Wilderness and Environmental Medicine, 20, 261 268 (2009)
ORIGINAL RESEARCH
Cognitive Assessment of a Trekking Expedition to 5100 m: A Comparison of Computerized and Written Testing Methods Gregory A. Harris, MBBS (Hons); Jennifer Cleland, PhD; Alex Collie, PhD; Paul McCrory, PhD From the Olympic Park Sports Medicine Centre, Olympic Boulevard, Melbourne, VIC, Australia (Dr Harris); Department of General Practice and Primary Care, University of Aberdeen, Foresterhill Health Centre, Westburn Road, Aberdeen, United Kingdom (Dr Cleland); Transport Accident Commission, Centre for Neuroscience, The University of Melbourne, Melbourne, VIC, Australia (Dr Collie); Centre for Health, Exercise and Sports Medicine, University of Melbourne, Parkville, VIC, Australia (Prof McCrory).*
Objective.—To assess a typical commercial trekking cohort for cognitive impairment after gradual ascent to 5100 m. Methods.—We performed a prospective, controlled, repeated-measures study within a trekking expedition to Nepal. A sample of expedition participants was studied; 36 were enrolled and 26 completed all testing. Additional normative data were sourced from sea level studies. Participants underwent cognitive assessment before travel with written, verbal, and computerized tests, then within 24 hours of arrival at 5100 m after an 18-day ascent from 400 m. Changes in performance in 6 written and 7 computerized tests were analyzed at an individual and group level using paired t tests. Effect size analysis was performed for individual performance. Results.—No individual demonstrated significant cognitive impairment at 5100 m. The subject group performed significantly better than the normative population in 3 of the 6 written tests. The group performed worse at 5100 m in 1 written test (digit span forwards, P ⬍ .01) and better in 2 written (digit-symbol substitution, P ⬍ .01; trail-making test, part B, P ⬍ .05) and 1 computerized test (monitoring test reaction time, P ⬍ .01). Performance was more variable in the written than the computerized tests. Conclusions.—Gradual ascent to high altitude causes no significant cognitive impairment in the majority of individuals. Computerized testing produced less variable results than written testing, but logistical difficulties are likely to preclude widespread use of such technology in the field. Key words: altitude, hypoxia, cognitive impairment, computerized testing.
Introduction The hypoxia of high altitude can produce measurable impairments in a number of cognitive domains. Rapid exposure to altitudes of 4000 m and above causes alterations in perceptual, learning, memory, and motor tasks.1,2 Impaired judgement and decision-making capacity may increase the risk of accident in dangerous terrain.3 Whether any long term cognitive damage is inCorresponding author: Gregory A. Harris, Senior Registrar, Australasian College of Sports Physicians, Olympic Park Sports Medicine Centre, Olympic Boulevard, Melbourne 3004, VIC, Australia (e-mail:
[email protected]). *For the duration of this study, Dr Collie was an employee and equity holder in CogState Ltd. The other authors claim no competing interests.
