exploring the evaluation of antiepileptic drug change in people with intellectual disabilities and...
TRANSCRIPT
Epilepsy&
Epilepsy & Behavior 5 (2004) 58–66
Behavior
www.elsevier.com/locate/yebeh
Exploring the evaluation of antiepileptic drug change in peoplewith intellectual disabilities and high-frequency epilepticseizures: seizure control and sustained responsiveness to
the environment
Clare Smith, Mike Kerr, David Felce,* Helen Baxter, Kathy Lowe, and Andrea Meek
Welsh Centre for Learning Disabilities Applied Research Unit, University of Wales College of Medicine, Cardiff, Wales, UK
Received 22 August 2003; revised 20 October 2003; accepted 21 October 2003
Abstract
Purpose. Optimum antiepilepsy medication should be successful in reducing seizures with minimal adverse effects on the patient�sability to concentrate or general level of awareness. The purpose was to investigate the potential of a method of measuring re-
sponsiveness to environmental events as a means of reflecting awareness levels among people with intellectual disabilities undergoing
review of medication for high-frequency epileptic seizures.
Methods. Observations of 22 participants referred to a specialist clinic were conducted three times a month over a 5-month period
following the initial baseline measures and clinical intervention. Behavioral responsiveness was measured by calculating the like-
lihood of appropriate activity occurring given the occurrence of staff interaction. This likelihood was represented by the statistic
Yule�s Q. Seizure frequency was also evaluated.
Results. Participant responsiveness after drug review was similar to baseline indicating an absence of long-term adverse effects.
Participants experienced a significant decrease in seizure frequency.
Conclusion. It was concluded that drug review led to seizure reduction while behavioral measurement confirmed no loss of
responsiveness.
� 2003 Elsevier Inc. All rights reserved.
Keywords: Intellectual disabilities; Medication; Outcome; Behavioral responsiveness
1. Introduction
There are a number of people within the general
population who experience difficulties in communica-
tion, such as very young children, people with Alzhei-mer�s disease, and people with intellectual disabilities.
Where such problems are acute, personal well-being and
state of mind are largely inferred from people�s ob-
servable behavior, with reliance on second-hand re-
porting from family or carers being common. Such
reliance can call into question the validity of the infor-
mation gathered on the efficacy of professional inter-
vention [1]. This problem has been particularly
* Corresponding author. Fax: +029-20610812.
E-mail address: [email protected] (D. Felce).
1525-5050/$ - see front matter � 2003 Elsevier Inc. All rights reserved.
doi:10.1016/j.yebeh.2003.10.018
highlighted where epilepsy and intellectual disabilities
occur together due to the high prevalence of the condi-
tion among people with intellectual disability and the
additional difficulties that this comorbidity causes.
The frequency of epilepsy occurring in people withintellectual disability is much higher than in the general
population and seems to increase with the severity of
disability [2–4]. Epidemiological studies suggest that as
many as one-fifth of the population of people with in-
tellectual disabilities have epilepsy [5–8] and reported
that people with such a ‘‘dual disability’’ showed poorer
life skills than their peers without epilepsy, and that
behavioral disturbances, such as aggression and self-in-jury, were particularly associated with frequent seizures
and with antiepileptic polytherapy. The coexistence of
epilepsy and intellectual disability also seems, as one
C. Smith et al. / Epilepsy & Behavior 5 (2004) 58–66 59
would expect, to be a factor associated with the impactof caring. Espie et al. [9] found that carers experience
strain and that family carers may be particularly prone
to experience clinically significant levels of anxiety and
depression.
Both conditions predispose to earlier mortality [10].
People with an intellectual disability are known to have
a lower life expectancy than the general population with
an estimated standardized mortality ratio (SMR) of 1.6[11]. The probability of survival decreases as the severity
of intellectual disability increases [12]. This higher risk
of mortality is increased for people with coexisting epi-
lepsy whose SMR may be as high as 5 [11,13,14]. With a
condition occurring with such frequency and with such
consequences for those concerned, it is possible to see
why investigation in this area is badly needed.
The introduction of new medication is a particularissue for professionals and families alike because of the
fear of undesirable side effects, such as suppression of
arousal [15]. In treating epilepsy, doctors strive to pre-
scribe the right type and level of medication to be suc-
cessful in reducing seizures but with minimal adverse
effects on other aspects of the patients� lives, such as
their level of awareness or ability to concentrate. How-
ever, achieving such a balance in managing epilepsy inpeople with intellectual disabilities is made more difficult
because of a number of problems.
It is common for there to be a higher than desirable
frequency and severity of seizures refractory to treat-
ment [16]. This may mean that patients have a greater
vulnerability to any cognitive side effects if dosages are
increased to maximize seizure control. The situation is
further complicated by the high prevalence of psychi-atric and behavioral disturbance in people with intel-
lectual disabilities [17]. A large proportion of people
with intellectual disabilities experience long-term expo-
sure to drug treatment for behavioral problems [18–21].
