL at higher concentrations were uncorrelated with dose prescribed.
Correlations Between Study Measures
Intercorrelations of the measures used in the study were VCT with ACT (r = .68, p < .001), both VCT and ACT with VGT (r = .40, p < .001) in the full sample. As shown in Table 8, only L-CPS scores on these short-term memory instruments demonstrated strong commonality among all three tests. Scores of R-CPS subjects were highly correlated on verbal tests only. Scores of GEN subjects showed an association between the VCT and ACT which approach significance (r = .40, p = .07), as well as a correlation between the VVT and TMT (r = .53, p = .01), two measures which have in common a task demand for monitoring and signal detection.
Correlations between study variables revealed expected, unexpected, and ambiguous associations. The differential association by groups between IQ variables revealed a significant relationship of PIQ with FSIQ only in the L-CPS subjects. A negative correlation of PIQ with CT scan was also seen. Performance on the STM component of the VGT was associated with PIQ, but only in the L-CPS group. However, in the full sample, the VCT and ACT short-term components showed associations




Table 7
Mean Dose and Phenytoin Concentration at Each Categorical Blood Level

Right CPS
12
Dose ABL

368 443 475

7.03 14.45 24.25

Left CPS
10
Dose ABL
250 4.79
392 13.67
450 22.55

Generalized
7
Dose ABL
388 5.98
414 13.63
350 28.13

Total Sample
29
Dose ABL

341 416 438

6.12 13.97 24.80

Note. Phenytoin dose in milligrams/day; ABL in micrograms/milliliter.

Group Number

LOW NORMAL TOXIC




Table 8
Intercorrelations of Total Test Scores by Group

VVT

n = 12
Visual Vigilance Visual Geometric Visual Consonants Auditory Consonants Tonal Memory
n = 10
Visual Vigilance Visual Geometric Visual Consonants Auditory Consonants Tonal Memory
n = 7
Visual Vigilance Visual Geometric Visual Consonants Auditory Consonants Tonal Memory

Right CPS
1.0 .19
1.0
Left CPS
1.0 .19
1.0
General i zed
1.0 .09
1.0

*** p < 0.001.

** p<0.01.

Variable

-.13
-.05
1.0
.11
.73***
1.0
.37
.10
1.0

-.07
.21 .65***
1.0
.15
. 66*** 83***
1.0
.31 .31
.40
1.0

.06
.29
-.01
.28
1.0
.19 .23 .08
.01 1.0
. 53**
-.16
.34 .27 1.0




with both VIQ and FSIQ. The association of longer seizure duration and age with better verbal STM performance was of interest. Correlations between the three STM tests were seen in the sample, a reflection of their strong associations in the L-CPS group. An absence of association between phenytoin dose and subsequent blood level at doses beyond the lowest recorded was also noted.
Group and Categorical Blood Level Effects on Test Scores
Univariate Analyses of Variance
Univariate groups (3) by categorical blood levels (3) repeated measures ANOVAs were performed using the GLM procedure of SAS for unbalanced data with VVT, TE, OE, CE, T71T, TMT3, TMT4, and TMT5 as dependent measure individually (Goodnight, 1979).
A trend toward a group effect was seen for VVT and Omission Errors (F(2, 26) = 3.03, p < .07, for both), and for Total Errors (F(2,26) =
3.13, p < .06). There was also a main effect for group (F(2, 26) =
3.54, p < .04) on Commission Errors. Duncan's Multiple Range Test for differences in means, performed post hoc, revealed that R-CPS and L-CPS errors, 6.1 and 4.9, respectively, were equal and fewer than those of the GEN group, 22.6 (Kirk, 1968).
A main effect for blood level was present in Total Errors (F(2,52) = 4.71, p < .01), and in Commission Errors (F(2,52) = 4.83, p < .01). Post hoc examination of means using Duncan's statistic showed errors at LOW and NORMAL were equal, Total Errors at LOW, NORMAL = 11.6, 7.1, in order, and fewer than those at TOXIC, 26.6. Commission Errors at LOW and NORMAL levels, 9.8 and 5.1, respectively, were also alike and fewer than those at TOXIC, 22.7.




Analyses for group and blood level effects on TMT, TMT3, TMT4, and TMT5 were nonsignificant. There were no interactions. Multivariate Analyses of Variance
Percentages of correct responses on immediate and short-term portions of the VGT, VCT, and ACT were dependent measures in a groups (3) by categorical blood levels (3) repeated measures MANOVA for unbalanced data (Goodnight, 1979; Shaffer, 1979; Tatsuoka, 1971).
The overall MANOVA was nonsignificant for a group effect (HotellingLawley Trace = 0.50, F approximation (12, 40) = 0.84, p < .61), as were all univariate ANOVAs. There was no significant blood level effect on these measures (Hotelling-Lawley Trace = 0.36, F approximation (12, 92) = 1.36, p < .20). One ANOVA, however, showed a main effect for blood level on the VCT-STM (F(2, 52) = 4.26, p < .02) in the direction of improved scores with phenytoin serum concentration increase. Scores at blood levels LOW through TOXIC were 62.3, 69.1, and 71.8, in order. No interaction between group and blood level was present (Hotelling-Lawley Trace = 0.39, F approximation (24, 182) = 0.75, p< .80). Univariate Analyses of Covariance
Five variables, duration of seizures, frequency of seizures, FSIQ,
VIQ, and PIQ, seen to be related to group performances, were each employed separately as a covariate in a GLM ANOVA with groups (3) as independent, test scores and errors as dependent measures (R. L. Carter, personal communication, April 26, 1984; Goodnight, 1979; Tatsuoka, 1971).
Covariance analysis of duration revealed a main effect for duration (F(i, 83) = 5.80, p < .02) and for group (F(2, 83) = 3.79, p < .03)




on Total Errors. Least squares mean errors of R-CPS and L-CPS groups, 7.8 for both, were equal to each other and fewer than those of the GEN group, 25.0. Much the same effect of duration on Commission Errors was seen (F(i, 83) = 5.20, p < .03) with enhancement of a group effect (F(2, 83) = 4.47, p < .01).
Covariance analyses of seizure frequency showed a main effect for frequency on VVT and Omission Errors scores (F(1, 83) = 11.82,
2 < .001, for both). Group effects on VVT and Omission Errors approached significance (F(2, 83) = p< .08). having been slightly enhanced with the addition of frequency as a covariate in the model. Least squares mean scores on the VVT were 86.4, 88.2, and 88.6 obtained by subjects in the L-CPS, GEN, and R-CPS groups, respectively. Corresponding Omission Errors were 3.6, 1.8, and 1.4.
Those analyses covarying FSIQ, VIQ, and PIQ showed FSIQ effects on VGT-IM (F(1, 83) = 5.95, p < .02) and on ACT-STM (F(1, 83) = 5.34,
2 < .02). Covariance effects of VIQ were shown on VGT-IM (F(1, 83) = 4.71, 2< .03), VCT-STM (F(I, 83) = 7.84, p < .01), and on ACT-STM (F(1, 83 = 8.68, p < .004). A group effect on Commission Errors (F(2, 83) = 3.92, < .02) was maintained with covariance of VIQ. No group differences appeared with covariance of PIQ, which showed an effect on Total Errors (F(1, 83) = 4.09, p < .05).
No covariance analyses for group effects on the TMT or its components achieved significance.




Group and Categorical Blood Level Effects on Within-Session Performance
Scores earned at varied response delay intervals during each testing with the VGT, VCT, and ACT were converted to regression lines representing performance within the individual session using the GLM procedure of SAS. The slopes of these lines served as dependent measures in GLM MANOVAs testing group effects on loss of information from short-term memory at each blood level separately (Campbell & Stanley, 1963; Goodnight, 1979).
None of these MANOVAs was significant: at LOW blood level (HotellingLawley Trace = 0.420, F approximation (6, 28) = 0.98, p < .46), at NORMAL blood level (Hotelling-Lawley Trace = 0.236, F approximation (6, 46) = 0.90, < .50), at TOXIC blood level (Hotelling-Lawley Trace = 1.377, F approximation (6, 14) = 1.61, p < .22). At the TOXIC level, one ANOVA showed a main effect for group on VGT (F(2, 10) = 5.02, p < .03). Mean VGT performance slopes at the TOXIC level were -1.41, 0, 46, and -0.72 for subjects in the R-CPS, L-CPS, and GEN groups, respectively.
The within-session regression slopes were also the dependent measures in tests for group (3) and blood level (2) effects on loss of information as two sets of subjects moved from LOW to NORMAL (n = 20), and from NORMAL to TOXIC (n = 13) phenytoin blood levels, using GLM repeated measures MANOVA (Campbell & Stanley, 1963; Goodnight, 1979; Shaller, 1979).
The MANOVA at LOW to NORMAL levels revealed no significant group effect (Hotelling-Lawley Trace = 0.697, F approximation (6, 28) = 1.63, p < .18) or blood level effect (Hotelling-Lawley Trace = 0.455, F approximation (3, 15) = 2.28, p < .12). One participating ANOVA approached significance (F(22, 17) = 2.02, p < .07) for VCT,




attributable to a blood level effect in the model (F(1, 17) = b.73, p < .02). Slopes of the VCT at LOW and NORMAL levels were -2.13 ana
-1.66, in order, suggesting that better retention occurred at phenytoin levels in the therapeutic range. There was no significant interaction between group and blood level (Hotelling-Lawley Trace = 0.258, F approximation (6, 28) = 0.60, p < .73).
A second MANOVA at NORMAL to TOXIC levels was nonsignificant for group effects (Hotelling-Lawley Trace = 2.57, F approximation (6, 12) = 2.57, p < .08), for blood level effects (Hotelling-Lawley Trace = 0.136, F approximation (3,7) = 0.32, p < .81) or for interaction (HotellingLawley Trace = 0.416, F approximation (6, 12) = 0.42, p < .85).
Overview of Mean Single and Combined Test Performances by Subjects
Mean scores earned by subjects over three testing occasions are presented in Table 9. The VVT and its error scores are expressed in absolute numbers, all others in percentage of correct responses. Scores on the VVT approached perfect performances of 90 correct. Groups did, however, distinguish themselves in numbers of commission errors made. Scores in the full sample were uniformly depressed on the remaining measures relative to normative and control groups. Short-term memory scores on the VGT, VCT, and ACT were lower by 9.6, 19.4, and 12.3 percentage points, respectively, when compared to those of older (M = 39.5 years) Boston VAH and community volunteer control subjects for whom educational level (M = 13.6 years), but not FSIQ, was reported. In contrast to postoperative anterior temporal lobectomy patients whose age was younger (M = 31.5 years), whose FSIQ (M = 104.9) and educational




Table 9
Mean Percentage Correct Responses Over All

Testing Sessions

Group Number

Visual Vigilance a
Omission Errors
Commission Errors Visual Geometric
Immediate
Delayed (STP)
Visual Consonants
Immedi ate
Delayed (STM)
Auditory Consonants
Immedi ate
Delayed (STri)

Right CPS
12
M SD
88.3 3.2
1.7 3.2
6.1 13.9

96.7
79.4
99. 1 71.0
100.0 70.7

6.8
15.4
3.3
14.4
0.0
14.0

Left CPS
10
M SO
86.4 5.9 3.6 5.9 4.9 12.9

92.8 83.8
97.0
64.0
97.7 62.6

Generalized7
M SD

88.3
1.4 22.6

96.8 82.3
99.2 68.6
99.2 69.9

13.6 17.6
7.0 21.6
1.5 16.3

3.5 3.0
45.2

6.8 11.5

Total Sample
29
M SD
87.7 4.4 2.3 4.3 9.7 25.8

95.4 81.6

2.4 17.2

98.4 68.0

2.4 15.9

99.7 67.7

9.8 15.3
4.8 17.9
1.5 15.6




Tonal Memory
3 Tones 4 Tones 5 Tones

71.4 56.9
43.1

24.5 25.4 27.1

70.7 58.3
42.3

18.0 26.9
34.0

72.4 54.3 42.9

19.7
24.4 27.2

71.4 21.1 56.8 25.4 42.8 29.4

Note. All subjects completed three sessions. a VVT scores and errors in absolute numbers.




level (M = 11.9 years) were reported, and who were tested some years after their surgical interventions (M = 6.5 years), the subjects in this sample scored higher only on the ACT, 67.7%, versus the patients' 62.6% (Samuels et al., 1972). Groups in this study showed a tendency toward diverging short-term memory performances on these instruments, the L-CPS subjects scoring lowest on the verbal measures, the VCT and ACT, the R-CPS subjects scoring lowest on the VGT.
Tonal Memory subtest scores did not differ appreciably across
groups. Mean performance in the sample of 57% correct identifications, however, resembled that of postoperative right temporal lobectomy patients, 53% correct, and corresponded to 9th percentile scores in the young adult student group whose normative data are provided by the authors (Milner, 1962; Seashore et al., 1960).
The following tables and figures describe group performances on individual measures in keeping with the clinical constraints of the investigation. Because so few subjects experienced three categorical phenytoin blood levels (n = 4), mean scores of each group are derived from varying numbers of sessions by members of that group at a given blood level.
Visual Vigilance Test
Changes in VVT performance at different phenytoin blood levels are presented in Table 10 in terms of error scores. Figure 2 shows that both CPS groups experienced a Commission Errors rate reduction at the NORMAL level, the L-CPS group showing improvement in Omission Errors as well. Subjects in the GEN group exhibited Commission Errors




Table 10
Mean Number of Omission and Commission Errors on Visual Vigilance Test at Varying Phenytoin Blood Levels

Group Number Error Type Sessions ABL LOW ABL NORMAL ABL TOXIC

Right CPS
12
Omit Commit

0.8 1.6 2.9

7.2
1.9 14.3

Left CPS
10
Omit Commit

5.0 2.1 8.3

11.1 2.2 7.3

General i zed
7
Omit Commit

0.1 2.4 1.5

12.0 17.7 55.0




D rnDc

10 ro U
9 L-CPS ] 60-CPS [- I
8 GEN A GEN A
S 7 50
6 40
4
V) 4 F 30
3
320
1 10
LOW NORMAL 'TOXIC LOW NORMAL 'TOXI
PHENYTOIN BLOOD LEVEL
Figure 2. Numbers of VVT errors made by groups at varying categorical phenytoin blood levels. LOW = 10 micrograms/milliliter of serum; NORMAL = > 10 and < 20 micrograms/milliliter of serum; TOXIC = > 20 micrograms/milliliter of serum.

C




increments, R-CPS subjects Omission Errors increments, with increasing phenytoin blood level.
Visual Geometric Test
Mean percentages of within-session correct responses to the VGT at
the three phenytoin blood levels are presented in Table 11. When phenytoin concentrations were NORMAL, as shown in Figure 3, R-CPS group subjects did less well than L-CPS subjects as response delay lengthened. Figure 4, which represents the mean of both LOW and TOXIC levels, shows a widening of this difference at 18 seconds, and Figure 5, at TOXIC level, further enhancement of the discrepancy between the CPS groups' performance on this task. Subjects in the GEN group, who showed little VGT information loss at NORMAL level, showed losses comparable to those of R-CPS subjects when blood levels were LOW or TOXIC and delay to response was the maximum 18 seconds.
Visual Consonants Test
A summary of mean within-session scores on the VCT is presented in Table 12. At the NORMAL level, as seen in Figure 6, all groups showed similar performances with the R-CPS and L-CPS subjects conforming closely to each other. However, at phenytoin levels which were extratherapeutic, as shown in Figure 7, R-CPS and L-CPS scores diverged at longer delay intervals with L-CPS subjects scoring lowest of the groups. When performance at the TOXIC level alone is examined (Figure 8), the R-CPS and L-CPS scores maintained their relative positions, but GEN group scores assumed the lowest values of the three groups.




Table 11

Visual Geometric Test Mean Percentage of Correct Responses at Increasing Delay Intervals
and Varying Phenyloin Blood Levels

Group Number

Delay in Seconds Sessions

Riqht CPS

Left CPS

0 3 9 18 0 3 9 18

Generalized
7
0 3 9 18

98.3 78.3 88.3 79.9 88.1 73.7 81.0

80.9

97.9 71.1 85.3 76.9

ABL NORMAL

94.3 84.2 85.2 74.1 92.9 84.2 93.8 79.8

96.2 85.2 85.1 83.2

100.0 75.0 72.9 75.0 100.0 75.0 83.3 100.0

ABL LOW

ABL TOXIC

95.8 83.3 87.3 75.0




L-CPS 0 R-CPS O GEN A

100 1-

90

80

~/ L

0 3 9 18
DELAY INTERVAL IN SECONDS
Figure 3. Mean within-session VGT performance by groups at NORMAL level.

| & I | | |




63
100 Q R-CPS 0
L-CPS
(I) GENA
90
. 80
C-)
70
0 3 9 18
DELAY INTERVAL IN SECONDS Figure 4. Mean within-session VGT performance by groups at blood levels outside NORMAL range.




100

R-CPS O L-CPS GEN A

90 -

80 E

'I

Figure 5. Mean

DELAY INTERVAL IN SECONDS
within-session VGT performance by groups at TOXIC level.




Table 12

Visual Consonants Test Mean Percentage of Correct Responses at Increasing Delay Intervals
and Varying Phenytoin Blood Levels

Group Number Delay in Seconds Sessions

Right CPS
12
0 3 9 18

Left CPS
10
0 3 9 18

Generalized

7
0 3 9 18

98.3 76.6 69.2

ABL NORMAL

99.1 74.5 69.1

57.5

95.3 67.7 47.7 40.6

60.7 97.4 69.8 71.9

60.5

100.0 73.0 66.6
9
99.1 76.9 68.4

73.6 98.0 75.0 68.8

ABL LOW

61.4

70.3

ABL TOXIC

100.0 86.9 82.3

64.5

98.0 70.8 64.8

58.3




100 R-CPS 0
L-CPS O
90 Gen A
0 80
70
8
60
50
I I I X I p
0 3 9 18
DELAY INTERVAL IN SECONDS Figure 6. Mean within-session VCT performance by groups at NORMAL level.