curred by altitude exposure remains unclear.4–9 Whereas simulated gradual ascents have found no significant cognitive impairments below 7000 m,2,6 there have been no field studies specifically designed to investigate the cognitive effects of a gradual ascent typical of commercial tours. One of the limitations to interpretation of altitude-related cognitive research is the vast diversity of methodologies reported.1,10 Virues-Ortega et al10 recently argued that a standardized neuropsychological testing battery for high altitude investigations should be based upon 4 criteria (viz, with known psychometric properties, being affected by altitude in at least 1 study, low practice and ceiling effects, and suitability of administration by nonexperts). This is particularly relevant at
262 moderate altitude, or with a gradual ascent, in which any cognitive changes could be expected to be subtle. Recent developments of computerized cognitive testing batteries have provided investigators with tools that are easy to use, have good psychometric properties for serial study, and that are designed to assess mild cognitive changes.11–13 Such batteries have been found to be sensitive to mild impairments in a variety of clinical settings and may be more sensitive than some of the written tests in common use.14 The use of computerized cognitive tests in a laboratory setting (eg, Operation Everest II, 1985) is well established.2 Use of such technology in the field is much less common, largely because of the logistical difficulties of transporting sensitive equipment and maintaining adequate power supply. Those studies that have used computerized tests in the field have investigated relatively rapid ascents such as to the Capanna Margherita (4559 m) or Pikes Peak (4300 m), where there is a standing laboratory. As the apparent strength of computerized testing seems to be in its sensitivity to mild impairment, it seems appropriate to utilize such a battery in a gradual ascent setting where cognitive changes would be expected to be mild at most. We assessed a cohort of subjects on a 6-week trek in Nepal, using both a computerized testing battery and a range of commonly used written tests. Our subjects were heterogeneous in age, previous high altitude experience, and general fitness. Some degree of pre-expedition aerobic training was recommended, as is typical of commercial guided tours. Our primary aim was to test the null hypothesis that there would be no difference in computerized measures of cognitive function with a gradual ascent to high altitude. Our secondary aims were to test the hypotheses that there would be no change in written test performance with gradual ascent and that there would be no difference between performance in computerized tests and written tests. In addition we examined the logistical issues related to field use of computerized tests. Methods Ethics permission for this study was granted by the Human Research Ethics Committee of the University of Melbourne, Australia, and Grampian Research Ethics Committee, Aberdeen, UK. PARTICIPANTS Participants in a medical research expedition (Medex Makalu 2003, Medical Expeditions [UK] Ltd, www. medex.org.uk) to eastern Nepal in April 2003 volun-
Harris et al
Figure. Ascent profile.
teered to participate in this study. All subjects were educated to tertiary level, and all spoke English as a primary language. All subjects were resident at low altitude, and none had traveled to altitudes of 3000 m or higher in the 6 months before the expedition. All participants had some previous trekking experience; about half had previously ascended to high altitudes (⬎4000 m). All subjects were naive to the cognitive tests utilized. Subjects provided written, informed consent before participation. Exclusion criteria were acute intercurrent illness including severe altitude illness (high altitude cerebral or pulmonary edema or severe Acute Mountain Sickness [AMS]). Self-reported symptoms of AMS were monitored at morning and evening mealtimes throughout the expedition using the Lake Louise AMS scoring system.15 EXPEDITION DESIGN Participants assembled in Kathmandu for 3 to 5 days before flying to Tumlingtar (410 m) in Eastern Nepal. Groups of 9 to 10 participants then underwent an 18day