This may lead to involuntary body movements, such as
tardive dyskinesia, which can look like seizure activity
[22].
Moreover, the assessment of the efficacy of medica-tion can be hampered by confusion between general
behavioral abnormalities and behavior problems caused
by epilepsy or the antiepileptic medication itself [3,23].
Early observations found that approximately two-thirds
of people with severe intellectual disabilities living in
institutions displayed stereotypical behavior including
head movements, rocking, and jerking [24,25]. Differ-
entiating this behavior from the pattern of behaviortypically seen in complex partial seizures, particularly
when there are associated ictal or postictal automatisms,
is complex.
Current measures used to assess the effectiveness of
medication, such as seizure diaries and seizure severity
scales, make it comparatively easy to judge whether im-
provements have occurred in the number and level of
seizures a person has experienced. However, it is oftenless easy for carers to notice other, more subtle changes in
the person�s state of alertness and behavior. It is unlikely
that the commonly used primary outcome measure, sei-
zure reduction, adequately assesses such quality-of-life
outcomes. A way of measuring a patient�s responsivenessto environmental stimuli would complement existing
evaluation approaches and give practitioners a more ac-
curate overview of medication effects.Observational methods for studying human behavior
and the interactions people have with others have made
a substantial contribution to the evaluation of services
for people with severe intellectual disability [26,27]. The
extent to which people are engaged in activity has been
measured reliably. The measure has been shown to
discriminate between institutional and community ser-
vices [26], day services and work [28], and people withlower and higher assessed adaptive behavior [29]. Ob-
served activity has also been shown to correlate signifi-
cantly with scores on staff reported activity scales [30].
In addition, observational methods have been used to
evaluate the impact on resident activity of changes in the
way staff attend to residents brought about by staff
training [31,32].
The application of palmtop computers to behavioralmeasurement has permitted real-time multiple-category
observation to be undertaken in real-world settings [33].
Sequential analysis [34] allows behavior occurring in a
period of time to be analyzed in relation to the envi-
ronmental states that precede or follow it. A person�sresponse to a defined environmental stimulus can be
represented by the conditional probabilities of the re-
sponse occurring or not given the occurrence and non-occurrence of the stimulus in question [35]. Such
probabilities divided one by the other constitute an odds
ratio, a measure of the size of effect that the stimulus has
on the person�s behavior. For example, Felce et al. [36]
were interested in whether staff became more effective at
assisting residents to engage in activities after they had
been given training. The odds ratio of resident engage-
ment in activity given the receipt of assistance aftertraining was compared with that before training. A
significant increase was interpreted as showing that staff
had become more effective at assisting residents. In other
words, a changed quality of assistance had caused an
increased probability of the resident response, engage-
ment in activity.
This approach to evaluating the responsiveness of
people with intellectual disabilities may have applicationbeyond analyzing the impact of changed environmental
stimuli. Responsiveness to unchanged environmental
stimuli but under different types or levels of medication
may provide a methodology for assessing the behavioral
impact of drug treatment in this population. The pur-
pose of this study was to explore the application of this
approach to studying the behavioral consequences of
60 C. Smith et al. / Epilepsy & Behavior 5 (2004) 58–66
antiepileptic medication review among adults withintellectual disabilities and severe epilepsy.
The aims of the present study were (1) to assess the
impact of antiepileptic drug change in terms of not only
seizure control but also sustained responsiveness to en-
vironment stimuli, and (2) to explore the association
between change in seizure frequency and such respon-
siveness.
2. Method
2.1. Participants and settings
A specialist outpatient epilepsy unit at the University
of Wales College of Medicine served a population of
approximately 1.5 million people in South Wales andreceived referrals of adults with intellectual disability
and severe epilepsy. It was intended to recruit 20 par-
ticipants, a number determined by practical consider-
ations concerning volume of data collection and power
to demonstrate statistical significance. Participants were
recruited prospectively during a 12-month period using
the following inclusion criteria: (1) a change in antiepi-
leptic medication was clinically indicated, and (2) in-formed consent and/or assent were given. Twenty-two
adults met these criteria. Full data sets were collected for
19; 3 people withdrew from treatment review during the
course of the study.
Six of the 19 participants lived in supported housing
and 13 lived in their family home. Ten participants were
male and the mean age of the sample was 32 years
(range, 18–46 years). Mean score on the Liverpool Sei-zure Severity Scale [37] at baseline was 31 (range, 8–69).
According to the Disability Assessment Schedule [38] 12
participants could not walk independently, 9 were
nonverbal, and 11 were not able to feed, wash, and dress
themselves. Six had severe challenging behavior and 11
were severely incontinent.