67
100 R-CPS O
L-CPS 90 GEN A
R
80
70
60
50
/ I
0 3 9 18
DELAY INTERVAL IN SECONDS Figure 7. Mean within-session VCT performance by groups at blood levels outside NORMAL range.




68
100 R-CPS 0
L-CPS 90
z
80
70
8
60
50
S I I I I II
0 3 9 18
DELAY INTERVAL IN SECONDS Figure 8. Mean within-session VCT performance by groups at TOXIC level.




Auditory Consonants Test
Mean scores on the ACT at the three phenytoin blood levels are
presented in Table 13. At NORMAL levels (Figure 9), all groups looked similar. The R-CPS and L-CPS groups scores diverged at extratherapeutic blood levels, with L-CPS group subjects consistently earning lowest scores at all blood levels. Scores outside NORMAL and at TOXIC levels are shown in Figures 10 and 11, respectively.
Tonal Memory Test
Mean performances by groups on the TMT within session at increasing levels of task difficulty and varied phenytoin blood levels are presented in Table 14. The maximum difference in scores was observed at NORMAL level of phenytoin concentration (Figure 12) with L-CPS subjects outperforming both R-CPS and GEN subjects at the higher levels of difficulty. Scores earned at levels other than NORMAL, shown in Figure 13, and at TOXIC level, shown in Figure 14, were highly similar across groups.
Relationships Between Memory Scores
and Categorical Phenytoin Levels
A test for the presence of a linear versus a quadratic relationship between memory performance and phenytoin blood level was performed using the CONTRAST statement within the GLM procedure of SAS (Goodnight, 1979). Dependent variables were the percentages of correct responses to immediate and short-term memory portions of the VGT, VCT, and ACT. One linear relationship between VCT-STM and level was significant (F(I, 27) =
4.64, p < .04) in the direction of improved scores with ascending blood level of phenytoin. As phenytoin levels moved from LOW through NORMAL




Table 13
Auditory Consonants Test Mean Percentage of Correct Responses at Increasing Delay Intervals
and Varying Phenytoin Blood Levels

Right CPS

Left CPS

Generalized

Delay in Seconds

0 3 9 18

0 3 9 18 0 3

Sessions

100.0 72.5 66.7

61.7 100.0 71.4 64.3

43.0 99.0 75.0 69.8 56.4

ABL NORMAL

100.0 77.9 70.4

63.4

99.6 73.3 66.7 53.9 99.1 79.8 68.6 67.7

100.0 83.5 75.0 68.8 100.0 72.8 62.3

Group Number

ABL LOW

9 18

ABL TOXIC

56.3 100.0 66.5 68.8 62.3




100 R-CPS 0
L-CPS 90 GEN
80
70
8
60
60
50
0 3 9 18
DELAY INTERVAL IN SECONDS Figure 9. Mean within-session ACT performance by groups at NORMAL level




R-CPS L-CPS GEN

DELAY INTERVAL IN SECONDS

Figure 10. Mean within-session ACT outside NORMAL range.

performance by groups at blood levels

100




100 R-CPS O
L-CPS [ 90 GEN A
80
70
60 50
0 3 9 18
DELAY INTERVAL IN SECONDS

Figure 11. Mean within-session ACT performance by groups at TOXIC level.

-l
j]




Table 14

Mean Percentage of

Tonal Memory Test
Correct Responses at Increasing Levels of Difficulty and Varying Phenytoin Blood Levels

Group Number Number of Tones Sessions

Right CPS
12
3 4 5

Left CPS
10
3 4 5

Generalized
7
3 4 5

ABL LOW

70.9

ABL NORMAL

71.7

61.0

58.3

37.0

44.4

67.1

59.1

58.6 40.0

73.2 60.5

80.0 57.5 51.3

68.9 50.0 38.9

47.5 65.0 47.5 40.0

ABL TOXIC 72.5 48.8

65.0 57.5 35.0




80 R-CPS O
L-CPS O GEN A 70
z
60
8 50
S 40
30
3O I, f I
3 4 5
NUMBER OF TONES IN SEQUENCE Figure 12. Mean TMT performance by groups at increasing levels of difficulty at NORMAL level.




76
80 R-CPS O
L-CPS O GEN A 70
60
50
a9~ 40
30
3 4 5
NUMBER OF TONES IN SEQUENCE Figure 13. Mean TMT performance by groups at increasing levels of difficulty at blood levels outside NORMAL range.




R-CPS 0 L-CPS [ GEN A

3 4

NUMBER OF TONES IN SEQUENCE Figure 14. Mean TMT performance by groups at increasing levels of difficulty at TOXIC level.

l

J
/




to TOXIC, VCT-STM increased in percentage correct from 62.3 to 69.1 then 71.8% in correspondence. No other relationships were significant.

Effects of Session Sequence

A test of session sequence effects using paired-difference t tests betweer first and third sessions, with overall correct scores as dependent measures. first and third sessions of the ACT was p. < .01). Mean ACT percentages correct 73.9, 74.7, and 78.6% in first, second,

;on memory measures was done first and second, then between VGT, VCT, and ACT percentage One difference, that between significant (t(28) = 3.47, scores in the full sample were and third sessions, respectively.




CHAPTER FOUR
DISCUSSION
The Experimental Hypotheses
Short-Term Memory in Complex Partial Versus Generalized Seizures
The study data do not support the hypothesis of greater short-term memory deficit in complex partial than in generalized major motor seizures. No significant group differences were observed on the visual consonant trigrams, auditory consonant trigrams, or the visual geometric figures short-term memory tasks. This finding is in agreement with Mayeux et al. (1980), who also employed auditory and visual consonant trigrams with complex partial and generalized seizure patient groups.
Neither do these data indicate differential performances between predominantly left-lateralized and predominantly right-lateralized complex partial seizure patients on the basis of material-to-be-remembered. Complex partial seizure patients with a left-sided focus earned lower scores on both visually and acoustically presented short-term memory measures having verbal content than did patients with a right-sided focus. Conversely, patients with a right-sided focus made lower scores on a visually presented short-term memory measure having geometric content than did patients with a left-sided focus. However, these differences were not statistically significant.
The present findings are in agreement with Glowinski's (1973) observations of patients with left- and right-lateralized complex partial




seizure foci. She found that although patients' scores on verbal and figural reproduction portions of the Wechsler Memory Scale were "in the predicted direction . the left temporal lobe epileptic group showed a greater verbal memory deficit and the right temporal lobe group showed a greater nonverbal memory deficit" (Glowinski, 1973, p. 133), they were not significantly different from each other. Short-Term Memory Functioning at Three Blood Levels of Phenytoin Effects on Absolute Scores
The second hypothesis of better short-term memory functioning at serum levels of phenytoin within the therapeutic range as contrasted with levels below and above that range was not confirmed. Delaney et al. (1980) found anticonvulsant levels uncorrelated with their long-term memory tasks in subjects with unilateral complex partial or focal frontal seizures. Dodrill (1975) and Matthews & Harley (1975) reported differences between toxic and nontoxic seizure patients in gross and fine motor performances, favoring the nontoxic patients, but not on the measures of "higher mental functions" (e.g., WAIS VIQ, the Category Test) (Dodrill, 1975, p. 596). Dekaban and Lehman (1975), who twice changed anticonvulsant doses of phenytoin, phenobarbital, and mysoline by 30-50% of initial dose, described a nonsignificant trend to better verbal recall at the "lowest dose of medication" (Dekaban & Lehman, 1975, p. 319) in a mixed epileptic sample of patients.
In contrast with Dekaban & Lehman (1975), and that of noncorrelation of blood level and long-term memory tasks by Delaney et al. (1980), patient performance in the present study differed. Although one must




cautiously interpret a significant ANOVA in the context of a nonsignificant MANOVA, subject scores were seen to improve significantly with increasing phenytoin blood level on the visually presented verbal measure. Seventy-two percent of this improvement (62.3% to 69.2% correct responses) occurred as the blood level moved from the subtherapeutic to the therapeutic range. The remaining 28% of the gain (69.1% to 71.8% correct responses) took place as the level moved from the therapeutic to the toxic range of phenytoin concentration. Mean phenytoin level in the toxic range for these patients was 24.8 micrograms/milliliter of serum. Inasmuch as patients began sessions at any serum phenytoin level, this improvement in score is not readily attributable to a learning effect.
The present results are consistent with the report of Dodrill &
Troupin (1977) that patients with therapeutic blood levels of phenytoin or carbamazepine showed improvements in test scores over scores made at less than therapeutic levels. Noninterference of a toxic phenytoin level in short-term verbal memory functioning is again in agreement with findings by Dodrill (1975) and Matthews & Harley (1975) in toxic patients.
Effects on Loss of Information from Short-Term Memory
Multivariate analyses of the combined short-term memory measures revealed no effects referable to seizure group membership or to blood level on the amount of information lost from short-term memory as delayto-report increased. However, two interesting task-related findings emerged. They are presented with the caveat of their being seen in significant ANOVAs within nonsignificant MANOVAs.
First, right-lateralized complex partial seizure patients showed greater losses of figural material than left complex partial seizure




patients or generalized seizure patients when all three groups had toxic phenytoin levels. A performance of this kind is in keeping with the hypothesis of memory deficit in complex partial seizure patients being material-related according to side of involvement. A patient with a right-sided complex partial seizure focus would thus be relatively more impaired in memory for nonverbal content. The way in which such a vulnerability might be expected to interact with a toxic serum concentration of anticonvulsant drug is a matter for further study.
Second, therapeutic blood level, as opposed to subtherapeutic blood level of phenytoin, significantly reduced the mean loss of visually presented consonant trigrams during the 18-second span of short-term memory measured. This finding applied to all three seizure groups, and is in agreement with Dodrill & Troupin's (1977) observations of test score improvements in phenytoin- or carbamazepine-treated patients versus untreated epileptics. Thus, there appears to be a complex relationship between phenytoin blood level, material and measure employed, and cognitive performances of epileptic patients. The Short-Term Memory Measures
The visual and auditory verbal measures differ in their level of
difficulty because response demand is not equivalent. The visual measure requires recognition and selection of the trigram letters from an array of consonants. The auditory measure task is one of productive recall. Interestingly, performance on the more difficult measure was seen to improve from first to third testing. Learning effects were not seen in university freshmen tested on three occasions. An alternate explanation




for this increase in scores is that of increasing effort by subjects to do the task, coupled with greater confidence in their ability to perform it. Subjects in the study commonly protested doing the task in the first session, claiming memory deficiencies, then returned for subsequent sessions saying they believed themselves more able to rely on their memories since beginning to take part in the study.
The material/modality relationships with seizure groups and measures employed were relatively complex. Short-term memory components of visual and auditory consonants measures correlated with (FSIQ and) VIQ in the full sample, suggesting that these instruments were eliciting a materialspecific, but not modality-specific, performance. The immediate memory aspect of the visual geometric measure also showed a strong association with (FSIQ and) VIQ. Looking at correlations by seizure groups, however, these relationships were differentially apportioned. Right complex partial and generalized seizure patients', but not left complex partial seizure patients', performances were characterized by use of verbal abilities in this figural task. Left complex partial seizure patients, on the other hand, showed no VIQ correlation, but a PIQ correlation, with the shortterm memory performance on the geometric figural measure. These associations suggested that left complex partial seizure patients, somewhat deficient in global verbal abilities, dealt with figural material using other resources than were used by right complex partial and generalized seizure patients. These latter two groups appeared to be compensating for their deficiency by tapping a verbal ability store, coping with figural material via a mechanism such as labeling figural stimuli presented.




Tonal Memory Test
There were no significant group or blood level differences on performance of the Tonal Memory subtest of the Seashore Measures of Musical Talents. Furthermore, scores on this task did not correlate with performances on any of the other dependent measures. Although the test was included in the present study as a measure of short-term auditory nonverbal memory, it clearly failed to contribute to any of the findings with regard to memory, seizure group, or blood level. Unlike the other shortterm memory tasks empl oyed i n thi s study, no di stracti on was i ntroduced between presentation of the tones and response. The failure to show any effect with this task may be attributable to this procedural difference.
Because accurate performance on this task depends more upon recognition of correct sequence than it does upon reproduction of the sequences or tonal qualities of the stimuli, it may be a less sensitive measure than the other short-term memory tasks used. The present findings are consistent with the preoperative performance of Milner's (1962) small sample of right and left temporal lobectomy patients, but not the postoperative differences observed.
Attention and Vigilance
Although the major hypotheses of the present study focused on
material-specific and modality-specific differences in short-term memory and phenytoin blood levels in three seizure groups, a measure of visual vigilance was introduced to examine the possible effects of attentional dysfunction on memory at each blood level. Performance on the Visual Vigilance Test was also examined with regard to its relationship with




other study measures, as well as with the subject variables of age of onset, duration, frequency, FSIQ, VIQ, and PIQ. Effects of Seizure Group
There were no differences between seizure groups on errors of
omission, which are interpreted to reflect errors of inattention. The group differences in numbers of errors of commission contributed to the overall group differences in total errors. Errors of commission occur when subjects respond to a nontarget stimulus.
Subjects with generalized major motor seizures made more responses to nontarget stimuli than did complex partial seizure patients. The present finding of more errors of commission than omission is not consistent with the earlier work of Fedio & Mirsky (1969) and Mirsky et al. (1960) who found that generalized seizure groups showed more inattention than complex partial seizure groups. However, reports in terms of absolute scores do not take commission errors into consideration. Their task was a more difficult one, requiring a two-step decision (respond to X only if it follows A) and a quicker response (0.1 to 0.2 sec presentation and 0.9 to 1.0 sec interstimulus interval versus this study's 1.0 sec presentation and 1.5 sec interstimulus interval). Differences in duration of task (10 min versus this study's approximately 40 min) do not serve to balance the two. This level of task difficulty may account for the discrepancy.
Effects of Phenytoin Blood Level
Performances by subjects at the three blood levels of phenytoin are significantly different in numbers of commission errors made at the toxic




level. All groups made more commission errors when phenytoin level was in the toxic range. This effect is consistent with that of Dekaban & Lehman (1975) reporting that chronic epileptics' performance on a vigilance measure is best at lowest doses of various combinations of phenytoin, phenobarbital, and primidone. The authors employed the more difficult vigilance task used by Mirsky et al. (1960), summing both error types to arrive at a "relative percentage correct" as described by the developers of the measure (Rosvold et al., 1956).
The level and similarity of groups' scores on this measure do not suggest impairment of attention in any group. As a consequence, differences in memory performances by groups are not a function of inattention. Rosvold et al. (1956) observed that commission errors do not necessarily represent attentional deficit. They proposed two scores for their measure, absolute and relative percentage correct, incorporating commission errors in the second of these, in the interest of providing ''a more complete description of (his) performance." Commission errors elicited in the context of attending, as instructed, to a recurring critical stimulus may represent a limited attentional capacity. On the other hand, commission errors may reflect disinhibition of central origin or, in this instance, one enhanced by intoxicating serum concentration of phenytoin.
Subject Variables and Group Relationships to Performance
on the Dependent Measures
Studies of epileptics impose a necessity for control, either experimentally or statistically, of numerous subject and seizure-related variables found relevant to these disorders. The literature in the area




has indicated the direction of expected influence of several of these variables. For example, longer duration and greater frequency of seizures are each believed associated with, if not causative of, performance deterioration. Efforts to delineate each type of seizure uniquely have perhaps obscured a middle ground of commonality, lying between specificto-type and general-to-epilepsy characteristics. The following observations made of this subject sample show both expectable and unanticipated relationships of variables to performance, commonalities across diagnostic groups, and a reciprocal mechanism which could partially account for types of errors seen in the context of performance of the vigilance task.
Unique to this study, the relationship of early age of onset to known etiology, as well as that of lower PIQ with an abnormal CT scan, seen in the right complex partial seizure group, is explicable. Patients in this group who developed seizures early in life often did so following head trauma with right-lateralized effects, impaired VIQ, and observable defect on the CT scan. Neither of the other two groups showed this pattern of correlations.
All subjects' performance on the VCT short-term memory portion improved with longer duration of seizure history. Right-lateralized complex partial seizure patients' scores improved on the ACT short-term memory portion as well. In this sample, the deleterious effects of more years of seizures may have been mitigated by access to continuing care with consequent better control of their disorder.
Both complex partial seizure groups showed similar associations of VIQ, and less comparable but positive associations, with short-term




memory performance on the ACT. Left-lateralized patients had correlations in common with the generalized patients' group only when these correlations were seen in all three groups. Right complex partial seizure patients, on the other hand, had commonalities with generalized seizure patients in several respects. The groups shared low correlation of PIQ with FSIQ; noncorrelation of PIQ with short-term performance on the VGT, but correlation of VIQ with immediate performance on that measure; and a correlation between seizure frequency and omission errors on the vigilance task. These resemblances between the two groups, provided that they are not idiosyncratic to this sample, could partially account for the difficulty often encountered in differentiating their performances on measures tapping PIQ abilities.
The effects of frequency and toxic blood levels of phenytoin on the types of errors made by subjects in this sample on the vigilance task appeared to operate reciprocally. With higher seizure frequency, omission errors were greater for right complex partial seizure and generalized seizure patients, but commission errors were fewer. At toxic phenytoin levels, commission errors increased in all three groups. Although this effect is not perfectly balanced, showing identical changes in all subjects, its observation stimulates the following conjecture. Because dyscontrol of seizures with increased frequency is commonly associated with insufficient serum anticonvulsant levels, these two states, high seizure frequency and toxic blood level, may be located distant from each other on a continuum. At one end is nil to low serum phenytoin, behaviorally accompanied by frequent seizures and omission errors on a vigilance task, at the other end, toxic phenytoin levels and commission errors.




