Dosage effects and individual responsitivity to Methylphenidate in Attention Deficit Disorder

/ ChUdPsychol. Psychiat. Vol. 29, No. 4, pp. 453-475, 1988 0021-9630/88 $3.00 + 0.00Printed in Great Britain. Pergamon Press pic

© 1988 Association for Child Psychology and Psychiatry.

DOSAGE EFFECTS AND INDIVIDUAL RESPONSIVITY TOMETHYLPHENIDATE IN ATTENTION DEFICIT DISORDER

V. I. DOUGLAS,* R . G . BARR,* K . AMIN,* M . E . O'NEILL* and B. G. BRITTON*

Abstract—Effects of three dosages of methylphenidate (0.15, 0.30 and 0.60 mg/kg) were assessed in19 ADD-H children on a variety of cognitive, academic and behavioral measures in the laboratoryand the classroom. A predominant linear pattern of improvement was found across almost all measures.A slight decrease between 0.3 and 0.6 mg/kg on one cognitive task leaves ojjen the possibility thathigher dosages reduce stimulant effectiveness or cause decrements on some kinds of "high-level/highload" tasks. Response patterns of individual children varied considerably across meaisures. All childrenimproved on at least several measures. Results were interpreted as evidence for stimulant activationof self-regulatory processes.

Keywords: Attention Deficit Disorder, Methylphenidate, dosage effects, individual responsivity

INTRODUCTION

A SERIES OF reports from the laboratory of Sprague and Sleator (Brown & Sleator,1979; Sprague and Sleator 1973, 1975, 1977) raised concern about prescribing highdoses of stimulants to children demonstrating Attention Deficit Disorder withHyperactivity (ADD-H). Sprague and Sleator (1977) reported ''peak enhancementof learning" at 0.3 mg/kg of methylphenidate (MPH) and "a decrement in leaming"in children given higher doses. Since optimal improvement in social behavior occurredat 1.0 mg/kg of MPH, Sprague and Sleator also concluded that behavioral andcognitive changes demonstrate different dose response curves.

These conclusions require further investigation. Sprague and Sleator did not exploreeffects of dosages between 0.3 and 1.0 mg/kg in their most cited (1977) study. A fewinvestigators have reported gains on cognitive and academic measures using MPHdoses up to 0.6 or 0.8 mg/kg (Ballinger, Varley & Nolen, 1984; Gharles, Schain &Zelniker, 1981; Pelham, Bender, Gaddell, Booth & Moorer, 1985; Rapport, DuPaul,Stoner, Birmingham, & Masse, 1985a; Rapport, Du Paul, Stoner & Jones, 1986;Rapport, Stoner, Birmingham & Tucker, 1985b; Sebrechts, Shaywitz, Shaywitz,Jatlow, Anderson, & Gohen, 1986; Walker, 1982). There have been occasional reports,as well, of improved performance on dosages higher than 1.0 mg/kg (Gittelman-Klein,Klein & Feingold, 1983).

Accepted manuscript received 18 September 1987

*Departments of Psychology and Pediatrics, McGill University—Montreal Children's Hospital ResezirchInstitute, Canada.

Requests for reprints to: Professor V. I. Douglas, Department of Psychology, McGill University,1205 Docteur Penfield Avenue, Montreal, Quebec H3A lBl, Canada.

453

4 5 4 V. I. DOUGLAS et al.

The "decrement in learning" reported by Sprague and Sleator (1977) has alsocaused confusion. First, the short-term memory task used in their study did not requirethe acquisition of new learning over trials. Secondly, in most studies using maxinicddosages up to 1.0 mg/kg, performance did not decline significantly below placebolevels. Rather, gains occurred at moderate dosages and failed to occur at higherdosages, or moderate doses produced greater improvement than high doses (Brown& Sleator, 1979; Gan & Gantwell, 1982; Peeke, Halliday, Gallaway, Prael & Reus,1984; Pelham et al, 1985; Rapport et al, 1986; Sprague & Sleator, 1973). Thereis also one report of increased cognitive perseveration at 1.0 mg/kg of MPH, but thisresult was obtained in a very small pilot study (Dyme, Sahakian, Golinko & Rebe,1982).

The issue of whether the shapes of behavioral and cognitive dose-response curvesdiffer also remains unsettled. In a recent review, Rapoport and Zametkin (in press)concluded that cognitive and behavioral effects of stimulants may be "temporally,pharmacokinetically and possibly biochemically independent" (p. 5). On the otherhand, studies from the laboratories of Pelham (Pelham et al, 1985) and Rapport(Rapport et al, 1986a,b) have yielded similar, linear dose-response curves acrossa majority of measures in both domains.

Further research is required to resolve these issues. In addition, if reducedeffectiveness or detrimental cognitive effects at high dosages are confirmed, it willbecome important to understand the underlying processes involved. This willnecessitate identifying the specific task parameters associated with these changes. Apromising strategy involves varying the quality and quantity of task demands. Anumber of investigators have suggested that "high level" or "high load" tasks areparticularly effective for discriminating between ADD-H and normal children,identifying unfavorable responders to the stimulants or assessing dosage effects(Ackerman & Dykman, 1982; Goons, Peloquin, Klorman, Ryan, Bauer, Perlmutter& Salzman, 1981; Douglas, Barr, O'NeiU & Britton, 1986; Dyme et al, 1982; Rapportetal, 1985b; Swanson, 1985; Swanson & Kinsbourne, 1979; Weiss & Laties, 1962).

Douglas has argued for the central importance of a self-regulatory defect inADD-H and has stressed the critical role of stimulants in activating self-regulatoryprocesses (Douglas, 1983, 1984 a,b; Douglas et al, 1986). In a recent paper, Douglas(in press) suggested methods for manipulating task parameters so as to increase self-regulatory demands. "High level/high load" tasks make heavy demands for complex,deliberate, controlled, flexible or exhaustive cognitive processing. Consequently,Douglas recommended manipulating such task variables as complexity, depth,flexibility, duration, amount or speed of cognitive processing. It is noteworthy inthis regard that Sprague and Sleator (1977) obtained clearest evidence of a decrementat 1.0 mg/kg on the largest matrix of their picture recognition test.

In the study to be reported, effects of three dosages of stimulant medication werecompared in the same group of ADD-H children on a large battery of measuresadministered in the laboratory and the classroom. This made it possible to comparedose-response curves from the same children within and across the behavioral,academic and cognitive domains. The tasks also differed on the depth, duration,amount and flexibility of processing required. Several tasks were chosen to reflectsuch "high level" processes as the deliberate and flexible application of mnemonic

METHYLPHENIDATE IN ADD-H 455

Strategies, the deployment of exhaustive, analytic, perceptual search strategies, carefulplanning of consecutive choices within a complex stimulus array and bimanual controlof fine motor movements. Three of the tasks also made it possible to study effectsof task load or duration by comparing performance over trials. One of these wasspecifically designed to assess the acquisition of new learning over trisils.

The three dosages of MPH studied were 0.15, 0.3 and 0.6 mg/kg. We were reluctantto prescribe doses above 0.6 mg/kg per dose, particularly since our design requiredtwo doses per day, a regime commonly used in clinical practice. Reports of side effectsat higher doses raised ethical concerns (Sprague & Sleator, 1977; Winsberg, Kupietz,Sverd, Hungund & Young, 1982). We were concerned, too, that drug-induced physicalsymptoms might interfere with the children's performance and thus be confoundedwith more direct cognitive effects of the medication.

We reasoned that increasing the discriminative power of our measures wouldfacilitate detection of any negative effects of the higher dose, as well as subtle differencesbetween dosages. We attempted to achieve this by developing tasks composed ofmultiple items, tailoring difficulty levels to each child's abilities, carrying out repeattesting in each drug condition, preparing carefully matched sets of tasks for repeattesting and providing a pre-assessment session in which children received practiceon the tests. Gare was taken to ensure that both laboratory and school measures wereassessed during the period when drug effects were optimal. All medication wasadministered by our research staff. Such precautions have sometimes been neglectedin other studies.

Another purpose of the study was to examine response patterns of individual childrento stimulant medication across the various measures in our assessment battery.Although investigators typically report that 30-40% of ADD-H children fail to respondto the stimulants (Gonners, 1972; Loney, 1986; Swanson & Kinsbourne, 1979;Ullmann & Sleator, 1986), this conclusion is usually based on a single behavioralor cognitive measure. Analysis of data from an earlier study (Douglas et al, 1986)revealed that, although 25-35% of our subjects failed to respond to any one measure,the individual children who failed to show improvement varied greatly from measureto measure.