Mean score on Part One of the Adaptive Behavior
Scale [39] at baseline was 92 (range, 29–242), and on theAberrant Behavior Checklist [40], 40 (range, 4–78).
Mean score on the Psychopathology Instrument for
Mentally Retarded Adults [41] at baseline was 9 (range,
2–17). This assessment has eight sections and a score of
4 or more on each suggests caseness in relation to the
diagnostic category in question. Overall, 8 participants
were assessed at threshold levels in relation to one or
more categories. In addition, 11 were assessed as havingthe triad of social impairments characteristic of autistic
spectrum disorder [42]. Lower Adaptive Behavior Scale
scores were significantly associated with the presence of
the triad of social impairments (r ¼ �0:61, P < 0:01).Otherwise, correlations between the behavioral assess-
ments or between them and psychiatric caseness were
not significant.
2.2. Research design
Recruitment for the study was staggered over 24
months and each participant was studied for 6 months
including an initial baseline assessment and five re-
peated assessments at monthly intervals after clinical
intervention began. Clinical consultants remained in
charge of the drug treatment for all the participants
throughout the study and they were free to change themedication at any time, including stopping the intro-
duction of a new drug or the changed level of a current
drug. During the course of the study three participants
had their current drugs reduced, one had them in-
creased, and 15 had new drugs introduced. Details of
the medication changes for each participant during the
5 months after clinical intervention began can be found
in Table 1.
2.3. Measurement
2.3.1. Participant characteristics
Information on each participant�s adaptive behavior,
social impairment, psychopathology, and challenging
behavior was collected using well-established reliable
and valid assessment scales. Such characteristics arecommonly employed to define intellectual disability re-
search populations and have been recommended as re-
quired subject descriptors in epilepsy research [23]. Data
were collected by interviewing family members or direct
care staff who knew the person well using the Disability
Assessment Schedule (DAS) [38], the Adaptive Behavior
Scale Part One (ABS) (Community and Residential
Version) [39], and the Aberrant Behavior Checklist [40].Presence of psychopathology was assessed using the
Psychopathology Instrument for Mentally Retarded
Adults [41]. Items from the DAS relating to participants�language and means of communication, the quality of
their social interaction, and the extents to which they
engaged in symbolic activities, stereotypic behavior, and
echolalia were used to assess whether participants had
the triad of social impairments characteristic of autisticspectrum disorder [42].
2.3.2. Seizure data
Data on seizure frequency and severity (using the
Liverpool Seizure Severity Scale) were obtained by ca-
rers completing structured seizure diaries from baseline
to 5 months after drug change.
2.3.3. Participant activity and receipt of attention from
carers
Responsiveness to the environment was conceptual-
ized as behavior given receipt of attention from carers.
The behavior of each participant and the attention he or
she received were therefore observed directly. Observa-
tions were conducted three times each month for 6
Table 1
Details of participant�s drug intervention and dosage during post intervention observations
Client Drug intervention Seizure types Month 1 Month 2 Month 3 Month 4 Month 5
01 Introduction of topiramate alongside
current medication: lamotrigine 575mg
Tonic–clonic, atonic,
complex partial
25mg twice
a day
25mg AMAM
50mg PMPM
50mg twice
a day
50mg twice
a day
50mg twice
a day
02 Reduction in tiagabine Tonic–clonic, atonic,
tonic
5mg twice
a day
5mg daily Discontinued Discontinued Discontinued
03 Introduction of lamotrigine alongside
current medication: primidone 125mg
Absence, tonic–clonic,
myoclonic
25mg every
other day
25mg twice daily 50mg twice daily 50mg twice daily 50mg twice
daily
04 Increase in lamotrigine Tonic, tonic–clonic,
unclassified
300mg AMAM
200mg PMPM
300mg AMAM
200mg PMPM
300mg AMAM
200mg PMPM
300mg AMAM
200mg PMPM
300mg AMAM
200mg PMPM
05 Introduction of topiramate alongside
current medication: lamotrigine 450mg
Tonic, tonic–clonic,
absence, unclassified
25mg twice
a day
25mg AMAM
50mg PMPM
75mg AMAM
75mg PMPM
75mg AMAM
75mg PMPM
75mg AMAM
100mg PMPM
06 Introduction of lamotrigine alongside
current medication: clonazepam 2mg
Myoclonic jerks 25mg every
other day
50mg daily 75mg daily 75mg daily 75mg daily
07 Introduction of lamotrigine alongside