ascent to the remote Chamlang Basecamp ([BC] 5100 m), staggered at 3 to 4 day intervals and supported by Sherpa guides and porters (the Figure). The ascent profile was consistent with recommended guidelines to prevent altitude illness.16 The greatest altitude attained during the trek was in traversing a 5400 m pass on Day 17. At no time before arrival at base camp was the sleeping altitude higher than that of base camp. Participants remained at BC for 4 to 10 days. Research equipment was packed into 120 L lockable plastic barrels in Kathmandu and transported to and from BC by helicopter, where it was met by the first trekking group. Computers were stored in insulated, lockable aluminum cases and kept in the lead researcher’s (GH) tent overnight to limit exposure to low overnight temperatures. Power supply for BC was provided by a combination of solar array 7⫻ 32W flexible solar panels (Unisolar
Cognitive Assessment on Expedition USF-32, United Solar Ovonic Europe GmbH, Frankfurt/ Main, Germany), 4⫻ 20W semi-rigid solar panels (Solarex MSX20L, BP Solar USA, Frederick, MD), with 2⫻ 20A regulators (BZ Products M20, BZ Products, St. Louis, MO), and wind turbine 1⫻ 250W (Wind Charger Rutland 913, Marlec Engineering Co Ltd, Corby, Northants, UK), run through 2⫻ 110 V, 1.3 kW mains inverters (Whistler PP1250, Whistler Group, Bentonville, AR). Power storage was provided by 6⫻ 50AHr sealed lead-acid batteries (Hawker SBS-60, Hawker Batteries, Chippenham, Wiltshire, UK). The power system was designed with redundancy: 2 entirely separate systems were set up, with one (using a single battery) providing power for the BC radio, and the other system, using the remainder of the batteries, providing power to the research tents. The system was designed such that it could easily be rewired in the event of any items failing. TESTING MATERIALS We utilized a range of cognitive function tests that have been previously found to be sensitive to the effects of altitude hypoxia: namely motor speed, verbal and visual learning and memory, verbal fluency, planning, and mental flexibility.1,10,17 The written testing battery took 30 minutes to perform, and the computerized test took 15 to 20 minutes to perform. Alternate forms of written tests were used to minimize learning effects. Written tests Digit Span Test: Subjects are asked to recall a list of random digits read out at a rate of one per second. In digit span forwards, subjects repeat the digits as presented; this measures the efficiency of attention. In digit span backwards the subjects repeat the digits in reverse order; this is a test of working memory.18 The score is the length of the longest string of digits correctly presented. A score greater than 6 (forwards) or 4 (backwards) is considered normal.18 Digit Symbol Substitution Test (DSST): Derived from the Wechsler Adult Intelligence Scale -Revised, this test measures attention, response, and processing speed and visuomotor coordination.18 The number of correct responses in 90 seconds is recorded. Trail-making Test, Part B (TMT-B): A test of visual and conceptual scanning, motor speed, and agility, where subjects must consecutively connect numbered and lettered circles as fast as possible.18,19 The time to completion is recorded. Rey’s Auditory-Verbal Learning Test (AVLT): Subjects are presented with and asked to recall a 15-word list on 5 occasions, then presented with one 15-word
263 intereference list followed by a sixth recall trial of the first list. Retention is assessed 30 minutes later. This is a well-validated test of verbal learning and memory.18 The sum of the 5 primary word list presentations (AVLTsum) and the delayed recall trial (AVLTrecall) were the primary measures assessed. Controlled Oral Word Association Test (COWA): Subjects are asked to provide as many words beginning with a given letter within a minute, excluding proper nouns, numbers, and the same word with a different suffix. Three trials comprise each test session, and normative data are available for 3 sets of letters: CFL, PRW, and FAS. The sum of all acceptable words produced in the three trials (COWAsum) was recorded. This is a test of verbal fluency and hence of frontal lobe function.18,20 Computerized test CogState (CogState