METHOD

SubjectsChildren were referred to the Hyperactivity Project at the Montreal Children's Hospital by pediatricians

or school personnel. Referral was based on the presence of inattentiveness, impulsivity, hyperactivity,restlessness, poor compliance and poor self-control. All children met criteria for a DSM-III diagnosisof Attention Deficit Disorder with Hyperactivity (American Psychiatric Association, 1980). Subjectsalso had to receive ratings from both mothers and teachers at or above 1.5 on the Hyperactivity Indexof the Revised Conners Rating Scales (Goyette, Conners & Ulrich, 1978). In all cases, ADD-H wasjudged to be the child's primary diagnosis. However, several subjects also met DSM-III criteria forother disorders. Of the 19 children, one met criteria for Conduct Disorder, eight for Oppositional Disorderand four for Overanxious Disorder.

Screening interviews were held with mothers and teachers to obtain information on additional criteria.Interviewers took detailed medical, developmental, academic and family histories to rule out childrenwith symptoms of psychosis and those with serious visual, auditory or language deficits. Also excluded

456 V. I. DOUGLAS et al.

were children diagnosed as brain-damaged and children whose restless behavior appeared to be primzurilyattributable to emotional problems or a stressful home environment.

Information was obtained on all medications taken by the children. One child had taken Ritalinpreviously and seven were receiving stimulsmt medication at the time of referral. Parent and teacherratings for these children were obtained when they were off medication. If a child met all criteria forthe study, the mother was given a detailed description of the project and informed consent was obtziined.

The sample chosen consisted of 17 males and 2 females. Ages varied from 7.0 to 13.1 yrs (mean8.72 yrs; SD=1.47). IQs ranged from 89 to 123 with a mean of 104.26 (SD = 8.35). Sixteen ofthe subjects attended regular classes; their grades varied from Grade 1 to 5. The remzdning three childrenwere attending special classes for children with leaming or behavioral problems. Social class variedfrom I to IV (Hollingshead & Redlich, 1958) with most families falling within the third or fourth levels.Mean score on the Teacher Hyperactivity Index was 2.12 (SD = 0.47); mean score on the ParentHyperactivity Index was 2.18 (SD = 0.23).

Experimental designChildren were seen in the laboratory for a screening session about a week before testing began. They

were given practice on all tasks in the battery and the appropriate level was established for each childon tasks graded for difficulty level. The seven children currently receiving stimulant medication weretaken off medication the day of screening and 48 hrs prior to the first testing day.

The total battery was administered on eight days over a two-week period. Individual testing tookplace in the mornings. Most children were tested at the laboratory, but because of transportationdifficulties, five were tested at their school. Each child worked with the same examiner, in the samesetting, for all eight sessions. For the other half of the testing day, subjects returned to their regularclassrooms, where teachers rated their behavior and administered two classroom tests (arithmetic andword discovery).

Drug treatment and determination of drug and test orderMedications were prepared in opaque gelatin capsules and administered in a double-blind manner.

Capsules containing placebo (100 mg lactose), 0.15, 0.30, and 0.60 mg/kg doses of methylphenidateHCL (Ritalin) were prepared individually for each subject, based on the child's weight. Drug orderwas determined by consecutive assignment to a randomly ordered list of 24 possible combinations ofthe four medication levels for each of the two testing weeks. No child received the same drug orderin both weeks.

Children ate breakfast at home, approximately 1 hr before taking the first capsule which wasadministered by the examiner. One hour after the child received the capsule, the examiner beganadministering the test battery which took approximately 2 to 2J4 hrs. The second capsule, identicalto the morning capsule, was administered by the examiner approximately 1 hr before the child returnedto his classroom for the afternoon. Average time between administration of the first and second capsulewas 3}/^-4hrs.

Prior to the study, 24 orders for the individually administered tasks were generated from a randomnumbers table, and subjects were consecutively assigned to a test order which remained the same forall testing days. All taisks were administered each day. The order of the two classroom tasks was alternatedacross children, and each child received the same order for all eight days.

Eight equivalent forms of the Paired Associates, Matching Familiar Figures and Word Discoverytasks, £ind 16 equivzdent forms of the Arithmetic tests (eight for laboratory testing and eight for schooltesting) were used in the experiment proper; one additional form of each of these tests was employedduring screening. Determination of the order by which equivalent forms were presented was decidedby consecutive assignment to one of the 24 random sequences generated a priori.

Procedure for classroom testingAn examiner met with each child's teacher to explain the importance of the classroom measures and

how to use them. The teacher was given eight Conners forms, eight ACTeRS forms, eight WordDiscovery tasks and eight Arithmetic tests, each labelled with the day it was to be used. The teacherwas asked to administer the two classroom tests at approximately the same time each afternoon at atime when children in the class were working at their desks. Importance of allowing the child to workindependently and observation of instructions regarding time allotted for each test were emphasized.


The teacher was asked to record the time the child began and finished each test. She was remindedto fill out the rating forms at the end of each afternoon and to base her ratings on the child's behaviorduring the entire afternoon. Since children spent mornings out of the classroom, teachers were ableto limit their observations to periods when drug effects were optimal. The examiner contacted the teacherseverzil times during the eight days of testing to ensure that instructions were being followed and thatforms were completed at the end of each day.

TESTS AND MEASURES

Behavioral measuresADD-H Comprehensive Teacher Rating Scale (ACTeRS). The ACTeRS was developed by Ullman, Sleater

and Sprague (1984) to measure ADD-H behaviors emphasized in DSM-III (American PsychiatricAssociation, 1980). The scale consists of 24 items with five to seven items assessing each of four categoriesof behavior (Attention, Hyperactivity, Social Skills and Oppositional Behaviors). Teachers are askedto rate the items on a 5-point scale (in which 1 represents "almost never" and 5 represents "almostalways"). Four additional items rated in the same manner provide information regarding the behaviorof classmates toward the child (Peer Acceptability) and the child's interactions with the teacher (Demandson Teacher's Time). Scores reflect mean ratings on each of the six ACTeRS categories.

Conners Hyperactivity Index (teacher and examiner administered). The Hyperactivity Index for the ConnersRevised Teacher Rating Scale (Goyette et al., 1978) was used to obtain behavioral ratings from eachchild's classroom teacher and from the examiner who tested the child. The teacher completed her ratingsat the end of each afternoon. The examiner completed hers at the end of each testing session. Thescore is based on the mean often items rated on a 4-point (0-3) scale. (The item "disturbs other children"was omitted from the examiner's ratings because it was not applicable to the individual testing situation.)

Effort ratings. At the end of each testing session the examiner made an overall effort rating based onthe amount of effort displayed by the child throughout the session. In addition, separate effort ratingswere assigned to reflect the child's effort on some of the individual tasks. Effort was rated on a 7-point(0-6) scale in which 0 indicated no effort, 3 indicated reasonable effort and 6 indicated excellent effort.

Academic tasksArithmetic task (teacher and examiner administered). The arithmetic tasks used by Douglas et al. (1986) were

lengthened to 60 items to avoid ceiling effects resulting from the higher dosages and multiple testingsessions used in the present study. Otherwise, the format and instructions for administering and scoringthe tasks were the same. Equivalent forms were developed for each of five difficulty levels. As in theearlier study, each child was assessed during a pretesting session to assess the highest level at which hecould solve the problems.

Word discovery task (teacher and examiner administered). This task was a slightly modified version of theword discovery task used by Douglas et al. (1986). The words from which the children were requiredto construct new words contained approximately 10 letters, of which three or four were vowels. Thetask was administered by the teacher in the classroom, using the procedures described by Douglas etal. (1986). During a pre-testing session the task was explained and the child was given practice formingnew words from longer words. Scoring procedures are described in Douglas et al. (1986).

Cognitive tasksReaction time task with delay interval (DR T). The child wore a set of earphones and was seated at a

table with a small response box directly in front of him. Mounted on the box was a 7.5 watt lightbulb which served as the reaction signal. The warning signal was a tone of 70 decibels. The child wasinstructed to press and hold down an acrylic response plate (4x6 cm). He was told that he would heara tone through the earphones and that this indicated that a light would soon come on. He was instructedto release the plate as quickly as possible when the light came on and then to press it down again until


the next light appeared. The examiner emphasized that the plate was to be released only for the light,not for the warning tone. The apparatus wais constructed so that the reaction signal would not be activatedunless the response plate was depressed.

The examiner sat across the table from the child behind a box containing a LCD quartz chronographand lights that indicated the onset and termination of the warning and reaction signals. The warningsigned marked the beginning of a 10 sec preparatory interval at the end of which the light (reactionsignal) came on. The chronograph was activated when the reaction signal went on and ran until thechild released the plate, measuring reaction time to the hundredth of a second. The examiner recordedreaction time and reset the timer after each trial. A total of 45 trials, with 5-sec intervals betweentrials, was administered.

Measures derived from the task included: mean reaction time over 45 trials; reaction time variability;number of amticipatory responses (responses made during the interval between the warning signal £indthe reaction signal); total numbers of "fiddling" responses (these usually involved fiddling with theequipment) and the examiner's effort rating.

Maze tracking task. This task was adapted from one developed by Humphries, Swanson, Kinsbourneand Yiu (1979). A maze pattern was drawn in red on a clear acetate sheet measuring 18 x 13 cm. Thepattern consisted of alternating ascending and descending alleyways, 2 mm in width, with an angleof incline of 45°. For scoring purposes, the maze was divided into six identical segments, each of whichconsisted of an 8 cm ascending and an 8 cm descending eilleyway. The acetate sheet was placed onthe screen of an "Etch-a-Sketch" toy (produced by Kelton Corporation for Peter-Austin Mfg). Situatedbelow the screen in the right and left comers were control knobs. The left knob controlled left-rightmovement of an etched line and the right knob controlled up and down movement of the same line.The child's task was to turn both knobs simultaneously in order to keep the line moving smoothly ina diagonal fashion through the alleyway. Turning the knobs alternately was not allowed. At the screeningsession, the child was given practice. He was told that, by turning the dials on the Etch-a-Sketch, hewas to make a line going through the zig-zag path from start to finish. He was told he could take asmuch time as he wished and that he should try to keep the line within the path. It was stressed thatif he went outside the path he had to get back inside right away. The examiner also explained thatthe two dials did different things and that the child had to use them both. With the dial on the right hecould only make the line go up and down. With the dial on the left he could only make the line go sideways.

An error was counted each time the line went outside the path drawn on the acetate. Time takento complete the maze and an effort rating were also recorded.

Matching Familiar Figures Test (MFF-20). In order to increase reliability, the 20-item revision (MFF-20),developed by Cairns and Cammock (1978) was used instead of Kagan's (1966) original task. Eightmatched versions of the MFF-20 were constructed. Versions for the first 4 days of testing consistedof the same target stimulus and the same choice stimuli as the original but location of the correct choiceamong the alternatives was changed each day. For the second 4 days, a modification of the MFF-20was developed by altering features in target and choice stimuli of the original version. Standardinstructions were used in administering the tasks.

Scores were based on total numbers of incorrect choices (over 20 cards) and mean latency of firstresponse choice. An effort rating was also recorded.

Self-Ordered Pointing Task. A shortened and modified version of the Self-Ordered Pointing Task developedby Petrides and Milner (1982) was administered on each of the eight days of testing. There were twoforms of the task, a Representational form and an Abstract form. Half of the children received theRepresentational form the first week and the Abstract form the second week; the other half receivedthe opposite order.

Four sets of each form were developed, one for each of the testing days. The Representational versionwas constructed from pictures from the Peabody Picture Vocabulary Test. The abstract version wasconstructed from doodle-like drawings, not representative of any recognizable object or shape. No pictureor drawing appeared in more than one set.

In practice sessions, the children were given practice with sets containing 6, 8, 10 and 12 pictures.On testing days, only 12-item sets were used. Subjects were presented with 12-page booklets. Each


page contained 12 pictures (or 12 abstract drawings), arranged in a 3 x 4 array. The same 12 pictures(or drawings) appeared on each of the 12 pages but their positions varied randomly on each page. Thechild received three trials each day; a different booklet containing a different random order of the picturesor drawings was used in each trial.

The child was instructed to touch each of the 12 pictures by touching one picture on each sheet.Pictures could be touched in any order, but each picture could be touched only once. The only restrictioninvolved repeated pointing to the same spatial location.

The experimenter recorded the total number of errors made over the three trials; an error was scoredwhen a child pointed to the same picture more than once during a trial. Total time taken to completethe three trials and an effort rating were also recorded.

Paired associates leaming task. The paired associates leaming task (PAL) consisted of nine lists of 18arbitrarily associated word pairs. The lists used by Douglas et al. (1986) were lengthened from 15 to18 pairs. If a child's performance during screening indicated that the 18-pair list was too difficult, thelist was shortened to 12 or 15 pairs.

As in the study by Douglas et al. (1986), one-syllable concrete nouns were randomly paired. Thepairs of words were not related in any obvious way and each list contained words from several differentcategories. Three leaming trizds were administered, using the instructions described by Douglas et al.(1986).

Numbers of word pairs correctly recalled was scored for each trial. Examiners also recorded an effortrating.

Statistical analysisIn order to compare dosage levels, a mean score was computed for each subject on each measure

at each dosage level. For all tasks except the Self-Ordered Pointing Task (on which different versionswere used on the two testing weeks), scores obtained by the subject in Weeks 1 and 2 were averaged.Subjects with missing data for a given measure were excluded from the statistical analysis of that measure.As seen in the tables, this occurred rarely and was usually due to the failure of a teacher to completethe rating forms correctly.

For each measure, a one-way repeated measures analysis of variance (ANOVA) was performed[Manova procedure of Statistical Package for the Social Sciences Version X (SPSS )] on the computedmean scores for the four dosage levels of the drug factor. Tests of trend, adjusted for unequal intervalsbetween levels of the drug factor, were also performed to establish whether a linear, quadratic or cubicrelationship existed. These trend analyses are reported only when the overall i^-value was significant.

Whenever the dependent variable was a raw proportion, analyses were also carried out using arcsinetransformed proportions. Since results for arcsine transformed data did not difYer from those for rawproportions, only analyses for raw proportions are reported.

To determine the location of differences between dosage levels, post hoc comparisons (Newman-Keuls)were carried out on all measures for which the overall F-value was significant. In addition, plannedcompairisons were performed on measures from the paired associates task. Reliability data on the measurescan be obtained from the first author.

RESULTS

Main effects for dosage levelsMean scores on all behavioral, academic and cognitive measures for the placebo

condition and each dosage level, together with F-values and probability levels, arereported in Tables 1, 2 and 3.

For variables on which significant JF-values were obtained, results from trend testsare also reported.


CO

s§

XaCQ

2

HO

o

SO

N

5OO

n

-d

-I

bo

o

bo6o

en

lUbO

sin

3CO

O O • * r^ • *o o -^ o • *o o o o od d d d d

CMoo

CM in 00 c i oin CT) 00 in_ CTjCT) d en •*- csi

en CM en ^ en

\ r ^ ' ^ ' ' — »

00 i n CM • ^0 0 CTo

.30

CT)O

.76

00

o

.59

00

o

.80

CTÔ

• ^ . -

.06

en CM en en

oT—4

.79

0000

o

.50

1-H t-..

o oT.H O

-< CM

enCOo

.82

en en CM CM

CO

OU

C/3CO fl

00

in

CO

98)

]

o_oo

00

dCT)in

CO

86)

1

1-H

oCM

o1-H

in

CO

,94)

]

oCT)

en

, —ên

d• ^ . - ên

00

CT)

w

inin

• ^—>

,T-

.08)

]

• w

enen

91)

oenO

00

o

CT)CM

CM

/"^

<£>in

CM

CT)

CM

O

en

Ocn

S"^ 2

o oo oo od d

en inCO ^

en d1-H en

CTI

inen

00 inTj-i CM

d d

o toin en

en in00 vod d

O

ooo

enCM

en

en

in

o

en

d

in

in

d

43)

o

CO

o

en

TJ< i n i nô i n •"—t

d d -CM ^

en

VX

SD

IH

CO

CO O "u "H rt fl) C

(U iH j H • :U

So

a Srt

-afl

IS

u

bo

-afl

IXI

13

(H

rt

fl

flrtu

flbo


Oi

CO

o

Q

O

CO

>

<CO

oQ

ZoCO

»—t

oia,OUCM

aCQ

U COIH U

bc

s

bcgo

en

flrt

be

sin

oV

(H

CO

V

O O CM I^ •rt oO O (£> O O Oq q q en_ q qd d d d d d

CM i n CT^ en o '-1o en in CM m oor>.' r ' CM -^ d VD

CT)

cn^

1-H1-H

CT^ CT^

CM • *

rt OT.H

CT^

09) ]

CM

CT^

O

o

CT^

CO00O

• s , -

00 •* O •* en ort en CO CT) q enr^ en d en d inCM en rt

CM encr>

d d d CM d

in <o -^ -rt en CMO CM 00 CT> O ' ^

CM e n

tere

d

ot—i

.18

(

ooo

.40

(

CM

o

.76

(

oCM

.81

(

1-H 1-H

o <oo -CM • *

. 2 CT) TJJ d en d eng T.H CM • ^

fl

•gX

^ IH-i IH ^

JJ IH UlC U V

be

g O rt-S fl ^

iiz; <;

O en f CM 00O O 00 00 OO O O (£) Od d d d d

f £5 O O enCT) 1-H en in in(£5 in CM d •*'

CTi CT)in in

r^ en ^ CO CMO CT) CM O O

en -^ d CM d

•rt o CT) en encp i^ f QQ C ^

't' CT^ o en oCM CM T-H

CM (£) T-H i n CM•rt •« CM O O

d d d CM d

•>ti en 00 o enCM 1-H r ^ CT) ^

d Ln' d en dCM CM -—I

CT) CT) o CO CMCT^ O CM O O•rt CM d CM d

i n CT) <£>

o r rCMO

00 CM1-1 CM o en o

CO CT) UD CO CMCO i n CM o oCM en d CM d

CO. -Hfl• PH

gXI

o

2mCOrt

. H

u

00in

CMo

in o o• r t CM

en o

XIV

CO

g l- o rt

fl

•g

a; j

flfl xs -D 2 __ V

lilillZ Z Hlilill< Z Z < H w

en en in1-H t~^ i nCD CT) C?

d d d

in 00 xjCT^ O r^en d CM

in in

CT) •r t

O r^ CMCT) CM OCM CT) d

CM CT)

O CMqd

CT) •<—ICMO

O CT) O

" CO CM

B6 ^

CT) CM

qCO

fl

qd

O CM•rt en_ od <y> d

o2enCO

rt1 3

CO

rt

ouCO

•^ -H .iJ

^ h w


CO

[il

zooo

oCO

oa

ISO

N

a.SoUenCQ

X3fl

-i

bo

o

bego

en

Q

flrt

be

gU")

Vrt

(H

CO

O 00 CM O OO rt rt O Oq q q q qd d d d d

00 CMq

00 en •*' 00 d

00

c?o

to

00

T-H

CM

o

00

CT)o

enen

CO

enCM

in

00

X*CT)00

o

o00

o o

o en o 00 oCM •rt o ^ e n CM

d d CM •rt T-

o in r -rt oun •.-H 1-H 1-H en

d d CM •^ • *

CT) v ort rt

d d

CM COi n •'-H

CM enCM •rt

r-> o• * CM

o o •"—I i-^

r>» CT)CM CM

in CM•rt (£)

d d in in •^

i ^ i n CM <x> t în CM CT) en_ qd d en •<*< en

UtoflOa,Vu

fla

befl

o •ô oq ind d

r oq 00

CT)

i n —<

rt COCM rt

CM

i n CT)

in dT-H

en en

in enCM

in om •rt

0 0 CM

CT) qd enen

• * CT)CM CM

0 0 CM

in in

TJ< e nen

en O•^ oq qd d

CT^ CT)

en d

CT)

CM

rt o • ^

CM

e n • ^

i nCM

CO CT)in en

i n 'îU2 CM

d -

in r

en en-I en

en t-»1-H inCM en

be

rt>H

CO

{H

• ^ > ^ <U

/,! tS "^be

• - > ^ ^

5 P o l

O CM O

o o oP "R *Pd d d

CT^ CM •rten T}^ 00

CT) i n CM

CT) CT)

t ^ m CMen CM CM

CM en

CM 00 r--

d in •«*<CM rt

T-H i n vo

CM CT) •rt

(.0 0 0r>. CO e ne n •rt T^CM -^

en en enO t^ tpin o •—!

00 CM Tj^q en COTJH T-H e nCM y-^

• * rt oin in 00CM i n •rt

<£> 0 0 • *(£5 q rt00 00 enCM

CO

oV1/3

o beIH fl C

tn G3 C\I

III

<£> CM Oin 00 oR •^ Rd d d

00 00 CT^CO

CM

in en

in en en

CM en

CO

rthbo.s

.g ••o 13

< <

t in o• * rt OCT) CM q

d d d

CM CM

00

en

m

00

o'-'

in

o

00

CM1-H

COCM

CT)

1-H

qen

1-H

CTI

,98)

]

o

inCT^

CT)

^—toen1-H

^ _

T.H1-H

O CM

2.47

)5.

89 (

1-H

CM

oin

0.60

)2.

58 (

1-H

d

o

CO1-H

3.97

(.0

9)

1-H

CMCT)

r 2.

709.

79 I

o

inCT)

f 0.

752.

70 \

.21)

o

r 1.

583.

50 1

, /—

T-H

en en

00

CM

T-H

d1-H

00

oin

^

oo

enin

CM

1-HCM

CTi

d

en CT)en

O CM CM

o«JH

CO

fl1V

g

be,fld 4-(H U

CO

oIH

V

Bfl 2 _ ^

13 13 b •{;; 13 13o o te .^ o oh H w < h H

befl

< iH

2

CO <

o o o oen 00 00 o<£) q • ^ qd d d d

COin

CO O

CM • r t •r-J

CT)1-H

^—^

inen—I

enin

CT)

oCM

en1-H

CT)1-H

• ^

CM

CO

o

CT)

. X—•

1-H

CT)

dCM00

CM

00T-H

COIT)

oen

1-H

inenen

CM

en

O

T-H

• s ^

CMCM

CM

en en en •rto to t~~ enCM CM CM •rt

•<+! CT^ ; O CM0 0 CM t-~ CO

C M •<*' en

CTI i n ••-<in•^ Ol Oi -^

<.O CT) ••-^ • '^CM CM • r t CM

CM en i n en

CO

-HooCOCO

PH

COIH

o I ?O • i - '

beqj fl2 rt

J H >H

H H


Behavioral measuresACTeRS scores. Significant drug effects and linear trends were found on all of the

ACTeRS scores, with behavioral improvement increasing with increasing dosage.A cubic trend was also found on one scale, the oppositional scale.

Conners Hyperactivity Index. Both teachers' ratings and examiners' ratings on theConners Hyperactivity Index yielded significant /^-values and linear trends withbehavior improving with increasing dosage. The quadratic trends also found probablyreflect relatively large improvement between the placebo and 0.15 mg/kg conditions.

Effort rating. Examiners' overall effort ratings yielded a highly significant i^-value.The ratings improved significantly in a linear fashion with increasing dosage.

Academic measuresArithmetic task administered by examiner. Significant i^-values were obtained on the

following measures: number correct, number attempted, efficiency and examiners'effort ratings; the accuracy score also approached significance. The only measurenot showing a significant /^-value was time taken to complete the task. Significantlinear trends were found on number correct, number attempted, efficiency and effortratings. The number attempted score and effort ratings also showed quadratic trends;these appear to reflect relatively large improvement between placebo and 0.15 mg/kg.In addition, significant cubic trends were found on the number correct, accuracyand efficiency scores. These result from slightly better performance at 0.15 mg/kgthan at 0.3 mg/kg. In all cases, however, best performance occurred at 0.6 mg/kg.

Arithmetic task administered in classroom. Significant F-values and linear trends werealso obtained on three measures from the classroom-administered arithmetic test:number correct, number attempted and efficiency. The /'-value on the accuracy scoreapproached significance and the trend test fell just short of a significant linear trend(P=0.06). The total time measure did not yield a significant effect.

Word discovery task administered in classroom. Significant i^-values and significant lineartrends were found on total number of words produced and the efficiency score. Thei -value on total time taken by the children to complete the task was not significant.

Cognitive measuresReaction time. Four of the measures on the DRT yielded both significant F-values

and significant linear trends. These were mean reaction time, reaction time variability,fiddling behaviors and effort ratings. In addition, significant quadratic trends werefound on the fiddling and effort variables. As on several previous variables, thesequadratic trends seem to reflect the fact that greatest improvement occurred betweenplacebo and 0.15 mg/kg. Anticipatory responses yielded a significant jp-value andsignificant quadratic and cubic trends; again most change occurred between placeboand 0.15 mg/kg.

In order to study the effects of dosage on performance over trials, a Dose (4) xBlock (3) repeated measures ANOVA was carried out to detect interactions between


dosage and trial blocks; for this analysis the 45 trials were divided into three blocksof 15 trials. There was a significant effect of trial block [F(2,34) = 8.16, P< O.OOl],as well as a significant linear trend [i^l,17)= 14.16, P < 0.001]. There was also asignificant effect of dose [JF1(3,51) = 8.78, P < O.OOl] and a significant linear trend[ = 15.71, P<O.OOl]. The block x dose interaction was not significant

P=0.66].

Maze tracking task. Three variables from the maze tracking task yielded significanti^-values and significant linear trends: total errors, error variability (over the sixsegments of the maze) and effort ratings. In addition, a significant quadratic trendwas found on error variability, reflecting a relatively large decrease between placebocind 0.15 mg/kg. Total time taken to complete the maze did not yield a significant effect.

A 2-way repeated measures ANOVA was carried out to check for interactionsbetween dosage (4 levels) and maze segment (3 levels: first two segments, middletwo segments and last two segments). A significant main effect was found for dosagelevel [i^3,54) = 7.07, P< 0.000], as well as a significant linear trend [i^l,18) = 13.06,P< 0.002]. Similcirly, a significant main effect was found for segment [i^2,36) = 5.05,P < 0.012], as well as a significant linear trend [i^l,18) = 7.33, P < 0.014]. Theinteraction was not significant [F(6,108) = 0.77, P=0.60].

MFF-20. All three measures from the MFF-20, including total errors, mean latencyand effort ratings, yield significant F-values and significant linear trends.

Self-Ordered Pointing Task. The F-value on the error score from the representationaltask approached significance {P = 0.06). The F-value on the abstract version was notsignificant. Effort ratings on both the representational and abstract versions showedsignificant JF-values and significant linear trends, with a quadratic trend also evidenton effort ratings for the representational task.

Paired associates leaming task. A significant /^-value and a significant linear trend werefound only on the effort rating. The /^-value on number of pairs learned on Triad2 approached significance (P = 0.08).

Pairwise comparisons between dosage levels (Neuman-Keuls)Pairwise comparisons between dosage levels for variables on which significant or

near-significant F-values were found appear in Table 4. To simplify presentation,only data from a representative sample of measures from each scale or task arepresented.

Behavioral measures. Significant drug-placebo differences were found at all three dosagelevels on all behavioral measures. Comparisons between dosage levels revealed thatthe 0.6 mg/kg dose produced greater behavioral improvement than 0.15 or 0.3 mg/kgon most measures. A significant difference between 0.15 and 0.3 mg/kg was foundonly on two ratings made by examiners: the Conners Index and the Overall Effortrating.


2o<OQ

nCO

oCO

0.SouCO

OH

bo

i n

bo

g ^ g V3 O "

bo be

bo5beg ^

in o

bo bo

bo beg ^ g V

^ ft^

bo4o c5

D b o , .

(Jrt

boo ^

jn be ,

s ^ gyE en

d

boo

Xi

o m-H qd d

ino I I

o in•^ 9d d

in m9 9d d

in T-^9 9d d

I Iin -19 9d d

inI 9

o

in

o I I I Io

d

o od d

q qd d

qd

q qd d

q qd d

in

o

q qd d

q qd d

qd

q qd d

in

o

in y^

q qd d

q qd d

qd

q qd d

„ bp bpbo g flfl "•a ••S^ i2 2at

enti

or

JH CO

> flo

• i - i '>]

o*^IHO

g ^S o

ne:

« .g U

u X '"

e: U >H I H

-3 U <U

I ill

i'Ss l-sCO fl 53^^ §13 l i O

rt UU >H IH

III•c - '

ooViVi

S ^.. oV WH

> o

l l

I I Im mo od d

I Iino in in

q qd d

I I Iino I I

1-H i n •'-Hq q qd d d

qd

q qd d

o

d

1-H i n •'-H

9 9 9d d d

qd

qd

i n i n ••—I

9 9 9d d d

qd

o

d

enflOa

^^ lao'u^.gCO rt bo ' Q

-H ^

^ <, ^ ta

boG

>flrt

rto

(UCO

V

a,V

•S 2

oTJ

oCO

boTJO

flo

TJVim

X!

mVi

rt


Academic measures. On the examiner-administered arithmetic test, all three dosagesshowed significant differences among the dosages.

On the classroom-administered arithmetic tasks only the 0.6 dosage yieldedconsistent drug-placebo differences, although some of the measures showed a trendtoward improvement at the 0.3 dosage. Comparisons between dosage levels alsorevealed that 0.6 was more effective than the two lower dosages.

On the classroom-administered word discovery task, drug-placebo differences werefound at 0.3 and 0.6 mg/kg. There were also trends suggesting better performanceat the two higher dosages than at 0.15 mg/kg.

Cognitive measures. On all measures derived from the DRT, all three dosage levelsproduced better performance than placebo. There was also a trend toward betterperformance at 0.6 mg/kg than at 0.15 mg/kg on mean reaction time.

On maize tracking the 0.3 and 0.6 mg/kg dosages improved performance significantlyover placebo levels. Significant differences also were found between the 0.15 doseand the 0.3 and 0.6 doses, with larger doses producing fewer errors.

On the MFFT-20, the error score showed significant drug-placebo differences atcill three dosage levels. Differences in favor of the two larger doses over 0.15 mg/kgalso were found. The latency score showed a significant drug-placebo difference onlyat 0.6 mg/kg. Between-dosage differences revealed longer latencies at 0.6 than at 0.15or 0.30 mg/kg.

On the representational form of the Self-Ordered Pointing Task there were trendstoward significant drug-placebo differences at all three drug levels, with fewer errorsoccurring in the drug conditions. There were no significant differences between dosagelevels.

Planned comparisons between dosage levels on paired associates taskThe paired associates task was not included in the Neuman-Keuls analyses because

JP-values did not reach significance. Because significant drug-placebo differences hadbeen found previously on this task at 0.3 mg/kg (Douglas et al., 1986), and becauseGan and Cantwell (1982) obtained a significant effect on a paired associates task at0.3 mg/kg but not at two higher doses, planned comparisons were used to investigateeffects of the 0.3 and 0.6 mg/kg dosages. As in our previous study (Douglas et al.,1986) significant drug-placebo differences occurred at 0.3 mg/kg both on Trial 2(P= 0.01) and on Trial 3 (P= 0.04). The difference between placebo and 0.6 mg/kg,however, was not significant on Trial 2, and on Trial 3 the difference only approachedsignificance (P< 0.10). All P values are for two-tailed tests, DF = 54.

Attempt to identify unfavorable respondersIn an attempt to identify consistent adverse responders, individual children were

selected who showed their best performance on each measure on the days they receivedplacebo, as opposed to the days they received one of the drug dosages.

Optimal response on placebo was relatively rare. On most measures a maximumof two or three children obtained their best scores in the placebo condition.Furthermore, the particular children classified as adverse responders by this criterionvaried greatly across the measures. For example, on the paired associates task, the


score of Subject 19 was substantially higher on placebo than on any of the threedrug dosages. Yet, this child showed excellent improvement on medication throughoutalmost all of the other measures, and there was no other measure on which he showedhis best performance on placebo. To take another example, the teacher of Subject13 gave him his best ratings on the Attention and Hyperactivity scales of the ACTeRSwhen he was receiving placebo and he performed best on the word discovery taskon his placebo days. Yet, this child showed his best performance on the reaction timetask, classroom arithmetic and paired associates on the days when he was receiving0.6 mg/kg.

In addition, all chUdren showed substantial improvement on at least severalmeasures. Thus, no child could be classified as a consistent adverse responder ornon-responder.

DISCUSSION

Linear dose-response trendsThe findings reveal a predominant pattern of linear dose-response trends with

performance improving up to 0.6 mg/kg of MPH, the highest dosage studied. Withthe possible exception of the paired associates learning task (PAL), the pattern wassimilar across behavioral, academic and cognitive measures. Thus these group findingsprovide little support for the hypothesis that dose response curves for behavioral andcognitive variables differ.

It is also evident that performance on most measures does not peak at 0.3 mg/kg,but continues to improve in a linear fashion up to at least 0.6 mg/kg. A few recentstudies employing smaller and less varied test batteries have yielded results consistentwith these findings (Pelham et al., 1985; Rapport et al., 1985a,b, 1986).

Dose-response curves from our cognitive and academic measures (Tables 2 and3) demonstrate that the impact of stimulant medication is not limited to "low level"tasks (Dyme et al, 1982; Swanson & Kinsbourne, 1979). Analyses from the relativelysimple reaction time task (DRT) yielded results comparable to those from mazetracking, the MFFT, arithmetic and word discovery tasks, all of which require quitecomplex skills.

The failure to find significant dose-by-trial interactions on the DRT and maze-tracking tasks was somewhat surprising, in view of previous studies suggesting thatstimulants help reduce deterioration in performance over time (Humphries etal., 1979;Strauss, Lewis, Klorman, Peloquin, Perlmutter & Salzman, 1984; Sykes, Douglas& Morgenstern, 1972). It is possible that the DRT was not sufficiently demandingto yield such effects. In the case of maze-tracking, our failure to replicate the differentialeffect on later trials reported by Humphries et al. (1979) may be attributable to theextended practice our subjects received on this task.

Interestingly, both the error and latency measures of the MFFT yielded significant^-values and linear trends. The significant findings on latency (reaction time to firstresponse choice) differ from those of Brown and Sleator (1979) and Rapport et al.(1985a), both of whom found significant drug effects on errors but not on latency.


In interpreting their findings, these investigators suggested that effects of the stimulantson ADD-H children's perceptual search strategies may be independent of effects onthe time they take to respond. It seems more likely, however, that a number ofpsychometric and statistical factors contributed tb the failure of these authors to obtaindrug differences on latency. These may have included psychometric problems withthe early version of the MFF used in the studies (Kogan, 1983); possible confoundingbetween time taken to reflect before responding and "off task" behaviors (Douglas,1983); small samples; and single assessments at each dosage level. Sebrechts et al.(1986) recently reported dosage effects similar to ours on both accuracy zind latency.Although these authors used early versions of the MFF, they administered both thechOd and adult versions to each subject. This would double the number of items eachchild received and may have improved task sensitivity. In any case, their latencymeasure was sufficiently discriminating to show a relationship between latency andpeak MPH concentration.

Our significant findings on MFF-20 latency contrast with the non-significant drugeffects found on time measures from a number of other tasks including maze tracking,self-ordered pointing, arithmetic and word discovery. On these tasks the medicationseemed to exert its effect, not by slowing the children down, or making them worklonger, but by enabling them to use their time more efficiently.

Sensitivity of measuresPairwise comparisons between dosage levels (Table 4) reveal that most of our

measures were highly sensitive to dosage changes. The degree of sensitivity to thelow 0.15 mg/kg dose and to relatively small changes in dosage is greater than thatachieved in several of the studies reviewed. This may reflect the care taken indeveloping and administering the measures, the use of a somewhat larger sampleand repeat testing at each dosage level. It is noteworthy that these differences wereobtained in drug trials in which dosage was changed on a day-to-day basis. The briefhalf-life of MPH evidently makes such research designs feasible (Gaultieri & Hicks,1985; Swanson, Kinsbourne, Roberts & Zucker, 1978).

Interestingly, all of the behavioral ratings showed improvement even at the 0.15dose. We had included the ACTeRS because this 24-item teacher-rating scale wasdeveloped to improve discrimination between the different behavioral symptomsassociated with ADD-H (Ullman et al., 1984). However, it appears that items fromthe various behavioral categories yield a very similar pattern. This means that teacherratings on the brief 10-item Conners Index provide as much information on drugeffects, at least when group data are being considered. Moreover, ratings by examinerson our single item reflecting amount of effort expended by the child during testingproved to be as sensitive to dosage changes as examiners' ratings using the ConnersIndex. It will be important to discover whether teachers' ratings using this singleitem prove to be equally sensitive. Our results suggest that raters may base theirjudgments on an overall impression of effortful or acceptable behaviors. If this is so,brief measures may discriminate stimulant-induced changes as accurately as morecomplex and time-consuming ones.

Evidence for the self-regulatory hypothesisThe pervasiveness and consistency of the linear relationships found across our


measures in both laboratory and classroom settings suggests that stimulants enableADD-H children to mobilize intact information processing skills by activating central,self-regulatory processes (Douglas, 1984a,b, 1988; Douglas et al., 1986). Dose-dependent enhancement of self-regulatory or "executive" control was demonstratedin a variety of ways. On the DRT, reaction times became faster and less variable,suggesting both more effective and more consistent deployment of attentional capacity.The accompanying decrease in "anticipatory" and "fiddling" responses indicatesthat the medication also strengthened inhibitory control. On maze tracking, thechildren showed improved capacity to plan and co-ordinate movements of both handsin order to control a line moving through the narrow zig-zag path of a maze. Theyalso showed more consistent performance over the segments of the maze. On theMFFT they slowed down and made fewer errors. This implies a more exhaustiveand analytic search of response alternatives. Results on the Self-Ordered PointingTask (SOP) are somewhat less clear. The reasons for this may be psychometric andstatistical. We shortened the task to save testing time; also, because different formsof this test were used in the two testing weeks, analyses had to be based on a singleassessment at each dosage. Nevertheless, the near-significant jp-value and consistenttrends toward improvement at all three dosages on the "representational" form suggestthat MPH helped the children organize their responses so as to avoid duplicationof response choices. The SOP was developed by Petrides and Milner (1982) to assessplanning and self-monitoring skills in patients with frontal lobe injuries. Shue andDouglas (in preparation) have shown that the representational form differentiatesbetween ADD-H and normal children.

Our academic tasks were also designed to assess relatively complex skills. Thearithmetic tasks were administered at a level that taxed each child's mathematicalabilities and they required shifting between different operations such as addition andsubtraction. The pattern of improvement suggests both enhanced effort and moreeffective deployment of effort. Finally, the improvement found on the word discoverytask reflects more persistent effort and improved ability to shift mental set productively.

These results demonstrate that it is unwise to dismiss stimulant-induced changesas "merely" reflecting improved production or test-taking behaviors. Thisinterpretation results from a failure to appreciate the central importance of defectiveself-regulation in ADD-H and the critical role of stimulants in activating self-regulatoryprocesses (Douglas et al., 1986).

It should be noted that the self-regulatory hypothesis predicts stimulant-inducedimprovement only to the extent that response deficits are attributable to defectiveself-regulation. We believe that the hypothesis may account for more modest "rate-dependent" effects that remain after mathematical problems associated with the rate-dependent phenomenon are eliminated (Gualtieri & Hicks, 1985; Swanson, 1988).The self-regulatory hypothesis also could explain the reductions in intra-subjectvariability that are sometimes found when subjects are placed on stimulants (Cohen,Douglas & Morgenstern, 1971; Gualtieri & Hicks, 1985; Peloquin & Klorman,1986; Sykes, Douglas & Morgenstern, 1972). In the present study, this occurred onthe two tasks in which consistency of performance over trials could be assessed, theDRT and maze tracking tasks. Finally, the self-regulatory hypothesis may help explainthe failure of some investigators to find significant correlations between stimulant-


induced change scores (Gittelman-Klein & Klein, 1975; Gualtieri & Hicks, 1985).According to the hypothesis, degree of improvement will depend upon the extentto which a child is performing below his own level of competence. In highly erraticADD-H children, this would be expected to vary considerably from moment to momentand from t2isk to task. Consequently drug-induced change scores would not be expectedto show high correlations.

Negative effects at high dosages?Although our results demonstrate linear dose-response patterns on several relatively

complex and demanding cognitive tasks, findings on the paired associates learningtask (PAL) leave open the possibility of decreased positive effects, or even detrimentaleffects, at high stimulant dosages on some cognitive tasks. Evidence of a decreasein performance between 0.3 and 0.6 mg/kg on the PAL is quite subtle. The numberof word pairs learned increased significantly over placebo levels at the 0.3 dosageon both trials 2 and 3. In contrast, the difference between placebo and 0.6 mg/kgwas not significant on trial 2 and showed only a trend on trial 3.

Considered alone, these findings could probably be ignored. However, we havefound more robust evidence of a decrement at 0.6 mg/kg in a replication study inwhich we increased the number of PAL trials from three to four (Douglas, Barr,O'Neill, Amin, Britton & Darling, in preparation). In addition, two other laboratorieshave reported similar findings using a version of the PAL developed by Swansonand Kinsbourne (1979). Gan and Cantwell (1982) obtained significant placebo-MPHdifferences on this task at 0.3 mg/kg but not at 0.5 or 1.0 mg/kg. Rapport et al. (1985b)found a levelling off in performance at their highest dosage (15 mg of MPH), evenwhen two children identified as non-responders by the PAL were dropped from theanalysis. They also obtained a significant drug effect on the PAL only when theyused data from each child's optimad dosage.

In defending their procedures of dropping PAL-identified non-responders fromsubject samples and using data from each child's best drug response. Rapport et al.(1985b) argued that the PAL has "special characteristics", because it was designedto differentiate responders from non-responders and to delineate the dose at whichindividual children perform optimally. As evidence for these two assumptions theycite a study in which Swanson, Kinsbourne, Roberts and Zucker (1978) found thatonly two-thirds of their hyperactive subjects showed a favorable response on the PAL.Moreover, when the dosage of a small group of favorable responders was increased,the "overdosed" favorable responders showed time-response curves on the PAL thatresembled the curves of adverse responders. In addition, Swanson and Kinsbourne(1979) have reviewed findings from severad studies in their laboratory whichdemonstrated that both favorable and unfavorable responders on the PAL show a goodresponse to stimulzmts on several other laboratory measures. Swanson and Kinsbourneargued that these other tasks seriously overestimate the beneficial response to stimulantdrugs, since we know "from the literature" that not all of these children will respondfavorably "in the clinicaJ sense". They concluded that task complexity and demandsfor learning over repeated trials account for the unusual discriminative power of thePAL.


Although our PAL task differs from the Swanson-Kinsbourne version in severalways, the two tasks share a number of characteristics. Both require the self-initiatedapplication of relatively sophisticated and imaginative rehears2il strategies. They alsoboth require accumulation of learning from the same stimulus set over successivetrials. It is noteworthy that a verbal learning task on which Peeke et al. (1984) foundimprovement on a 10 mg dose of MPH, but not on a 21 mg dose, also shares someof these characteristics.

Although the picture recognition task used by Sprague and Sleator (1977) did notassess learning in the usual sense, it is interesting that the drop in their dose-responsecurve at 1.0 mg/kg of MPH occurred on their largest, most difficult matrix. Sincethis matrix was also administered last, the children were probably being pushed towardthe limits of their capabilities. The two tasks on which Rapport et al. (1986) foundevidence of a slight decrement do not fit the "high level/high load" criteria as clearly.However, difficult classroom assignments are known to be particularly taxing for ADD-H children and the version of the continuous performance task used by Rapport etal. was relatively long, demanding and fast-paced. Unfortunately, Rapport et al. didnot report dosage-by-trial interactions for this task.

We believe that the self-regulatory hypothesis of stimulant action may help explainthese few reports of decreased effectiveness or decrements at high doses. Critical tothis explanation is the assumption that stimulants improve performance only to theextent that poor performance is caused by faulty self-regulation. Consequently, asdosage is increased, individuals should approach the limits of their ability to improve.This may only produce a levelling off in performance. However, if self-regulatoryprocesses are activated to supra-optimal levels, deleterious effects might be expected.Clinicians who administer psychological tests to children on stimulants occasionallysee the behavior of an overdosed ADD-H child shift from a free-wheeling, impulsiveapproach to one that is slow, overly cautious and agonizing. It is possible that occasionalreports of drops in dose-response curves reflect the deleterious effects of such excessiveself-regulation. This formulation overlaps somewhat with theories that describestimulants as causing "overfocusing", "overactivation", or "perseveration" (Dymeet al., 1982; Robbins & Sahakian, 1979; Swanson & Kinsbourne, 1979).

Before theories involving either disruptive levels of self-regulation or overfocusingcan be accepted, however, it will be necessary to pursue possible negative effects ofhigh dosages with greater care and precision. Two major strategies are possible:investigating effects of dosages higher than 0.6 mg/kg and/or increasing the sensitivityof our measures. If higher dosages are employed, it will be important to monitorside effects closely and to consider their impact on cognitive functions. A child whois experiencing debilitating physical symptoms is unlikely to perform optimally,particularly on demanding cognitive tasks. Using a maximum dosage of 0.6 mg/kgof MPH, we have found only occasional evidence of side effects on ratings by teachersand examiners, mood scales and measures of heart rate and blood pressure. It isclear, however, that a variety of side effects become evident at higher dosages (Charles,Schain & Zelniker, 1981; Sprague & Sleator, 1977; Werry & Sprague, 1974; Winsberget aL, 1982). Future research designs should facilitate the early identification of childrenshowing such symptoms.

In attempting to design more sensitive tasks, our most promising clues rest on the


fact that the most consistent evidence of performance decrements at high dosageshas been found on PAL tasks involving arbitrary associations. Beyond the first fewstimulus pairs, which are probably encoded relatively automatically, these tasks requiresubjects to engage in sophisticated and imaginative mnemonic strategies (Douglas etal., 1986). Excessive self-regulation may be particularly harmful to this kind of flexibleproblem solving. In addition, it may be useful to study performance on these andother complex learning tasks over several trials. It is true that most evidence pointsto beneficial stimulant effects on later trials (Conners, Eisenberg & Sharpe, 1964;Humphries i a/., 1979; Strauss et al., 1984; Sykes etal., 1971). However, increasingdemands for high level processing, combined with high dosage levels, may over-activateand over-tax self-regulatory processes and thus lead to impaired performance.

Identifying adverse responders and non-respondersWe found it impossible to classify any of the children in our study as consistent adverse

responders. In addition, every child made substantial gains on at least several measures.Thus, our results do not support the widely held assumption that at least 30% ofADD-H children fall to respond or show an adverse response to the stimulants(Conners, 1972; Loney, 1986; Sprague & Sleator, 1975; Swanson & Kinsbourne,1979; Ullman & Sleator, 1986).

It has been argued that our failure to find unfavorable responders is due to thefact that subject samples from our laboratory are more carefuUy screened than othersamples for non-ADD-H symptoms (Swanson, 1988). Certainly, this possibility shouldbe investigated. Although we attempt to choose children whose primary diagnosisis ADD-H, some of our subjects also meet DSM-III criteria for other disorders (seedescription of subject sample). It is, of course, difficult to make definitive statementsabout the response patterns of this small group; however we were unable to detectany differences in the response of these particular children.

A more likely explanation for our failure to identify unfavorable responders restson the fact that previous figures on individual responsivity have been based on a singlebehavioral or cognitive measure. In contrast, we studied the response of each childacross several behavioral, cognitive and academic measures.

Our failure to demonstrate that any task, including the PAL, predicted a child'sresponse on other measures calls into question the practice of using the PAL as anindicator of a child's response to stimulant medication, either for selecting researchsamples or for making clinical predictions. Data from the two children who wereidentified as non-responders by the PAL in the Rapport et al. (1985b) study are relevantto this issue. These two children showed positive dose-response curves resemblingthose of "responders" on the other measures studied. Surprisingly, Rapport et al.concluded that it is "to the credit of the PAL" that it "erred on the conservativeside" by producing only two false-negative classifications.

A recent study by Wender and Kinsbourne (1985) raises further doubts. Wenderand Kinsbourne found that all five of their subjects who were identified as adverseor non-responders on the PAL showed a favorable response on home and school trials.Furthermore, all children who responded poorly on the home and school trials showeda favorable response on the PAL, even at the highest dosage used (0.75 mg/kg).

These findings suggest that physicians prescribing stimulant medication may be


forced to make clinical decisions in the face of conflicting evidence from differentmeasures. Certainly, a negative response on the PAL, or any other measure, suggestscaution. Nevertheless, other factors, such as a child's behavior or productivity in theclassroom, may require urgent amelioration. Thus, in attempting to arrive at anoptimal dosage, the clinician must consider each child's pattern of response acrossbehavioral, academic and cognitive measures.

Acknowledgements—This research was supported by Grant No. MA 6913 from the Medical ResearchCouncil of Canada and by the W. R. Crant Foundation Faculty Scholar Program.

REFERENCES

Ackerman, P. T. & Dykman, R. A. (1982). Automatic and effortful information-processing deficitsin children with leaming and attention disorders. Topics in Leaming and Leaming Disabilities, July,12-22.

American Psychiatric Association (1980). Diagnostic and Statistical Manual of Mental Disorders (3rd edn).Washington, DC.

Ballinger, C. T., Varley, C. K. & Nolen, P. A. (1984). Effects of methylphenidate on reading in childrenwith Attention Deficit Disorder. American Joumal of Psychiatry, 141(12), 1590-1593.

Brown, R. T. & Sleator, E. K. (1979). Methylphenidate in hyperactive children: Difference in doseeffects on impulsive behavior. Pediatrics, 64, 401-411.

Cairns, E. & Cammock, T. (1978). Development of a more reliable version of the matching familiarfigures test. Developmental Psychology, 14(5), 555-560.

Charles, L., Schain, R. & Zelniker, T. (1981). Optimal dosages of methylphenidate for improvingthe leaming and behavior of hyperactive children. Developmental and Behavioral Pediatrics, 2:3, 78-81.

Cohen, N. J., Douglas, V. I. & Morgenstern, G. (1971). The effects of methylphenidate on attentivebehaviour and autonomic activity in hyperactive children.Psychopharmacologia, 22, 282-294.

Conners, C. K. (1972). Psychological effects of stimulant drugs in children with minimal braindysfunction. Pediatrics, 49(5), 702-708.

Conners, C. K., Eisenberg, L. & Sharpe, L. (1964). Effects of methylphenidate on paired-associateleaming and Porteus maze performance in emotionally disturbed children, foumal of ConsultingPsychology, 28, 14-22.

Coons, H. W., Peloquin, L. J., Klorman, R., Ryan, R. M., Bauer, L. O., Perlmutter, R. A. & Salzman,L. F. (1981). Effects of methylphenidate on young adults' event related potentials. Electro-encephalography and Clinical Neurophysiology, 51, 373-387.

Douglas, V. I. (1983). Attentional and cognitive problems. In M. Rutter (Ed.), Developmentalneuropsychiatry, pp. 280-328. New York: Guilford Press.

Douglas, V. I. (1984a). The psychological processes implicated in ADD. In L. M. Bloomingdale (Ed.),Attention Deficit Disorder. Diagnostic, cognitive, and therapeutic understanding, pp. 147-162. Jamaica, NY:Spectrum Publications Inc.

Douglas, V. I. (1984b). Comments on citations of paper by author. Citation Classics, Current Contents,16(44), 16.

Douglas, V. I . (1988). Cognitive deficits in children with Attention Deficit Disorder with Hyperactivity.Monograph supplement, foumal of Child Psychology and Psychiatry (in press).

Douglas, V. I., Barr, R. G., O'Neill, M. E. & Britton, B. G. (1986). Short term effects ofmethylphenidate on the cognitive, learning and academic performance of children with AttentionDeficit Disorder in the laboratory and the classroom, foumal of Child Psychology and Psychiatry,27, 191-211.

Douglas, V. I., Barr, R. G., O'Neill, M. E., Amin, K., Britton, B. G. & Darling, M. (in preparation).Effects of methylphenidate on learning in Attention Deficit Disorder.

Douglas, V. I. & Peters, K. G. (1979). Toward a clearer definition of the attentional deficit of hyperactivechildren. In G. A. Hale and M. Lewis (Eds), Attention and the development of cognitive skills, pp. 173-247.New York: Plenum Press.


Dyme, I. Z., Sahakian, B. J., Golinko, B. E. & Rebe, E. F. (1982). Perseveration induced bymethylphenidate in children: Preliminary findings. Progress in Neurological Psychopharmacotogy andBiological Psychiatry, 6, 269-273.

Gan, J . & Cantwell, D. P. (1982). Dosage effects of methylphenidate on paired associate leaming:Positive/negative placebo responders. Joumat of Abnormal Chitd Psychology, 6(149), 237-242.

Gittelman, R., Klein, D. F. & Feingold, I. (1983). Children with reading disorders—II. Effects ofmethylphenidate in combination with reading remediation. Joumal of Child Psychology and Psychiatry,24, 193-212.

Gittelman-Klein, R. & Klein, D. F. (1975). Are behavioral and psychometric changes related inmethylphenidate-treated children? International Joumal of Mental Health, 4, 182-198.

Goyette, C. H., Conners, C. K. & Ulrich, R. F. (1978). Normative data on revised Conners Parentand Teacher Ratings Scales. Joumal of Abnormat Child Psychology, 6(2), 221-236.

Gualtieri, C. T. & Hicks, R. E. (1985). Neuropharmacology of methylphenidate. InJ . H. Beitchman(Ed.), Psychiatric clinics of North America, Vol. 8, No. 4. Pennsylvania: W. B. Saunders Company.

Hollingshead, A. & Redlich, F. (1958). Social class and mental illness. New York: John Wiley.Humphries, T., Swanson, J., Kinsbourne, M. & Yiu, L. (1979). Stimulant effects on persistence of

motor performance of hyperactive children. Joumal of Pediatric Psychology, 4(1), 55-67.Kagan, J. (1966). Reflection-impulsivity: The generality of dynamics of conceptual tem^po. Joumal of

Abnormat Psychology, 7, 17-24.Kogan, N. (1983). Stylistic variation in childhood and adolescence: Creativity, metaphor zmd cognitive

styles. In P. Mussen (Ed.), Handbook of Child Psychiatry, 4th edn. Vol. 13, pp. 630-706. New York:John Wiley.

Loney, J. (1986). Predicting stimulant drug response among hyperactive children. Psychiatric Annals,16:1, 16-19.

Peeke, S., Halliday, R., Callaway, E., Prael, R. & Reus, V. (1984). Effects of two doses ofmethylphenidate on verbal information processing in hyperactive children. Joumal of ClinicalPsychopharmacotogy, 4:2, 82-88.

Pelham, W. E., Bender, M. E., Caddell, J., Booth, S. & Moorer, S. A. (1985). Methylphenidate andchildren with Attention Deficit Disorder. Dose effects on classroom academic and social behavior.Archives of General Psychiatry, 42, 948-952.

Peloquin, L. J. & Klorman, R. (1986). Effects of methylphenidate on normal children's mood, event-related potentials, and performance in memory scanning and vigilance, youma/ of Abnormat Psychology,95(1), 88-98.

Petrides, M. & Milner, B. (1982). Deficits on subject-ordered tasks after frontal and temporal lobelesions in man. Neuropsychologia, 20(3), 249-262.

Rapoport, J. L. & Zametkin, A. (1988). Drug treatment of attention deficit disorder. MonographSupplement, Joumal of Child Psychotogy and Psychiatry (in press).

Rapport, M. D. DuPaul, G. J., Stoner, G., Birmingham, B. K. & Masse, G. (1985a). Attention deficitdisorder with hyperactivity: Differential effects on methylphenidate on impulsivity. Pediatrics, 76(6),938-943.

Rapport, M. D., DuPaul, G. J., Stoner, G. & Jones, T. J. (1986a). Comparing classroom and clinicmeasures of attention deficit disorder: Differential! idiosyncratic, and dose-response effects ofmethylphenidate. Joumat of Consulting and Clinical Psychotogy, 54(3), 334-341.

Rapport, M. D., Stoner, G. J., Birmingham, B. K. & Tucker, S. (1985b). Methylphenidate inhyperactive children: Differential effects of dose on academic, leaming and socizil behavior, youma/of Abnormal Child Psychology, 13(2), 227-244.

Robbins, T. W. & SahaJcian, B. J. (1979). "Paradoxical" effects of psychomotor stimulzmt dmgs inhyperactive children from the standpoint of behavioural pharmacology. Neuropharmacotogy, 18,931-950.

Sebrechts, M. M., Shaywitz, S. E., Shaywitz, G. A., Jatlow, P., Anderson, G. M. & Cohen,D. J. (1986). Components of attention, methylphenidate dosage and blood levels in children withattention deficit disorder. Pediatrics, 77(22), 222-228.

Shue, K. & Douglas, V. I. (in preparation). Peirallels between cognitive deficits in children demonstratingAttention Deficit Disorder and patients with frontal lobe injuries.

Sprague, R. L. & Sleator, E. K. (1973). Effects of psychopharmacologic agents on leaming disorders.Pediatric Clinics of North America, 20(3), 719-735.


Sprague, R. L. & Sleator, E. K. (1975). What is the proper dose of stimulant drugs in children?International Joumal of Mental Health, 75-118.

Sprague, R. L. & Sleator, E. K. (1977). Methylphenidate in hyperkinetic children: Differences in doseeffects on leaming and social behavior. Science, 198, 1274-1276.

Strauss, J . Lewis, J . L., Klorman, R., Peloquin, L.-J., Perlmutter, R. A. & Salzman, L. F. (1984).Effects of methylphenidate on young adults' performance and event-related potentials in a vigilanceand a paired-associates leziming test. Psychophysiology, 21(6), 609-621.

Swanson, J. M. (1985). Measures of cognitive functioning appropriate for use in pediatricpsychopharmacology research studies. Psychopharmacology Bulletin, 21:4, 887-890.

Swanson, J. M. (1988). What do psychopharmacological studies tell us about information processingdeficits in ADD-H/hyperactivity? Monograph supplement, Joumal of Child Psychology and Psychiatry(in press).

Swanson, J. M. & Kinsbourne, M. (1979). The cognitive effects of stimulant drugs on hyperactivechildren. In G. A. Hale and M. Lewis (Eds), Attention and cognitive development pp. 249-274. NewYork: Plenum Press.

Swanson, J. M., Kinsbourne, M., Roberts, W. & Zucker, K. (1978). Time-response analysis of theeffect of stimulant medication on the leaming ability of children referred for hyperactivity. Pediatrics,61:1, 2-29.

Sykes, D. H., Douglas, V. I. & Morgenstern, G. (1972). The effect of methylphenidate (Ritalin) onsustained attention in hyperactive children. Psychopharmacotogia, 25, 262-274.

Ullmann, R. K. & Sleator, E. K. (1986). Responders, non-responders, zmd placebo responders zimongchildren with Attention Deficit Disorder: Importance of a blinded placebo evaluation. ClinicalPediatrics, 25(12), 594-599.

Ullman, R. K., Sleator, E. K. & Sprague, R. L. (1984). A new rating scale for diagnosis and monitoringof ADD children. Psychopharmacotogy Bulletin, 160-164.

Walker, W. M. (1982). Psychostimulants. In S. E. Breuning and A. Poling (Eds), Drugs and mentalretardation, pp. 250-256. Springfield, IL: Charles C. Thomas.

Weiss, B. & Laties, V. G. (1962). Enhancement of human performance by caffeine and amphetamines.Pharmacology Review, 14, 1-36.

Wender, E. H. & Kinsbourne, M. (1985). Validation of the Paired Associates Task as a Measure of StimulantMedication Response in Attentional Deficit Disorder. Paper presented at the Annual Meeting of AmbulatoryPediatric Association, Washington, DC.

Werry, J. S. & Sprague, R. L. (1974). Methylphenidate in children: Effect of dosage. Australia andNew Zealand Joumal of Psychiatry, 8, 9-19.

Winsberg, B. G., Kupietz, S. S., Sverd, J., Hungund, B. L. & Young, N. L. (1982). Methylphenidateoral dose plasma concentrations and behavioral response in children. Psychopharmacology, 76,329-332.

Dosage effects and individual responsitivity to Methylphenidate in Attention Deficit Disorder

Documents

Transcript of Dosage effects and individual responsitivity to Methylphenidate in Attention Deficit Disorder