current medication: vigabatrin 2 g,
carbamazepine 200mg
Tonic–clonic, myoclonic,
tonic, unclassified
25mg daily 75mg daily 100mg AMAM
25mg PMPM
100mg AMAM
75mg PMPM
100mg AMAM
100mg PMPM
08 Introduction of lamotrigine alongside
current medication: carbamazepine
700mg, diazepam 20mg, pericyazine
20mg
Tonic–clonic, complex
partial, unclassified
25mg daily 75mg daily 100mg AMAM
25mg PMPM
100mg AMAM
75mg PMPM
100mg AMAM
100mg PMPM
09 Introduction of topiramate alongside
current medication: sodium valproate
2 g and lamotrigine 300mg
Tonic–clonic, atonic,
unclassified
25mg PMPM 25mg AMAM
25mg PMPM
25mg AMAM
50mg PMPM
25mg AMAM
50mg PMPM
25mg AMAM
50mg PMPM
10 Introduction of lamotrigine Tonic, complex partial 25mg every
other day
25mg daily 25mg daily 25mg daily 25mg daily
11 Introduction of topiramate alongside
current medication: sodium valproate
1400mg, gabapentin 2700mg, clobazam
10mg, diazepam 2mg
Tonic–clonic, tonic,
absence, secondary
generalized
25mg daily 25mg twice
a day
50mg daily 50mg twice
a day
50mg twice
a day
12 Introduction of topiramate alongside
current medication: lamotrigine 600mg
and phenobarbital 30mg
Tonic–clonic, tonic,
absence, complex
partial
25mg PMPM 25mg AMAM
50mg PMPM
50mg twice
a day
50mg twice
a day
50mg twice
a day
13 Reduction in gabapentin Tonic–clonic, tonic,
myoclonic jerks, atonic
800mg
400mg
800mg
400mg
400mg
800mg
400mg
400mg
400mg
400mg PMPM Discontinued
14 Introduction of topiramate alongside
current medication: carbamazepine
1200mg, vigabatrin 2 g, clobazam 10mg
Tonic–clonic, tonic,
unclassified
25mg daily 25mg AMAM
50mg PMPM
50mg twice daily 50mg twice daily 50mg twice
daily
15 Reduction in phenytoin alongside current
medication: carbamazepine 400mg, 600mg:
primidone 3mg; phenytoin 300mg
Tonic–clonic 275mg daily 250mg daily 225mg daily 200mg daily 175mg daily
16 Introduction of lamotrigine alongside
current medication: carbamazepine 200mg;
400mg; sodium valproate 2 g
Tonic–clonic 25mg every
other day
50mg daily 100mg daily 100mg daily 100mg daily
C.Smith
etal./Epilep
sy&
Behavio
r5(2004)58–66
61
Table
1(continued)
Client
Drugintervention
Seizure
types
Month
1Month
2Month
3Month
4Month
5
17
Introductionoflevetiracetam
alongside
currentmedication:sodium
valproate
1700mg,lamotrigine225mg
Tonic–clonic,
tonic,
atonic
250mgdaily
250mg
AM
AM
500mg
PM
PM
500mgtw
ice
aday
500mgtw
ice
aday
500mgtw
ice
aday
18
Introductionoflamotrigine
alongsidecurrentmedication:
phenytoin
200mg,sodium
valproate
500mg,chlorpromazine75mg
Tonic–clonic,absence
25mgevery
other
day
25mgdaily
50mgdaily
75mgdaily
100mgdaily
19
Introductionoftopiramate
alongside
currentmedication:sodium
valproate
2g,
lamotrigine250mg,Diazepam
5mg
Generalizedtonic–clonic
25mgevery
other
day
50mgdaily
100mgdaily
100mgdaily
100mgdaily
62 C. Smith et al. / Epilepsy & Behavior 5 (2004) 58–66
months, including an initial baseline assessment and 5monthly assessments following clinical intervention.
Observations took place within the participant�s servicesettings (residence or day placement). Each observation
session lasted 2 hours. The location and time of day of
each observation session were held constant for each
participant but varied across them. The observations
themselves were conducted using Psion palmtop com-
puters programmed for real-time, multiple-category,data capture [43]. It was thus possible to produce a
complete record of participant behavior and the carer
attention each received during the observed period. One
key was designated for each observational code. In
general, keys acted independently allowing for concur-
rency of behavior. One key depression signaled the onset
of a code and a second depression signaled its termi-
nation. The data were stored within the computer in anarray that gave the time (in seconds) of the onset and the
termination of each behavior. Data were transferred to
an IBM-compatible personal computer for analysis.
Four carer and five participant behaviors were
measured.
Carer attention:
(i) Verbal instruction: spoken or signed instruction to
perform an activity.(ii) Prompting/demonstration: gestural or physical
prompting to perform an activity.
(iii) Physical guidance: direct physical help to do an
activity.
(iv) Conversation: spoken or signed interactions nei-
ther encouraging nor discouraging of activity (e.g.,
pleasantries).
Participant behaviors:(i) Social engagement: spoken, signed, gestured, or
other attempts to gain or retain the attention of another
person (except by challenging behavior) or the giving of
attention, as evidenced by eye contact or orientation of
the head, to another person who is reciprocally inter-
acting.
(ii) Domestic engagement: getting ready for, doing, or
clearing away a household or gardening activity.(iii) Personal engagement: getting ready for, doing, or
clearing away a self-help or personal activity.
(iv) Other engagement: getting ready for, doing, or
clearing away a recreational activity or educational ac-
tivity.
(v) Other: all other behavior including no activity,
unpurposeful activity, and challenging behavior (self-
injury, aggression to others, damage to property, ste-reotypy, and all other inappropriate behavior).
2.4. Analysis
The three observation sessions per participant each
month were arranged as a single file set. The following
operations were performed on each of the six file sets
C. Smith et al. / Epilepsy & Behavior 5 (2004) 58–66 63
(baseline and months 1-5 post medication review) foreach participant using a program for sequential analysis
called Harclag [43]:
1. the category total participant engagement in activity
was created by combining all occurrences of social,
domestic, personal and other engagement and its extent
calculated
2. the category receipt of carer attention was created
by combining all occurrences of verbal instruction,prompting or demonstration, physical guidance and
other conversation, and its extent calculated
3. the conditional probability of total participant
engagement in activity occurring within a 5-second lag
after receipt of carer attention was calculated.
The quantities in 1-3 above were then used to con-
struct a 2� 2 contingency table for the occurrence or
otherwise of total participant engagement in activitygiven the receipt or otherwise of carer attention (see
Fig. 1), as a prelude to the calculation of odds ratios and
Yule�s Q.
The application of odds ratios to the analysis of re-
lationships between events in observational data has
been described by Bakeman [44]. For a generic 2� 2
contingency table as illustrated in Fig. 1 with first row
quantities a and b and second row quantities c and d, theodds ratio for the variable defining the first column
occurring given the variable defining the first row oc-
curring is given by a=b divided by c=d which equals
ad=bc. This may be investigated as a concurrent asso-
ciation (lag zero) or lagged in time.
Odds ratios vary from 0 (perfect negative relation-
ship) through 1 (no relationship) to infinity (perfect
positive relationship). Yule�s Q is an arithmetic trans-formation [ðad � bcÞ=ðad þ bcÞ] which preserves the
rank ordering of the data and establishes a more con-
ventional range to the index so that 1 depicts a perfect
negative relationship, 0 no relationship, and +1 a perfect
positive relationship [45].
Yule�s Q was not calculable where there was no
participant engagement in activity or interaction from a
carer. Differences between baseline and postbaselinewere tested for significance using a Wilcoxon matched
pairs signed-ranks test [46].
Fig. 1. Two-by-two contingency table for total participant engagement
in activity given receipt of carer attention.
2.5. Reliability
Interobserver agreement was assessed by two re-
searchers observing 42 sessions simultaneously.
Agreement was defined by observers having the same
times for the onset and offset of codes within a tol-
erance of 4 seconds. Level of agreement was calcu-
lated for each code by using Cohen�s j modified to
allow for the setting of the tolerance [47]. This pro-vides an estimate of agreement corrected for chance
agreement. Summary j values were calculated as an
average weighted for the occurrence of the behavioral
category in question. j values for verbal instruction,
prompting/demonstration, physical guidance, and
other conversation were 0.78, 0.71, 0.85, and 0.70,
respectively. j values for participant engagement in
personal engagement, other engagement, and socialengagement were 0.87, 0.83, and 0.83, respectively.
There were insufficient amounts of domestic activity to
calculate a j score. Suen and Ary [48] suggest that a jvalue of 0.6 or higher is acceptable for observational
research. The above figures suggest that the occur-
rence of verbal instruction, prompting, physical
guidance, and other conversation from carers could
be reliably distinguished, as could participantengagement in personal, other nonsocial, and social
activity.
3. Results
At baseline, Yule�s Q for engagement occurring gi-
ven carer interaction was positive for 18 of the 19participants, indicating that carer attention made par-
ticipant engagement in activity more likely (baseline
mean, 0.77; range, )0.20–0.99). Five months following
drug intervention Yule�s Q had increased for 10 of the
19 participants, stayed the same for 2, and decreased
for 6. Yule�s Q was not calculable for 1 person.
Overall, change in Yule�s Q 5 months after antiepi-
lepetic drug change was not significant. However, therewas a significant increase in Yule�s Q for participants
with a baseline Yule�s Q below the baseline mean
(z ¼ �1:997, P < 0:05).The mean number of seizures experienced per month
at baseline was 75 (range, 2–577). Sixty-eight percent of
participants had in excess of an average of 2 seizures a
week. Five months postintervention, 16 people experi-
enced a decreased seizure frequency. The postinterven-tion mean (59; range, 0–499) was significantly lower
(z ¼ �2:68, P < 0:01).Across the entire sample, there was no significant
association between increases in Yule�s Q and decreases
in seizure number. However, there was a significant as-
sociation between these quantities for people with
challenging behavior and for people with threshold
64 C. Smith et al. / Epilepsy & Behavior 5 (2004) 58–66
levels of mental illness (r ¼ �0:808, P < 0:005, andr ¼ �0:708, P < 0:05, respectively).