Summary and Conclusions
Neither experimental hypothesis of differences in seizure groups' performances in short-term memory or differences in these same performances at therapeutic versus nontherapeutic levels of serum phenytoin received support at a significant level. One dependent measure, errors of commission on a test of vigilance, did distinguish generalized seizure patients from those having complex partial seizures. The same measure was seen to increase significantly when blood levels of phenytoin were in the toxic range.
Results obtained indicate that the groups of complex partial seizure and generalized major motor seizure patients studies had a similar level of short-term memory dysfunction. Although similar in level, the impairments appeared to be nonsignificantly different in each group, an impression received from observations of performances on short-term memory measures. Performances of complex partial seizure patients were nonsignificantly different according to hypothesized material effects on their memories, with left-lateralized patients performing less well than others on two verbal measures and right-lateralized patients performing less well than others on one geometric measure. Generalized seizure patients' performances tended to resemble, on each measure, that of the complex partial seizure group performing better. No observations made in the study were better understood by invoking a modality-of-presentation hypothesis.
The small size of the sample clearly restricts both conclusions
drawn from these data and their generalizability to other, similar groups.




That small size, the use of a generalized seizures contrast group, and constraints imposed by clinical treatment responsibilities are likely contributors to the failure to support or disconfirm the existence of short-term memory deficits in complex partial seizure patients. Larger samples and contrast with other focal seizure patients are recommended.
Although a unitary effect of phenytoin blood level was not seen,
inclusion of anticonvulsant level among controlled variables is advised as a means of clarifying its role in cognitive functioning of epileptics. Its differential effects on a visual verbal and a visual geometric measure suggest that, in larger samples, it may be found to influence specific types of performances with differing ranges of therapeutic effectiveness than are utilized as guidelines for medical management of seizures.




APPENDIX A
SAMPLE VISUAL GEOMETRIC TEST STIMULI




APPENDIX B
VISUAL GEOMETRIC TEST DISTRACTION TASK
________________________DATE___________

1 2 5

Copy these figures as quickly and as well as you can. Begin another row when one is finished. Continue until you are asked to STOP.

NAME

SESSION

4 z

ru




APPENDIX C
VISUAL CONSONANTS TEST RESPONSE FORM
DATE

NAME

SESSION:

H M J N
K P H M
H M J N
K P

TEST:

N S
P T L Q M R
P D L G M H N S
P T L Q M R N J
P K L G

16. Z T P K F V Q L G B W R M H C X S N J D




APPENDIX D
PATIENT INFORMATION FORM




Full Text

PAGE 1

SHORT-TERM MEMORY IN EPILEPTIC PATIENTS BEING TREATED WITH PHENYTOIM MARIANNE L. D. MCGUIGAN 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 IIVERSITY OF FLORIDA 1984

PAGE 2

ACKNOWLEDGMENTS To the following people, who have contributed to my education and my life, I am deeply grateful: Suzanne Johnson, your enthusiasm for scholarly research, magnanimity, and faith in a nonstandard prospective student fanned a spark and helped to make graduate study possible; Roger Blashfield, encountering your sensitivity and devotion to intellectual rigor, graced with a sense of humor and irony, made me know I was in the best of company; Cynthia Belar, the discovery that my nobly professional and intellectual mentor could as well be tender friend brought warmth and renewed hope; Joe Wilder, getting to know you better, gentleman physician who permitted me to work with the patients you serve so well and value so highly, was a pleasure and a compliment; Eileen Fennel 1, the generosity of spirit, the model of professional excellence, the wit, intelligence, and warm friendship you have shared with me have done me honor; Finally, my thanks go to L. James Willmore on whose hypothesis of anticonvulsant effects on memory a portion of this study was based. n

PAGE 3

TABLE OF CONTENTS PAGE ACKNOWLEDGMENTS ii ABSTRACT v CHAPTER ONE INTRODUCTION 1 Anatomy and Connections of the Temporal Lobe .... 2 Investigation of Memory in Epilepsy 4 The Issue of Antiepileptic Medications 12 Rationale For This Study 19 TWO METHOD 22 Subjects 22 Pilot Studies 24 Materials and Apparatus 25 Procedure 28 Instructions to the Subjects 29 Statistical Analyses 31 THREE RESULTS 33 Overview of Results 33 Description of Groups and Relationships of Variables. 34 Group and Categorical Blood Level Effects on Test Scores 50 Group and Categorical Blood Level Effects on Within-Session Performance 53 Overview of Mean Single and Combined Test Performances by Subjects 54 Relationships Between Memory Scores and Categorical Phenytoin Levels 69 Effects of Session Sequence 78 FOUR DISCUSSION 79 The Experimental Hypotheses 79 Attention and Vigilance 84 Summary and Conclusions 89 ; n

PAGE 4

APPENDICES A SAMPLE VISUAL GEOMETRIC TEST STIMULI 91 B VISUAL GEOMETRIC TEST DISTRACTION TASK 92 C VISUAL CONSONANTS TEST RESPONSE FORM 93 D PATIENT INFORMATION FORM 95 E PATIENT INFORMED CONSENT TO BE A RESEARCH SUBJECT ... 98 REFERENCES 100 BIOGRAPHICAL SKETCH 107 TV

PAGE 5

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 SHORT-TERM MEMORY IN EPILEPTIC PATIENTS BEING TREATED WITH PHENYTOIN By Marianne L.D. McGuigan August 1984 Chairman: Eileen B. Fennell Major Department: Clinical Psychology Studies have confirmed the presence of long-term, but not shortterm, memory dysfunction in complex partial seizures. A controversy exists regarding the nature of these deficits. One view is that stimulus material, verbal or nonverbal, interacts with hemisphere of seizure focus, leading to verbal deficits in left-lateralized and nonverbal deficits in right-lateral ized complex partial seizures. The other view holds modality is relevant, resulting in impaired memory for any material presented acoustically. Other studies have reported that antiepileptic drugs do/do not have deleterious effects on cognitive-intellectual performances in epileptics. This study tested the hypotheses that (a) complex partial seizures are accompanied by greater short-term memory impairment than are

PAGE 6

generalized seizures, and (b) phenytoin levels outside the therapeutic range are detrimental to memory performance. A minor hypothesis is that deficits are not the result of inattention. Twenty-nine male epileptics having unilateral complex partial or generalized seizures were tested three times with measures of short-term memory and attention at three blood levels of phenytoin. Results of analyses did not support the hypothesis of greater memory deficit in complex partial seizures. All patients showed memory impairments. Left focus subjects scored lowest on verbal measures; right focus patients scored lowest on a figural measure. These findings were in the predicted direction, but not significant. Generalized seizure patients differed from complex partial seizure patients in number of commission errors made on the test of attention. The hypothesis of greater impairment of memory at nontherapeutic blood levels of phenytoin was not statistically confirmed. Two findings in accord with the hypothesis must be cautiously interpreted because of a nonsignificant MANOVA. These were (a) improvement in a verbal score as blood level became therapeutic, and (b) greater loss of information from short-term memory on a figural measure when blood level was toxic. When blood level was toxic, significantly more commission errors were made by all subjects. The conclusions were that all patients had memory deficits, but were not inattentive. The modality hypothesis did not enhance understanding of any observations. Nonsignificant findings in the predicted directions leave issues open to further study. vi

PAGE 7

CHAPTER ONE INTRODUCTION Cognitive-intellectual deficits accompanying epilepsy, described in the nineteenth century as leading in some patients to global mental deterioration, remain a topic of controversy regarding their specific nature and causation. One of these deficits, dysfunctional memory, is often associated with the presence of complex partial seizures (CPS, formerly called psychomotor), a type of paroxysmal cerebral dysrhythmia, arising in one or both temporal lobes. Greater attention has been given this seizure type since it came to be better differentiated clinically and electroencephalographically (Gibbs, Gibbs, & Fuster, 1948; Gibbs, Gibbs, & Lennox, 1938). Although the disruptive effects of other focal and generalized seizures have not been overlooked, the proximity of a periodically discharging temporal lobe focus to the limbic system and the hippocampal structures forming the anatomical substrate for memory processing has implications for that function (Hecaen & Albert, 1978; Walsh, 1978). Moreover, patients with seizures, their families, and their caretakers provide anecdotal support of dysfunction in epileptics. This dysfunction is characterized as failure to retain recently acquired information. A number of factors have been shown to influence general and specific deficits among epileptics. These have included, in addition to heredity, coexisting brain damage and seizure-related variables 1

PAGE 8

such as seizure type, age at onset of seizures, frequency of seizures, duration of seizure history, laterality of a discrete focus, electroencephalographic patterns, and the effects of antiepileptic medications (Dikmen & Matthews, 1977; Dikmen, Matthews, & Harley, 1977; Dodrill, 1975, 1981; Guey, Charles, Coquery, Roger, & Soulayrol, 1967; Klove & Matthews, 1966; Lennox, 1960; MacLeod, Dekaban, & Hunt, 1978; Matthews & Klove, 1967). Studies of heterogeneous groups of epileptics have often yielded uninterpretable or nonsignificant results which did not discriminate experimental subjects from nonepileptic controls on measures of memory (Loiseau, Strube, Broustet, Battel lochi, Gomeni, & Morsel! i, 1980; Scott, Moffett, Mathews, & Ettlinger, 1967; Tomlinson, Stirling, Merrifield, & Reynolds, 1981). Separation of subjects into diagnostic groups to assess memory in CPS as compared to other seizure groups has become more common since the advent of surgical treatment of refractory CPS disorder, as well as the publication of data from the study of patients before and after temporal lobe resection. Anatomy and Connections of the Temporal Lobe The temporal lobe is the mass of cerebral tissue below the Sylvian fissure, extending posteriorly to approximately Brodmann's areas 40, 39, and 19 (see Figure 1). It is composed on its lateral surface of neocortex which forms its major gyri On its medial unexposed surface it consists of phyloqenetical ly older cortex, archicortex and paleocortex, These genetically older tissues form not only its medial gyri, but the subcortical hippocampus and amygdala as well. Together, these three

PAGE 9

1T6Figure 1. Left cerebral hemisphere, lateral view. Positions of subcortical amygdala (A) and hippocampus (H) are illustrated below Sylvian fissure (SF), with superior (STG) and inferior (ITG) lateral temporal gyri Bottom figures. AC = anterior coronal section at approximately dotted lin AC; PC = posterior coronal section at approximately dotted line PC. Structure locations labeled. A = amygdala; C = caudate nucleus; DM = dorsomedial nucleus of the thalamus; H = hippocampus; ITG = inferior temporal gyrus; LT lateral thalamus; LV = lateral ventricle; P = putamen; STG = superior temporal gyrus.

PAGE 10

distinguishable cortical tissues provide the temporal lobe with auditory, association, and limbic cortex. The connections of the temporal lobe are similarly diversified. It receives afferent projections from the sensory system and projects efferently to the parietal and frontal association regions, limbic system, and basal ganglia. The left and right lobes communicate via the corpus callosum (neocortex) and the anterior commissure (archi cortex) (Kolb & Whishaw, 1980). Investigation of Memory in Epilepsy The terminology used to describe memory processes varies across disciplines and studies. Reviewing the studies below, which typically refer to a storage model of memory, I shall define several relevant terms (Russell 1981). The earliest stage of the memory process, that of input, is called sensory memory, which begins to decay in approximately 250 msec. In terms of its storage capacity, more information can be held than the subject is able to report. The visual form of sensory memory is termed "iconic" memory, the auditory form "echoic" memory. Immediate memory is a clinical term defined as that quantity of material reproducible directly after presentation. Short-term memory (STM) and long-term memory (LTM) refer to hypothetical processes mediating retention, each with its own storage capacity. In STM a limited number of items, generally seven plus or minus two, are maintained for about 20 to 30 seconds without rehearsal. Material in STM is subject to displacement by new input of information. Overlapping with STM from about 0.5 seconds after input, LTM is that process believed to consolidate, store, and

PAGE 11

make information available for future retrieval. The storage capacity of LTM is presumed to be indefinitely large and relatively permanent, although there is some loss over time (Waugh & Norman, 1965). A final process, retrieval, refers to the locating of stored material in LTM by means of an STM probe, whereby the material is returned to STM for active use. Finally, recent and remote memory refer to the personal time scale of the patient or examinee. Temporal Lobectomy Studies Temporal lobe resection produces differential performance deficits related to side of operation and the extent of hippocampus removed (Kolb & Whishaw, 1980). The following studies are representative of those reporting effects of these procedures. Milner (1962, 1967) reported that a left temporal lobe seizure focus interfered with verbal memory performance as tested with the Logical Memory subtest of the Wechsler Memory Scale (WMS). This deficit appeared to be material-specific rather than being related to the mode of presentation, appearing in the visual as well as the auditory presentational mode. Temporal lobectomy markedly worsened this verbal memory impairment, also seen for paired-associate words and recognition of numbers or letters. Considerable recovery took place during the extended postoperative period. Auditory, but not visual, verbal deficits were found by Meyer (1959; and Meyer & Yates (1955) in left temporal lobectomy patients. The authors described their left temporal lobectomy subjects as doing less well than right temporal lobectomy subjects, if not significantly so,

PAGE 12

on all measures employed. At follow-up testing 11-15 months later, this auditory verbal deficit was often still present. Cherlow & Serafetinides (1976), comparing right-handed groups of left and right temporal lobectomy patients, found left-lobectomized subjects significantly impaired in learning verbal material. These patients needed more repetitions to learn somewhat less material than did right lobectomy patients. Percentages of learned material recalled, however, did not differ between groups. Deficits found in patients with right temporal lobe foci both preand postoperatively have been documented by Milner (1962, 1965, 1968), Kimura (1963), and Corkin (1965), among others. Memory for visual material not easily verbally encoded was impaired in right temporal lobectomy patients on measures consisting of nonsense figures, random dot patterns, and photographs of human faces. Although preoperative comparisons with left lobectomy patients in performance on the Tonal Memory subtest of the Seashore Measures of Musical Talents were small, postoperative testing revealed highly significant differences representing losses by right-lobectomized patients (Kimura, 1963; Milner, 1962, 1968). Right lobectomy subjects did poorly in tactual maze-learning, relative to left lobectomy subjects, making significantly more errors and requiring more trials to criterion (Corkin, 1965). Milner (1965) had reported parallel findings in subjects with bilateral hippocampal damage on a visual maze-learning task. Samuels, Butters, & Fedio (1972) attempted to differentiate perceptual from memory components of these deficits, as well as to evaluate

PAGE 13

the material/modality specificity issue. They presented verbal and nonverbal stimuli in the visual, and verbal stimuli in the auditory, mode to patients having unilateral temporal lobe excisions. Their method was that of Peterson & Peterson (1959) in which either immediate or delayed (3, 9, or 18 seconds) recall is tested after tachistoscopic presentation. When recall was delayed, backwards counting was introduced to prevent rehearsal or verbal labeling. They considered impairment on the immediate recall evidence for perceptual deficit, on the delayed recall alone as supportive of STM disturbance. Both right and left temporal lobectomy patient groups performed as well as normal controls on the visual tasks. This was consistent with earlier studies that had provided evidence for major participation of the parietal lobe in the mediation of visual retention (Butters & Barton, 1970; Butters, Barton, & Brody, 1970; Butters & Brody, 1968; Butters, Samuels, Goodglass, & Brody, 1970; Samuels, Butters, & Goodglass, 1971). The authors suggested that the differences in their results from those of Milner, who used delays of 45 seconds or more, might be explained by the existence of a double dissociation in these processes relative to material/modality specificity and anatomical separation. Thus, in addition to the mediation of visual input by the parietal and auditory by the temporal areas, STM storage might be modality-specific and LTM storage material-specific. In general, lobectomy studies found verbal deficits in left temporal lobectomy patients and nonverbal deficits in right temporal lobectomy patients with respect to their memory functioning. Inconsistencies with

PAGE 14

these findings have yet to be resolved. There is, however, agreement that bilateral damage to both hippocampi leads to substantially greater impairment than does unilateral lesion or removal. Artificial Stimulation Studies Ojemann and his associates, performing brain stimulation in the context of surgically treating movement disorders, reported having observed effects on speech. They elicited anomia with stimulation of the left pulvinar of the thalamus and of the deep parietal paracallosal white matter of both hemispheres in right-handers. The right-sided findings were interpreted as being due to extension of the stimulating current to the speech area by way of afferent callosal fibers from both hemispheres. Later, using a Peterson & Peterson (1959) technique variant, they found differences dependent on stimulation time. Results showed significant interference with memory with stimulation on recall. There were no effects with stimulation during distraction or presentation plus recall. Slight nonsignificant improvement occurred with stimulation during presentation. No manipulation interefered with performance on recognition tasks (Ojemann & Fedio, 1968; Ojemann, Fedio, & Van Buren, 1968). Ojemann (1978) also reported phenomena produced by temporal cortical stimulation in the course of mapping the area anatonomically during resection for intractable CPS. He concluded that the areas of the temporal lobe where stimulation affects language, interfering with naming of common objects, are adjacent to, but not identical with, the areas where it affects short-term verbal memory (STVM). He proposed that storage of

PAGE 15

short-term verbal memory is in the posterior language area cortex, and that anterior language cortex plays a part in retrieval from short-term memory. Data from studies of electroconvulsive shock lend further support to temporal laterality effects on memory (Hal li day, Davison, Browne, & Kreeger, 1968). Unilateral application of shock in treatment of depressed patients showed interference with nonverbal material retention with right-sided electrode placement. Left-sided electrode placement produced deficits in verbal learning and retention. Overview of the Psychological Characteristics of Epileptics Numerous studies of epileptics which have compared this patient population with normals or with other organically impaired groups have reported differences in epileptics' cognitive-intellectual performances (Dodrill, 1981). In addition to this considerable literature, another body of work evaluating epileptics' proneness to various forms of psychopathology has developed. Patients with CPS are seen as being at greatest risk, compared to other epileptics, for a variety of sequelae ranging from undesirable personality change to psychosis (Hermann & Whitman, 1984). Studies of Medically Treated Epileptics Three types of studies of medically treated epileptics will be reviewed. The first of these has examined general cognitive and/or memory functioning in heterogeneous CPS samples, often including right, left, and bilateral CPS patients. A second set of studies, employing CPS patients grouped by side of focus, has provided some data to support