Ltd, Carlton, VIC, Australia, www.cogstate.com) is a computerized cognitive test battery that is designed to assess changes in cognitive function and has good psychometric properties for serial study.11,12 It is easily administered by nonexperts and has been shown to be sensitive to mild cognitive impairment from a variety of causes.21–26 Seven subtests of the CogState battery were used: simple reaction time ([SRT] measuring psychomotor speed) and choice reaction time ([ChRT] measuring decision making and psychomotor speed), monitoring task reaction time and accuracy (measuring attention), working memory task reaction time and accuracy, and learning task accuracy. These tasks assess cognitive domains previously found to be affected by high altitude.1,10 Power calculations As we could not identify any directly comparative effect size analyses, we chose to assume that the cognitive impairment at 5000 m altitude would be similar to that occuring after 24 hours of sustained wakefulness. Maruff found that 24 hours of sustained wakefulness induced a clinically and statistically significant change with an effect size of Cohen’s d ⫽ 1.20 on the CogState simple reaction time task and greater than 1.0 on the CogState choice and working memory tasks with 40 subjects.22 In an analysis performed before the recruitment of participants, we calculated that 26 participants should be sufficient to detect a significant effect (Cohen’s d ⫽ 1.0, SRT task). TESTING PROTOCOLS Participants underwent baseline testing in a quiet room at sea level in London, January 2003. Written tests were
264 administered according to standard protocols. CogState testing took place in a separate, quiet room. Two subjects were tested simultaneously on laptop computers (Toshiba Tecra 8100, Intel Celeron 597 MHz processor, 256MB RAM, Microsoft Windows 98SE) with CogState (version 2.2.0). A practice test was performed before a formal baseline test. This has been found to provide a stable baseline from which there are no significant learning effects on repeated testing at intervals of up to 2 years.27,28 The format of the test battery has been described in detail previously.13,28 All tests were readministered within the first 24 hours of arrival at BC. Written tests were performed in a 3 ⫻ 3.5 m A-frame tent erected at a short distance from the main camp to ensure privacy and minimize distractions. CogState testing took place in a separate tent, again with 2 subjects tested simultaneously on the same machines as at baseline. All tests were administered by 1 of 2 of the authors (JC or GH) at baseline and on expedition. Blinding was achieved through coded deidentification of participants throughout the study. This code was revealed postanalysis. STATISTICAL METHODS Written tests Baseline function was compared to published age-, sex-, and education-matched normative data.29 Day 1 BC performance for each test was compared to baseline using a series of paired, 1-tailed t tests with significance set at P ⬍ .05. An individual change score was standardized against the group baseline performance for each test. Performance changes greater than 2 SD were considered significant.
Harris et al Table 1. Demographic data* Subjects
Norms
26 15:11
411 379:32
34.9 (23–53) 32.5 (25–40)
34.2 (25–55) 34.2 (25–40)
100%
97% 3%
n M:F age (y), mean (range) M F Education Tertiary/higher Secondary *M indicates male; F, female.
data to provide a reliable change index for individual response to altitude. For each subject, the magnitude of any observed changes from baseline to BC was expressed as a z-score, where the difference was divided by the within-subject standard deviation (WSD) of the normative group (WSDnorm).25,26,31 Z-scores less than ⫺1.64 (P ⬍ .05 1-tailed) were considered a significant impairment. Using the binomial probability tables of Ingraham and Aitken, subjects were classified as cognitively impaired at BC if 2 or more variables demonstrated a significant impairment at that time.25,32 To enable assessment of a global cognitive change across a number of variables, a composite z-score was also calculated for each subject at BC. This composite z-score was calculated by summation of the 7 individual z-scores, divided by the SD of the normative composite scores.25 Composite z-scores less than ⫺1.64 (P ⬍ .05 1-tailed) were considered a significant impairment. Comparison of computerized and written tests