Across the 6 months of study, change in Yule�s Q
and seizure frequency for each participant showed one
of three patterns (see Fig. 2). At baseline, subgroup 1
had marked seizure frequencies associated with low
behavioral responsiveness (i.e., low Yule�s Q). Behav-
ioral responsiveness increased through Months 2–5
accompanied by decreased seizure frequency. Bothreturned to baseline levels in Month 6. Overall, this
illustrates a short-term beneficial effect from antiepi-
leptic drug change. Subgroup 2 showed stability over
time. They had initially low seizure frequency associ-
ated with high responsiveness. Both were unaltered by
drug change. Subgroup 3 had an initial level of be-
havioral responsiveness inbetween those of the other
two groups associated with an intermediate seizurefrequency. Yule�s Q fell immediately after drug change
but returned to baseline by Month 6. Seizure fre-
quency fell progressively to low levels. Overall, this
illustrates a beneficial long-term effect of antiepileptic
drug change emerging after a short-term lowering of
responsiveness.
Fig. 2. Change in mean Yule�s Q and seizure frequency for three
groups: (1) participants 4, 10, 12, 14, and 20 whose Yule�s Q was
raised in Months 2–5; (2) participants 7, 9, 13, 15, and 17 whose
Yule�s Q stayed fairly constant throughout Months 2–6; and (3)
participants 2, 5, 6, 11, 16, and 22 whose Yule�s Q decreased in
Months 2–5.
4. Discussion
The aim of this study was to explore whether the
response of adults with severe learning disabilities and
high-frequency epileptic seizures to receipt of attention
from carers could be used as a sensitive means of rep-
resenting their responsiveness to the environment when
change in antiepileptic medication was undertaken to
review the effectiveness of seizure control. Sequentialanalysis was used to calculate the likelihood of partici-
pant engagement in activity following receipt of carer
attention, using a variant of the odds ratio, Yule�s Q.
The analyses showed a number of interpretable findings
that chime with clinical experience.
First, participant responsiveness after drug review
overall was similar to baseline, while the decrease in the
number of seizures participants experienced was signif-icant. This suggests that successful seizure reduction
following drug review was achieved with no loss of re-
sponsiveness. It is important to emphasize here that
treatment effectiveness need not result in an increase in
responsiveness among people whose responsiveness
is already high. The value of the behavioral analysis is
in providing a means of monitoring that there is not
an undue deterioration in functioning when medicationis changed.
Second, there was no association overall between
measures of responsiveness and seizure frequency but an
association was found for certain individuals. For peo-
ple regarded as having threshold levels of mental illness
and challenging behavior, reduced seizure frequency was
associated with increased responsiveness. Given that
these may be additional indicators of neurological ab-normalities, this finding is worth further investigation.
Third, the three patterns of change in responsiveness
and seizure frequency demonstrated reflect clinical ex-
perience. Responsiveness appeared lowered among
people with very frequent seizures refractory to treat-
ment. Drug change brought temporary beneficial change
to both quantities. However, improvements had disap-
peared by the sixth month. Responsiveness was at ceil-ing levels among people with infrequent seizures.
Change in medication did not affect either quantity.
Responsiveness was at a slightly lower level for people
experiencing an intermediate level of seizures. Increasing
control over seizures following medication change ap-
peared to be associated with a short-term deterioration
in responsiveness.
This methodology may be useful in the future in re-search studies aimed at differentiating effectiveness and/
or side effects of drugs or other therapeutic interventions
in people with severe communication problems. It,
however, has a number of limitations currently for
routine clinical practice: the volume of direct behavioral
observation is considerable and its analysis is time
consuming despite the availability of custom-made
C. Smith et al. / Epilepsy & Behavior 5 (2004) 58–66 65
computer programs. The lag analyses performed hererequire a certain minimal level of carer interactions from
which the conditional probability of participant re-
sponse can be calculated. Rather than relying on the
vagueries of natural environment assessment, it may be
possible to achieve this with a much lower level of ob-
servation by structuring a clinical assessment in such a
way as to ensure that the required carer interactions
occur. A more definitive demonstration of the method-ology utility would be required for such a development
to be worthwhile.
In conclusion, this study sought to demonstrate how
a behavioral measure can be employed to investigate
maintenance of behavioral functioning after antiepilep-
tic drug change in patients with intellectual disabilities.
In the past, measures such as seizure diaries and seizure
severity scales have made it comparatively easy to judgewhether improvements have occurred in the number and
level of seizures a person is experiencing. However, as-
sessing well-being and state of mind in this population
has typically been a problem. This study has shown that
a measure of behavioral responsiveness may be a useful
aid to interpretation when antiepileptic drug change is
undertaken in this population. Moreover, this investi-
gation has the validity associated with being conductedwithin normal clinical practice. What may work in this
context may prove more powerful in more tightly con-
trolled situations.
Acknowledgment
This study was funded by Wellcome Trust ProjectGrant #M/98/366.
References
[1] Espie C, Paul A. Epilepsy and learning disabilities. In: Cull C,
Goldstein LH, editors. The clinical psychologist�s handbook of
epilepsy. London: Routledge; 1997. p. 184–202.
[2] Richardson SA, Katz M, Koller H, McLaren J, Rubenstine B.
Some characteristics of a population of mentally retarded young
adults in a British city. J Ment Defic Res 1979;23:275–87.
[3] Corbett J. Epilepsy and mental handicap. In: Laidlaw J, Richens
A, Oxley JA, editors. Textbook of epilepsy. 3rd ed. Edinburgh:
Churchill Livingstone; 1993. p. 631–6.
[4] Kerr M, Fraser W, Felce D. Primary health care for people with a
learning disability: a keynote review. Br J Learn Disabil
1996;24:2–8.
[5] Welsh health survey 1995. Cardiff: Welsh Office; 1995.
[6] Espie CA, Pashley AS, Bonham KG, Sourindhrin I, O�Donovan
M. The mentally handicapped person with epilepsy: a comparative
study investigating psychosocial functioning. J Ment Defic Res
1989;33:123–35.
[7] Espie CA, Gillies JB, Montgomery JM. Anti-epileptic polyphar-
macy, psychosocial behaviour and locus of control orientation
among mentally handicapped adults living in the community. J
Ment Defic Res 1990;34:351–60.
[8] Gillies JB, Espie CA, Montgomery JM. The social and behav-
ioural functioning of people with mental handicaps attending
adult training centres: a comparison with those with and without
epilepsy. Ment Handicap Res 1989;2:129–36.
[9] Espie CA, Paul A, Graham M, Sterrick M, Foley J, McGarvey C.
The epilepsy outcome scale: the development of a measure for use
with carer�s of people with epilepsy plus intellectual disability. J
Intellect Disabil Res 1998;42:90–6.
[10] Morgan C, Scheepers M, Kerr M. Mortality in patients with
intellectual disability and epilepsy. Curr Opin Psychiatry
2001;14:471–5.
[11] Forgren L, Edvinsson S, Nystrom L, Blomquist HK. Influence of
epilepsy on mortality in mental retardation: an epidemiological
study. Epilepsia 1996;37:956–63.
[12] Herbst DS, Baird PA. Survival rates and causes of death among
persons with nonspecific mental retardation. Biomedical Aspects.
In: Berg JM, editor. Perspectives and progress in mental retarda-
tion vol II: biomedical aspects. Baltimore, MD: Univ. Park Press;
1984. p. 3–15.
[13] Brorson LO, Wranne L. Long-term prognosis in childhood
epilepsy: survival and seizure prognosis. Epilepsia 1987;28:
324–30.
[14] Wieseler NA, Hanson RH, Nord G. Investigation of mortality
and morbidity associated with severe self-injurious behavior. Am J
Ment Retard 1995;100:1–5.
[15] Espie CA, Kerr M, Paul A, et al. Learning disability and epilepsy:
a review of available outcome measures and position statements
on developmental priorities. Seizure 1997;6:337–50.
[16] Tobias ES, Brodie AF, Brodie MJ. An outcome audit at the
epilepsy clinic: results from 1000 consecutive patients. Seizure
1994;3:37–43.
[17] Brown S, Betts T, Chadwick D, Hall W, Shorvon S, Wallace S. An
epilepsy needs document. Seizure 1993;2:91–103.
[18] Aman M, Singh N. Pharmacological intervention. In: Matson J,
Mulick J, editors. Handbook of mental retardation. New York:
Pergamon; 1983. p. 317–37.
[19] Jacobson J. Problem behavior and psychiatric impairment with a
developmentally disabled population. III. Psychotropic medica-
tion. Res Dev Disabil 1988;9:23–38.
[20] Clarke DJ, Kelly S, Thinn K, Corbett JA. Psychotropic drugs and
mental retardation. J Ment Defic Res 1990;34:385–95.
[21] Branford D. A study of the prescribing for people with learning
disabilities living in the community and in national health service
care. J Intellect Disabil Res 1994;38:577–86.
[22] Paul A. Epilepsy or stereotypy? Diagnostic issues in people with
learning disabilities. Seizure 1997;6:111–20.
[23] Kerr M, Espie C. Learning disability and epilepsy: towards
common outcome measures. Seizure 1997;6:331–6.
[24] Berkson G, Davenport RK. Stereotyped movements of mental
defectiveness. 1. Initial survey. Am J Ment Defic 1962;66:849–52.
[25] Kaufman ME, Levitt HA. A study of three stereotyped behav-
iours in institutionalised mental defectives. Am J Ment Defic
1965;69:467–73.
[26] Emerson E, Hatton C. Deinstitutionalization in the UK and
Ireland: outcomes for service users. J Intellect Dev Disabil
1996;21:17–37.
[27] Felce D, Emerson E. Observational methods in assessment of
quality of life. In: Thompson T, Felce D, Symons F, editors.
Computer-assisted behavioral observation in developmental dis-
abilities. Baltimore, MD: Paul H. Brookes; 2000. p. 317–33.
[28] Kilsby M, Beyer S. Engagement and interaction: a comparison
between supported employment and day service provision. J
Intellect Disabil Res 1996;40:348–57.
[29] Felce D, Perry J. The extent of support for ordinary living
provided in staffed housing: the relationship between staffing
levels, resident characteristics, staff:resident interactions and
resident activity pattern. Social Sci Med 1995;40:799–810.
66 C. Smith et al. / Epilepsy & Behavior 5 (2004) 58–66
[30] Perry J, Felce D. Objective indicators of the quality of life: how
much do they agree with each other. J Community Appl Social
Psychol 1995;5:1–19.
[31] Jones E, Perry J, Lowe K, et al. Opportunity and the promotion of
activity among adults with severe mental retardation living in
community residences: the impact of training staff in active
support. J Intellect Disabil Res 1999;43:164–78.
[32] Jones E, Felce D, Lowe K, et al. Evaluation of the dissemination
of active support training in staffed community residences. Am J
Ment Retard 2001;106:344–58.
[33] Repp A, Felce D. A microcomputer system use for evaluative and
experimental behavioural research in mental handicap. Ment
Handicap Res 1990;3:21–32.
[34] Yoder PJ, Feurer ID. Quantifying the magnitude of sequential
association between events or behaviors. In: Thompson T, Felce
D, Symons F, editors. Computer-assisted behavioral observation
in developmental disabilities. Baltimore, MD: Paul H. Brookes;
2000. p. 317–33.
[35] Bakeman R, Quera V. Analysing interaction: sequential analysis
with SDIS and GSEQ. New York: Cambridge Univ. Press;
1995.
[36] Felce D, Bowley C, Baxter H, Jones E, Lowe K, Emerson E. The
effectiveness of staff support: evaluating active support training
using a conditional probability approach. Res Dev Disabil
2000;21:243–55.
[37] Baker GA, Smith DF, Dewey M, Morrow J, Crawford PM,
Chadwick DW. The development of a seizure severity scale as an
outcome measure in epilepsy. Epilepsy Res 1991;8:245–51.
[38] Holmes N, Shah A, Wing L. The disability assessment schedule: a
brief screening device for use with the mentally retarded. Psychol
Med 1982;2:879–90.
[39] Nihira K, Leland H, Lambert N. AAMR Adaptive Behavior
Scale—residential and community. 2nd ed. Austin, TX: Pro-Ed;
1993.
[40] Aman MG, Singh NN. The Aberrant Behavior Checklist. New
York: Slosson Educational Publications; 1986.
[41] Matson JL. The Psychopathology Instrument for mentally
retarded adults. Orland Park, IL: International Diagnostic Sys-
tems; 1988.
[42] Wing L, Gould J. Systematic recording of behaviors and skills of
retarded and psychotic children. J Autism Childhood Schizophre-
nia 1978;8:79–97.
[43] Emerson E, Reeves D, Felce D. Palmtop computer technologies
for behavioral observation research. In: Thompson T, Felce D,
Symons F, editors. Computer-assisted behavioral observation
methods for developmental disabilities. Baltimore, MD: Paul H.
Brookes; 2000. p. 47–59.
[44] Bakeman R. Behavioral observation and coding. In: Reis HT,
Judd CK, editors. Handbook of research methods in social
psychology. New York: Cambridge Univ. Press; 2000. p. 138–59.
[45] Bakeman R, McArthur D, Quera V. Detecting group differences
in sequential association using sampled permutations: log odds,
kappa, and phi compared. Behav Res Methods Instrum Comput
1996;28:446–57.
[46] Siegel S. Nonparametric statistics for the behavioral sciences.
Tokyo: McGraw-Hill Kogakusha; 1956.
[47] Reeves D. Calculating inter-observer agreement and Cohen�skappa on time-based observational data allowing for �natural�measurement error. Manchester: Hester Adrian Research Centre,
University of Manchester; 1994.
[48] Suen HK, Ary D. Analyzing quantitative behavioral observation
data. Hillside, NJ: Laurence Erlbaum; 1989.