PAGE 16

10 differential deficits between left-lateralized and right-lateralized CPS. Neither of the two types of studies described has included medication as a relevant factor. The third type of study, in which antiepileptic drugs have been considered or controlled, will be reviewed following a brief general discussion of the use of these agents in obtaining seizure control Heterogeneous group studies Those investigations which have utilized mixed groups of CPS subjects, often contrasted with a group of other focal or generalized seizure patients, have not found right CPS versus left CPS differences. A disparity in attentional capacity favoring the CPS group has been reported, as well as a variety of verbal and memory impairments seen on the Wechsler Adult Intelligence Scale (WAIS) and WMS favoring the subjects with generalized seizure types (Glowinski, 1973; Matthews & Klove, 1967; Mirsky, Primac, Marsan, Rosvold, & Stevens, 1960; Quadfasel & Pruyser, 1955). Matthews and Klove (1967) used multiple groups including (a) brain-damaged without seizures, (b) major motor, and (c) mixed seizures, each group being again divided into known and unknown etiology, to control for demonstrable organic pathology. They ranked their CPS groups with unknown and known etiology of seizure disorder second and third, respectively, after their nonneurological control group. Rankings were based on integrity of neuropsychological functioning, measured by the Hal stead Impairment Index. The authors' earlier conclusions (Klove & Matthews, 1966) that epileptic seizures were associated with

PAGE 17

11 impaired neuropsychological status regardless of etiology were amended following this investigation. Subjects having major motor or mixed (CPS and major motor) seizures continued to perform less well than controls. Their most impaired groups were (a) major motor of known etiology, and (b) brain-damaged without seizures, which did not differ from each other. Homogeneous left and right group studies A comparison of performance in 6 to 14 year old children with either CPS or generalized absence (formerly called petit mal ) seizures by Fedio & Mirsky (1969) revealed laterality differences. Their left CPS and right CPS subjects performed differently on the verbal and nonverbal components of a derived "supraspan" score. Verbal deficits were prominent in left CPS subjects; nonverbal deficits characterized performances of right CPS subjects. Children with absence seizures differed from both CPS groups in earning lowest scores on a measure of attention, the Continuous Performance Test. Other investigators studying adult groups of divided left and right CPS epileptics have also reported laterality effects. Reynolds (1974) was able to classify an overall 79% of his right and left CPS subjects. A stepwise discriminant analysis of responses on verbal and nonverbal memory instruments provided the prediction equation he used. In comparing CPS groups to lateralized groups with frontal foci, Ladavas, Umieta, and Provincial (1979) referred to their findings as supporting long-term but not short-term deficits in CPS subjects, according to side of involvement. Among frontal subjects, those with

PAGE 18

12 a left-sided focus were impaired in judging the recency of presentation of stimuli Mayeux, Brandt, Rosen, and Benson (1980) used a variety of verbal and nonverbal measures to contrast groups of CPS and generalized seizure patients. They found that left-focus subjects differed from both other groups on a confrontation naming task, whereas right-focus subjects did not differ in any interpretable way. The authors proposed that leftlateralized CPS patients have a specific anomia as distinct from a memory deficit, and suggested that this impairment, with its accompanying word-finding difficulties, might explain the circumstantiality of speech so often attributed to left CPS patients. Delaney, Rosen, Mattson, and Novel ly (1980) compared right and left CPS patients with frontal focal epileptics and with normal controls. Right CPS subjects performed worst on one visual nonverbal measure; left CPS subjects earned the lowest derived verbal score. On other measures of verbal and nonverbal memory both CPS groups were alike, but differed from contrast groups, with left CPS subjects showing verbal and right CPS subjects nonverbal deficits. The authors reported low correlations between anticonvulsant serum levels and performances on psychometric measures, but a consistent correlation between longer duration of seizure history and declining memory scores. The Issue of Antiepileptic Medications Consideration of anticonvulsant blood level by Delaney et al (1980) reflected a concern many have had for the participation of this variable in epileptics' psychological

PAGE 19

13 functioning. Studies of epilepsy which do not control these agents must at least acknowledge the omnipresence of anticonvulsants in patient treatment. Until rather recently, only enumerating of drugs, quantifying of dose, and seizure-frequency monitoring have been available as methods. Evaluation of the role antiepileptic medications have in psychological performances has become possible in the last 10 to 12 years as laboratory methods of quantitative analysis have been developed and made widely accessible (Wilder & Brum', 1981). An invaluable tool for the neurological clinician, ready measurement of anticonvulsant blood level (ABL) has been a methodological aid to the researcher as well. Drug Treatment in Epilepsy Antiepileptic drugs may be used alone or in combination, the choice of agent being in part dictated by the clinical seizure type (Millichap, 1972; Wilder & Brum', 1981). Coatsworth and Penry (1972), writing on the general effectiveness of antiepileptic drugs, estimated that 70 to 80 percent of chronic epileptics obtain some seizure reduction with their use. Astute management is essential after careful selection of an appropriate drug. Pharmacokinetic variation, i.e., differences in serum concentration produced by a given dose, is common and may be seen when the same quantity of a given drug is administered to different patients (Richens, 1976). As with most effective drugs, there is a potential for undesired side effects which may be manifested in several body systems. These unlooked-for outcomes extend as far as toxicity, which is sometimes expressed psychologically in alterations in mental functioning ranging from mild dulling, through memory and learning

PAGE 20

14 problems, to relatively rare toxic psychosis (Glaser, 1972; Wilder & Bruni, 1981; Wilder & Ramsay, 1974). Pharmacological effects are believed difficult to separate from those of seizure activity itself, particularly when several medications are simultaneously in use (Glaser, 1972, 1974, 1980). The pharmacological properties of the antiepileptic drugs continue to be clinically studied, descriptions of their efficacy and characteristics forming an extensive literature. One area of interest has been that of determining the range of serum concentration for each agent within which seizure control is enhanced without production of symptoms of intoxication. Less is known of the psychological effects of these medications, but it is evident that both extremes of the dosage continuum require monitoring to secure an optimal clinical treatment effect. S tudies of Cognitive-Intellectual Effects of Antiepileptic Drugs Guey et al. (1967) studied the effects of ethosuximide on 25 referred children with severe petit mal seizures, 15 of whom were mentally retarded. They found intellectual deterioration manifested by memory disorders, speech problems, and emotional disturbances in their subjects. Anticonvulsants already prescribed were continued, the test drug being added to the regimen. The authors, testing with the Wechsler Intelligence Scale for Children (WISC), and the Benton Visual Motor and Retention tests an average of seven months later, reported declining scores on all these measures. Serum levels of druqs were not measured.

PAGE 21

15 Booker (1972) reported the results of an attempt to relate seizure control achieved to serum concentration of anticonvulsants. He found no significant differences in the serum levels of controlled and uncontrolled patients. Seizure control status was seen to be associated with type of clinical seizure, measured IQ, Halstead-Reitan Impairment Index, and the nature of background electroencephalogram (EEG). Prospects for effective control were better in those persons having major motor seizures, less evidence of encephalopathy, and normal (8-12 Hz) EEG background activity. The WAIS, WMS, and Beck Depression Inventory were given to a general sample of 118 male and female epileptics by Reynolds and Travers (1974). Serum anticonvulsant concentrations were monitored as well. Those patients who had clear signs of drug toxicity, mental retardation from birth, or evidence of gross cerebral lesions were excluded. The investigators looked at such dependent variables as psychomotor slowing, intellectual deterioration, psychiatric diagnosis, and personality change. These were evaluated from records, prior psychiatric reports, relatives' responses to inquiry, a brief mental state examination, and the psychometric measures. The authors found that those patients with psychomotor slowing, intellectual deterioration, psychiatric diagnosis, or personality change had, as a group, higher serum concentrations of phenytoin and phenobarbitone than the group showing no evidence of change. The mean serum values of those persons showing mental changes fell within a recommended therapeutic range, although there was a wide scatter both below and above that range.

PAGE 22

16 Dodrill (1975), evaluating the neuropsychological status of 70 diagnosed major motor and psychomotor seizure patients receiving phenytoin, found differences between groups designated High and Low on the basis of blood level. The High group did less well on tests of gross and fine motor performance. Measures of higher mental functioning, as measured by the Halstead-Reitan Battery, showed no significant intergroup differences. As part of a larger study of cognitive performance in epileptic patients receiving phenobarbital at two therapeutic dosage levels, MacLeod et al. (1978) reported short-term but not long-term memory retrieval problems at the higher dose. Their tasks consisted of (a) sequential presentations of single digits followed by a probe which the subject responded to as in/not in the set, and (b) presentation of pairs of letters, e.g., AA, Aa, AB, Ab, to be identified as same/ different in terms of the letter name. The experimenters proposed these as measures of speed of access to STM and LTM, respectively, using reaction time (assuming accuracy to be perfect) as the dependent variable. Observing increases in reaction time to the digit task on the higher dosage, MacLeod et al concluded that increased phenobarbital concentration reduced efficiency of STM scanning. No report was made of efforts to separate attentional mechanisms from those of memory. Matthews and Harley (1975) administered 15 measures of intellectual and adaptive abilities, sensory discrimination, and motor proficiency to 63 subjects currently being treated with phenytoin, phenobarbital, and primidone. Serum concentrations of the drugs were seen to be in the

PAGE 23

17 toxic range of one or two of the agents for 35 subjects. Comparing these toxic patients with the 28 whose serum levels were nontoxic, the investigators found significantly poorer performance for the former group on WAIS Coding, the Seashore Rhythm, Knox Cube, Maze Coordination, and Static Steadiness tests. They concluded that these findings reflected impairment of attention and concentration, as well as motor coordination and steadiness in those individuals with serum concentrations on the toxic range. They noted, however, that the groups had not been equated for seizure control status, which obscured the source of the deficits found. Danish researchers Sommerbeck, Thielgaard, Rasmussen, Lohren, Gram, and Wu Iff (1977) tested a subset of the patients participating in a triple blind study of valproate sodium to assess reported psychotropic effects of the drug. They found differences only of a negative kind in psychomotor tempo and visuospatial analytic and synthetic functioning. Other anticonvulsants being used by these subjects, who had therapyresistant seizures, were held constant. These agents averaged 3.5 per patient. Interpretation of the authors' results is difficult in view of a 33 percent sample attrition, failure of most subjects to complete most assessment measures, and the possibility that the introduction of valproate sodium produced an additive effect rather than one reflecting its unique characteristics. Dekaban and Lehman (1975) adjusted the anticonvulsant doses of 15 epileptic patients upward twice by 30 to 50 percent of their initial doses. Before each dose change, tests of vigilance, reaction time, and verbal memory were given. The authors reported that "the greatest

PAGE 24

18 number of patients performed best on the lowest dose of medication" (Dekaban & Lehman, 1975, p. 319) on the vigilance and reaction time measures. Verbal recall scores showed similar but nonsignificant trends. Serum levels of anticonvulsants, various combinations of phenytoin, phenobarbital and primidone, were monitored but not reported. Thompson and Trimble (1982) studied 15 epileptics receiving mutiple drugs, either reducing or withdrawing their medications and adding or introducing carbamazepine. Two other groups of 20 and 10 patients had reductions only or no change, respectively. Measures of cognitive and motor functioning improved from the initial to the third session six months later for subjects in the medication change and reduction groups, but no differences were observed in the control group. Gannaway and Mawer (1981) also seeking to reduce polypharmacy (the use of several drugs of the same or different classes), transferred 18 patients, poorly controlled on their anticonvulsants, to phenytoin therapy alone. After patients achieved blood levels of 15 micrograms/milliliter of serum the physicians increased doses by 25 mg increments until seizures were completely suppressed or clinical intoxication appeared. Tests of speech, motor coordination, verbal memory, and vigilance were given as blood levels ascended. Only the test of vigilance proved sensitive to clinical intoxication states. Not all studies of anticonvulsant effects on psychological functioning are cautionary. Dodrill and Troupin (1977) reported results of a double-blind, crossed comparison study of the anticonvulsants carbamazepine and phenytoin. All seizure patients spent four months

PAGE 25

19 on each drug. As opposed to an unmedicated state, the presence of therapeutic blood levels of anticonvulsants coincided with improvements in test scores. During phenytoin dosage periods, patients showed significant reduction of errors on tests of receptive aphasia, the Stroop test (a test of perceptual interference which also measures word-reading fluency), and a brief self-administered mental ability test. While taking carbamazepine, error reduction was seen in a test of constructional dyspraxia. Greater improvements were seen in patients with initially lower average IQ, longer duration of seizures, and more emotional disturbance on the MMPI. The authors saw these effects as being related to attention maintenance and to the fostering of performance of higher-level complex skills requiring manipulation. Instances of detrimental psychological effects stemming from antiepileptic drug treatment appear to receive more investigative attention. These effects tend to be associated with high or toxic serum concentrations, although idiosyncratic responses in individuals to amounts within recommended bounds are not unusual. It is also evident that therapeutic dosage which reduces seizure activity can provide some concomitant improvement in neuropsychological performance (Dodrill & Troupin, 1977; Thompson, Huppert, & Trimble, 1980, 1981; Thompson & Trimble, 1980, 1981; Trimble, 1981; Trimble & Reynolds, 1976; Trimble & Thompson, 1981). Rationale For This Study The preceding review lends support to the existence of memory deficits in complex partial seizure patients. These impairments are more prominent in them than are other kinds of cognitive deficits and,

PAGE 26

20 in LTM performance, often reflective of the hemisphere of involvement. The point in memory processing at which these difficulties arise is obscure, although current evidence supports the view that it occurs beyond sensory memory. The material -specific effects on LTM according to side of focus have not been shown to be due to problems of retrieval or as taking place at earlier stages during the registration process. Early studies using heterogeneous CPS groups reported general memory deficits in patients in this seizure group, but described no impairments related to side of involvement. The view that side of focus might well be an issue has stimulated investigators to look at left and right CPS groups separately. However, these studies have presented some problems in interpretation of their results because of inappropriate contrast groups, lack of adequate control of relevant variables (e.g., anticonvulsant medication], limitation of measures to those assessing LTM, and unclear reporting of or incorrectly used statistical analyses. The material -to-be-remembered versus modal ity-of-presentati on controversy continues. In part, this is due to the difficulty inherent in developing genuinely nonverbal measures and to the omission of any nonverbal auditory measure. Studies taking into consideration or controlling antiepileptic drugs have often used patient samples taking multiple medications, which obscures the source of effects seen. These studies have more usually employed LTM measures or, in using STM measures, have not evaluated STM with inclusion of rehearsal prevention. This study was designed to examine STM functioning in CPS patients with respect to material -to-be-remembered versus modal ity-of-presentati on.

PAGE 27

21 Patients having generalized major motor seizures served as a contrast group. Antiepileptic medication, limited as much as possible to the drug phenytoin, was controlled. A variety of tests of attention and memory were compared at subtherapeutic, therapeutic, and toxic levels of blood phenytoin to assess the effects of seizure medication on performance in these three groups. This design would allow the investigation of the extent to which serum concentration of phenytoin and its therapeutic range, arrived at by criteria of optimum seizure reduction without clinical toxicity, is meaningful to STM functioning. The following hypotheses were tested. First, groups will perform differentially with right and left CPS groups showing greater STM impairment than the generalized major motor seizure group. Of the CPS subjects, those with a predominantly left-lateral ized temporal focus will perform less well than subjects with a predominantly rightlateralized temporal focus on verbal memory measures. Subjects with a right-lateral ized temporal focus will show more impairment in memory for less readily verbalized "nonverbal" material than will subjects with a left-lateralized focus. Second, subjects in the sample will show better STM performance at phenytoin serum concentrations within the therapeutic range than at subtherapeutic or toxic blood levels. Such effects on performance are likely to be particularly detrimental at toxic levels, resulting in the enhancement of the material-specific vulnerabilities according to side of involvement.