Computerized tests Reaction times were normalized using a logarithmic (base 10) transformation and accuracy scores with an arcsine transformation.30 These are standard transformations for presenting such data in previously published studies.25,26 Subjects’ baseline test performance was compared to age- and sex-matched normative data drawn from the CogState database (unpublished data, CogState, 2008) (n ⫽ 411, M ⫽ 379, F ⫽ 32). These data had been collected at sea level from healthy subjects educated to either secondary or tertiary level. The group BC performance for each subtest was compared to baseline using a series of paired, 1-tailed t tests with significance set at P ⬍ .05. Additional analysis was performed with the CogState
Coefficients of variation were calculated for each test at baseline to provide some indication of measurement error.31 Results PARTICIPANTS Thirty-six subjects completed baseline tests; 26 were tested at Day 1 at BC (Table 1). Drop-outs were a result of the staggered nature of the trek, with 10 subjects reaching BC before the investigators. No subjects were excluded because of altitude illness or other exclusion criteria. The subject group performed significantly better at baseline than age-, sex-, and education-matched normative data in digit span tests forwards and backwards
Cognitive Assessment on Expedition
265
Table 2. Normative, baseline, and base camp scores (mean [SD]) and coefficients of variation at baseline for written and CogState tests, categorized by cognitive domain tested (n ⫽ 26)†‡
Cognitive domain Motor speed Attention Decision making Working memory
Learning Frontal (language/ executive)
Cognitive test
Test type
TMT-B (s) SRT (RT) Digit span forwards Monitoring (RT) DSST ChRT (RT) Digit span backwards AVLTrecall Working memory (RT) Working memory (Acc) AVLTsum Learning (Acc) COWAsum Monitoring (Acc)
Written CogState Written CogState Written CogState Written Written CogState CogState Written CogState Written CogState
Population normative values 56.84 2.44 6.81 2.49 68.85 2.66 4.98 11.73 2.82 1.18 54.67 0.95 47.56 1.09
(21.92) (0.10) (1.11)*** (0.24) (11.37)*** (0.09) (1.33)*** (2.37) (0.15) (0.17) (8.41) (0.18) (11.18) (0.11)
Subject Baseline mean Base camp mean Baseline (SD) (SD) COV 53.23 2.43 12.08 2.56 77.69 2.67 8.69 11.58 2.81 1.23 55.19 1.01 49.81 1.12
(27.81) (0.04) (2.62) (0.09) (12.59) (0.10) (3.04) (2.53) (0.10) (0.20) (10.42) (0.13) (16.13) (0.14)
44.38 2.44 9.92 2.49 87.38 2.66 8.15 10.85 2.79 1.20 53.69 0.99 51.92 1.10
(16.03)* (0.05) (2.26)*** (0.09)** (13.57)** (0.08) (2.65) (2.80) (0.08) (0.16) (9.12) (0.14) (15.49) (0.16)
52% 1.70% 22% 3.50% 16% 3.60% 35% 22% 3.60% 16% 19% 13% 32% 12%
†COV indicates coefficient of variation; TMT-B, Trail-making test (Part B); SRT, simple reaction time; RT, log10(reaction time); DSST, DigitSymbol Substitution Test; ChRT, choice reaction time; AVLT, Rey’s Auditory Verbal Learning Test; Acc, arcsin(%accuracy); COWA, Controlled Oral Word Association; ‡TMT-B and CogState RT scores: decreased score ⫽ improvement; Digit span, DSST, AVLT, and CogState Acc scores: increased score ⫽ improvement. *Significantly better than baseline, P ⬍ .05; **Significantly better than baseline, P ⬍ .01; ***Significantly worse than baseline, P ⬍ .01.
and DSST (Table 2) There were no differences between the subject and normative groups for any other baseline tests. PRIMARY OUTCOMES: COMPUTERIZED TEST VS ALTITUDE Group performance Significant improvements in score from baseline to BC were detected in the CogState monitoring test reaction time (P ⬍ .01) (Table 2). Individual performance No subjects demonstrated significant overall cognitive changes from baseline to BC. Five subjects recorded worsened cognitive performance in 1 CogState test only (ie, z-score ⬍ ⫺1.64): 4 subjects (3 male, 1 female) performed worse in learning test accuracy. One other male subject performed worse in working memory accuracy. None of these subjects reported any illness or significant AMS symptoms at BC or during the trek, nor were they using any medications, including acetazolamide. No subjects improved from baseline to BC on any CogState tests.
