cognitive testing in early phase clinical trials: outcome according to adverse event profile in a...

Cognitive testing in early phase clinical trials: outcomeaccording to adverse event profile in a Phase I studyy

Alex Collie1,2*, Paul Maruff1,3, Peter J. Snyder4, Miss Amanda Darekar5 and John P. Huggins5

1CogState Limited, Melbourne, Australia2Centre for Neuroscience, The University of Melbourne, Australia3Department of Psychology, La Trobe University, Melbourne, Australia4Department of Psychology, The University of Conneticut, USA5Pfizer Global Research & Development, Sandwich, United Kingdom

Background It has been proposed that objective cognitive testing provides additional information to that collected viaadverse event (AE) recordings. However, in clinical trials of compounds with potentially negative effects on cognition, theresults of cognitive testing may overlap with AE recordings.Aims To examine cognitive function in subjects who do and do not report sedation-related AEs in a Phase I clinical trial.Methods Five computerized cognitive tasks were administered to 28 healthy male volunteers enrolled in a simulated PhaseI study using midazolam to induce sedation-related AEs and cognitive dysfunction. For each subject, the magnitude ofcognitive change between pre-dose and 1 hr post-dose assessments was calculated. Group and individual level cognitiveoutcome was compared between subjects who did and did not report sedation-related AEs following administration of 1.75and 5.25mg midazolam.Results At both doses of midazolam, cognitive dysfunction was observed in both subject groups (i.e., thosewho did and didnot report AEs). Analysis of individual outcomes identified consistent cognitive dysfunction among subjects who reportedsedation-related AEs. Further, in the 5.25mg condition a subset of individuals (66.7%) who did not report sedation-relatedAEs nevertheless displayed substantial cognitive dysfunction.Conclusions Following administration of oral midazolam, there is a dissociation between sedation-related AE recordingsand performance on computerized cognitive tests of motor function, attention, strategy use and problem solving, learning anddelayed recall. Inclusion of computerized cognitive tests in early phase trials may allow identification of subtle cognitivechange, beyond that which is possible by self-report and clinical observation. Copyright # 2006 John Wiley & Sons, Ltd.

key words—Phase I; adverse event; midazolam; cognition; cogstate; sedation

INTRODUCTION

Inclusion of cognitive testing early in clinical drugdevelopment may allow early identification of

clinically meaningful CNS effects (which in principlemay be adverse or beneficial) and a greater under-standing of the pharmacokinetic/pharmacodynamicrelationship prior to entering pivotal later-phase trials(Beglinger et al., 2005a). We have recently reportedthat it is possible to repeatedly administer a 12 mincomputerized cognitive test battery within a Phase Iclinical trial without affecting the collection of safetyand pharmacokinetic data (Collie et al., submitted).The extent to which compounds in early phase trialscan effect thinking is generally determined on thebasis of subjective adverse event (AE) reports.However, inclusion of objective cognitive testing in

human psychopharmacologyHum Psychopharmacol Clin Exp 2006; 21: 481–488.

Published online 4 September 2006 in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/hup.799

*Correspondence to: Dr A. Collie, Level 7, 21 Victoria Street,Melbourne, Vic. 3000, Australia. Tel: þ613 9664 1300. Fax: þ6139664 1301. E-mail: [email protected] the duration of this study, Dr Collie and Dr Maruff were staffscientists at CogState Ltd., a cognitive test development companythat maintains several contractual arrangements with Pfizer Inc., forthe provision of their clinical technologies. All other authors werestaff scientists at Pfizer Inc. throughout the design and conduct ofthis study.

Copyright # 2006 John Wiley & Sons, Ltd.