PAGE 28

CHAPTER TWO METHOD Subjects The subjects in the study were 37 male and 2 female epileptic patients attending the Seizure Clinic of the Gainesville Veterans Administration Medical Center or of the Shands Teaching Hospital of the University of Florida, Gainesville. They were classified into groups of predominantly right-lateralized complex partial seizures (R-CPS), predominantly left-lateralized complex partial seizures (L-CPSJ, and generalized major motor seizures (GEN) on the basis of clinical appearance of the ictus and electroencephalographic recording (EEG). All subjects had complete neurological examinations, computerized axial tomography (CT) scans of the head, and periodic laboratory studies of hepatic and other physiological functioning, in addition to serum anticonvulsant level determinations as needed. They were seen during the investigation either as outpatients or as inpatients if their medical status warranted hospitalization. Minimum inclusion criteria for the study were presence of a documented unilateral-focus complex partial or generalized major motor seizure disorder, current or intended treatment with phenytoin (diphenylhydantoin, DPH, Dilantin), age between 18 and 62 years, inclusive, and a minimum Full Scale 1Q of 85. Mental retardation, gross brain damage, or other major neurological disease were exclusion criteria. 22

PAGE 29

23 Ten subjects, two women and eight men, did not complete the study. Two women subjects did not return for the final testing session. Treatment requirements of three men led to use of an alternate anticonvulsant medication; two men did not achieve serum concentrations of phenytoin within the therapeutic range over repeated testings. Of the three remaining subjects, one withdrew, one moved out of the state, and one was disqualified because of prominent clinical symptoms of narcolepsy. The final sample numbered 29 male subjects of whom 12 had a predominantly right-lateralized temporal seizure focus, 10 a predominantly left-lateralized temporal focus, and 7 evidence of generalized epileptogenic activity on EEG. Sample means for these subjects were age, 45.6 years (SD_ = 10.2); education, 12.7 years (SD = 3.3); age at onset of seizures, 31.9 years (SD = 14.3); daily frequency of seizures, 0.27 (STJ = 0.66); and duration of seizure disorder, 13.7 years (SD_ = 11.4). Mean 10 scores for the sample were Full Scale IQ, 107.2 (S_D = 10.2); Verbal IQ, 108.3 (SD = 12.6); and Performance IQ, 104.7 (SD = 10.5). Within the sample racial balance was 90% white, 10% black; handedness, 93% right-handed, 1% left-handed; etiology of seizures, 59% unknown, 41% known; and CT scans, 59% normal, 41% abnormal. Twenty-one of the subjects (72%) were taking phenytoin as their sole anticonvulsant. The other eight subjects (28%) were taking one additional anticonvulsant drug as follows: methsuximide 3, valproic acid 3, carbamazepine 1, and gamma vinyl GABA 1. Anticonvulsants other than phenytoin were monitored to maintain therapeutic range serum concentrations. With the exception of gamma vinyl GABA, an experimental agent for which a range

PAGE 30

24 of efficacy was not yet established, all medications were at therapeutic or lower concentrations during subject testing. Subjects received no payment in return for their participation. They were given feedback on their anticonvulsant serum concentrations, and questions concerning the testing were answered. Subjects were offered the opportunity to discuss their perceived memory problems with the examiner at the end of their participation. Pi lot Studies Three measures of short-term memory, which variously combined modality of presentation and material to be remembered, had displayed differences in degree of difficulty in an earlier study with one visual test producing higher scores (Samuels et al 1972). The use of a verbal distraction task across the three tests appeared to interfere negligibly with the instrument composed of geometric figures presented visually. Seven staff members and graduate students of the University of Florida were tested individually on separate occasions and found to exhibit greater score variability when a visual task employing unfamiliar figures was introduced as a distractor. This task, rather than the former verbal one, was adopted for use with the measure described. A second pilot study was done to examine the properties of these three measures of memory processes to look for the presence of learning effects, specifically, improved scores upon readministration. Six University of Florida freshmen students, three men and three women, were individually given the measures of visual memory for geometric figures, visual memory for consonant trigrams, and auditory memory for consonant

PAGE 31

25 trigrams on three separate occasions. Four of the subjects were given the three tests within five days. The remaining two subjects were tested three times over 10 and 20 days. Pair-difference t tests were used to compare mean differences in scores between first and second, then first and third testing sessions. All t tests were nonsignificant at £ < .05. Because mean scores increased with each testing session on the visual consonant trigrams measure, three forms of this instrument using different sets of trigrams were prepared for use in repeated testing. Materials and Apparatus The following test instruments were employed: The Wechsler Adult Intelligence Scale (WAIS) A short form was given which includes all of the subtests of this measure. Two subtests are left intact and every other or every third item of the remaining eight are used (Satz & Mogel 1962). A validation study using this abbreviated version for retesting of schizophrenic patients revealed correlations of r. = .98 with Verbal IQ (VIQ), _r = .93 with Performance IQ (PIQ), and r = .97 with Full Scale IQ (FSIQ) with the full WAIS (Mogel & Satz, 1963). The Visual Vigilance Test (VVT) This is a measure of attention in which a series of pairs of simple visual geometric figures is presented in random order. A Terak Model 8532-1 Monitor computer is employed in presenting stimuli. The examinee's response is a bar press to the target stimulus, two vertically displayed circles on the left or right of the screen. Presentation time is 1.0 second; interstimulus

PAGE 32

26 interval is 1.5 seconds. Target stimuli comprise 10% of 900 presentations. Scoring includes omission and commission errors, separately and summed, in addition to correct responses. The task is analogous to the Continuous Performance Test described by Rosvold, Mirsky, Sarason, Bransome, and Beck (1956) which has been used to evaluate the effect of phenothiazine medication on schizophrenics, showing score improvements in treated versus untreated groups (Kornetsky, 1972). The following three instruments are adaptations of measures used in earlier investigations of short-term memory (Butters et al., 1970; Peterson & Peterson, 1959; Samuels et al., 1972; Witmer, 1935). The Visual Geometric Test (VGT) This is a measure of short-term visual memory. An unfamiliar geometric line drawing on a 4-inch by 6-inch card (see Appendix A for examples) is manually presented briefly, followed by either immediate or delayed (3, 9, or 18 seconds) selection from nine individually presented patterns. In the delay condition a distraction task, consisting of copying unfamiliar graphics, is interposed between stimulus presentation and response (see Appendix B). Percentages of trials correct in the immediate and in each delay condition are the scores. The Visual Consonants Test (VCT) This is a measure of short-term visual memory. A consonant trigram of less than 21% association value is manually presented briefly, in horizontal array, on a 4-inch by 6-inch card followed by either immediate or delayed (3, 9, or 18 seconds) recognition and selection from 10 consonants (Appendix C). In the delay condition a distraction task, counting backward from a three-digit number presented by the examiner, is interposed. Percentages of letters

PAGE 33

27 correct in the immediate and in each delay condition are the scores. Three forms are available for use in repeated testing. The Auditory Consonants Test (ACT) This is a measure of short-term auditory memory. A consonant trigram of less than 21% association value is verbally presented at the rate of 1 second per letter, followed by either immediate or delayed (3, 9, or 18 seconds) verbal report of letters recalled. The distraction task and scoring are the same as those used in the VCT. The Tonal Memory subtest (TMT) of the Seashore Measures of Musical Talents This instrument is composed of 30 pairs of tone sequences. The examinee is required to identify the number of the single tone changed in the second playing of each pair (Seashore, Lewis, & Saetviet, 1960). Tone sequences are 3, 4, and 5 in length for 10 trials of each. The subtest is recorded and was presented on an Ampex cassette tape using a Panasonic RQ-212DS recorder/player. Scores are based on percentage of correct responses per 10 trials and over 30 trials. This measure has proven sensitive to memory problems in right temporal lobectomy as compared with left temporal lobectomy patients, but not to preoperative differences in the small groups examined (Milner, 1962). A Patient Information Form This is an instrument developed for this study and designed to record demographic and seizure-related medical information (see Appendix D). Used in an interview context, this instrument provides an ongoing record of the subject's anticonvulsant dose, serum concentration, and dates of testing sessions, as well as relevant medical record data. Seizure frequency estimations recorded here were based on reported seizure activity during the preceding 90 days.

PAGE 34

28 Procedure Patients were invited to take part in this investigation if one of three conditions was present: 1.) seizure dyscontrol believed due to subtherapeutic anticonvulsant serum concentration whether by noncompliance, altered dose requirements, or new onset of seizure disorder; 2.) clinical symptoms of phenytoin toxicity; or 3.) expressed interest on the part of a patient in perceived memory problems with or without a documented treatment history of difficulty in maintaining serum concentrations of anticonvulsants within the therapeutic range. Thus, subjects began participation at any categorical blood level of phenytoin: subtherapeutic (LOW), to < 10, therapeutic (NORMAL), 10 to < 20, or supratherapeutic (TOXIC), j^ 20 micrograms per milliliter of serum concentration. A blood sample for measurement of anticonvulsant serum level was drawn within hours of each study testing session. Testing was done blind to serum concentration of phenytoin or other medication because of an approximately 48-hour delay in receiving laboratory reports on routine requests for this procedure. After reading and signing the consent form (see Appendix E), subjects were interviewed to obtain a history of their seizure disorder, then given the set of measures described. The abbreviated WAIS was given during a session when the subject was able to undertake the task, i.e., not clinically drug-intoxicated or otherwise appearing disadvantaged. The neurologist then adjusted the dose of phenytoin if present symptoms and/or history warranted it, or awaited the laboratory results to evaluate the amount and direction of change needed.

PAGE 35

29 Subsequent sessions were scheduled for a time when the anticonvulsant level was expected to have changed in response to the new dose. A minimum of three testing sessions took place until at least one was concurrent with a phenytoin blood level within the therapeutic range and one concurrent with a level outside the range. The memory measures were presented in differing orders on each occasion of testing. The VVT was given first or last because of the necessity for use of stationary equipment located at a distance from the usual testing site. No data collected within 24 hours of a seizure were used in analyses; the data were discarded and the session rescheduled. Instructions to the Subjects V i sual Vigilance Test Subjects were instructed by means of programmed instructions, reinforced verbally by the examiner, to watch the screen, pressing the space bar only at the appearance of two circles, vertically displayed, on the left or right side. After a brief practice session, the 900 stimuli of the measure were presented. Average length of testing was 40 minutes. Visual Geometric Test In the immediate response condition subjects were shown a line drawing of a figure, then asked to identify it by stopping the examiner during a sequential review of the nine stimuli. The delay condition featured presentation of the target stimulus, followed by the examiner's saying, "Copy." The subject then copied unfamiliar graphics for either 3, 9, or 18 seconds until hearing, "Stop," and again selecting the most recently presented drawing from those reviewed. The number of trials was 24, 6 immediate and 18 delayed

PAGE 36

30 with the intervening distraction task. Immediate responses were grouped at the beginning as a perceptual check; delayed responses were arranged randomly. A practice period employing six nontest stimuli with both immediate and delayed response conditions preceded the actual test. Visual Consonants Test Subjects were given instructions similar to those above, except for the response format and distraction task. The response requested was that of selecting the letters in the displayed trigram from an array of 20 different letters and circling them. Correct order of letters was not a requirement. The distraction task consisted of counting backward from a three-digit number supplied by the examiner immediately after withdrawing the stimulus. The number of trials was 16, 4 immediate and 12 with intervening distraction. A practice session using nontest trigram stimuli preceded the test. Auditory Consonants Test Subjects were instructed to listen to the three letters spoken by the examiner. In the immediate response condition they were to repeat the letters heard. The delay condition imposed the backward-counting task before a verbal response was given. In all other respects this measure was identical to the VCT. Tonal Memory Test Subjects were given the standard instructions and practice experience described in the manual. The examiner told subjects they would hear several tones, a pause, then the same tones repeated except for one which would be changed. Subjects were told their task was to decide which note was different in the second playing and report it by number. After listening to several randomly selected pairs from the measure, subjects began the actual test. Response consisted of circling the number of the changed tone after hearing a pair of sequences.

PAGE 37

31 Statistical Analyses The independent variables were type of diagnosed seizure disorder (R-CPS, L-CPS, GEN) and phenytoin blood level, either absolute (ABL) or categorical (LOW, NORMAL, TOXIC J Dependent variables were the absolute scores obtained on the measure of attention (VVT) and its omission, commission, and total errors (OE, CE, TE, respectively); percentage correct responses on the tests of short-term memory, both immediate and delayed (VGT-IM and -STM, VCT-IM and -STM, ACT-IM and -STM) ; and, for the TMT, both total (TMT) and partial (TMT3, TMT4, TMT5) percentages correct. Scores obtained at increasing delay intervals within the testing session on each short-term memory test were converted to regression lines representing loss of information from short-term memory over time, Each session thus provided 3 lines for a total of 87 in the sample. The slopes of these lines (SLOPEVGT, SLOPEVCT, SLOPEACT) became dependent measures in multivariate analyses for group and blood level effects. Univariate one-way analyses of variance (ANOVA) were employed to test for heterogeneity of seizure diagnostic groups. Relationships between subject variables and between dependent measures were examined using Pearson product-moment correlations. The research hypotheses of differential performance of memory tasks by seizure groups and of impairment of this performance at phenytoin blood levels outside the therapeutic range were tested with two-way repeated measures analysis of variance and two-way repeated measures multivariate analysis of variance (MANOVA) using the GLM procedure of the Statistical Analysis System (SAS) for unbalanced data (Goodnight, 1979; Shaffer, 1979). Dependent measures in these procedures were

PAGE 38

32 absolute scores, numbers of errors, and percentage of correct responses as appropriate to the instruments administered. The contributions to group effects of each of five uncontrolled memoryand seizure-related variables were evaluated post hoc by means of GLM analyses of covariance. Tests of the above hypotheses as they relate to loss of information from short-term memory over time were performed on within-session scores converted to regression slopes. Because few subjects (n_ = 4) experienced all three possible phenytoin blood level categories, group effects were first tested by applying the MANOVA procedure at each categorical blood level. Repeated measures MANOVA was used to test both seizure group and blood level effects upon movement from LOW to NORMAL levels and from NORMAL to TOXIC levels separately. Tests for the presence of a linear versus quadratic relationship between phenytoin blood level and short-term memory performance were done using the CONTRAST statement within the GLM procedure. With the exception of hand-calculated paired t tests comparing mean scores of short-term memory tests across sessions in order of occurrence to test for the presence of a sequence effect, all statistical analyses were performed at the Northeast Regional Data Center, University of Florida, Gainesville.

PAGE 39

CHAPTER THREE RESULTS Overview of Results The hypothesis that epileptic subjects with focal complex partial seizures have greater short-term memory impairment than subjects with generalized major motor seizures was not supported at a significant level. Although short-term memory test data tended in the predicted direction, and to be consistent with earlier investigators' reports of material-specific deficits according to hemisphere of involvement, test behavior related to attentional functioning was a better predictor of seizure type in this sample. Similarly, the hypothesis that phenytoin serum concentrations outside the accepted therapeutic range of 10 to 20 micrograms per milliliter of serum exert detrimental effects on short-term memory was not confirmed. Once again, attention-related performance was more responsive than measures of memory functioning to alterations in anticonvulsant level Subjects in the study demonstrated short-term memory impairments across seizure types and anticonvulsant serum levels relative to control and normative groups. Material -related rather than modality-related information loss problems were seen in subjects with focal seizures. 33

PAGE 40

34 Description of Groups and Relationships of Variables The three subject groups, R-CPS, L-CPS, and GEN, did not differ in age, education, age at seizure onset, duration of seizure history, frequency of seizures, FSIO, or VIQ. They were also alike v/ith regard to racial composition, handedness, etiology of seizure disorder, evaluation of CT scan, and anticonvulsant medications being taken. They were, however, dissimilar in PIQ, F[2, 26) = 3.49, p_ < .05, the two CPS groups being equal and scoring higher than the GEN group. Group and sample characteristics are presented in Table 1. Correlations Between Subject Variables Somewhat differing strengths of correlations between subject variables among groups can be seen in Table 2. The R-CPS and GEN groups showed less association of PIQ with FSIO than did the L-CPS group. Both CPS groups exhibited positive association between age and age of seizure onset, a reflection of those groups' older members' later development of seizure disorder. In the R-CPS group the relationship of age to etiology, _r = .70, p_ < .01, revealed an association between earlier-developed seizures and known etiology, as well as later-developed seizures and unknown etiology in these subjects. This group also showed a negative correlation between PIQ and CT scan, r = .75, £ < .01, indicating that lower PIQ scores were associated with abnormal CT scans of the head, higher PIQ scores with normal scans. Correlations Between Subject Variables and Dependent Measures Correlations between subject descriptor variables and test scores presented in Table 3 show relationships between FSIQ and VIQ and

PAGE 41

35 cl c| c| •Ito >>

PAGE 42

36 c

PAGE 43

37 c|

PAGE 44

38

PAGE 45

39 -Q i— rs I—

PAGE 46

40 CM

PAGE 47

41 performance on the short-term memory aspect of the two tasks employing verbal materials, the VCT and ACT. Although FSIO was associated with better performance on both immediate and short-term memory aspects of the VGT, which is composed of line drawings of geometric stimuli, VIQ was related only to the immediate memory portion of this instrument. The verbal measures were also positively correlated with longer seizure duration, the visually presented VCT with age as well. These and other relationships seen in the full sample are better traced to sources in the groups by examining Table 4 where age, age of onset, and duration can be seen to be differentially related to test scores within each group. For example, better scores on short-term memory components of verbal measures by R-CPS subjects were associated with earlier onset and longer duration. Conversely, however, the scores of GEN subjects on short-term memory components of the geometric test were associated with later onset and shorter duration. The L-CPS group showed relationships between better VCT scores and VIQ (r = .60, p_ < .001), and between better VGT scores and PIQ (r_ = .35, p_ < .05), suggesting their use of abilities other than verbal skills in VGT performance. Correlations Between Frequency and Duration of Seizure Disorder and Dependent Measures Several correlations between the subject variables of duration, frequency, and 10 with errors made on the VVT were found to be differentially distributed across groups. As Table 4 shows, the negative relationship of duration to total errors became nonsignificant in any single group. Seizure frequency remained associated with omission

PAGE 48

42 cnn o cm CO CO O CM

PAGE 49

43

PAGE 50

44 errors for the R-CPS and GEN groups (for both, p_ < .001). However, a negative relationship of FSIQ and PIQ to total errors and of PIQ to commission errors continued only in the GEN group. Because duration and frequency of seizures have often emerged as relevant variables in studies of epileptics' neuropsychological functioning, these relationships were inspected by dividing the groups on these variables to further examine their effect on error rates. Subjects were separated into those having fewer and more than 10 years seizure history, and those reporting fewer and more than four seizures a month (B. J. Wilder, personal communication, February 17, 1984). Errors on the VVT by group and duration are presented in Table 5. Subjects in the L-CPS and GEN groups showed decreased VVT commission errors with more than 10 years duration, whereas the R-CPS group showed an increase in this type of error. Omission errors also differed by groups according to duration. The R-CPb group showed little change, the L-CPS group showed an increase with longer duration as compared with the GEM group whose mean error rate decreased with longer seizure history. Table 6 presents the VVT error rates by group and frequency of seizures. As shown, greater numbers of omission errors were made by R-CPS and GEN subjects, but L-CPS subjects having more than four seizures a month had fewer errors. All groups made fewer commission errors at the higher seizure frequency.