SECONDARY OUTCOMES Written tests vs altitude Group performance: There was a significant worsening in digit span forwards test score from baseline to BC (P ⬍ .01), but the subject group remained better-performed than the normative data. There were significant improvements in the DSST (P ⬍ .01) and and trail-making test (Part B) (P ⬍ .05) (Table 2). Individual performance: One male subject recorded an isolated, significantly worsened digit span forwards test. One female subject recorded an isolated, significantly worsened AVLTrecall test. Four subjects significantly improved in DSST and one subject significantly improved in the trail-making test (Part B). Computerized vs written tests The written tests displayed much greater variability within both the normative population and the study group, with high coefficients of variation compared to the CogState tests. There was no apparent congruency in group performance between testing modalities within cognitive domains (Table 2), and of the subjects displaying significant improvements in written test perfor-
266 mance none showed any change from baseline in any CogState test. Discussion Recent increased interest and ease of access to high altitude destinations brings an increasing number of people into a potentially dangerous environment. The risk to cognitive function in such an environment remains poorly understood, in part through a lack of investigations specifically designed to assess the exposure typical of commercial tours. This study aims to address some of the shortfall in relevant data. Our subjects demonstrated no significant cognitive impairments with CogState upon arrival at 5100 m, either as a group or individually. This is consistent with the majority of published investigations into simulated altitude, in which gradual exposure has not produced cognitive deficits below 7000 m.2,6 Of the comparable field studies, Kramer et al assessed subjects after a significantly different ascent profile (testing at 4360 m after ascent to 6194 m).33 Pagani et al found deficits in a visual learning task in 9 subjects after a 10 to 13 day trek to 5350 m, but did not report any baseline investigations nor the subjects’ previous altitude exposure.34 Whereas the group performed worse at BC in the digit span forwards task, they remained better performed than both the normative population and recommended ‘‘normal’’ values. As an isolated finding this seems to have little clinical relevance, particularly with the concurrent improvements in another test of attention (CogState monitoring RT) and highlights the difficulty in interpreting single tests. The concurrent significant improvements detected in the DSST and trail-making test (Part B) serve further to highlight the difficulty in interpretation of these tests in the setting of mild cognitive impairment. Similar paradoxical improvements in the DSST and trail-making test (Part B) have been detected following athletic concussion, with the authors suggesting this as evidence of a significant learning effect despite the use of alternate forms and a practice test prebaseline.12 Previous investigations into hypoxia with these tests have also produced conflicting findings. Evans and Witt reported deficits in the DSST at 4300 m, whereas Kennedy et al found no changes with gradual decompression up to 8845 m.2,35 Berry et al found a failure to improve in the DSST in hypoxic subjects when compared to normoxic controls, but found no difference in the digit span or trail-making test (Part B).36 Although not powered to detect correlations between testing methods, this study found no apparent overlap in performance in written and computerized tests. It is likely that these findings are a manifestation of the inherent
Harris et al variability of the tests, which was greater for the written tests, both at a group and an individual level. Our group performed equivalently at baseline to the normative population for all CogState tests, but better than the normative population in 3 written tests. This may be because the majority of participants were highly educated fellow investigators or again may be a result of the inherent variability of the tests. Given that assessment was made at an individual level, where no subjects were impaired at BC with CogState and only 2 with a single written test, this group finding should not limit the generalizability of the findings to a population with a wider range of educational attainment. It does, however, highlight the importance of individualized assessment and baseline testing. An ideal test of cognitive function in a remote environment would be portable, easy to interpret, rapid, and provide clinically relevant, individualized information. Although a comprehensive discussion of the psychometric properties of the computerized and written tests is beyond the scope of this article, the statistical treatment of the computerized data arguably enables a more robust and more easily interpreted measure of individual performance.10,18,37 At a practical level, the immediate access to results is essential for test utility. Whereas written tests may provide immediate results, the clinical interpretation of a change in score may be difficult, particularly in a situation where the assessor may also be at risk of some cognitive impairment. The currently available computerized testing systems either require an internet connection to upload tests and download reports or are not available for a hand-held platform, and consequently are not entirely appropriate for use in remote locations. Other logistical difficulties limiting the use of computers in the field include power supply, performance of hardware in extreme cold, and transport of electronic equipment. The sun reached BC at around 7:30 AM, and research generally was not started until 9 AM or 10 AM, allowing some charge into the camp batteries before any power was used. Peak power from the solar panels was about 400 watts, dropping rapidly when afternoon clouds rolled in. The wind charger typically provided about 10 to 15 watts during the normally light winds that continued for most of the 24 hours. A smaller, more mobile expedition may not have the luxury of being able to wait for climatic conditions to be suitable for power supply. Due to the cold conditions, computer batteries failed to maintain charge for more than 30 minutes, which was barely long enough to perform and save one test. Consequently, power supply was needed for all testing. Despite our computers being stored overnight within the investigators’ sleeping bags, they still required 30
Cognitive Assessment on Expedition minutes of direct sunlight each morning to start functioning properly. Again, a nonresearch expedition may not have this time nor have the benefits of a helicopter drop of equipment. In summary, we have found that a gradual ascent to high altitude causes no clinically significant cognitive impairment in the majority of individuals. Although advances in computer-based cognitive testing are promising, logistical difficulties continue to preclude a widespread use. The ideal of a simple, portable, reliable, and relevant cognitive test for field use is not yet available. Acknowledgments The authors would like to thank Medex (UK) for selecting this project to be part of the ‘‘Makalu 2003’’ Expedition and for extensive logistical and administrative support, and CogState Ltd for supplying data analysis and normative data. No funding was received for this study. CogState Ltd provided data analysis free of charge. References 1. Bahrke MS, Shukitt-Hale B. Effects of altitude on mood, behaviour and cognitive functioning: a review. Sports Med. 1993;16:97–125. 2. Kennedy RS, Dunlap WP, Banderet LE, et al. Cognitive performance deficits in a simulated climb of Mount Everest: Everest Operation II. Aviat Space Environ Med. 1989;60:99–104. 3. Nelson TO, Dunlosky J, White DM, et al. Cognition and metacognition at extreme altitudes on Mount Everest. J Exp Psychol Gen. 1990;119:367–374. 4. Regard M, Oelz O, Brugger P, et al. Persistent cognitive impairment in climbers after repeated exposure to extreme altitude. Neurology. 1989;39:210–213. 5. Hornbein TF, Townes BD, Schoene RB, et al. The cost to the central nervous system of climbing to extremely high altitude. N Engl J Med. 1989;321:1714–1719. 6. Townes BD, Hornbein TF, Schoene RB, et al. Human cerebral function at extreme altitude. In: West JB, ed. High Altitude and Man. Bethesda, MD: American Psychological Society; 1984:31–36. 7. Clark CF, Heaton RK, Weins AN. Neurophysiological functioning after prolonged high altitude exposure in mountaineering. Aviat Space Environ Med. 1983;54:202– 207. 8. Cavaletti G, Moroni R, Garavaglia P, et al. Brain damage after high-altitude climbs without oxygen. Lancet. 1987; 1:101. 9. Hornbein T. Long term effects of high altitude on brain function. Int J Sports Med. 1992;13:S43–S45. 10. Virue´s-Ortega J, Buela-Casal G, Garrido E, et al. Neuropsychological functioning associated with high-altitude exposure. Neuropsychol Rev. 2004;14:197–224.
267 11. Collie A, Darby D, Maruff P. Computerized cognitive assessment of athletes with sports related head injury. Br J Sports Med. 2001;35:297–302. 12. Makdissi M, Collie A, Maruff P, et al. Computerized cognitive assessment of concussed Australian Rules footballers. Br J Sports Med. 2001;35:354–360. 13. Westerman R, Darby DG, Maruff P, et al. Computer-assisted cognitive function assessment of pilots. Australian Defence Force Health. 2001;2:29–36. 14. Collie A, Maruff P, Makdissi M, et al. CogSport: reliability and correlation with conventional cognitive tests used in post-concussion medical examinations. Clin J Sports Med. 2003;13:28–32. 15. Roach RC, Ba¨rtsch P, Hackett PH, et al. The Lake Louise mountain sickness scoring system. In: Sutton JR, Houston CS, Coates G eds. Hypoxia and Mountain Medicine. Burlington VT: Queen City Printers; 1993:272–274. 16. Ward MP, Milledge JS, West JB. Altitude Acclimatisation. In: Ward MP, Milledge JS, West JB, eds. High Altitude Medicine and Physiology. 3rd ed. London: Arnold Publishers; 2000:44–49. 17. Lieberman P, Protopapas BS, Kanki BG. Speech production and cognitive deficits on Mt. Everest. Aviat Space Environ Med. 1995;66:857–864. 18. Lezak MD, Howieson DB, Loring DW, et al. Neuropsychological Assessment. 4th ed. New York, NY: Oxford University Press; 2004. 19. Schear JM, Sato SD. Effects of visual acuity and visual motor speed and dexterity on cognitive test performance. Arch Clin Neuropsychol. 1989;4:25–32. 20. Benton AL, Hamsher K deS, Sivan AB. Multilingual Aphasia Examination (kit). 3rd ed. Iowa City: AJA Assocs; 1994. 21. Darby D, Maruff P, Collie A, et al. Mild cognitive impairment can be detected by multiple assessments in a single day. Neurology. 2002;59:1042–1046. 22. Maruff P, Falleti MG, Collie A, et al. Fatigue-related impairment in the speed, accuracy and variability of psychomotor performance: comparison with blood alcohol levels. J Sleep Res. 2005;14:21–27. 23. Snyder PJ, Werth J, Giordani B, et al. A method for determining the magnitude of change across different cognitive functions in clinical trials: the effects of acute administration of two different doses alprazolam. Hum Psychopharm Clin Exp. 2005;20:263–273. 24. Maruff P, Snyder P, McStephen M, et al. Cognitive deterioration associated with an expedition in an extreme desert environment. Br J Sports Med. 2006;40:556–560. 25. Mollica C, Maruff P, Vance A. Development of a statistical approach to classifying treatment response in individual children with ADHD. Hum Psychopharm Clin Exp. 2004; 19:445–456. 26. Silbert BS, Maruff P, Evered LA, et al. Detection of cognitive decline after coronary surgery: a comparison of computerised and conventional tests. Br J Anaesth. 2004; 92:814–820. 27. Falleti MG, Maruff P, Collie A, et al. Practice effects as-
268
28.
29.
30. 31. 32.
Harris et al sociated with the repeated assessment of cognitive function using the CogState battery at 10-minute, one week and one month test-retest intervals. J Clin Exp Neuropsychol. 2006;28:1095–1112. Collie A, Maruff P, Darby DG, et al. The effects of practice on the cognitive test performance of neurologically normal individuals assessed at brief test-retest intervals. J Int Neuropsychol Soc. 2003;9:419–428. Mitrushina M. Handbook of Normative Data for Neuropsychological Assessment. 2nd ed. New York, NY: Oxford University Press; 2005. Winer BJ ed. Statistical Principles in Experimental Design. London: McGraw-Hill; 1971. Bland M. An Introduction to Medical Statistics. 3rd ed. Oxford: Oxford University Press; 2000. Ingraham L, Aitken CB. An empirical approach to deter-
33. 34. 35.
36.
37.
mining criteria for abnormality in test batteries with multiple measures. Neuropsychology. 1996;10:120–124. Kramer AF, Coyne JT, Strayer DL. Cognitive function at high altitude. Hum Factors. 1993;35:329–344. Pagani M, Ravagnan G, Salmaso D. Effect of acclimatisation to altitude on learning. Cortex. 1998;34:243–251. Evans WO, Witt NF. The interaction of high altitude and psychotropic drug action. Psychopharmacologia. 1966;10: 184–188. Berry DTR, McConnell JW, Phillips BA, et al. Isocapnic hypoxemia and neuropsychological functioning. J Clin Exp Neuropsychol. 1989;11:241–251. Collie A, Maruff P, McStephen M, et al. Psychometric issues associated with computerized neuropsychological assessment of concussed athletes. Br J Sports Med. 2003; 37:556–559.