Received 27 March 2006

Accepted 26 July 2006

early phase trials may allow investigators to be moreconfident about the presence or absence of CNSeffects of their compounds. Cognitive tests may alsoallow detection of both improvement and decline inbrain function, whereas recording AEs detect pre-dominantly negative outcomes. Even if investigatorsare interested in showing that their compounds aresafe, it is well known that individuals can rarelydetermine reliably whether their own cognitivefunction has declined, especially where the magnitudeof change is subtle or if they expect that cognitivedecline will occur. Therefore the current study soughtto determine the nature of the relationship betweencognitive change detected objectively and the AEreports of subjects after treatment with a sedativecompound whose association with cognitive deteriora-tion is well known.Midazolam is a commonly used and widely studied

benzodiazepine with CNS sedating properties. Side-effects of midazolam are well-recognized (Nordst andClark, 1997) and include somnolence, confusion,amnesia, and fatigue. Midazolam has also been shownto affect performance on cognitive tests. For example,Thompson et al. (1999) compared the pre- and post-dosing cognitive outcome in 18 healthy patientsundergoing routine dental procedures. All participantsreceived 3.0mg midazolam administered intrave-nously. Cognitive testing was initiated 5 min post-dosing and identified significant decline on tests ofsimple and choice reaction time, immediate anddelayed recall, as well as word and picture recognitiontasks when compared to a no-drug control condition.Other recent studies have observed similar results(Girdler et al., 2002; Rogers et al., 2002). The time-course of oral midazolam is well-established, withtime to maximum plasma concentration (Cmax)occurring within 1 hr post-dosing, and a goodcorrelation observed between plasma levels andclinical effects (Kanto, 1985).We recently investigated the magnitude and time-

course of cognitive dysfunction following oral mid-azolam administration in a simulated Phase I environ-ment (Collie et al., submitted). This study demonstratedthat it was practical to include cognitive tests in a Phase Ienvironment using a compound that is rapidlybioavailable with a relatively brief elimination half-life. The cognitive data generated from this studysuggested that single doses of both 1.75 and 5.25mgmidazolam produced large and significant declinesacross a range of cognitive functions. Further,qualitative analysis of the AE data suggested thatcognitive decline occurred despite a large proportion ofsubjects failing to subjectively report AEs.

In the current manuscript we re-analyze the datagenerated from Collie et al. (submitted) to specificallyinvestigate the relationship between sedation-relatedAEs and cognitive outcome. The aim of this analysesis to compare outcome on a series of computerizedcognitive tests in subjects reporting AEs and subjectsnot reporting sedation-related AEs following admin-istration of two doses of oral midazolam which havebeen shown to cause both cognitive dysfunction andsubjective AEs. It was hypothesized that there wouldbe a dissociation between the results of cognitive testsand the results of AE recordings in the sample studied.

METHODS

Subjects

Twenty-eight healthy male volunteers between theages of 18 and 55 (mean 30.6� 8.4 years) wererecruited for the study. Subjects were recruited via twodifferent clinical research units in Singapore (N¼ 12)and Brussels (N¼ 16). All subjects were in goodhealth as determined by medical history, physicalexamination, vital signs, ECG, and clinical laboratorymeasurements. All subjects had a body weight ofgreater than 50 kg (group mean 72.0� 9.1 kg), and abody mass index of approximately 18–30 kg/m2

(group mean 23.6� 2.7 kg/m2). The exclusion criteriaare summarized as follows: (1) evidence or history ofclinically significant hematological, renal, endocrine,pulmonary, gastrointestinal, cardiovascular, hepatic,psychiatric, neurologic, or allergic disease (includingdrug allergies, but excluding untreated, asymptomatic,seasonal allergies at time of dosing); (2) any conditionpossibly affecting drug absorption (e.g., gastrectomy);(3) smoking beyond predefined limits; (4) routine useof sedative or other psychoactive medication; (5)participation in any studies of investigational ormarketed drugs during the 30-day period before thestart of the study; (6) a 12-lead ECG demonstratingQTc >430msec; (7) use of prescription or non-prescription drugs, vitamins and dietary supplementswithin 14 days prior to the first dose of trialmedication; (8) blood donations within 56 days priorto dosing, or intention to donate blood or bloodproducts during the study or for 2 months followingcompletion; (9) history of sensitivity to heparin orheparin-induced thrombocytopenia; (10) any othersevere acute or chronic medical or psychiatriccondition or laboratory abnormality though to increasethe risk associated with participation in this study. Thestudy design was approved by institutional ethicscommittees and all subjects gave written informed

Copyright # 2006 John Wiley & Sons, Ltd. Hum Psychopharmacol Clin Exp 2006; 21: 481–488.DOI: 10.1002/hup

482 a. collie ET AL.

consent to participate. The study was conducted inaccordance with the Declaration of Helsinki and GoodClinical Practice.

Study design

This simulated Phase I study employed a double-blind, four-way crossover design using three escalat-ing doses of midazolam (0.6, 1.75, and 5.25mg) withplacebo insertion. Subjects were randomly assigned toone of four treatment sequences. We have previouslydescribed the study design and methodology in detail(Collie et al., submitted).

Each subject completed the cognitive test batterytwice at a screening assessment conducted 28 daysprior to baseline, and again upon admission to theclinical research unit to reduce influence of taskfamiliarity effects. Cognitive assessments were thenconducted immediately pre-dosing (baseline) and at 1,2, 3, 4, 6, and 24 hr post-dosing. The present analysisutilized performance data at baseline and at theassessment closest to Cmax for midazolam (1 hr post-dosing) for those conditions at which both cognitivedecline and sedation-related AEs were previouslyidentified (1.75 and 5.25mg). The study blind wasmaintained by mixing the trial medication in its liquidform with 240ml of ambient temperature water.Sodium benzoate powder for oral solution (POS)supplied with water for irrigation was reconstituted asplacebo. Study medication was administered orally.

Apparatus

All observed and volunteered AEs were recorded byexperienced clinical research unit staff on standar-dized case report forms, regardless of the treatmentcondition or the suspected causal relationship to thestudy treatment. The study was run in a standardclinical research environment with staff who wereexperienced at identifying AEs. In addition, allsubjects were recruited from a large pool of subjects

who had prior experience of identifying AEs inclinical research settings.The cognitive test battery consisted of five different

tests of psychomotor function, attention, learningmemory, problem solving/strategy use, and delayedrecall. All were presented on tablet computers (i.e., aportable computer with a touchscreen) with head-phones to minimize distracting noise. The tests wereselected because of their brevity, demonstrated utilityfor within-subjects experimental designs and for theparametric properties of their outcome measures(Falleti et al., 2003; Snyder et al., 2005). Thecognitive testing required between 8 and 12 min toadminister, with a time-out limit set at 12 min. Taskswere selected from the CogState battery (Collie et al.,2003; Falleti et al., 2003), with the addition of theGroton Maze Learning Test (GMLT; Snyder et al.,2005). The five individual tasks in this battery anddetails of their administration have been described indetail previously (Collie et al., submitted; Snyderet al., 2005, Snyder et al., in press). Table 1 includes abrief summary of the test and the primary outcomemeasure drawn from each test.

Data analysis

Classification of subjects according to adverse eventrecordings. For both treatment conditions, subjectswere classified as either AE-positive (those subjects inwhich an AE was noted) or AE-negative (thosesubjects not recording AEs). While all AEs wererecorded, for the purposes of this analysis AEs weredefined as those that may indicate a sedating effect ofmidazolam (e.g., somnolence, fatigue, confusion,dizziness). AEs of a ‘non-cognitive’ or ‘non-sedative’nature were excluded from this classification algor-ithm (e.g., bruising at venpuncture site, acne,dermatitis).

Cognitive data pre-processing. For the Detect andIdentify tasks, the mean reaction time of correctresponses was computed for each subject and a

Table 1. Summary of cognitive tests utilised in this study, in the order of administration

Cognitive task Cognitive function measured Measure type Metric

Groton Maze Learning Test (GMLT) Executive function(problem solving/spatialworking memory)

Speed Moves per second

One-card learning task (Learn) Learning and memory Accuracy Arcsine percentage accuracyIdentification task (Identify) Simple attention Speed Log10 reaction timeDetection task (Detect) Psychomotor Speed Log10 reaction timeGMLT—delayed recall task Delayed recall Accuracy Number of errors


cognition versus adverse events 483

logarithmic (base10) transformation was then applied.For the Learn task, the number of correct responseswas expressed as a percentage of the total number ofresponses and an arcsine transformation applied(Collie et al., 2003; Falleti et al., 2003; Snyderet al., 2005). For the GMLT, the number of correctchoices on each trial was expressed as a ratio of thetime required to learn these (correct moves/sec)and averaged over the five trials completed in eachsingle assessment. For the GMLT-delayed recalltask, the total number of errors made were calculated.All transformations were performed to ensure thatdata was normally distributed prior to statisticalanalysis.

Effect of midazolam on cognition: computation ofcognitive change scores. The magnitude of thedifference between baseline and post-dose meansfor each individual subject on each of five outcomemeasures was computed using a standardized changescore (Snyder et al., 2005). This change score wascalculated by dividing the mean difference by thewithin-subject standard deviation (WSD) using thefollowing formula:

ðMean post� dose scoreÞ � ðMeanBaseline scoreÞWSD

The WSD was calculated from the placebocondition and according to the method described byBland and Alman (1996). The resulting differencescore may be interpreted as follows: changes of0.2–0.4 are considered small, 0.5–0.7 consideredmoderate and >0.8 considered large (e.g., Cohen,1988).

Cognitive outcome classified according to adverseevent profile. For all cognitive tasks, group meancognitive change scores for AE-positive and AE-negative groups were then calculated under bothtreatment conditions.Cognitive change scores were also used to classify

individuals as showing midazolam related cognitivedecline in two ways. First, previous simulationsindicated that for 5 outcome measures, a change inperformance of 1.5 or greater on two or more testswould retain the Type I error for classification ofimpairment at less than 5% under a two tailedhypothesis (Ingraham and Aiken, 1996). Therefore,individuals who showed cognitive change scores of�1.5 or greater on two or more outcome measureswere classified as showing midazolam relatedcognitive change. The second method defined mid-

azolam-related cognitive change by identifyingwhether subtle impairment in cognitive functionoccurred across a range of measures (Rasmussenet al., 2001; Maruff et al., in press). A compositecognitive change score was computed by averagingthe cognitive change scores across the 5 performancemeasures for each individual. An individual with acomposite cognitive change score of greater than�1.65 was classified as showing midazolam-relatedcognitive impairment. This criteris also retains a TypeI error rate of 5% given a one-tailed hypothesis (i.e.,only impairment was expected). The number ofindividuals in the AE-positive and AE-negative groupswho met either criterion for abnormality wascalculated.

RESULTS

Adverse events

A total of 39 study sedation-related AEs were recorded(Table 2). In the 1.75mg condition, 14 subjects (50%)reported a total of 16 AEs. The remaining 14 subjectsreported no sedation-related AEs at this dose. A totalof 23 sedation-relatedAEs were reported in the 5.25mgcondition by 19 subjects (67.9%), with the remaining9 subjects reporting no AEs at this dose. The majority(30: 76.9%) of reported AEs involved either fatigue orsomnolence. All study drug-related AEs were classifiedby clinical researchers as mild, with the exception of10 instances of moderate somnolence (3 occurring atthe 1.75mg dose and 7 at the 5.25mg dose).

Table 2. Description of ‘cognition-related’ adverse eventsrecorded in 28 healthy male subjects at two different doses ofmidazolam

1.75mg midazolam 5.25mg midazolam

Total N subjects 28 (100.0) 28 (100.0)AE-negative 14 (50.0) 9 (32.1)AE-positive 14 (50.0) 19 (67.9)

Adverse eventsSomnolence 11 19Fatigue 3 0Confusion 0 1Dizziness 1 2Pain 1 1Headache 0 0Total 16 23

Note: All figures are reported as number (percentage of total). AEpositive¼ those subjects for whom adverse events were recorded;AE negative¼ those subjects for whom no adverse events wererecorded. Note that each subject may have reported more than oneadverse event.



Cognitive outcome according to adverseevent profile

Group mean standardized cognitive change score (CS)for AE-positive and AE-negative groups at the5.25mg dose of midazolam are shown in Figure 1.At this dose, the AE-positive group displayed verylarge negative changes on all cognitive tests(CS<�1.90 for all tests) and also on the compositecognitive change score (CS¼�2.10). One-sample t-tests revealed that these effects were all statisticallysignificant (see Figure 1). The AE-negative groupdisplayed a very large decline on the GMLT-D task(CS¼�2.13) in addition to large changes on theGMLT (CS¼�1.12) and the Identify (CS¼�1.29)tasks. Smaller changes were observed for the Detect(CS¼�0.40) and Learn (CS¼�0.52) tasks. Acomposite cognitive change score of �1.09 wasidentified in this group. Despite the magnitude of thesechanges, no significant differences were observed onone-sample t-tests, although the Identify task(p¼ 0.07) approached significance.

One-way Analysis of Variance (ANOVA) identifieda significant between-groups difference for themagnitude of the standardized cognitive changescores on the Learn task [F(1,26)¼ 9.299,p¼ 0.005], while the COMP score approachedsignificance (p¼ 0.07).

Group mean standardized cognitive change score(CS) for AE-positive and AE-negative groups at the1.75mg dose of midazolam are shown in Figure 2. Atthis dose, the AE-positive group displayed large andsignificant negative changes on the GMLT (CS¼�1.25) and Learn (CS¼�1.00) tasks and the COMPscore (CS¼�0.81). Moderate negative changes on theIdentify (CS¼�0.52), Detect (CS¼�0.57) andGMLT-D (CS¼�0.70) tasks were also observed.The AE-negative group displayed large and significantnegative change on the Detect task (CS¼�1.32), withmoderate negative changes on the Learn (CS¼�0.48),Identify (CS¼�0.78), GMLT-D (CS¼�0.64) taskand COMP score (CS¼�0.71), and a small change onthe GMLT task (CS¼�0.32). No significant between-group differences were observed.Table 3 displays the number of individual subjects

who displayed significant cognitive decline under boththe 5.25 and 1.75mg conditions categorised by AEgroup. Chi-square analysis revealed two significantresults from this analysis. For the 5.25mg condition, 6of 9 (66.7%) AE-negative subjects met the criteria forsignificant cognitive decline (p¼ 0.003). For the1.75mg condition, 7 of 14 (50.0%) AE-positivesubjects met the criteria for significant cognitivedecline (p¼ 0.002).Figure 3 displays individual change scores for the

composite cognitive index for the AE-positive and

-2.50

-2.00

-1.50

-1.00

-0.50

0.00

COMPGMLT-DDETECTIDENTIFYLEARNGMLTS

tan

dar

diz

ed c

og

nit

ive

chan

ge

sco

re

AE-negative

AE-positive***

***

***

**

*****

Figure 1. Group mean change from baseline on five cognitive tasks in subjects reporting adverse events (AE-positive;N¼ 19) and subjectsnot reporting AEs (AE-negative; N¼ 9) following administration of 5.25mg midazolam, ��p< 0.01; ��p< 0.001



AE-negative groups in the 5.25mg condition. Thisfigure shows that all subjects reporting sedation-related AEs in this condition displayed a cognitivedecline of some magnitude, while 6 of 9 subjects whodid not report sedation-related AEs neverthelessdisplayed a decline in cognition (CS<�1.0).

DISCUSSION

Cognitive testing is becoming more commonly used inearly phase clinical trials (e.g., van der Meyden et al.,1992; Meyers et al., 2000; Grundman et al., 2002;

Beglinger et al., 2004; Beglinger et al., 2005a). Anumber of groups have proposed that inclusion ofcognitive testing in early phase trials may lead togreater understanding of the CNS effects of newchemical entities (Collie et al., submitted; Beglingeret al., 2005b). Cognitive tests are more objective andpotentially more sensitive than AE ratings. If designedappropriately, cognitive tests may detect bothimprovement and decline in brain function (Mollicaet al., 2004), meaning that they may provide moreinformation than that gained from purely ‘negative’AE ratings. One important issue yet to be addressed inthis area is the extent to which the results of cognitivetests overlap with AE recordings. The current studyexamined this issue via re-analyses of data collectedfrom a simulated Phase I study using the benzo-diazepine midazolam (Collie et al., submitted).

The sedation-related cognitive effects of midazo-lam are well recognized. For example, our group hasrecently reported large and significant cognitivechanges for up to 2 hr following administration of5.25mg oral midazolam (Collie et al., submitted).Other groups have reported similar findings withmidazolam administered intravenously (Thompsonet al., 1999; Rogers et al., 2002). The AE profileobserved in the current study is also very similar tothose reported previously. The results of the current

-2.50

-2.00

-1.50

-1.00

-0.50

0.00

COMPGMLT-DDETECTIDENTIFYLEARNGMLT

Sta

nd

ard

ized

co

gn

itiv

e ch

ang

e sc

ore

AE-negative

AE-positive

**

**

**

**

Figure 2. Group mean change from baseline on five cognitive tasks in subjects reporting adverse events (AE-positive; N¼ 14) and subjectsnot reporting AEs (AE-negative; N¼ 14) following administration of 1.75mg midazolam, �p< 0.05; ��p< 0.01

Table 3. Number of subjects fulfilling criteria for cognitivedysfunction following administration of 1.75 or 5.25mg oralmidazolam

Condition AE-positive AE-negative Total

1.75mg midazolamCognitive dysfunction 7 2 9No cognitive dysfunction 7 12 195.25mg midazolamCognitive dysfunction 16 6 22No cognitive dysfunction 3 3 6

Note: AE positive¼ those subjects for whom adverse events wererecorded; AE negative¼ those subjects for whom no adverse eventswere recorded.



study extend these findings and add to our under-standing of the interaction between sedation-relatedAEs and cognitive dysfunction. Cognitive testing andAE data from two doses of midazolam administered to28 healthy males were analyzed. Subjects whoreported sedation-related AEs following a dose of5.25mg oral midazolam also displayed very large andhighly significant changes across a range of cognitivefunctions including motor function, attention, strategyuse and problem solving, learning and delayed recall(Figure 1). Subjects who did not report sedation-related AEs at this dose nevertheless displayedsubstantial cognitive decline across all cognitivedomains assessed, although the small group sizeensured that this change was not sufficient to reachstatistical significance. At the lower dose of 1.75mg asimilar pattern was observed, however the magnitudeof cognitive changes was reduced (Figure 2). Thesefindings suggest that, at a group level, individuals whofeel subjectively that they are not experiencing anyadverse cognitive function nevertheless display sub-stantial cognitive deterioration following adminis-tration of midazolam.

Analysis of data at an individual level also producedvery interesting results. The majority (84.2%) ofsubjects reporting sedation-related AEs following

administration of 5.25mg midazolam also displayedcognitive deterioration. However, 66.7% of subjectswho did not report sedation-related AEs neverthelessdisplayed cognitive impairment at this dose (Table 3).In this case, the dissociation between AE reports andcognitive impairment was significant when assessedwith chi-square analysis. This is consistent withfindings from clinical studies. For example inneuropsychiatric studies of diseases such as HIV,Hep C, early Alzheimer’s disease, individual patientswho report that their cognition has declined rarelyhave detectable impairment on formal neuropsycho-logical testing (e.g., Kliegel et al., 2005). In the Phase Isetting, where healthy young men are typicallystudied, it is possible that some individuals are lessprone to report AEs despite subjective feelings ofimpairment. In such cases, objective cognitive testingmay help to identify such individuals.Importantly the current data indicates that cognitive

impairment will occur in individuals who do not reportAEs. This finding has implications for the conduct ofearly-phase clinical trials, especially those assessingthe safety of potentially CNS-active compounds. AErecordings are used routinely to make decisions aboutdose-escalation in early phase trials, and are one of thekey pieces of information considered by pharmaceutical

-5.00

-4.00

-3.00

-2.00

-1.00

0.00

1.00

2.00

0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3 2.5

Co

mp

osi

te c

og

nit

ive

chan

ge

sco

re

AE negative AE positive

Figure 3. Scatterplot of individual change from baseline on the composite cognitive score in subjects reporting adverse events(AE-positive; N¼ 19) and subjects not reporting AEs (AE-negative; N¼ 9) following administration of 5.25mg midazolam



regulatory authorities assessing compounds for market-ing approval. While larger studies with more statisticalpower will be required to confirm the result observedhere, the findings of this study suggest that AErecordings are not an adequate measure of cognitiveimpairment in early phase clinical trials. Inclusion ofcomputerized cognitive tests in such trials may allowidentification of cognitive dysfunction or improvement,beyond that which is possible by subject self-report andclinical observation.

ACKNOWLEDGEMENTS

This research was funded by Pfizer Inc. The authorswould like to thankCarolineWooldridge andDikNg forexpert operational support and the Erasme and Singa-pore Pfizer Phase I Unit staff for performing this study.

REFERENCES

Beglinger LJ, Gaydos BL, Kareken DA, Tangphao-Daniels O,Siemers ER, Mohs RC. 2004. Neuropsychological test perform-ance in healthy volunteers before and after donepezil adminis-tration. J Psychopharm 18(1): 102–108.

Beglinger LJ, Gaydos B, Tangphao-Daniels O, et al. 2005a. Practiceeffects and the use of alternate forms in serial neuropsychologicaltesting. Arch Clinl Neuropsychol 20(4): 517–529.

Beglinger LJ, Tangphao-Daniels O, Kareken DA, Zhang L, Mohs R,Siemers ER. 2005b. Neuropsychological test performance in healthyelderly volunteers before and after donepezil administration: arandomized, controlled study. J Clin Psychopharm 25(2): 159–165.

Bland M, Alman D. 1996. Measurement error. BMJ 313: 744.Cohen J. 1988. Statistical power analysis for the behaviouralsciences, 2nd edn. Erlbaum: Hillsdale, NJ.

Collie A, Maruff P, Darby DG, McStephen M. 2003. The effects ofpractice on the cognitive test performance of neurologicallynormal individuals assessed at brief test-retest intervals. J IntNeuropsychol Soc 9(3): 419–428.

Collie A, Darekar A,Weissgerber G, et al. 2005.Cognitive testing inearly phase clinical trials: Development of a rapid computerisedtest battery and application in a simulated Phase I study. Manu-script submitted for publication.

Falleti MG, Maruff P, Collie A, Darby DG, McStephen M. 2003.Qualitative similarities in cognitive impairment associated with24 h of sustained wakefulness and a blood alcohol concentrationof 0.05%. J Sleep Res 12: 265–274.

Girdler NM, Lyne JP, Wallace R, et al. 2002. A randomized,controlled trial of cognitive and psychomotor recovery frommidazolam sedation following reversal with oral flumazenil.Anaesthesia 57(9): 868–876.

GrundmanM, FarlowM, Peavy G, et al. 2002. A phase I study of AIT-082 in healthy elderly volunteers. J Mol Neurosci 18(3): 283–293.

Ingraham LJ, Aiken CB. 1996. An empirical approach to determin-ing criteria for abnormality in test batteries with multiplemeasures. Neuropsychology 10: 120–124.

Kanto JH. 1985. Midazolam: the first water-soluble benzodiazepine.Pharmacology, pharmacokinetics and efficacy in insomnia andanesthesia. Pharmacotherapy 5(3): 138–155.

Kliegel M, Zimprich D, Eschen A. 2005. What do subjectivecognitive complaints in persons with aging-associated cognitivedecline reflect? Int Psychogeriatr 17(3): 499–512.

Meyers CA, Hess KR, Yung WK, Levin VA. 2000. Cognitivefunction as a predictor of survival in patients with recurrentmalignant glioma. J Clin Oncol 18(3): 646–650.

Mollica CM, Maruff P, Vance P. 2004. Development of a statisticalapproach to classifying treatment response in individual childrenwith ADHD. Hum Psychopharm 19(7): 445–456.

Nordst SP, Clark RF. 1997. Midazolam: a review of therapeutic usesand toxicity. J Emerg Med 15(3): 357–365.

Rasmussen LS, Larsen K, Houx P, et al. 2001. The assessment ofpostoperative cognitive function. Acta Anaesthes Scand 45: 275–289.

Rogers JF, Morrison AL, Nafziger AN, Jones CL, Rocci ML, Jr,Bertino JS, Jr. 2002. Flumazenil reduces midazolam-inducedcognitive impairment without altering pharmacokinetics. ClinPharmacol Ther 72(6): 711–717.

Snyder PJ, Werth J, Giordani B, Caveney AF, Feltner D, Maruff P.2005. A method for determining the magnitude of changeacross different cognitive functions in clinical trials: theeffects of acute administration of two different doses of alpra-zolam. Hum Psychopharm 20(4): 263–273.

Thompson JM, Neave N, Moss MC, Scholey AB,Wesnes K, GirdlerNM. 1999. Cognitive properties of sedation agents: comparison ofthe effects of nitrous oxide and midazolam on memory and mood.Brit Dent J 187: 557–563.

van der Meyden CH, Bartel PR, Sommers DK, et al. 1992. Effect ofacute doses of controlled-release carbamazepine on clinical,psychomotor, electrophysiological, and cognitive parameters ofbrain function. Epilepsia 33(2): 335–342.



cognitive testing in early phase clinical trials: outcome according to adverse event profile in a...

Documents