PAGE 51

45 cl c|

PAGE 52

46 cl c|

PAGE 53

47 Correlations Between Phenytoin Dose, Blood Levels, and Dependent Measures Mean phenytoin doses ordered and serum concentrations measured at each categorical level in the three subject groups are presented in Table 7. There was a significant association between ABL and CE in the GEN seizure group {r_= .52, £ < .05). Although dose of phenytoin and ABL were related at the lowest serum concentration measures in each subject (_r = .43, £< .02), subsequent measurements of phenytoin ABL at higher concentrations were uncorrelated with dose prescribed. Correlations Between Study Measures Intercorrelations of the measures used in the study were VCT with ACT (r = .68, £ < .001), both VCT and ACT with VGT (r = .40, £ < .001) in the full sample. As shown in Table 8, only L-CPS scores on these short-term memory instruments demonstrated strong commonality among all three tests. Scores of R-CPS subjects were highly correlated on verbal tests only. Scores of GEN subjects showed an association between the VCT and ACT which approach significance [r_ = .40, £ = .07), as well as a correlation between the VVT and TMT (r. = .53, £ = .01), two measures which have in common a task demand for monitoring and signal detection. Correlations between study variables revealed expected, unexpected, and ambiguous associations. The differential association by groups between IQ variables revealed a significant relationship of PIQ with FSIQ only in the L-CPS subjects. A negative correlation of PlQ with CT scan was also seen. Performance on the STM component of the VGT was associated with PIQ, but only in the L-CPS group. However, in the full sample, the VCT and ACT short-term components showed associations

PAGE 54

48 3: cc

PAGE 55

Table 8 Intercorrelations of Total Test Scores by Group 49 Variable VVT VGT VCT ACT TMT n = 12

PAGE 56

50 with both V1Q and FSIQ. The association of longer seizure duration and age with better verbal STM performance was of interest. Correlations between the three STM tests were seen in the sample, a reflection of their strong associations in the L-CPS group. An absence of association between phenytoin dose and subsequent blood level at doses beyond the lowest recorded was also noted. Group and Categorical Blood Level Effects on Test Scores Univariate Analyses of Variance Univariate groups (3) by categorical blood levels (3) repeated measures ANOVAs were performed using the GLM procedure of SAS for unbalanced data with VVT, TE, OE, CE, TMT, TMT3, TMT4, and TMT5 as dependent measure individually (Goodnight, 1979). A trend toward a group effect was seen for VVT and Omission Errors (£(2, 26) = 3.03, p_ < .07, for both), and for Total Errors (£(2,26) = 3.13, p < .06). There was also a main effect for group (£(2, 26) = 3.54, £ < .04) on Commission Errors. Duncan's Multiple Range Test for differences in means, performed post hoc, revealed that R-CPS and L-CPS errors, 6.1 and 4.9, respectively, were equal and fewer than those of the GEN group, 22.6 (Kirk, 1968). A main effect for blood level was present in Total Errors (£(2,52) = 4.71, p < .01), and in Commission Errors (£(2,52) = 4.83, p_ < .01). Post hoc examination of means using Duncan's statistic showed errors at LOW and NORMAL were equal, Total Errors at LOW, NORMAL = 11.6, 7.1, in order, and fewer than those at TOXIC, 26.6. Commission Errors at LOW and NORMAL levels, 9.8 and 5.1, respectively, were also alike and fewer than those at TOXIC, 22.7.

PAGE 57

51 Analyses for group and blood level effects on TMT, TMT3, TMT4, and TMT5 were nonsignificant. There were no interactions. Multivariate Analyses of Variance Percentages of correct responses on immediate and short-term portions of the VGT, VCT, and ACT were dependent measures in a groups (3) by categorical blood levels (3) repeated measures MANOVA for unbalanced data (Goodnight, 1979; Shaffer, 1979; Tatsuoka, 1971). The overall MANOVA was nonsignificant for a group effect (HotellingLawley Trace = 0.50, £ approximation (12, 40) = 0.84, £ < .61), as were all univariate ANOVAs. There was no significant blood level effect on these measures (Hotel ling-Lawley Trace = 0.36, £ approximation (12, 92) = 1.36, p_ < .20). One AN0VA, however, showed a main effect for blood level on the VCT-STM (F(2, 52; = 4.26, p < .02) in the direction of improved scores with phenytoin serum concentration increase. Scores at blood levels LOW through TOXIC were 62.3, 69.1, and 71.8, in order. No interaction between group and blood level was present (Hotel ling-Lawley Trace = 0.39, F approximation (24, 182) = 0.75, £< .80). Univariate Analyses of Covariance Five variables, duration of seizures, frequency of seizures, FSIQ, VIQ, and PIQ, seen to be related to group performances, were each employed separately as a covariate in a GLM AN0VA with groups (3) as independent, test scores and errors as dependent measures (R. L. Carter, personal communication, April 26, 1984; Goodnight, 1979; Tatsuoka, 1971). Covariance analysis of duration revealed a main effect for duration (F(l, 83) = 5.80, p < .02) and for group (F(2, 83) = 3.79, £< .03)

PAGE 58

52 on Total Errors. Least squares mean errors of R-CPS and L-CPS groups, 7.8 for both, were equal to each other and fewer than those of the GEN group, 25.0. Much the same effect of duration on Commission Errors was seen (£(1, 83) = 5.20, £_ < .03) with enhancement of a group effect (F(2, 83) = 4.47, p_ < .01). Covariance analyses of seizure frequency showed a main effect for frequency on VVT and Omission Errors scores (JF(1, 83) = 11.82, _p_ < .001, for both). Group effects on VVT and Omission Errors approached significance (£(2, 83) = p_ < .08). having been slightly enhanced with the addition of frequency as a covariate in the model. Least squares mean scores on the VVT were 86.4, 88.2, and 88.6 obtained by subjects in the L-CPS, GEN, and R-CPS groups, respectively. Corresponding Omission Errors were 3.6, 1.8, and 1.4. Those analyses covarying FSIQ, VIQ, and PIQ showed FSIQ effects on VGT-IM (F(l, 83) = 5.95, p < .02) and on ACT-STM (F(l, 83) = 5.34, p_ < .02). Covariance effects of VIQ were shown on VGT-IM (F(l, 83) = 4.71, p < .03), VCT-STM (F(l, 83) = 7.84, p < .01), and on ACT-STM (F(l, 83 = 8.68, p < .004). A group effect on Commission Errors (F_(2, 83) = 3.92, p_ < .02) was maintained with covariance of VIQ. No group differences appeared with covariance of PIQ, which showed an effect on Total Errors (F(l, 83) = 4.09, p_ < .05). No covariance analyses for group effects on the TMT or its components achieved significance.

PAGE 59

53 Group and Categorical Blood Level Effects on Within-Session Performance Scores earned at varied response delay intervals during each testing with the VGT, VCT, and ACT were converted to regression lines representing performance within the individual session using the GLM procedure of SAS. The slopes of these lines served as dependent measures in GLM MANOVAs testing group effects on loss of information from short-term memory at each blood level separately (Campbell & Stanley, 1963; Goodnight, 1979). None of these MANOVAs was significant: at LOW blood level (HotellingLawley Trace = 0.420, F approximation (6, 28) = 0.98, £ < .46), at NORMAL blood level (Hotel! ing-Lawley Trace = 0.236, £ approximation (6, 46) = 0.90, £ < .50), at TOXIC blood level (Hotel 1 ing-Lawley Trace = 1.377, F approximation (6, 14) = 1.61, £ < .22). At the TOXIC level, one AN0VA showed a main effect for group on VGT (F(2, 10) = 5.02, £ < .03). Mean VGT performance slopes at the TOXIC level were -1.41, 0, 46, and -0.72 for subjects in the R-CPS, L-CPS, and GEN groups, respectively. The within-session regression slopes were also the dependent measures in tests for group (3) and blood level (2) effects on loss of information as two sets of subjects moved from LOW to NORMAL (n = 20), and from NORMAL to TOXIC (n = 13) phenytoin blood levels, using GLM repeated measures MAN0VA (Campbell & Stanley, 1963; Goodnight, 1979; Shaller, 1979). The MAN0VA at LOW to NORMAL levels revealed no significant group effect ( Hotel 1 ing-Lawley Trace = 0.697, F approximation (6, 28) = 1.63, £ < .18) or blood level effect (Hotelling-Lawley Trace = 0.455, F approximation (3, 15) = 2.28, £ < .12). One participating ANOVA approached significance (£(22, 17) = 2.02, £ < .07) for VCT,

PAGE 60

54 attributable to a blood level effect in the model (£(1, 17) = 6.73, p_ < .02). Slopes of the VCT at LOW and NORMAL levels were -2.13 ana -1.66, in order, suggesting that better retention occurred at phenytoin levels in the therapeutic range. There was no significant interaction between group and blood level (Hotel ling-Lawley Trace 0.258, £ approximation (6, 28) = 0.60, p^ < .73). A second MANOVA at NORMAL to TOXIC levels was nonsignificant for group effects (Hotel ling-Lawley Trace = 2.57, £ approximation (6, 12) = 2.57, £< .08), for blood level effects (Hotel ling-Lawley Trace = 0.136, £ approximation (3,7) = 0.32, p < .81) or for interaction (HotellingLawley Trace = 0.416, £ approximation (6, 12) = 0.42, p_ < .85). Overview of Mean Single and Combined Test Performances by Subjects Mean scores earned by subjects over three testing occasions are presented in Table 9. The VVT and its error scores are expressed in absolute numbers, all others in percentage of correct responses. Scores on the VVT approached perfect performances of 90 correct. Groups did, however, distinguish themselves in numbers of commission errors made. Scores in the full sample were uniformly depressed on the remaining measures relative to normative and control groups. Short-term memory scores on the VGT, VCT, and ACT were lower by 9.6, 19.4, and 12.3 percentage points, respectively, when compared to those of older (M = 39.5 years) Boston VAH and community volunteer control subjects for whom educational level (M = 13.6 years), but not FSIQ, was reported. In contrast to postoperative anterior temporal lobectomy patients whose age was younger (M = 31.5 years), whose FSIQ (M = 104.9) and educational

PAGE 61

55 1 CD

PAGE 62

56

PAGE 63

57 level (M = 11.9 years) were reported, and who were tested some years after their surgical interventions (M = 6.5 years), the subjects in this sample scored higher only on the ACT, 67.7%, versus the patients' 62.6% (Samuels et al 1972). Groups in this study showed a tendency toward diverging short-term memory performances on these instruments, the L-CPS subjects scoring lowest on the verbal measures, the VCT and ACT, the R-CPS subjects scoring lowest on the V6T. Tonal Memory subtest scores did not differ appreciably across groups. Mean performance in the sample of 57% correct identifications, however, resembled that of postoperative right temporal lobectomy patients, 53% correct, and corresponded to 9th percentile scores in the young adult student group whose normative data are provided by the authors (Milner, 1962; Seashore et al 1960). The following tables and figures describe group performances on individual measures in keeping with the clinical constraints of the investigation. Because so few subjects experienced three categorical phenytoin blood levels (n_ = 4), mean scores of each group are derived from varying numbers of sessions by members of that group at a given blood level Visual Vigilance Test Changes in VVT performance at different phenytoin blood levels are presented in Table 10 in terms of error scores. Figure 2 shows that both CPS groups experienced a Commission Errors rate reduction at the NORMAL level, the L-CPS group showing improvement in Omission Errors as well. Subjects in the GEN group exhibited Commission Errors

PAGE 64

58 = 1 c| cl

PAGE 65

59 >, 00 C E CD ro r— SO -iStfOHKH NOISSIWWOD fO A | cu lo qcu CO s+-> sCD •o +J i— S-r-rSi— r— CU !r— o oo r-^o m
PAGE 66

60 increments, R-CPS subjects Omission Errors increments, with increasing phenytoin blood level. Visual Geometric Test Mean percentages of wi thin-session correct responses to the VGT at the three phenytoin blood levels are presented in Table 11. When phenytoin concentrations were NORMAL, as shown in Figure 3, R-CPS group subjects did less well than L-CPS subjects as response delay lengthened. Figure 4, which represents the mean of both LOW and TOXIC levels, shows a widening of this difference at 18 seconds, and Figure 5, at TOXIC level, further enhancement of the discrepancy between the CPS groups' performance on this task. Subjects in the GEN group, who showed little VGT information loss at NORMAL level, showed losses comparable to those of R-CPS subjects when blood levels were LOW or TOXIC and delay to response was the maximum 18 seconds. Visual Consonants Test A summary of mean wi thin-session scores on the VCT is presented in Table 12. At the NORMAL level, as seen in Figure 6, all groups showed similar performances with the R-CPS and L-CPS subjects conforming closely to each other. However, at phenytoin levels which were extratherapeutic, as shown in Figure 7, R-CPS and L-CPS scores diverged at longer delay intervals with L-CPS subjects scoring lowest of the groups. When performance at the TOXIC level alone is examined (Figure 8), the R-CPS and L-CPS scores maintained their relative positions, but GLN group scores assumed the lowest values of the three groups.

PAGE 67

en to c: i— •iO) 61 u

PAGE 68

62 3 2 2 8 w 100 90 80 70 ^ 3 9 18 DELAY INTERVAL IN SECONDS Figure 3. Mean within-session VGT performance by groups at NORMAL level

PAGE 69

63 100 90 70 R-CPS O L-CPS GEN //13 DELAY INTERVAL IN SECONDS Figure 4. Mean within-session VGT performance by groups at blood levels outside NORMAL range.

PAGE 70

64 s pi n 8 Ed 2 100 90 70 // 3 9 18 DELAY INTERVAL IN SECONDS Figure 5. Mean within-session VGT performance by groups at TOXIC level.

PAGE 72

66 in W CO 5 pi W Pi Pi U 100 90 70 60 50 y A 3 9 18 DELAY INTERVAL IN SECONDS Figure 6. Mean within-session VCT performance by groups at NORMAL level.

PAGE 73

67 100 90 70 60 50 yj18 DELAY INTERVAL IN SECONDS Figure 7. Mean within-session VCT performance by groups at blood levels outside NORMAL range.

PAGE 74

63 Q 100 90 70 60 50 //3 9 18 DELAY INTERVAL IN SECONDS Figure 8. Mean within-session VCT performance by groups at TOXIC level.

PAGE 75

69 Auditory Consonants Test Mean scores on the ACT at the three phenytoin blood levels are presented in Table 13. At NORMAL levels (Figure 9), all groups looked similar. The R-CPS and L-CPS groups scores diverged at extratherapeutic blood levels, with L-CPS group subjects consistently earning lowest scores at all blood levels. Scores outside NORMAL and at TOXIC levels are shown in Figures 10 and 11, respectively. Tonal Memory Test Mean performances by groups on the TMT within session at increasing levels of task difficulty and varied phenytoin blood levels are presented in Table 14. The maximum difference in scores was observed at NORMAL level of phenytoin concentration (Figure 12) with L-CPS subjects outperforming both R-CPS and GEN subjects at the higher levels of difficulty. Scores earned at levels other than NORMAL, shown in Figure 13, and at TOXIC level, shown in Figure 14, were highly similar across groups. Relationships Between Memory Scores and Categorical Phenytoin Levels A test for the presence of a linear versus a quadratic relationship between memory performance and phenytoin blood level was performed using the CONTRAST statement within the GLM procedure of SAS (Goodnight, 1979). Dependent variables were the percentages of correct responses to immediate and short-term memory portions of the VGT, VCT, and ACT. One linear relationship between VCT-STM and level was significant (£(1, 27) = 4.64, £ < .04) in the direction of improved scores with ascending blood level of phenytoin. As phenytoin levels moved from LOW through NORMAL

PAGE 76

c| co 70 en in C r— •iOJ i— c o c o

PAGE 77

71 2 C-l b a pes s w 100 90 70 60 50 ^n 3 9 18 DE1AY INTERVAL IN SECONDS Figure 9. Mean within-session ACT performance by groups at NORMAL level

PAGE 78

C/3 W W 2 W OS G PU 100 r DELAY INTERVAL IN SECONDS Figure 10. Mean within-session ACT performance by groups at blood levels outside NORMAL range.

PAGE 79

73 100 90 70 60 50 // / 3 9 18 DELAY INTERVAL IN SECONDS Figure 11. Mean within-session ACT performance by groups at TOXIC level.

PAGE 80

74 ci ilO o O -M E w>, O) ai c c .c .— o o. (Q Q_ C W CI o > o

PAGE 81

75 o CO 2 i s o S Ea CJ 3 70 60 50 40 30 R-CPS O L-CPS GEN NUMBER OF TONES IN SEQUENCE Figure 12. Mean TMT performance by qroups at increasing levels of difficulty at NORMAL level.

PAGE 82

76 70 60 50 40 30 // NUMBER OF TONES IN SEQUENCE Figure 13. Mean TMT performance by groups at increasing levels of difficulty at blood levels outside NORMAL range.

PAGE 83

77 70 ~ 60 50 40 30 R-CPS O L-CPS GEN A NUMBER OF TONES IN SEQUENCE Figure 14. Mean TMT performance by groups at increasing levels difficulty at TOXIC level. of

PAGE 84

73 to TOXIC, VCT-STM increased in percentage correct from 62.3 to 69.1 then 71.8% in correspondence. No other relationships were significant. Effects of Session Sequence A test of session sequence effects on memory measures was done using paired-difference t tests between first and second, then between first and third sessions, with overall VGT, VCT, and ACT percentage correct scores as dependent measures. One difference, that between first and third sessions of the ACT was significant (t_(28) = 3.47, p_ < .01). Mean ACT percentages correct scores in the full sample were 73.9, 74.7, and 78.6% in first, second, and third sessions, respectively.

PAGE 85

CHAPTER FOUR DISCUSSION The Experimental Hypotheses Short-Term Memory in Complex Partial Versus Generalized Seizures The study data do not support the hypothesis of greater short-term memory deficit in complex partial than in generalized major motor seizures. No significant group differences were observed on the visual consonant trigrams, auditory consonant trigrams, or the visual geometric figures short-term memory tasks. This finding is in agreement with Mayeux et al. (1980), who also employed auditory and visual consonant trigrams with complex partial and generalized seizure patient groups. Neither do these data indicate differential performances between predominantly left-lateralized and predominantly right-lateralized complex partial seizure patients on the basis of material-to-be-remembered. Complex partial seizure patients with a left-sided focus earned lower scores on both visually and acoustically presented short-term memory measures having verbal content than did patients with a right-sided focus. Conversely, patients with a right-sided focus made lower scores on a visually presented short-term memory measure having geometric content than did patients with a left-sided focus. However, these differences were not statistically significant. The present findings are in agreement with Glowinski's (1973) observations of patients with leftand right-lateralized complex partial 79

PAGE 86

80 seizure foci. She found that although patients' scores on verbal and figural reproduction portions of the Wechsler Memory Scale were "in the predicted direction ... the left temporal lobe epileptic group showed a greater verbal memory deficit and the right temporal lobe group showed a greater nonverbal memory deficit" (Glowinski, 1973, p. 133), they were not significantly different from each other. Short-Term Memory Functioning at Three Blood Levels of Phenytoin Effects on Absolute Scores The second hypothesis of better short-term memory functioning at serum levels of phenytoin within the therapeutic range as contrasted with levels below and above that range was not confirmed. Delaney et al. (1980) found anticonvulsant levels uncorrected with their long-term memory tasks in subjects with unilateral complex partial or focal frontal seizures. Dodrill (1975) and Matthews & Harley (1975) reported differences between toxic and nontoxic seizure patients in gross and fine motor performances, favoring the nontoxic patients, but not on the measures of "higher mental functions" (e.g., WAIS VIQ, the Category Test) (Dodrill, 1975, p. 596). Dekaban and Lehman (1975), who twice changed anticonvulsant doses of phenytoin, phenobarbital and mysoline by 30-50% of initial dose, described a nonsignificant trend to better verbal recall at the "lowest dose of medication" (Dekaban & Lehman, 1975, p. 319) in a mixed epileptic sample of patients. In contrast with Dekaban & Lehman (1975), and that of noncorrelation of blood level and long-term memory tasks by Delaney et al. (1980), patient performance in the present study differed. Although one must

PAGE 87

81 cautiously interpret a significant ANOVA in the context of a nonsignificant MANOVA, subject scores were seen to improve significantly with increasing phenytoin blood level on the visually presented verbal measure. Seventy-two percent of this improvement (62.3% to 69.2% correct responses) occurred as the blood level moved from the subtherapeutic to the therapeutic range. The remaining 28% of the gain (69.1% to 71.8% correct responses) took place as the level moved from the therapeutic to the toxic range of phenytoin concentration. Mean phenytoin level in the toxic range for these patients was 24.8 micrograms/milliliter of serum. Inasmuch as patients began sessions at any serum phenytoin level, this improvement in score is not readily attributable to a learning effect. The present results are consistent with the report of Dodrill & Troupin (1977) that patients with therapeutic blood levels of phenytoin or carbamazepine showed improvements in test scores over scores made at less than therapeutic levels. Noninterference of a toxic phenytoin level in short-term verbal memory functioning is again in agreement with findings by Dodrill (1975) and Matthews & Harley (1975) in toxic patients. Effects on Loss of Information from Short-Term Memory Multivariate analyses of the combined short-term memory measures revealed no effects referable to seizure group membership or to blood level on the amount of information lost from short-term memory as delayto-report increased. However, two interesting task-related findings emerged. They are presented with the caveat of their being seen in significant AMOVAs within nonsignificant MANOVAs. First, right-lateral ized complex partial seizure patients showed greater losses of figural material than left complex partial seizure

PAGE 88

82 patients or generalized seizure patients when all three groups had toxic phenytoin levels. A performance of this kind is in keeping with the hypothesis of memory deficit in complex partial seizure patients being material-related according to side of involvement. A patient with a right-sided complex partial seizure focus would thus be relatively more impaired in memory for nonverbal content. The way in which such a vulnerability might be expected to interact with a toxic serum concentration of anticonvulsant drug is a matter for further study. Second, therapeutic blood level, as opposed to subtherapeutic blood level of phenytoin, significantly reduced the mean loss of visually presented consonant trigrams during the 18-second span of short-term memory measured. This finding applied to all three seizure groups, and is in agreement with Dodrill & Troupin's (1977) observations of test score improvements in phenytoinor carbamazepine-treated patients versus untreated epileptics. Thus, there appears to be a complex relationship between phenytoin blood level, material and measure employed, and cognitive performances of epileptic patients. The Short-Term Memory Measures The visual and auditory verbal measures differ in their level of difficulty because response demand is not equivalent. The visual measure requires recognition and selection of the trigram letters from an array of consonants. The auditory measure task is one of productive recall. Interestingly, performance on the more difficult measure was seen to improve from first to third testing. Learning effects were not seen in university freshmen tested on three occasions. An alternate explanation

PAGE 89

83 for this increase in scores is that of increasing effort by subjects to do the task, coupled with greater confidence in their ability to perform it. Subjects in the study commonly protested doing the task in the first session, claiming memory deficiencies, then returned for subsequent sessions saying they believed themselves more able to rely on their memories since beginning to take part in the study. The material /modality relationships with seizure groups and measures employed were relatively complex. Short-term memory components of visual and auditory consonants measures correlated with (FSIQ and) VIQ in the full sample, suggesting that these instruments were eliciting a materialspecific, but not modality-specific, performance. The immediate memory aspect of the visual geometric measure also showed a strong association with (FSIQ and) VIQ. Looking at correlations by seizure groups, however, these relationships were differentially apportioned. Right complex partial and generalized seizure patients', but not left complex partial seizure patients', performances were characterized by use of verbal abilities in this figural task. Left complex partial seizure patients, on the other hand, showed no VIQ correlation, but a PIQ correlation, with the shortterm memory performance on the geometric figural measure. These associations suggested that left complex partial seizure patients, somewhat deficient in global verbal abilities, dealt with figural material using other resources than were used by right complex partial and generalized seizure patients. These latter two groups appeared to be compensating for their deficiency by tapping a verbal ability store, coping with figural material via a mechanism such as labeling figural stimuli presented.

PAGE 90

34 Tonal Memory Test There were no significant group or blood level differences on performance of the Tonal Memory subtest of the Seashore Measures of Musical Talents. Furthermore, scores on this task did not correlate with performances on any of the other dependent measures. Although the test was included in the present study as a measure of short-term auditory nonverbal memory, it clearly failed to contribute to any of the findings with regard to memory, seizure group, or blood level. Unlike the other shortterm memory tasks employed in this study, no distraction was introduced between presentation of the tones and response. The failure to show any effect with this task may be attributable to this procedural difference. Because accurate performance on this task depends more upon recognition of correct sequence than it does upon reproduction of the sequences or tonal qualities of the stimuli, it may be a less sensitive measure than the other short-term memory tasks used. The present findings are consistent with the preoperative performance of Milner's (1962) small sample of right and left temporal lobectomy patients, but not the postoperative differences observed. Attention and Vigilance Although the major hypotheses of the present study focused on material-specific and modality-specific differences in short-term memory and phenytoin blood levels in three seizure groups, a measure of visual vigilance was introduced to examine the possible effects of attentional dysfunction on memory at each blood level. Performance on the Visual Vigilance Test was also examined with regard to its relationship with

PAGE 91

85 other study measures, as well as with the subject variables of age of onset, duration, frequency, FSIQ, VIQ, and PIQ. Effects of Seizure Group There were no differences between seizure groups on errors of omission, which are interpreted to reflect errors of inattention. The group differences in numbers of errors of commission contributed to the overall group differences in total errors. Errors of commission occur when subjects respond to a nontarget stimulus. Subjects with generalized major motor seizures made more responses to nontarget stimuli than did complex partial seizure patients. The present finding of more errors of commission than omission is not consistent with the earlier work of Fedio & Mirsky (1969) and Mirsky et al (1960) who found that generalized seizure groups showed more inattention than complex partial seizure groups. However, reports in terms of absolute scores do not take commission errors into consideration. Their task was a more difficult one, requiring a two-step decision (respond to X only if it follows A) and a quicker response (0.1 to 0.2 sec presentation and 0.9 to 1.0 sec interstimulus interval versus this study's 1.0 sec presentation and 1.5 sec interstimulus interval). Differences in duration of task (10 min versus this study's approximately 40 min) do not serve to balance the two. This level of task difficulty may account for the discrepancy. Effects of Phenytoin Blood Level Performances by subjects at the three blood levels of phenytoin are significantly different in numbers of commission errors made at the toxic

PAGE 92

36 level. All groups made more commission errors when phenytoin level was in the toxic range. This effect is consistent with that of Dekaban & Lehman (1975) reporting that chronic epileptics' performance on a vigilance measure is best at lowest doses of various combinations of phenytoin, phenobarbital and primidone. The authors employed the more difficult vigilance task used by Mirsky et al. (1960), summing both error types to arrive at a "relative percentage correct" as described by the developers of the measure (Rosvold et al 1956). The level and similarity of groups' scores on this measure do not suggest impairment of attention in any group. As a consequence, differences in memory performances by groups are not a function of inattention. Rosvold et al. (1956) observed that commission errors do not necessarily represent attentional deficit. They proposed two scores for their measure, absolute and relative percentage correct, incorporating commission errors in the second of these, in the interest of providing "a more complete description of (his) performance." Commission errors elicited in the context of attending, as instructed, to a recurring critical stimulus may represent a limited attentional capacity. On the other hand, commission errors may reflect disinhibition of central origin or, in this instance, one enhanced by intoxicating serum concentration of phenytoin. Subject Variables and Group Relationships to Performance on the Dependent Measures Studies of epileptics impose a necessity for control, either experimentally or statistically, of numerous subject and seizure-related variables found relevant to these disorders. The literature in the area

PAGE 93

87 has indicated the direction of expected influence of several of these variables. For example, longer duration and greater frequency of seizures are each believed associated with, if not causative of, performance deterioration. Efforts to delineate each type of seizure uniquely have perhaps obscured a middle ground of commonality, lying between specificto-type and general-to-epilepsy characteristics. The following observations made of this subject sample show both expectable and unanticipated relationships of variables to performance, commonalities across diagnostic groups, and a reciprocal mechanism which could partially account for types of errors seen in the context of performance of the vigilance task. Unique to this study, the relationship of early age of onset to known etiology, as well as that of lower PIQ with an abnormal CT scan, seen in the right complex partial seizure group, is explicable. Patients in this group who developed seizures early in life often did so following head trauma with right-lateralized effects, impaired VIQ, and observable defect on the CT scan. Neither of the other two groups showed this pattern of correlations. All subjects' performance on the VCT short-term memory portion improved with longer duration of seizure history. Right-lateralized complex partial seizure patients' scores improved on the ACT short-term memory portion as well. In this sample, the deleterious effects of more years of seizures may have been mitigated by access to continuing care with consequent better control of their disorder. Both complex partial seizure groups showed similar associations of VIQ, and less comparable but positive associations, with short-term

PAGE 94

memory performance on the ACT. Left-lateralized patients had correlations in common with the generalized patients' group only when these correlations were seen in all three groups. Right complex partial seizure patients, on the other hand, had commonalities with generalized seizure patients in several respects. The groups shared low correlation of PIQ with FSIQ; noncorrelation of PIQ with short-term performance on the VGT, but correlation of VIQ with immediate performance on that measure; and a correlation between seizure frequency and omission errors on the vigilance task. These resemblances between the two groups, provided that they are not idiosyncratic to this sample, could partially account for the difficulty often encountered in differentiating their performances on measures tapping PIQ abilities. The effects of frequency and toxic blood levels of phenytoin on the types of errors made by subjects in this sample on the vigilance task appeared to operate reciprocally. With higher seizure frequency, omission errors were greater for right complex partial seizure and generalized seizure patients, but commission errors were fewer. At toxic phenytoin levels, commission errors increased in all three groups. Although this effect is not perfectly balanced, showing identical changes in all subjects, its observation stimulates the following conjecture. Because dyscontrol of seizures with increased frequency is commonly associated with insufficient serum anticonvulsant levels, these two states, high seizure frequency and toxic blood level, may be located distant from each other on a continuum. At one end is nil to low serum phenytoin, behaviorally accompanied by frequent seizures and omission errors on a vigilance task, at the other end, toxic phenytoin levels and commission errors.

PAGE 95

89 Summary and Conclusions Neither experimental hypothesis of differences in seizure groups' performances in short-term memory or differences in these same performances at therapeutic versus nontherapeutic levels of serum phenytoin received support at a significant level. One dependent measure, errors of commission on a test of vigilance, did distinguish generalized seizure patients from those having complex partial seizures. The same measure was seen to increase significantly when blood levels of phenytoin were in the toxic range. Results obtained indicate that the groups of complex partial seizure and generalized major motor seizure patients studies had a similar level of short-term memory dysfunction. Although similar in level, the impairments appeared to be nonsignificantly different in each group, an impression received from observations of performances on short-term memory measures. Performances of complex partial seizure patients were nonsignificantly different according to hypothesized material effects on their memories, with left-lateralized patients performing less well than others on two verbal measures and right-lateral ized patients performing less well than others on one geometric measure. Generalized seizure patients' performances tended to resemble, on each measure, that of the complex partial seizure group performing better. No observations made in the study were better understood by invoking a modal ity-of-presentation hypothesis. The small size of the sample clearly restricts both conclusions drawn from these data and their general izability to other, similar groups.

PAGE 96

90 That small size, the use of a generalized seizures contrast group, and constraints imposed by clinical treatment responsibilities are likely contributors to the failure to support or disconfirm the existence of short-term memory deficits in complex partial seizure patients. Larger samples and contrast with other focal seizure patients are recommended. Although a unitary effect of phenytoin blood level was not seen, inclusion of anticonvulsant level among controlled variables is advised as a means of clarifying its role in cognitive functioning of epileptics. Its differential effects on a visual verbal and a visual geometric measure suggest that, in larger samples, it may be found to influence specific types of performances with differing ranges of therapeutic effectiveness than are utilized as guidelines for medical management of seizures.

PAGE 97

APPENDIX A SAMPLE VISUAL GEOMETRIC TEST STIMULI tr 91

PAGE 98

APPENDIX B VISUAL GEOMETRIC TEST DISTRACTION TASK NAME DATE SESSION 1 2 5 Copy these figures as quickly and as well as you can. Begin another row when one is finished. Continue until vou are asked to STOP. £^JA,^b. 5> U Q N o3 t 92

PAGE 99

APPENDIX C VISUAL CONSONANTS TEST RESPONSE FORM NAME DATE SESSION: 1. BGLQVCHMRWDJNSXFKPTZ 2. CHMRWDJNSXFKPTZBGLOV 3. DJNSXFKPTZBGLQVCHMRW 4. BGLQVCHMRWDKNSXFKPTZ TEST: 1. BGLQVCHMRWDJNSXFKPTZ 2. CHMRWDJNSXFKPTZBGLQV 3. DJNSXFKPTZBGLQVCHMRW 4. FKPTZBGLQVCHMRWDJNSX 5. VWLGBWRMHCXSNJDZTPKF 6. WRMHCXSNJDZTPDFVQLGB 7. XSNJDZTPKFVQLGBWRMHC 8. ZTPKFVQLGBWRMHCXSNJD 9. BGLQVCHMRWDJNSXFKPTZ 10. CHMRWDJNSXFKPTZBGLOV 11. DJNSXFKPTZBGLQVCHMRW 12. FKPTZBGLQVCHMRWDJNSX 13. VQLGBWRMHCXSNJDZTPKF 14. WRMHCXSNJDZTPKFVQLGB 15. XSMKDZTPLFVQLGBWRMJC 16. ZTPKFVQLGBWRMHCXSNJD 93

PAGE 100

APPENDIX D PATIENT INFORMATION FORM

PAGE 101

NAME ADDRESS

PAGE 102

96 ANTICONVULSANT BLOOD LEVELS SESSION 1 SESSION 2 SESSION 3 DATE NOTES :

PAGE 103

APPENDIX E PATIENT INFORMED CONSENT TO BE A RESEARCH SUBJECT

PAGE 104

PATIENT INFORMED CONSENT TO BE A RESEARCH SUBJECT NAME ADDRESS PHONE Area Code Number DATE OF BIRTH / / SEX S.S.# Month Day Year I, Case No. agree to be a subject in an experimental study which will examine memory processes in people taking anticonvulsants. I understand that: 1. The study will be conducted by B.J. Wilder, M.D., and M.L. McGuigan, M.S., at the Veterans Administration Medical Center and the University of Florida Shands Teaching Hospital in Gainesville, Florida. 2. My condition has been diagnosed as a form of epilepsy. 3. I will be tested during my routine outpatient clinic visits or as a hospital inpatient should my condition make hospitalization advisable. 4. The study will proceed as follows: On the first visit my complete medical history will be taken, and I will receive a complete physical and neurological examination. An electroencephalogram (EEG) will have been performed in the past year. Appropriate clinical laboratory tests will be performed. No hazards, but minor discomfort from placement of scalp surface electrodes with a water-soluble glue are to be expected from the EEG. I will also be given a standard mental ability test at some time during the study. While in the clinic, I will receive a prescription for (an) anticonvulsants ) to be taken by mouth. My serum levels of anticonvulsants will be measured with periodic blood tests. There is some discomfort such as local pain or skin discoloration connected with the taking of blood samples from my vein. I will be individually given several tests of memory and attention three (3) times while returning to the clinic for my routine medical care or while an inpatient if hospitalization is advisable. These tests are experimental and, although there is no risk in taking them, they cannot be guaranteed to provide all the information I might wish to have about my memory. 5. Taking part in this study will provide me with an evaluation of my condition and medication needs to reduce seizures. I will also learn about what effects my medication has on me and get some information about how my memory is working. 6. I have been informed that the side-effects of anticonvulsants are those related to coordination and balance. Dr. Wilder has informed me that rarely patients feel sleepy and not much inclined to be physically active when they are taking this medication. In addition, I understand 98

PAGE 105

99 that stomach upset may rarely be experienced. Side-effects of anticonvulsants also include skin rashes, mild acne, extra growth of body hair, swelling of the gums, occasional allergic reaction, change in white blood cell count or in liver function. 7. In addition to the drug phenytoin, other drugs such as phenobarbital primidone, carbamazepine, and valproic acid are alternative treatments for my seizures, and will be made available to me if required. 8. Questions I may have about the procedures in the study will be answered. 9. Any information obtained in this investigation will be strictly confidential. I may withdraw from the project at my request without that decision having any effect upon my medical care. I will not be paid to participate in this study. In the event of my sustaining a physical injury which is proximately caused by this experiment, professional medical care received at the J. Hillis Miller Health Center, exclusive of hospital expenses, will be provided me without charge. This exclusion of hospital expenses does not apply to patients at the Veterans Administration Medical Center (VAMC) who sustain physical injury during participation in VAMC-approved experimental studies. I have read and understand the above-described medical procedure in which I am to participate and have received a copy of this description. DATE SIGNATURE Patient DATE SIGNATURE Witness I, the undersigned, have defined and fully explained this medical procedure to the above patient and/or the person authorized to consent for the patient. DATE SIGNATURE Investigator

PAGE 106

REFERENCES Booker, H. E. (1972). Phenobarbital mephobarbital and metharbital: Relationship of plasma levels to clinical control. In D. M. Woodbury, J. K. Penry, & R. F. Schmidt (Eds.), Antiepileptic drugs (pp. 329334). New York: Raven Press. Butters, N., & Barton, M. (1970). Effect of parietal lobe damage on the performance of reversible operations in space. Neuropsychologic 8, 205-214. Butters, N. Barton, M. & Brody, B. A. (1970). Role of right parietal lobe in the mediation of cross-modal associations and reversible operations in space. Cortex 6_, 174-190. Butters, N., & Brody, B. A. (1968). The role of the left parietal lobe in the mediation of intraand cross-modal associations. Cortex 4, 328-343. Butters, N., Samuels, I., Goodglass, H. & Brody, B. (1970). Short-term visual and auditory memory disorders after parietal and frontal lobe damage. Cortex 6_, 440-459. Campbell, D. T. & Stanley, J. C. (1963). Experimental and quasi experimental designs for research Chicago: Rand McNally College Publishing Company. Coatsworth, J. J., & Penry, J. K. (1972). General principles: Clinical efficacy and use. In D. M. Woodbury, J. K. Penry, & R. P. Schmidt (Eds.), Antiepileptic drugs (pp. 87-96). New York: Raven Press. Cherlow, D. G. & Serafetinides, E. A. (1976). Speech and memory assessment in psychomotor epileptics. Cortex 12 21-26. Corkin, S. (1965). Tactually guided maze learning in man: Effects of unilateral cortical excisions and bilateral hippocampal lesions. Neuropsychologia 3_, 339-352. Dekaban, A. S. & Lehman, E. J. B. (1975). Effects of different dosages of anticonvulsant drugs on mental performance in patients with chronic epilepsy. Acta Neurol ogica Scandinavica, 52 319-330. 100

PAGE 107

101 Delaney, R. C, Rosen, A. J., Mattson, R. H., & Novelly, R. A. (1980). Memory function in focal epilepsy: A comparison of nonsurgical, unilateral temporal lobe and frontal lobe samples. Cortex 16 103-117. Dikmen, S., & Matthews, C. G. (1977). Effect of major motor seizure frequency upon cognitive-intellectual functions in adults. Epilepsia 18, 21-29. Dikmen, S., Matthews, C. G. & Harley, J. P. (1977). Effect of early versus late onset of major motor epilepsy on cognitive-intellectual performance: Further considerations. Epilepsia 18 31-36. Dodrill, C. B. (1975). Di phenyl hydantoin serum levels, toxicity, and neuropsychological performance in patients with epilepsy. Epilepsia 16, 593-600. Dodrill, C. B. (1981). Neuropsychology of epilepsy. In S. B. Filskov & T. J. Boll (Eds.), Handbook of clinical neuropsychology (pp. 366395). New York: John Wiley & Sons. Dodrill, C. B., & Troupin, A. S. (1977). Psychotropic effects of carbamazepine in epilepsy: A double-blind comparison with phenytoin. Neurology 27, 1023-1028. Fedio, P., & Mirsky, A. F. (1969). Selective intellectual deficits in children with temporal lobe or centrencephalic epilepsy. Neurospychologia ]_, 287-300. Gannaway, D. J., & Mawer, G. D. (1981) Serum phenytoin concentration and clinical response in patients with epilepsy. British Journal of Clinical Pharamacology 12, 833-829. Gibbs, E. L., Gibbs, F. A. & Fuster, B. (1948). Psychomotor epilepsy. Archives of Neurology and Psychiatry 60, 331-339. Gibbs, F. A., Gibbs, E. L. & Lennox, W. G. (1938). Cerebral dysrhythmias of epilepsy. Archives of Neurology and Psychiatry 39 298-314. Glaser, G. H. (1972). Diphenyl hydantoin: Toxicity. In D. M. Woodbury, J. K. Penry, & R. P. Schmidt (Eds.), Antiepileptic drugs (pp. 219226). New York: Raven Press. Glaser, G. H. (1974). Epilepsy: Neuropsychological aspects. In S. Arieti (Ed.), American handbook of psychiatry (pp. 330-336). New York: Basic Books. Glaser, G. H. (1980). Treatment of intractable temporal lobe-limbic epilepsy (complex partial seizures) by temporal lobectomy. Annals of Neurology 8, 455-459.

PAGE 108

102 Glowinski, H. (1973). Cognitive deficits in temporal lobe epilepsy: An investigation of memory functioning. Journal of Nervous and Mental Disease' 157 129-137. ""'" Goodnight, J. H. (1979). ANOVA procedure. GLM procedure: Introduction and reference. In J. T. Hellwig,& K. A. Council (Eds.), SAS user's guide Raleigh, NC: SAS Institute, Inc. Guey, M., Charles, C, Coquery, C. Roger, J., & Soulayrol, R. (1967). Study of psychological effects of ethosuximide (Zarontin) on 25 children suffering from petit mal epilepsy. Epilepsia 8_, 129-141. Halliday, A. M. Davison, K. Browne, M. W. & Kreeger, L. C. (1968). A comparison of the effects on depression and memory of bilateral E.C.T. and unilateral E.C.T. to the dominant and nondominant hemispheres. British Journal of Psychiatry 114, 997-1012. Hecaen, H. & Albert, M. L. (1978). Human neuropsychology New York: Wiley. Hermann, B. P., & Whitman, S. (1984). Behavioral and personality correlates of epilepsy: A review, methodological critique, and conceptual model. Psychological Review 95 451-497. Kimura, D. (1963). Right temporal lobe damage: Perception of unfamiliar stimuli after damage. Archives of Neurology 8_, 264-271. Kirk, R. E. (1963). Experimental design: Procedures for the behavioral sciences Belmont, CA: Brooks/Cole Publishing Company. Klove, H., & Matthews, C. G. (1966). Psychometric and adaptive abilities in epilepsy with differential etiology. Epilepsia 7_, 330-338. Kolb, B., & Whishaw, I. Q. (1980). Fundamentals of human neuropsychology San Francisco: W. H. Freeman & Company. Kornetsky, C. (1972). The use of a simple test of attention as a measure of drug effects in schizophrenic patients. Psychopharmacologia 24, 99-106. Ladavas, E., Umieta, C. & Provincial!', L. (1979). Hemisphere-dependent cognitive performances in epileptic patients. Epilepsia 20 493-502. """ Lennox, W. G. (1960). Epilepsy and related disorders Boston, MA: Little, Brown. Loiseau, P., Strube, E., Broustet, D. Battellochi, S., Gomeni, C. & Morselli, P. L. (1980). Evaluation of memory function in a population of epileptic patients and matched controls. Acta Neurol ogica Scandinavica (Supplementum) 62 58-61.

PAGE 109

103 MacLeod, CM., Dekaban, A. S. & Hunt, E. (1978). Memory impairment in epileptic patients: Selective effects of phenobarbital concentration. Sc ience 202, 1102-1104. Matthews, C. G. & Harley, P. L. (1975). Cognitive and motor-sensory performance in toxic epileptic subjects. Neurology 25, 184-188. Matthews, C. G., & Klove, H. (1967). Differential psychological performances in major motor, psychomotor, and mixed seizure classifications of known and unknown etiology. Epilepsia 8, 117-128. Mayeux, R. Brandt, J., Rosen, J., & Benson, D. F. (1980). Interictal memory and language impairment in temporal lobe epilepsy. N eurolo gy, 30, 120-125. Meyer, V. (1959). Cognition changes following temporal lobectomy for relief of temporal lobe eDileDsy. AHA Archives o f Neuro logy an d Psychi atry, 81, 199-309. ~ Meyer, V., & Yates, A. J. (1955). Intellectual chanaes following temporal lobectomy for psychomotor epilepsy. Journal o f Neurology, Neuro surger y and Psychi atry, 18, 44-62. Millichap, J. G. (1972). General principles: Clinical efficacy and use. In D. M. Woodbury, J. K. Penry, & R. P. Schmidt (Eds.), Anti epileptic drugs (op. 97-101) New York: Raven Press. Milner, B. (-962). Laterality effects in audition. In VB. Mountcastle Inte rhemi spheric relations and cerebral dom inanc e (pp. 177-195;. Baltimore", MD: Johns HopTons Prsss, Milner, B. (1965). Visually guided maze learning in man: Effects of bilateral hippocampal, bilateral frontal, and unilateral cerebral lesions. Neuropsychologia 3_, 317-338. Milner, B. (1967). Brain mechanisms suggested by studies of temporal lobes. In F. L. Darley (Ed.), Bra in mecha n isms underlying speech and language. New York: Grune & Stratton. Milner, B. (1968). Visual recognition and recall after right temporal lobe excision in man. Neuropsychologia 6, 191-209. Mirsky, A. F. Primac, D. W. Marsan, C. A., Rosvold, H. E., & Stevens, J. R. (1960). A comparison of the psychological test performance of patients with focal and nonfocal epilepsy. Experimental Neurology 2, 75-89. Mogel S., & Satz, P. (1963). Abbreviation of the WAIS for clinical use: An attempt at validation. Journal of Clinical Psychology 19, 298300.

PAGE 110

104 Ojemann, G. A. (1978). Organization of short-term verbal memory in language areas of human cortex: Evidence from electrical stimulation. Brain and Language 5, 331-340. Ojemann, G., & Fedio, P. (1968). Effect of stimulation of the human thalamus and parietal and temporal white matter on short-term memory. Journal of Neurosurgery 29 51-59. Ojemann, G., Fedio, P., & Van Buren, J. (1968). Anomia from pulvinar and subcortical stimulation. Brain _91, 99-116. Peterson, L. R., & Peterson, M. J. (1959). Short term retention of individual verbal items. Journal of Experimental Psychology 58, 193-198. Quadfasel, A. F. & Pruyser, P. W. (1955). Cognitive deficit in patients with psychomotor epilepsy. Epilepsia 4, 80-90. Reynolds, G. E. (1974). Memory functions in temporal lobe epileptics: Comparison between groups with right and left unilateral foci. Dissertation Abstracts International 36, 921B. Reynolds, E. H., & Travers, R. D. (1974). Serum anticonvulsant concentrations in epileptic patients with mental symptoms: A preliminary report. British Journal of Psychiatry 124 440-445. Richens, A. (1976). Clinical pharmacology and medical treatment. In J. Laidlaw & A. Richens (Eds.), A textbook of epilepsy (pp. 185-247). New York: Churchill Livingstone. Rosvold, H. E., Mirsky, A., Sarason, I., Bransome, E. D., Jr., & Beck, L. H. (1956). A continuous performance test of brain damage. Journal of Consulting Psychology 20, 343-350. Russell, E. W. (1981). The pathology and clinical examination of memory. In S. B. Filskov & T. J. Boll (Eds.), Handbook of clinical neuropsychology (pp. 287-319). New York: John Wiley & Sons. Samuels, I., Butters, N., & Fedio, P. (1972). Short term memory disorders following temporal lobe removals in humans. Cortex 8, 283-298. Samuels, I., Butters, N., & Goodglass, H. (1971). Visual memory deficits following cortical and limbic lesions: Effect of field of presentation. Physiology and Behavior 6, 447-452. Satz, P., & Mogel, S. (1962). An abbreviation of the WAIS for clinical use. Journal of Clinical Psychology 18 77-79. Scott, D. F., Moffett, A., Mathews, A., & Ettlinger, G. (1967). The effect of epileptic discharges on learning and memory in patients. Epilepsia 8, 188-194.

PAGE 111

105 Seashore, C. E., Lewis, D. & Saetviet, J. G. (1960). Seashore measures of musical talents: Manual Psychological Corporation. Shaffer, J. W. (1979). On the analysis of repeated measures over time in medical, pharmacological, and behavioral research. Journal of Behavioral Medicine 2, 221-238. Sommerbeck, K. W., Thielgaard, A., Rasmussen, K. E., Lohren, V., Gram, L., & Wulff, K. (1977). Valproate sodium: Evaluation of so-called psychotropic effect: A controlled study. Epilepsia 18 159-167. Tatsuoka, M. M. (1971). Multivariate analysis: Techniques for educational and psychological research. New York: John Wiley & Sons. Thompson, P., Huppert, F., & Trimble, M. (1980). Anticonvulsant drugs, cognitive function, and memory. Acta Neurologica Scandinavica (Supplementum) 62 75-81. Thompson, P., Huppert, F. A., & Trimble, M. (1981). Phenytoin and cognitive function: Effects on normal volunteers and implications for epilepsy. British Journal of Clinical Psychology 20 155-162. Thompson, P., & Trimble, M. R. (1980). Comparative effect of anticonvulsant drugs on the memory test performance of non-epileptic subjects. Acta Neurologica Scandinavica (Supplementum) 62 26. Thompson, P. J., & Trimble, M. R. (1981). Further studies on anticonvulsant drugs and seizures. Acta Neurologica Scandinavica (Supplementum) 64, 51-57. "~ Thompson, P. J., & Trimble, M. R. (1982). Anticonvulsant drugs and cognitive functions. Epilepsia 23 531-544. Tomlinson, L., Stirling, N., Merrifield, E., & Reynolds, E. H. (1981). Recognition memory in treated epileptics. Acta Neurologica Scandinavica (Supplementum) 64 43-50. Trimble, M. (1981). Anticonvulsant drugs, behavior, and cognitive abilities. In W. B. Essman & L. Valzelli (Eds.), Current developments in psychopharmacoloqy New York: Spectrum Publications. Trimble, M. R., & Reynolds, E. H. (1976). Anticonvulsant drugs and mental symptoms: A review. Psychological Medicine 6, 169-178. Trimble, M. R., & Thompson, P. J. (1981). Memory, anticonvulsant drugs and seizures. Acta Neurologica Scandinavica (Supplementum) 64 31-39.

PAGE 112

106 Walsh, K. W. (1978). Neuropsychology: A clinical approach New York: Churchill Livingstone. Waugh, N. C. & Norman, D. A. (1965). Primary memory. Psychological Review 72, 89-140. Wilder, B. J., & Brum', J. (1981). Seizure disorders: A pharmacological approach to treatment New York: Raven Press. Wilder, B. J. & Ramsay, R. E. (1974). Psychological aberrations associated with antiepileptic drugs. Clinical Electroencephalography 5, 199-200. Witmer, L. R. (1935). The association value of three-place consonant syllables. Journal of Genetic Psychology 47 337-359.

PAGE 113

BIOGRAPHICAL SKETCH Marianne L. D. McGuigan was born on February 21, 1935, in St. Louis, Missouri. She attended and graduated from Incarnate Word Academy and from St. John's Hospital School of Nursing, an affiliate institution of St. Louis University. After marrying, she continued to practice as a registered nurse in Pennsylvania and Washington state, where she is currently licensed. For some years she shared a domestic odyssey, enriched along the way by the births of three children. She resumed her education in Gainesville, Florida, receiving a B.A. with high honors from the University of Florida in 1977. Since completing the requirements for the M.S. degree in the University of Florida graduate program in clinical psychology in 1980, she has served a clinical internship and added research activity to her earlier experience in psychological assessment and therapy. She has also returned to single life, now enriched by three young adults. She is a member of Phi Beta Kappa, Phi Kappa Phi, and several professional organizations. 107

PAGE 114

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. ': J^CS. U IEileen B. Fennel 1, Chairman Associate Professor of Clinical Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. /$/$
PAGE 115

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. This dissertation was submitted to the Graduate Faculty of the College of Health Related Professions and to the Graduate School, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August 1984 (UL 6 KU&L***J£ Dean, College of Health Related Professions Dean for Graduate Studies and Research

PAGE 116

UNIVERSITY OF FLORIOA 3 1262 08666 359 7


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E8XTYA9HH_O3KDJ6 INGEST_TIME 2017-07-11T22:09:22Z PACKAGE UF00101071_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES