Author's personal copy...ghost rider (last best ride). During Month 3, participants were instructed...

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Exergaming and Older Adult CognitionA Cluster Randomized Clinical Trial

Cay Anderson-Hanley, PhD, Paul J. Arciero, DPE, Adam M. Brickman, PhD,Joseph P. Nimon, BS, Naoko Okuma, BS, Sarah C. Westen, BS, Molly E. Merz, BS,

Brandt D. Pence, BA, Jeffrey A. Woods, PhD, Arthur F. Kramer, PhD, Earl A. Zimmerman, MD

Background: Dementia cases may reach 100 million by 2050. Interventions are sought to curb orprevent cognitive decline. Exercise yields cognitive benefıts, but few older adults exercise. Virtualreality–enhanced exercise or “exergames” may elicit greater participation.

Purpose: To test the followinghypotheses: (1) stationary cyclingwith virtual reality tours (“cybercycle”)will enhance executive function and clinical status more than traditional exercise; (2) exercise effort willexplain improvement; and (3) brain-derived neurotrophic growth factor (BDNF) will increase.

Design: Multi-site cluster randomized clinical trial (RCT) of the impact of 3months of cybercyclingversus traditional exercise, on cognitive function in older adults. Data were collected in 2008–2010;analyses were conducted in 2010–2011.

Setting/participants: 102 older adults from eight retirement communities enrolled; 79 wererandomized and 63 completed.

Interventions: A recumbent stationary ergometer was utilized; virtual reality tours and competi-tors were enabled on the cybercycle.

Main outcome measures: Executive function (Color Trails Difference, Stroop C, Digits Back-ward); clinical status (mild cognitive impairment; MCI); exercise effort/fıtness; and plasma BDNF.

Results: Intent-to-treat analyses, controlling for age, education, and cluster randomization, re-vealed a signifıcant group X time interaction for composite executive function (p�0.002). Cyber-cycling yielded a medium effect over traditional exercise (d�0.50). Cybercyclists had a 23% relativerisk reduction in clinical progression toMCI. Exercise effort and fıtness were comparable, suggestinganother underlyingmechanism.A signifıcant groupX time interaction for BDNF (p�0.05) indicatedenhanced neuroplasticity among cybercyclists.

Conclusions: Cybercycling older adults achieved better cognitive function than traditional exercis-ers, for the same effort, suggesting that simultaneous cognitive and physical exercise has greaterpotential for preventing cognitive decline.

Trial registration: This study is registered at Clinicaltrials.gov NCT01167400.(Am J Prev Med 2012;42(2):109–119) © 2012 American Journal of Preventive Medicine

Introduction

Dementia is a growing global epidemic with sub-stantial personal, social, and economic costs1

and has led to calls for interventions to preventor slow cognitive decline.2,3 Cross-sectional research sug-

gests physical exercisemay prevent or delay dementia,4–6

andmeta-analyses demonstrate that physical exercise im-proves cognitive function in normal aging7,8 and in de-mentia.9 Recent research has extended these fındings toolder adults with mild cognitive impairment10–12 whosedefıcits are beyond those expected for their age but that

From the Healthy Aging and Neuropsychology Lab, Department of Psychol-ogy (Anderson-Hanley, Arciero, Nimon, Westen, Merz), Union College,Schenectady; the Health and Exercise Sciences Department (Anderson-Hanley, Arciero, Okuma), Skidmore College, Saratoga Springs; the TaubInstitute for Research onAlzheimer’sDisease and theAgingBrain,Depart-ment of Neurology (Brickman), College of Physicians and Surgeons, Co-lumbia University, New York; Albany Medical Center, Department ofNeurology (Zimmerman), Albany, New York; and the Department of

Kinesiology (Pence, Woods), the Beckman Institute for Advanced Scienceand Technology (Kramer), University of Illinois, Urbana-Champaign,Illinois

Address correspondence to: Cay Anderson-Hanley, PhD, Departmentof Psychology, Union College, 807 Union Street, Schenectady NY 12308.E-mail: [email protected].

0749-3797/$36.00doi: 10.1016/j.amepre.2011.10.016

© 2012 American Journal of Preventive Medicine • Published by Elsevier Inc. Am J Prev Med 2012;42(2):109–119 109


do not interfere with daily living and yetmay be a precur-sor to dementia.

Further, evidence is accumulating that cognitive bene-fıts may be achieved by way of improved neuronal func-tions, including neurogenesis, shown by concomitantstructural and functional changes in the brain,13–17 im-pacts on biomarkers of Alzheimer’s disease,18,19 and in-creases in brain-derived neurotrophic growth factor(BDNF).10,14,19,20 Cognitive benefıt from exercise isfound primarily in executive control and frontal lobefunctions, such as planning, divided attention, and inhi-bition of responses.8,21,22 These abilities are often im-paired in dementia and are key to maintaining indepen-dence and delaying institutionalization.

The demonstrated cognitive and health benefıts of ex-ercise are such that the American College of Sports Med-icine (ACSM) and the American Heart Association(AHA) upgraded recommended daily exercise.23 Yet datafrom the CDC Healthy People 2010 Database indicatethat only 14% of adults aged 65–74 years and 7% ofthose aged �75 years reported regular exercise. Physi-cian prescription of exercise24 has not been shown tosubstantially increase participation;�4% of patients inone study complied.25 These data suggest the need formore-compelling interventions to increase the moti-vation of older adults to exercise, as well as multimodalinterventions that address the multiple defıcits fromphysical inactivity.26

Virtual reality–enhanced exercise or “exergames”combine physical exercise with computer-simulated en-vironments and interactive videogame features and havebecome popular as a means to promote healthy behav-iors27 and increase the appeal of exercise (e.g., theWii Fitand PlayStation Move).28 Exergames have the potentialto increase exercise by shifting attention away from aver-sive aspects and towardmotivating features such as com-petition and three-dimensional (3D) scenery. Participa-tion in exergaming compared with traditional exercisecan lead to greater frequency and intensity29 and en-hanced health outcomes.28,30,31 A recent study32 reportedthat compared with traditional stationary cycling, olderadults preferred cycling with interactive gaming.

Although promising, there are limited published dataon whether interactive exergaming technologies are reli-ably associated with enhanced physical and cognitivehealth outcomes, and more-controlled research on theeffects of health games is needed.27,33 One early study34

investigated virtual reality–enhanced stationary cyclingusing virtual tours and on-screen competition (referredto here as “cybercycling”) and found cognitive improve-ment in patients with traumatic brain injury. However,without a traditional exercise control group, it is unclear

whether cybercycling yielded cognitive benefıt beyondphysical exercise alone.

While there are reports of the psychological benefıts ofcybercycling,29,30,35 no previous randomized clinical trial(RCT) has evaluated the cognitive benefıts of virtualreality–enhanced exercise. Presented herein are results ofthe Cybercycle Study, a multi-site cluster RCT in whichthe cognitive benefıt of cybercycling was compared withtraditional stationary cycling, for older adults living inde-pendently. On the basis of prior research8,21,22 showingprimarily executive function gains from exercise, it washypothesized that cybercycling would yield greater exec-utive function. Further, it was hypothesized that anychange would be due to increased exercise effort spurredon by engaging interactive virtual tours, competition, andadded mental challenge. Secondary analyses examinedchange in BDNF as a biomarker indicating possible neu-roplasticity, which has been implicated as amechanismofchange linking exercise to cognition.10,14,18–20

MethodsDesign

This cluster RCT (2008–2010) compared the impact on executivefunction of two exercise interventions: physical exercise alone andphysical plus mental challenge as combined in an exergame.

Setting and Participants

Participants were recruited by fliers and information sessions ateight independent living facilities. The facilities were chosen be-cause of proximity to investigator institutions; similarity in size(average 100–200 residents); and presence of contiguous livingareas to ensure indoor access to a study bike (to minimize barriersassociated with travel). Participants volunteered based on demon-strations of cybercycle functionality, not knowing which conditionthey would be randomized to, but aware that all could use thecybercycle after the 3-month intervention. Volunteers aged �55years were screened; exclusion criteria were known neurologicdisorders (e.g., Alzheimer’s or Parkinson’s) and functional disabil-ities that would substantially restrict participation in cognitivetesting or exercise. Written physician approval was required.Union and Skidmore Colleges’ IRBs approved the study; partic-

ipants provided written informed consent. A priori sample sizeestimates were calculated based on published effect sizes for cogni-tive (d�0.48)8 and physiologic (d�0.41)36 outcomes fromphysicalexercise. An a priori power analysis had found that in a 2 � 2(group � time) design, a sample of 100 would achieve 0.82 powerto detect a signifıcant effect (p�0.05). However, for logistic rea-sons, the study design was changed from individual to clusterrandomization. Post hoc statistical power is reported in the Resultssection.

Interventions

Participants in the cybercycle and control conditions rode identicalrecumbent stationary bikes, except for the virtual reality displaythat was enabled on the cybercycle (Appendixes A and B, availableonline at www.ajpmonline.org). Participants were trained in the

110 Anderson-Hanley et al / Am J Prev Med 2012;42(2):109–119

www.ajpmonline.org


use of the bike, log-in procedures, and paper log for recording ridestatistics as a backup to the computer. Participants were given atarget heart rate range to maintain during exercise using the HeartRateReserve (HRR)method23;mid-intervention adjustmentsweremade to maintain a relative HRR of 60%.A 1-month familiarization period allowed participants to learn

to attend to continuous biofeedback information for safety (e.g.,heart rate), before introducing distracting virtual tours in the cy-bercycle condition. Participants were instructed to gradually in-crease exercise frequency to 45 minutes per session fıve times perweek consistent with the ACSM and AHA recommendations.23

Individual progress reports and leaderboards were posted weeklyto control goal-setting and competition across interventions. Par-ticipants were asked to hold constant other lifestyle factors (e.g.,diet and other physical activity) during their study participation toisolate the effect of the interventions. The minimum threshold for“completers” was 25 rides during the intervention period; thus“completers” rode an average of three rides per week minus 2weeks’ allowance for illness, holidays, or equipment repair.

Cybercycle group. After 1month of familiarization, cybercycleparticipants experienced 3D tours and competed with their own“ghost” rider (last best ride). During Month 3, participants wereinstructed to outpace on-screen riders.

Control group. After 1 month of familiarization, controls con-tinued to ride the traditional stationary bike viewing biofeedbackinformation (e.g., heart rate and mileage). Each month, placebotraining (e.g., hydration and stretching) matched the attentiongiven to the cybercycle group.

Randomization. A priori plans were for individual randomassignment through software controls, but equipment problems,combined with limited funding and space, led to cluster assign-ment in order to limit cross-condition contamination. Sites wereselected by random draw. Cluster random assignment achievedsimilar levels of cognitive function and physiologic status at pre-test, although the groups differed in age and education, which wereentered as covariates in analyses (p�0.002 and p�0.001, respec-tively; Table 1).

Main Outcome Measures

Cognitive assessment. Cognitive testing was done at enroll-ment (baseline), 1 month later (pre-intervention), and 3 monthslater (post-intervention). Analyses were conducted using pre- andpost-scores. Baseline testing minimized the impact of practice andlearning effects associated with serial assessments and provided amore stringent test of the main hypothesis.37 Blinded ratings wereachieved inmost cases. The primary cognitive outcome of interest,executive function, was assessed via Color Trails 2-1 differencescore (time to connect alternating color and number dots, minustime to connect only numbered dots)38; Stroop C (time to namecolor of ink of contrasting color word)39; and Digit Span Back-wards (number of correct trials repeating a string of numbers inreverse order).40

To reduce the number of statistical comparisons, an executivefunction composite score was obtained by converting raw scoreson each test to z-scores using the grand mean and SD across bothgroups for each time point, then averaging the three measures(Cronbach’s ��0.67). Timed tasks were reversed; a positive valueon the composite indicates a score above the mean. Secondary

Table 1. Baseline characteristics of trial participants, M(SD) unless otherwise indicated

Cybercycle(n�38)

Control bike(n�41)

Age (years)a 75.7 (9.9) 81.6 (6.2)

Women (n [%]) 33 (70.7) 29 (86.8)

Education (years)a 12.6 (2.2) 14.8 (2.3)

Physiologic factors

Weight (kg) 75.0 (13.1) 72.1 (15.9)

BMI 29.0 (4.7) 27.4 (6.3)

Fat mass (kg) 31.8 (8.0) 28.0 (11.7)

Lean mass (kg) 40.6 (6.3) 41.9 (6.8)

Abdominal fat (%) 47.4 (8.4) 39.9 (12.4)

Insulin (uU/mL) 10.7 (5.0) 9.9 (8.0)

Glucose (mM/L) 6.4 (2.0) 5.5 (0.6)

Physical activity level(daily kcal)b

301.3 (218.0) 307.2 (215.3)

NEUROPSYCHOLOGIC MEASURES

Intelligence proxy (NAART), IQ 117.6 (8.7) 120.6 (5.2)

Executive function

Color Trails Difference(2-1; s)

55.2 (30.7) 75.6 (64.8)

Stroop C (s) 67.3 (35.7) 68.7 (35.8)

Digits Backward(sum score)

5.8 (1.9) 6.5 (2.1)

Attention

LDST (sum score) 29.2 (7.1) 29.1 (6.6)

Verbal fluency

COWAT (sum score) 33.1 (15.5) 37.8 (12.4)

Categories (sum score) 15.9 (4.2) 16.1 (4.6)

Verbal memory (immediate)

RAVLT (sum 5 trials score) 36.1 (12.1) 38.9 (9.5)

RAVLT immediate recall(score)

7.2 (2.9) 7.2 (3.8)

Verbal memory (delayed)

RAVLT delayed recall (score) 6.9 (3.6) 6.8 (3.9)

Fuld delayed recall (score) 7.6 (2.7) 7.2 (1.8)

Visuospatial skill

Figure copy (sum score) 26.3 (5.8) 27.1 (7.2)

Clock (sum score) 5.8 (1.4) 6.1 (1.3)

(continued on next page)

Anderson-Hanley et al / Am J Prev Med 2012;42(2):109–119 111

February 2012


cognitive outcomes were included to characterize the sample (e.g.,clinical status below); no changes were expected on these tests(Appendix C, available online at www.ajpmonline.org). At thecompletion of the study, participants’ clinical status pre- and post-intervention was classifıed according to “typical” diagnostic crite-ria41,42 formild cognitive impairment (MCI; performance�1.5 SDon at least one subtest in the domains of executive function, verbalfluency, verbal memory, visuospatial skill, and visuospatial mem-ory compared to normative data).40 MCI incidence was compara-ble to prior research (Table 1).43

Physiologic Assessment

Baseline and post-exercise measurements included: weight (kilo-grams); height (centimeters); BMI; total and abdominal body com-position (fat and lean mass) using the iDXA (GE Lunar, Inc.);muscle strength of quadriceps and hamstrings using the HUMACCybex Dynamometer (CSMI Solutions, Inc.); and insulin and glu-cose (Millipore, Inc.).

Assessment of Exercise Behavior

During the fırst year, daily physical activity (kilocalories) wasmeasured using the Aerobics Center Longitudinal Study Physi-cal Activity Questionnaire (ACLS-PAQ).44 METs were used tocompute energy expended in activities. In the second year,additional resources allowed measurement of daily physicalactivity (kilocalories) using an accelerometer (Actical; PhillipsRespironics, Inc). Ride behaviors (frequency, intensity, and du-ration) were recorded on the bike computer and by participantsin a paper log.

Neuroplasticity Assessment

Fasting morning plasma samples were collected during pre- andpost-evaluations, not after exercise. Brain-derived neurotrophicfactor (BDNF) levels were analyzed via enzyme-linked immu-nosorbent assay (ELISA; Chemicon, Millipore, Billerica, MA; seeAppendix D, available online at www.ajpmonline.org).

Statistical Analysis

Data were analyzed using SPSS, version 19.0. For normally distrib-uted continuous variables, arithmetic Ms and SDs were calculated.For comparisons between groups of categoric baseline data, chi-square analyses were conducted. For comparisons of continuouslydistributed baseline and demographic variables, t tests were per-formed. Intent-to-treat analysis was conducted using the last ob-servation carried forward (LOCF). Four analytic strategies wereemployed to examine between-group changes in outcomes: intentto treat, complete case, age matched, and comparison of com-pleters and noncompleters.Mixed linear modeling, including fıxed and random effects,

estimated the impact of the interventions on executive functioncomposite scores, when adjusted for age, education, and nestedvariability in clusters (eight sites). A likelihood ratio test was con-ducted to compare the full and restrictedmodels, with and withoutsites nested. Follow-up repeated measures general linear models(GLMs) examined the group X time interaction effect, fırst byexamining the multivariate omnibus test (to control Type I error),then examining the univariate results for the three executive func-tion measures. To test whether between-group differences in cog-nitive outcomes were due to differential exercise effort, t-tests wereused. Effect sizes were computed using Cohen’s d formula withpooled SDs. Tests of signifıcance used a two-sided alpha of p�0.05.

ResultsA CONSORT flow chart (Figure 1) shows that 102 inde-pendent-living, older adults from eight retirement com-munities met criteria and consented to participate; 79began exercise training and were randomized by site (av-erage cluster n�10, SD�3.6; Figure 1). Sixty-three olderadults, ranging in age from 58 to 99 years, completed thestudy (80% of randomized).

Effect of the Intervention on CognitiveFunction, Physical Health, and ExerciseBehaviorsThe group X time interaction effects for executivefunction of the full and restricted mixed linear modelswere highly similar (F[1, 51.8]�10.4, p�0.002; F[1,76.2]�10.4, p�0.002, with and without sites nested,respectively; Figure 2). There was no benefıt of addingthe cluster random effect (LR �2[1]�3.16, p�0.93); thus,in order tomaximize df in this relatively small sample, theleast-restrictive fıtting model was selected and subse-quent parsimonious analyses were chosen. A differencebetween groups in change in executive function over 3months was indicated by a group X time interaction in a

Table 1. Baseline characteristics of trial participants, M(SD) unless otherwise indicated (continued)

Cybercycle(n�38)


Visuospatial memory(delayed)

Figure delayed recall(score)

8.8 (6.2) 9.6 (4.7)

Motor function

Pegboard dominanthand (s)

120.7 (50.1) 130.0 (44.6)

Pegboard nondominanthand (s)

136.1 (85.7) 139.3 (47.1)

Clinical status (n [%])

MCI (�1 domain: � �1.5SD of norm)

16 (42.1) 14 (34.1)

aGroup difference at baseline on age (p�0.002) and education(p�0.001)

bPhysical activity level (daily kilocalories) was estimated in Year 1 viaquestionnaire and Year 2 via Actical (see Methods)

COWAT, Controlled Oral Word Association Test; IQ, intelligencequotient; LDST, Letter Digit Symbol Test; MCI, mild cognitive impair-ment; NAART, North American Adult Reading Test; RAVLT, ReyAuditory Verbal Learning Test


www.ajpmonline.org


multivariate repeatedmeasures GLMof Color Trails Dif-ference, Stroop C, and Digits Backward, simultaneouslyand revealing a large effect (F[3, 62]�5.50, p�0.002,�p

2�0.21, power�0.93). Given the signifıcant omnibustest, univariate group X time interactions were examinedand found signifıcant for all three measures of executivefunction (Table 2).

Planned simple effects analyses controlled for age, ed-ucation, and cognitive performance at baseline and re-vealed an increase in performance on the Color TrailsDifference (p�0.01) and Stroop C (p�0.05) tests for cy-bercyclists, with no change for controls. Cybercyclistsmaintained a steady performance on Digits Backward,whereas the control group declined (p�0.01). No inter-action effects were found on physiologic or secondarycognitive outcomes (Table 2). Analyses were repeatedusing age-matched and complete-case subsamples andresults were similar (Appendixes E–H, available online atwww.ajpmonline.org). No differences in exercise fre-quency, intensity, or duration were found between thecybercyclists and controls (Table 3). While the averageenergy expended was relatively low (approximately 100calories/ride), research has shown that even low-intensity

exercise (100 calories) can serve as an adequate trainingstimulus among sedentary older adults.45

Cybercycling yielded a medium average effect size forexecutive function that was over and above the averageeffect for traditional exercise (d�0.50), contrasted withprior research that showed a small effect size for aerobicexercise over and above nonaerobically exercising con-trols (d�[0.48–0.16]�0.32).51 Cybercyclists experi-enced a 23% reduction in risk of clinical progression toMCI compared with traditional exercisers (nine controlsversus three cybercyclists converted to MCI). That is,using the “typical” diagnostic criteria for MCI,41,42 theseparticipants began the trial with performances in thenormal range, but experienced a decline to �1.5 SD be-low normative data on at least one test within thosedomains.

Adherence to prescribed exercise (79.7%) was compa-rable with prior research (78.2%).12 Consistent withCONSORT standards, a comparison of study completersand noncompleters is reported. Similar rates on noncom-pletion were found in both conditions; at baseline, non-completers were more compromised than completers onsome cognitive and physiologic measures that may haveled to greater diffıculty completing the study (Appendix G,available online at www.ajpmonline.org). Appendix I(available online at www.ajpmonline.org) shows the 13adverse events in the study.

Biomarker Evidence of PossibleNeuroplasticity: Brain-Derived NeurotrophicGrowth Factor ResultsPlasma BDNF data from 30 participants were available(ages 66–89 years). A signifıcant group (cycle condition)

Figure 2. Change in executive function composite beforeand after 3 months of exerciseNote: n�79; mixed linear model (random effects: age, education, and cluster)group X time interaction is significant (p�0.002).

Figure 1. CONSORT diagram showing flow of participantsfrom screening to post-exercise evaluation


February 2012


Table 2. Neuropsychologic and physiologic outcomes after 3 months of exercise (intent-to-treat analysis)a

Mean difference from baseline (95% CI)

Cybercycle (n�38) Control bike (n�41) p-value (df)b

PRIMARY COGNITIVE OUTCOMES

Executive function

Color Trails Difference (2-1) (s) �15.94 (�16.21, 15.66) 9.74 (9.48, 10.00) 0.007 (1, 73)

Stroop C (s) �6.59 (�6.67, �6.51) 0.56 (0.49, 0.64) 0.05 (1, 73)

Digits Backward (sum score) 0.36 (0.34, 0.38) �0.83 (�0.85, �0.82) 0.03 (1, 73)

SECONDARY COGNITIVE OUTCOMESc

Attention

LDST (sum score) 0.79 (0.62, 0.95) 0.73 (0.57, 0.89) 0.95 (1, 72)

Verbal fluency

COWAT (sum score) 3.51 (2.77, 4.25) 2.33 (1.62, 3.03) 0.63 (1, 73)

Categories (sum score) �0.03 (0.11, �0.18) 1.18 (1.32, 1.04) 0.22 (1, 73)


RAVLT (sum 5 trials score) �0.73 (�1.27, �0.19) 0.85 (0.33, 1.37) 0.50 (1, 73)

RAVLT immediate recall (score) 0.77 (0.60, 0.94) 0.06 (�0.10, 0.22) 0.32 (1, 73)


RAVLT delayed recall (score) 0.71 (0.62, 0.79) 0.10 (0.01, 0.18) 0.43 (1, 73)

Fuld delayed recall (score) 0.15 (0.13, 0.17) 0.39 (0.37, 0.41) 0.61 (1, 73)

Visuospatial skill

Figure copy (sum score) 3.27 (3.56, 2.98) 3.69 (3.97, 3.40) 0.81 (1, 72)

Clock (sum score) 0.07 (0.07, 0.07) �0.19 (�0.19, �0.19) 0.45 (1, 72)

Visuospatial memory (delayed)

Figure delayed recall (score) 0.07 (0.22, �0.08) 1.66 (1.80, 1.52) 0.28 (1, 72)

Motor function

Pegboard dominant hand (s) 10.61 (8.64, 12.57) 6.13 (4.22, 8.03) 0.56 (1, 72)

Pegboard nondominant hand (s) 7.76 (5.86, 9.65) 13.79 (11.95, 15.63) 0.36 (1, 72)

PHYSIOLOGIC OUTCOMES

Weight (kg) �0.63 (�0.75, �0.52) �0.04 (�0.15, 0.07) 0.24 (1, 72)

BMI �0.26 (�0.29, �0.23) �0.03 (�0.06, 0.00) 0.26 (1, 67)

Fat mass (kg) �1.04 (�0.95, �1.13) �0.76 (�0.67, �0.84) 0.50 (1, 72)

Lean mass (kg) 0.39 (0.31, 0.47) 0.56 (0.48, 0.63) 0.65 (1, 72)

Abdominal fat (%) �1.79 (�1.97, �1.61) �0.94 (�1.11, �0.78) 0.32 (1, 66)

Leg extension 60° (s�1) �2.96 (�3.00, �2.92) 11.09 (11.05, 11.13) 0.04 (1, 71)

Leg flex 60° (s�1) �2.79 (�3.26, �2.31) 5.70 (5.25, 6.15) 0.07 (1, 71)

Insulin (uU/mL) 2.75 (2.39, 3.12) 1.53 (1.16, 1.90) 0.46 (1, 67)

Glucose (mM/L) �0.09 (�0.01, �0.16) �0.06 (0.01, �0.13) 0.90 (1, 68)

aMarginal mean differences and CIs reported, based on repeated measures ANCOVA controlling for age and educationbFor ANCOVA, repeated measures, group X time; the first df in parentheses refers to the effect (group X time) and the second refers to the error termcNo significant changes expected given prior research literatureCOWAT, Controlled Oral Word Association Test; LDST, Letter Digit Symbol Test; RAVLT, Rey Auditory Verbal Learning Test


www.ajpmonline.org


X time (pre- and post-intervention) interaction, with ageand education as covariates, was found, revealing thatcybercyclists experienced a greater increase in BDNFthan traditional exercise (Appendix J, available online atwww.ajpmonline.org; F[1, 25]�4.89; p�0.05).

DiscussionThis cluster RCT provides preliminary evidence that ex-ergaming can yield greater cognitive benefıt, bufferingagainst decline, more so than traditional exercise alone.Older adults in an independent-living facility who exer-cised on a virtual reality–enhanced cybercycle for3months had signifıcantly better executive function thanthose expending similar effort on a traditional stationarybike. In contrast with prior research showing a smalleffect of exercise over and above controls,8 cybercyclingproduced a medium effect that was over and above tradi-tional exercise, with average improvements in perfor-mance of one half SD. Additionally, fewer cybercyclistsconverted to MCI, suggesting a reduction in risk ofprogression to MCI; however, the incidence and rate ofconversion to MCI herein might be higher than in acommunity-dwelling sample, and further research isneeded to establish replicability and generalizability.

Contrary to expectations, effort and fıtness did notappear to be the factors behind differential cognitive ben-efıts found in the cybercycle group. Perhaps because thiswas a prescriptive intervention, most participants acrossboth groups were compliant with the regimen, and fur-ther research is needed to evaluate whether naturalisticuse would lead to greater effort by cybercyclists. These

fındings are consistent with some assertions in the litera-ture that the cognitive benefıt derived from exercise is notnecessarily tied to fıtness outcomes, although the debatecontinues.46,47

Future research will be needed to tease apart the con-tributions of a variety of factors in the cybercycling con-dition. Consistency across conditions for goal setting andcompetition suggests virtual reality imagery and interac-tive decision making might be the potent factors of thecybercycle. Exit interviews provided anecdotal evidenceof the value of these unique features. Participants com-mented on their enjoyment of visual stimulation and thechallenge of outpacing avatars. One woman, aged86 years, noted that she felt healthier and attributed thisto actively maneuvering to “compete with that fellowahead of me!” Cybercycling provides a different experi-ence than other cognitive stimulation such as TV, be-cause it is interactive.

One explanation for the greater cognitive benefıt foundwith cybercycling compared with traditional cyclingcould be that the effect is due directly to the addedmentalexercise required. Given that both exercise interventionsamples exerted similar effort over 3 months, the maindifference between the two interventions was the virtualreality experience. Navigating a 3D landscape, anticipat-ing turns and competing with others, requires additionalfocus, expanded divided attention, and enhanced deci-sion making. These are activities that depend in part onexecutive function, which was signifıcantly affected. Adirect impact of cognitive stimulation herein does reso-nate with a growing, but formative literature on the ef-

Table 3. Exercise behavior outcomes after 3 months of exercise: cybercycle vs control bikea

M (SD)

Exercise behavioroutcomes

Cybercycle(n�30)


Difference betweeninterventions (M [95% CI]) p-value (df)

Frequency of rides (n) 51.3 (3.32) 53.3 (3.14) �1.96 (�2.31, �1.61) 0.68 (1, 59)

Power (watts)b 36.3 (3.28) 32.1 (3.15) 4.20 (3.93, 4.46) 0.44 (1, 31)

Energy expended (kcal) 107.9 (8.05) 93.6 (7.63) 14.32 (13.47, 15.17) 0.23 (1, 59)

Duration (min) 35.5 (1.81) 33.8 (1.72) 1.61 (1.42, 1.80) 0.54 (1, 59)

Distance average (miles) 5.4 (0.40) 4.8 (0.38) 0.65 (0.61, 0.69) 0.27 (1, 59)

Distance total (miles) 283.9 (28.80) 261.4 (27.29) 22.51 (19.47, 25.54) 0.59 (1, 59)

Speed average (mph)b 7.4 (0.38) 8.3 (0.37) �0.83 (�0.86, �0.80) 0.19 (1, 31)

Speed peak (mph)b 10.7 (0.39) 9.8 (0.37) 0.97 (0.94, 1.00) 0.13 (1, 31)

Physical activity daily (kcal) 324.4 (32.91) 304.2 (32.22) 20.22 (0.94, 1.00) 0.66 (1, 43)

aMarginal Ms and SDs reported, based on ANCOVA controlling for age and educationbSample sizes: cybercycle (n�17) and control bike (n�18) because of enhanced ride data available in Year 2mph, miles per hour


February 2012


fects of cognitive training.48 While research is mixed andtransfer is debatable, some research supports the utility ofmental exercise to facilitate cognitive health in olderadults.49–52 Future research should measure the amountof cognitive stimulation participants engage in during theperiod of an exercise intervention to clarify the potentialadded benefıt of activities beyond physical exercise.

Another explanation for the greater cognitive benefıtfound for cybercycling could be that the effect is due tothe interactive nature of combined physical and cogni-tive exercise. Perhaps cybercyclists benefıt from a dual-exercise experience, accruing the positive effects of inter-twined cognitive and physical exercise.When comparingaverage effect sizes in the literature,51 controls demon-strate test–retest growth (0.16), cognitive stimulationalone yields a comparable negligible effect (0.13), physicalexercise yields a small effect over and above controls(0.32), while combined cognitive and physical exerciseherein produced a medium effect beyond that tradition-ally found for exercising controls (0.50). It is interestingthat the combined effect of cognitive and physical exer-cise exceeds the sum of effects noted in the literatureabove, perhaps indicating a compounding or synergisticeffect of cybercycling. Future research could evaluate thisby comparing cognitive stimulation alone, physical exer-cise alone, and the combination of the two.

Compounding cognitive benefıt from a combined taskdoes fıt with the evolving understanding of the mecha-nisms of brain plasticity and the role of exercise andenriched environments in inducing angiogenesis, neuro-genesis, and other changes that foster neurovascular in-tegrity.15,53 A combined effect would be consistent withthe animal literature, where cognitive benefıt from phys-ical exercise and mental stimulation has been found tooccur by differentmechanisms (cell proliferation and cellsurvival, respectively).53–55 This combined-effect hy-pothesis expands on prior research in humans, which hasfound enhanced cognitive benefıts of physical and cogni-tive exercise interventions administered in tandem.56,57

Similarly, these fındings fıt with prior research that indi-cates cognitive benefıt beyond that from traditional exer-cise, when physical exercise is cognitively challenging(e.g., Tai Chi or dancing).58–60

To further illuminate possible mechanisms linking ex-ercise to cognitive change, alternative measures of inter-mediary physiologic or brain “fıtness” (e.g., neurotrophicgrowth factors), may be needed beyond cardiovascularfıtness outcomes typically assessed.61 In this study, it wasfound that cybercyclists experienced a signifıcantlygreater increase in BDNF than traditional exercisers, sug-gesting that exercisemay lead to cognitive benefıts bywayof biomarkers linked to neurotrophic effects. The litera-ture on BDNF change with physical exercise is mixed,

and researchers continue to evaluate possiblemoderatorssuch as age, gender, and type of exercise.10,14,20 Cyber-cyclists exhibited a signifıcant change in BDNF, whichdoes fıt with research that has shown a signifıcant in-crease in BDNF after computerized cognitive training.62

Compared with prior research on the effects of physi-cal exercise alone, the effect of the cybercycle interventionadds to the growing consensus that exercise has a consis-tent effect on executive functions.8,21,22 However, thecontrol group herein was also an exercising group (con-sistent with recommendations),63 but did not show pre-to post-test improvement on executive function. It ap-pears the added rigor of using an additional pre-test forfamiliarization did “wash-out” practice advantages37 ev-ident in prior studies. While traditional exercise did notyield “improvement” in cognition, it may have sloweddecline, which would be consistent with some prior re-search which found that in a similarly aged sample, thecontrol group declined on cognitive function.64

Limitations of this study include unequal representationof age and education in the groups despite randomization,and while statistical controls were used and age- and edu-cation-matched post hoc analyses were conducted, futureresearch could prospectively match on these variables.Also, participants had a relatively high level of education,and ethnic variability was limited; additional research isneeded to test generalizability. Noncompleters per-formed worse on some cognitive and physiologic mea-sures; thus, screening for minimum levels of functionmay be advisable.

Several strengths of this study are noteworthy. Thisstudy addresses a gap in the literature as no prior RCThascompared cognitive benefıts for older adults of virtualreality–enhanced exercise with traditional exercise. Theobserved effect exceeds that typically reported in tradi-tional exercise research. The intervention should be ap-plicable to a wide range of older adults in an independentliving context given the ease of using a recumbent bikeand increasing availability of exergaming technologies.The fınding that cognitive outcomes could be improvedwith cybercycling over and above those from traditionalexercising is surprising in light of similar exercise effort,but this also provides an intriguing issue to explore infuture research.

Follow-up studies could aim to replicate prior researchby using neuroimaging to examine the impact of exer-gaming on the brain for further evidence of neuroplastic-ity.13–16 With a refıned experimental design, future re-search could clarify if cognitive exercise alone is suffıcientto produce the observed cognitive change, or if exergam-ing leads to added benefıt by synergistic neurophysiologicadvantages when mental challenges are linked to physio-logic movements. Future research could compare out-


www.ajpmonline.org


door street-cycling with cybercycling, since the naturalworld, street obstacles, other cyclists, and way-fındingwould similarly create cognitive challenge. It would alsobe interesting to evaluate biophilia factors, degree of cog-nitive stimulation, and social presence. Additionally,some labs have full-surround audio-visual virtual realityenvironments, that could allow controlled testing of “out-door” factors while ensuring safety.65 Last, a cost–benefıtanalysis would be useful, in light of reports that physicalactivity interventions for inactive older adults can be costeffective.66

In summary, this cluster RCT indicates that for olderadults, virtual reality–enhanced interactive exercise or“cybercycling” two to three times per week for 3 monthsyielded greater cognitive benefıt and possibly added pro-tection from progression to MCI, compared with a simi-lar dose of traditional exercise. Additional research isneeded to examine the cause of this curious fınding,which may be due to the presence of unique mentalstimulation in virtual reality, or the interactive combina-tion of cognitive and physical challenges wielding dualimpacts, perhaps promoting neuroplasticity via multiplepathways.53,54 The implication is that older adults whochoose exergaming with interactive physical and cogni-tive exercise, over traditional exercise, may garner addedcognitive benefıt and perhaps prevent decline, all for thesame exercise effort.

This study was funded by a grant from the Pioneer Portfolio ofthe Robert Wood Johnson Foundation, through the HealthGamesResearch national program (#64449); and by faculty andstudent grants fromUnion and Skidmore Colleges. The RobertWood Johnson Foundation had no role in the design and con-duct of the study, analysis and interpretation of the data, orpreparation or approval of the manuscript.We acknowledge important technical assistance from: Brian

Button of Interactive Fitness Holdings, LLC regarding use ofthe Expresso platform; BruceWinkler and Ivjot Kholi from RASports, LLC and Web Racing, LLC, regarding use of theirNetAthalon virtual reality cycling software and sensor kits; MarkMartens regarding our pilot of the FitClub riding software fromPantometrics, Ltd.; and John Cowan of Neurosciences AdvancedImaging Research Center, AlbanyMedical Center.We greatly appreciate the participation of the residents and

essential facilitation of the site administrators from: BeltroneLiving Center, Glen Eddy, Hightpointe Apartments, KingswayVillage, Prestwick Chase, Schaffer Heights,WesleyHealth Care(Embury Apartments and Woodlawn Commons), and West-view Apartments.This research would not have been possible without the

dedication of many research assistants; in particular, we wouldlike to acknowledge: Stephen Brink, Lyndsay De Matteo, ElliotHarmon, Veronica Hopkins, Eric Hultquist, Dinesh Kom-

mareddy, Darlene Landry, Shi Feng Lin,Mariale Renna, TraceyRocha, Nick Steward, Amanda Snyder, and Vadim Yerokhin.We are grateful for the thoughtful critiques and helpful com-ments of those who reviewed the manuscript: ChristinaBrueggeman, MD, Jeffrey Cummings, MD, Lissy Jarvik, MD,Andrew Leuchter, MD, Loretta Malta, PhD, Timothy Nichol-son, MD, and Molly Shuland, MD, as well as several anony-mous reviewers.Earlier versions of these data were presented at the annual

meetings of the American College of Sports Medicine, theAmerican Psychological Association, and the Society of Behav-ioral Medicine.No fınancial disclosures were reported by the authors of this

paper.

References1. Plassman BL, Langa KM, Fisher GG, et al. Prevalence of dementia in

the United States: the aging, demographics, and memory study. Neu-roepidemiology 2007;29:125–32.

2. Larson E. Prospects for delaying the rising tide of worldwide, late-lifedementias. Int Psychogeriatr 2010;22(8):1196–202.

3. Morrison-Bogorad M, Cahan V, Wagster M. Brain health interven-tions: the need for further research. Alzheimers Dement 2007;3(2S):S80–S85.

4. LarsonE. Physical activity for older adults at risk forAlzheimer disease.JAMA 2008;300(9):1077–9.

5. Chang M, Jonsson P, Launer L, et al. The effect of midlife physicalactivity on cognitive function among older adults: AGES—ReykjavikStudy. J Gerontol (A Bio Sci Med Sci) 2010;65(12):1369–74.

6. Scarmeas N, Luchsinger J, Stern Y, et al. Physical activity, diet, and riskof Alzheimer disease. JAMA 2009;302(6):627–37.

7. Angevaren M, Aufdemkampe G, Verhaar HJ, Aleman A, Vanhees L.Physical activity and enhanced fıtness to improve cognitive function inolder people without known cognitive impairment. Cochrane Data-base Syst Rev 2008;(2):CD005381;ISSN:1469–93X.

8. Colcombe S, Kramer, AF. Fitness effects on the cognitive function ofolder adults: a meta-analytic study. Psychol Sci 2003;14(2):125–30.

9. Heyn P, Abreu BC, Ottenbacher KJ. The effects of exercise training onelderly persons with cognitive impairment and dementia: a meta-analysis. Arch Phys Med Rehabil 2004;85(10):1694–704.

10. Baker LD, Frank LL, Foster-Schubert K, et al. Effects of aerobic exerciseon mild cognitive impairment: a controlled trial. Arch Neurol 2010;67(1):71–9.

11. Geda YE, Roberts RO, KnopmanDS, et al. Physical exercise, aging, andmild cognitive impairment. Arch Neurol 2010;67(1):80–6.

12. Lautenschlager N, Cox K, Flicker L, et al. Effect of physical activity oncognitive function in older adults at risk for Alzheimer disease: arandomized trial. JAMA 2008;300(9):1027–37.

13. Colcombe S, Erickson KI, Scalf PE, et al. Aerobic exercise trainingincreases brain volume in aging humans. JGerontol (ABio SciMedSci)2006;61(11):1166–70.

14. Erickson K, Voss MW, Prakash RS, et al. Exercise training increasessize of hippocampus and improves memory Proc Natl Acad Sci U S A2011;108(7):3017–22.

15. Kramer A, Erickson K. Capitalizing on cortical plasticity: Influence ofphysical activity on cognition and brain function. Trends Cog Sci2007;11(8):342–8.

16. Pajonk FG, Wobrock T, Gruber O, et al. Hippocampal plasticity inresponse to exercise in schizophrenia. Arch Gen Psychiatry 2010;67(2):133–43.


February 2012


17. Voss MW, Erickson KI, Prakash RS, et al. Functional connectivity: asource of variance in the association between cardiorespiratory fıtnessand cognition? Neuropsychologia 2010;48(5):1394–406.

18. Liang K, Mintun M, Head D, et al. Exercise and Alzheimer’s diseasebiomarkers in cognitively normal older adults. AnnNeurol 2010;68(3):311–8.

19. Yaffe K. Biomarkers of Alzheimer’s disease and exercise: one stepcloser to prevention. Ann Neurol 2010;68(3):275–6.

20. Knaepen K, Goekint M, Heyman E, Meeusen R. Neuroplasticity—exercise-induced response of peripheral brain-derived neurotrophicfactor: a systematic review of experimental studies in human subjects.Sports Med 2010;40(9):765–801.

21. Etnier JL, Chang YK. The effect of physical activity on executive func-tion: a brief commentary on defınitions, measurement issues, and thecurrent state of the literature. J Sport Exerc Psychol 2009;31(4):469–83.

22. Hillman C, Erickson K, Kramer A. Be smart, exercise your heart:exercise effects on brain and cognition. Nat Rev Neurosci 2008;9(1):58–65.

23. American College of Sports Medicine, Chodzko-Zajko W, Proctor D,Skinner J, et al. American College of Sports Medicine position stand.Exercise and physical activity for older adults. Med Sci Sports Exerc2009;41(7):1510–30.

24. ReedBD, Jensen JD,GorenfloDW.Physicians and exercise promotion.Am J Prev Med 1991;7(6):410–5.

25. Grandes G, Sanchez A, Sanchez-Pinilla RO, et al.; PEPAF Group.Effectiveness of physical activity advice and prescription by physiciansin routine primary care: a cluster randomized trial. Arch Intern Med2009;169(7):694–701.

26. Sallis J. New thinking on older adults’ physical activity. Am J PrevMed2003;25(3S2):110–1.

27. Read JL, Shortell SM. Interactive games to promote behavior change inprevention and treatment. JAMA 2011;305(16):1704–5.

28. Lieberman DA. Designing serious games for learning and health ininformal and formal settings. In: RitterfeldU,CodyM,Vorderer P, eds.Serious games: mechanisms and effects. New York: Routledge,2009:117–30.

29. Annesi JJ, Mazas J. Effects of virtual reality-enhanced exercise equip-ment on adherence and exercise-induced feeling states. Percept MotSkills 1997;85(3 Pt 1):835–44.

30. LangeBS, Requejo P, Flynn SM, et al. The potential of virtual reality andgaming to assist successful agingwith disability. PhysMedRehabil ClinN Am 2010;21(2):339–56.

31. Chuang TY, Sung WH, Chang HA, Wang RY. Effect of a virtualreality–enhanced exercise protocol after coronary artery bypass graft-ing. Phys Ther 2006;86(10):1369–77.

32. van Schaik P, Blake J, Pernet F, Spears I, Fencott C. Virtual augmentedexercise gaming for older adults. CyberPsychol Behav 2008;11(1):103–6.

33. Baranowski T, Buday R, Thompson D, Baranowski J. Playing for real:video games and stories for health-related behavior change. Am J PrevMed 2008;34(1):74–82.

34. Grealy MA, Johnson DA, Rushton SK. Improving cognitive functionafter brain injury: the use of exercise and virtual reality. Arch PhysMedRehabil 1999;80(6):661–7.

35. Plante T, Aldridge A, Su D, Bogdan R, Belo M, Kahn K. Does virtualreality enhance the psychological benefıts of exercise? J HumanMove-ment Stud 2003;45:485–507.

36. RAND. Exercise programs for older adults: a systematic review andmeta-analysis. Santa Monica CA: Southern California Evidence-BasedPractice Center, 2003.

37. Yang L, ReedM, Russo F,WilkinsonA. A new look at retest learning inolder adults: learning in the absence of item-specifıc effects. J Gerontol(B Psychol Sci Soc Sci) 2009;64B(4):470–3.

38. D’Elia LG, Satz P,UchiyamaCL,White T. Color Trails Test. Odessa FL:Psychological Assessment Resources, 1996.

39. van der ElstW, van BoxtelMPJ, van BreukelenGJP, Jolles J. The StroopColor-Word Test: influence of age, sex, and education; and normativedata for a large sample across the adult age range. Assessment 2006;13(1):62–79.

40. Strauss E, Sherman EMS, Spreen O. A compendium of neuropsycho-logical tests. 3rd ed. New York: Oxford University Press, 2006.

41. Petersen R,Morris J. Mild cognitive impairment as a clinical entity andtreatment target. Arch Neurol 2005;62(7):1160–3.

42. Jak A, Bondi M, Delano-Wood L, et al. Quantifıcation of fıve neuro-psychological approaches to defıningmild cognitive impairment. Am JGeriatric Psychiatry 2009;17(5):368–75.

43. Saxton J, Snitz BE, Lopez OL, et al.; GEM Study Investigators. Func-tional and cognitive criteria produce different rates of mild cognitiveimpairment and conversion to dementia.NeurolNeurosurgPsychiatry2009;80(7):737–43.

44. Kohl H, Blair S, Paffenbarger R,Macera C, Kronenfeld J. Amail surveyof physical activity habits as related to measured physical fıtness. Am JEpidemiol 1988;127(6):1228–39.

45. Foster V, Hume G, Byrnes W, Dickinson A, Chatfıeld S.Endurancetraining for elderly women: moderate vs low intensity. J Gerontol1989;44(6):M184–8.

46. Etnier JL, Nowell PM, Landers DM, Sibley BA. A meta-regression toexamine the relationship between aerobic fıtness and cognitive perfor-mance. Brain Res Rev 2006;52(1):119–30.

47. Smiley-Oyen AL, Lowry KA, Francois SJ, Kohut ML, Ekkekakis P.Exercise, fıtness, and neurocognitive function in older adults: The“selective improvement” and “cardiovascular fıtness” hypotheses. AnnBehav Med 2008;36(3):280–91.

48. Owen A, Hampshire A, Ballard C, et al. Putting brain training to thetest. Nature 2010;465(7299):775–8.

49. Studenski S, Carlson MC, Fillit H, Greenough WT, Kramer A, RebokGW. From bedside to bench: Does mental and physical activity pro-mote cognitive vitality in late life? Sci Aging Knowledge Environ2006;2006(10):pe21.

50. Unverzagt F, Smith D, Rebok GW, et al. The Indiana Alzheimer Dis-ease Center’s Symposium on Mild Cognitive Impairment. Cognitivetraining in older adults: lessons from the ACTIVE Study. Curr Alzhei-mer Res 2009;6(4):375–83.

51. Valenzuela M, Sachdev P. Can cognitive exercise prevent the onset ofdementia? Systematic review of randomized clinical trials with longi-tudinal follow-up. Am J Geriatr Psychiatry 2009;17(3):179–87.

52. Papp K,Walsh S, Snyder P. Immediate and delayed effects of cognitiveinterventions in healthy elderly: a review of current literature andfuture directions. Alzheimers Dement 2009;5(1):50–60.

53. van Praag H. Neurogenesis and exercise: past and future directions.Neuromolecular Med 2008;10(2):128–40.

54. Fabel K, Wolf SA, Ehninger D, Babu H, Leal-Galicia P, KempermannG. Additive effects of physical exercise and environmental enrichmenton adult hippocampal neurogenesis inmice. Front Neurosci 2009;3:50.

55. Olson AK, Eadie BD, Ernst C, Christie BR. Environmental enrichmentand voluntary exercise massively increase neurogenesis in the adulthippocampus via dissociable pathways. Hippocampus 2006;16(3):250–60.

56. Fabre C, Chamari K, Mucci P, Massé-Biron J, Préfaut C. Improvementof cognitive function by mental and/or individualized aerobic trainingin healthy elderly subjects. Int J Sports Med 2002;23(6):415–21.

57. OswaldW, Gunzelmann T, Rupprecht R, Hagen B. Differential effectsof single versus combined cognitive and physical training with olderadults: the SimA study in a 5-year perspective. Euro J Ageing2006;3:179–92.

58. Taylor-Piliae R, Newell K, Cherin R, LeeM, King A, HaskellW. Effectsof Tai Chi andWestern exercise on physical and cognitive functioningin healthy community-dwelling older adults. J Aging Phys Act2010;18(3):261–79.

59. HoganM. Physical and cognitive activity and exercise for older adults:a review. Int J Aging Hum Dev 2005;60(2):95–126.


www.ajpmonline.org


60. Verghese J. Cognitive andmobility profıle of older social dancers. J AmGeriatr Soc 2006;54(8):1241–4.

61. Nation D, Hong S, Dimsdale J, et al. Stress, exercise, and Alzheimer’sdisease: a neurovascular pathway.MedHypotheses 2011;76(6):847–54.

62. Vinogradov S, FisherM, Holland C, ShellyW,Wolkowitz O,Mellon S.Is serum brain-derived neurotrophic factor a biomarker for cognitiveenhancement in schizophrenia? Biol Psychiatry 2009;66(6):549–53.

63. Booth FW, Lees SJ. Physically active subjects should be the controlgroup. Med Sci Sports Exerc 2006;38(3):405–6.

64. Hill R, Storandt M, Malley M. The impact of long-term exercise trainingon psychological function in older adults. J Gerontol 1993;48(1):P12–7.

65. Kwon DS, Yang G-H, Park Y, et al. KAIST interactive bicycle racingsimulator: the 2nd version with advanced features. Intelligent Robots andSystem 2002 IEEERSJ International Conference. 2002;3:2961–6.

66. Sevick M, Dunn A, Morrow M, Marcus B, Chen G, Blair S. Cost-effectiveness of lifestyle and structured exercise interventions insedentary adults: results of project ACTIVE. Am J Prev Med2000;19(1):1–8.

Appendix

Supplementary data

Supplementary data associatedwith this article can be found, in theonline version, at doi:10.1016/j.amepre.2011.10.016.

A pubcast created by the authors of this paper can be viewed athttp://www.ajpmonline.org/content/video_pubcasts_collection.

Did you know?The latest AJPM news is available online.

Visit www.ajpmonline.org to see the “News fromAJPM” section on the homepage.


February 2012

Am J Prev Med 2012; 42(2) A-1

Exergaming Improves Older Adult Cognition

A Cluster Randomized Clinical Trial

Cay Anderson-Hanley, PhD, Paul J. Arciero, DPE, Adam M. Brickman, PhD, Joseph P. Nimon, BS, Naoko Okuma, BS, Sarah C. Westen, BS, Molly E. Merz, BS, Brandt D. Pence, BA,

Jeffrey A. Woods, PhD, Arthur F. Kramer, PhD, Earl A. Zimmerman, MD

Appendix A

Cybercycle demonstration image and screen shot of virtual terrain and avatars

Am J Prev Med 2012;42(1) A-2

Am J Prev Med 2012;42(1) A-3

Appendix B

Exercise equipment specifications

Bicycles in both years of the study were recumbent, with a “walk-through” design (affording an open space to step in between the seat and steering mechanism so that participants did not have to lift their leg over a center bar), and all bikes had gear shifts so participants could adjust the pedaling resistance. The cycles were identical across the cybercycle and traditional bicycle conditions except for the additional interactive virtual reality display that was enabled in the cybercycle condition. In Year 1, a recumbent Tunturi stationary bike (e60r) was utilized and interfaced with Netathalon riding software (v. 2.0) on an Acer laptop. Many participants were novice computer users and some had functional issues (e.g., arthritis in their hands) that made it difficult to use the computer components (e.g., learning to “mouse”). In Year 2, a different exergame setup was sought and the recumbent Expresso bike (S3R) was chosen since the computer components were more seamlessly integrated and included a touchpad that was easier to operate. In both years, participants in the cybercycle condition observed their avatar on the screen as a virtual rider traveling a virtual 3D terrain. There were no significant differences between years in ride frequency, intensity, or duration.

Am J Prev Med 2012;42(1) A-4

Appendix C

Details of cognitive measures

The primary cognitive domain of executive function was assessed by: Color Trails 2-1 difference score (time to connect dots alternating colors and numbers, minus time to connect numbers)1; Stroop C (time to state the ink color while suppressing the contrasting typed color name)2; and Digit Span Backwards (number of correct trials repeating a string of numbers in reverse order).3

Secondary cognitive domains included: (1) attention: Letter Digit Symbol Test (LDST; number of correct matches for digit-letter pairs within 60s; range 0–25)4; (2) verbal fluency: Controlled Word Association Test (COWAT; number of words within 1 minute that start with a given letter)3; (3) category fluency (number of words within 1 minute that fit a given category)3; (4) verbal memory, immediate: Rey Auditory Verbal Learning Task (RAVLT; sum of five trials to learn a list of 15 words; range 0–75)3 and RAVLT recall (number recalled of the practiced list after a distracter list)3; (5) verbal memory, delayed: RAVLT delay (recall of the practiced list after a 20-minute delay)3 and Fuld Object Memory Evaluation (Fuld; number of objects from a bag recalled after a 20- minute delay; range 0–10)5; (6) visuospatial skill: Complex Figure copy (score assigned by at least two raters for copying an arrangement of geometric shapes; range 0–36)3 and Clock (score assigned by at least two raters for drawing a clock with hands at a given time; range 0–8)6; (7) visuospatial memory: Complex Figure recall (scored by at least two raters for reproducing the previously copied figure after a 30-minute delay)3; and (8) motor function: Grooved Pegboard (time to fill a pegboard with pegs; with dominant and then nondominant hand).3

Alternate forms of tests were used at each subsequent evaluation.

References for Appendix C

1. D’Elia LG, Satz P, Uchiyama CL, White T. Color trails test. Odessa, FL: Psychological Assessment Resources, 1996.

2. van der Elst W, van Boxtel MPJ, van Breukelen GJP, Jolles J. The Stroop color-word test: influence of age, sex, and education; and normative data for a large sample across the adult age range. Assessment 2006;13:62–79.

3. Strauss E, Sherman EMS, Spreen O. A compendium of neuropsychological tests, 3rd Ed., New York: Oxford University Press, 2006.

4. van der Elst W, van Boxtel MPJ, van Breukelen GJP, Jolles J. The letter digit substitution test: normative data for 1,858 healthy participants aged 24–81 from the Maastricht Aging Study (MAAS): influence of age, education, and sex. J Clin Exp Neuropsychol 2006;28:998–1009.

5. Fuld PA, Masur DM, Blau AD, Crystal H, Aronson MK. Object-memory evaluation for prospective detection of dementia in normal functioning elderly: predictive and normative data. J Clin Exp Neuropsychol 1990;12:520–8.

6. LaRue A, Romero LJ, Ortiz IE, Liang HC, Lindeman RD. Neuropsychological performance of Hispanic and non-Hispanic older adults: an epidemiologic survey. Clin Neuropsychol 1999;13:474–86.

Am J Prev Med 2012;42(1) A-5

Appendix D

Details of enzyme-linked immunosorbant assay (ELISA) analyses

Intra-and inter-assay coefficients of variation were 3.7% and 8.5%, respectively. Ethylenediaminetetraacetic acid (EDTA) plasma samples were diluted 2–4X and assayed against a standard curve with a 2000 pg·ml–1 highest concentration. The supplied kit reagents were used as described in the manufacturer’s instructions, and the plate was read at 450 nm on a spectophotometric plate reader (BioTek, Winooski VT).

Am J Prev Med 2012;42(1) A-6

Appendix E

Neuropsychological and physiologic outcomes: age- and education-matcheda

Mean difference from baseline (95% CI)

Cybercycle Control bike (n=26) (n=26) p-value (df)b


Executive function

Color trails difference (2-1), seconds –15.67 (–28.22, –3.12)

14.08 (0.93, 27.23)

0.008 (1, 46)

Stroop C, seconds –6.57 (–15.01, 1.88) 0.96 (–7.53, 9.45) 0.05 (1, 46)

Digits backwards, sum score 0.23 (–0.53, 0.99) –1.38 (–2.14, –0.63) 0.01 (1, 47)

SECONDARY COGNITIVE OUTCOMESc

Attention, sum score

LDST 1.22 (–1.11, 3.56) 0.90 (–1.55, 3.35) 0.77 (1, 46)

Verbal fluency, sum score

COWAT 3.82 (–1.18, 8.81) 2.10 (–2.97, 7.17) 0.57 (1, 47)

Categories 0.66 (–0.90, 2.22) 1.26 (–0.33, 2.85) 0.61 (1, 47)


RAVLT, sum 5 trials score –0.61 (–3.92, 2.70) 0.01 (–3.32, 3.33) 0.82 (1, 46)

RAVLT immediate recall, score 0.97 (–0.18, 2.11) –0.17 (–1.29, 0.95) 0.22 (1, 45)

Verbal memory (delayed), score

RAVLT delayed recall 0.60 (–0.71, 1.92) –0.14 (–1.43, 1.15) 0.46 (1, 45)

Fuld delayed recall 0.14 (–0.58, 0.86) 0.32 (–0.40, 1.05) 0.73 (1, 48)

Visuospatial skill, sum score

Figure copy 4.16 (1.96, 6.36) 4.70 (2.40, 6.99) 0.79 (1, 46)

Clock –0.18 (–0.59, 0.23) –0.29 (–0.71, 0.13) 0.06 (1, 48)

Visuospatial memory (delayed), score

Figure delayed recall 0.46 (–1.58, 2.49) 2.90 (0.72, 5.08) 0.11 (1, 45)

Motor function, seconds

Pegboard, dominant hand 11.35 (–4.98, 27.68) 8.30 (–8.46, 25.07) 0.69 (1, 47)

Pegboard, nondominant hand 7.38 (–9.88, 24.64) 13.31 (–3.88, 30.50) 0.43 (1, 46)

Physiologic outcomes

Weight, kg –0.70 (–7.02, 5.62) 0.03 (–6.29, 6.35) 0.16 (1, 48)

BMI –0.26 (–2.60, 2.09) –0.01 (–2.35, 2.34) 0.20 (1, 48)

Fat mass, kg –1.15 (–5.54, 3.24) –0.92 (–5.32, 3.49) 0.61 (1, 48)

Lean mass, kg 0.43 (–2.36, 3.22) 0.73 (–2.06, 3.52) 0.41 (1, 48)

Abdominal fat, % –1.86 (–6.36, 2.64) –0.93 (–5.45, 3.60) 0.35 (1, 45)

Leg extension 60°, s–1 –2.06 (–12.66, 8.54) 14.91 (3.98, 25.84) 0.05 (1, 46)

Leg flex 60°, s–1 –1.75 (–8.73, 5.24) 9.23 (1.96, 16.49) 0.07 (1, 46)

Insulin, uU/mL 3.48 (0.58, 6.38) 1.87 (–1.14, 4.89) 0.43 (1, 46)

Glucose, mM/L 0.04 (–0.29, 0.37) –0.14 (–0.50, 0.21) 0.24 (1, 43)

Note: When more than one participant of the same age was available for matching, the decision was made based on matching education and/or gender. a Marginal Ms and SDs reported based on repeated-measures ANCOVA controlling for age and education

Am J Prev Med 2012;42(1) A-7

b For ANCOVA, repeated measures, group X time; the first df in parentheses refers to the effect (group X time) and the second refers to the error term. c No significant changes were expected, given prior research literature.

COWAT, controlled oral word association test; LDST, letter digit symbol test; RAVLT, Rey auditory verbal learning test

Am J Prev Med 2012;42(1) A-8

Appendix F

Exercise behavior outcomes after 3 months of exercise: age- and education-matcheda

M (SD) Difference between interventions, M (95% CI)

p-value (df)

Cybercycle (n=30)

Control bike (n=33)

Exercise behavior outcomes

Frequency of rides, n 51.3 (3.32) 53.3 (3.14) –1.96 (–2.31, –1.61) 0.68 (1, 59)

Power, wattsb 36.3 (3.28) 32.1 (3.15) 4.20 (3.93, 4.46) 0.44 (1, 31)

Energy expended, kcal 107.9 (8.05) 93.6 (7.63) 14.32 (13.47, 15.17) 0.23 (1, 59)

Duration, minutes 35.5 (1.81) 33.8 (1.72) 1.61 (1.42, 1.80) 0.54 (1, 59)

Distance average, miles 5.4 (0.40) 4.8 (0.38) 0.65 (0.61, 0.69) 0.27 (1, 59)

Distance total, miles 283.9 (28.80) 261.4 (27.29) 22.51 (19.47, 25.54) 0.59 (1, 59)

Speed average, mphb 7.4 (0.38) 8.3 (0.37) –0.83 (–0.86, –0.80) 0.19 (1, 31)

Speed peak, mphb 10.7 (0.39) 9.8 (0.37) 0.97 (0.94, 1.00) 0.13 (1, 31)

Physical activity daily, kcal 324.4 (32.91) 304.2 (32.22) 20.22 (0.94, 1.00) 0.66 (1, 43) a Marginal Ms and SDs reported based on ANCOVA controlling for age and education b Sample sizes: cybercycle (n=17) and control bike (n=18) due to enhanced ride data available in Year 2

Am J Prev Med 2012;42(1) A-9

Appendix G

Characteristics of completers versus noncompleters, M (SD) unless otherwise indicated

Cybercycle Control bike

Completers (n=30)

Noncompleters (n=8) p-value

Completers (n=33)

Noncompleters (n=8) p-value

Age, years 76.1 (10.3) 74.0 (8.3) 0.62 81.6 (6.3) 81.8 (5.7) 0.94

Women, n (%) 25 (83.3) 8 (100) 0.27 21 (63.6) 8 (100) 0.06

Education, years 12.7 (2.4) 11.9 (0.9) 0.35 14.8 (2.3) 14.5 (2.1) 0.72

Physiologic factors

Weight, kg 74.8 (13.7) 75.9 (10.9) 0.86 73.3 (17.0) 66.8 (8.7) 0.33

BMI 28.6 (4.7) 31.2 (5.0) 0.23 27.8 (6.6) 25.7 (4.5) 0.44

Fat mass, kg 31.4 (8.4) 33.9 (5.9) 0.49 28.0 (12.4) 28.1 (8.8) 0.98

Lean mass, kg 40.9 (6.7) 39.1 (4.5) 0.53 43.0 (6.9) 36.7 (2.8) 0.02

Abdominal fat, % 46.6 (8.2) 55.3 (5.4) 0.09 39.5 (12.5) 41.9 (12.8) 0.65

Insulin, uU/mL 11.1 (5.3) 8.8 (1.9) 0.31 9.9 (8.2) 9.9 (7.9) 0.99

Glucose, mM/L 6.0 (0.9) 8.6 (4.0) 0.002 5.5 (0.7) 5.4 (0.6) 0.68

Physical activity level, daily kcal 286.9 (234.3) 339.7 (180.3) 0.63 312.8 (226.2) 288.2 (195.3) 0.83

NEUROPSYCHOLOGICAL MEASURES

Intelligence proxy (NAART), IQ 118.9 (8.9) 112.6 (6.3) 0.09 120.9 (5.0) 119.5 (6.1) 0.52

Executive function

Color trails difference (2-1), seconds 58.4 (30.2) 84.3 (90.8) 0.19 59.1 (28.2) 79.7 (47.1) 0.12

Stroop C, seconds 58.2 (21.8) 72.0 (50.0) 0.27 55.5 (17.3) 77.9 (50.0) 0.04

Digits backwards, sum score 6.2 (1.5) 5.0 (1.6) 0.08 6.9 (2.2) 6.1 (1.8) 0.33

Attention, sum score

LDST 30.1 (7.2) 27.0 (7.1) 0.31 30.7 (5.7) 25.5 (6.3) 0.03

Verbal fluency, sum score

COWAT 33.0 (14.1) 32.4 (11.7) 0.93 39.8 (11.8) 33.8 (12.3) 0.20

Categories 12.1 (3.4) 13.0 (2.4) 0.51 11.6 (4.2) 13.0 (3.4) 0.38


RAVLT, sum 5 trials score 36.7 (9.5) 28.0 (5.7) 0.03 35.8 (9.2) 35.9 (9.3) 0.97

Am J Prev Med 2012;42(1) A-10

RAVLT immediate recall, score 6.3 (3.0) 4.0 (3.2) 0.09 6.0 (2.9) 5.6 (2.2) 0.71


RAVLT delayed recall, score 6.1 (3.2) 2.9 (2.9) 0.02 5.6 (3.4) 6.1 (2.2) 0.66

Fuld delayed recall, score 7.3 (1.5) 6.9 (0.7) 0.43 7.0 (1.9) 6.9 (1.5) 0.86


Figure copy 25.9 (4.3) 25.5 (5.3) 0.82 26.8 (6.0) 24.1 (6.5) 0.29

Clock 6.5 (1.1) 5.3 (2.1) 0.05 6.7 (1.0) 6.0 (1.2) 0.11


Figure delayed recall 12.6 (5.3) 10.1 (4.2) 0.26 11.8 (5.6) 11.7 (5.2) 0.96

Motor function, seconds Pegboard, dominant hand 109.2 (47.0) 103.3 (19.7) 0.75 112.9 (32.1) 132.4 (33.3) 0.14 Pegboard, nondominant hand 120.4 (59.8) 118.7 (26.7) 0.94 126.2 (37.5) 143.1 (40.2) 0.27

COWAT, controlled oral word association test; LDST, letter digit symbol test; NAART, North American adult reading test; RAVLT, Rey auditory verbal learning test

Am J Prev Med 2012;42(1) A-11

Appendix H

Neuropsychological and physiologic outcomes after 3 months of exercise (complete-case analysis)a

Mean difference from baseline (95% CI) Cybercycle Control bike

(n=30) (n=33) p-value (df)b


Executive function

Color trails difference (2-1), seconds –17.61 (–19.70, 15.52) 10.85 (8.79, 12.90) 0.005 (1, 57)

Stroop C, seconds –6.18 (–7.37, –4.98) 1.07 (–0.06, 2.20) 0.04 (1, 57)

Digits backwards, sum score 0.39 (0.29, 0.49) –1.12 (–1.21, –1.02) 0.007 (1, 57) SECONDARY COGNITIVE OUTCOMES, SUM SCOREc

Attention

LDST, sum score 0.88 (0.65, 1.11) 0.55 (0.33, 0.77) 0.78 (1, 57)

Verbal fluency

COWAT, sum score 4.04 (3.17, 4.91) 2.37 (1.53, 3.20) 0.54 (1, 58)

Categories, sum score 0.54 (0.68, 0.41) 1.27 (1.40, 1.14) 0.49 (1, 58)


RAVLT, sum 5 trials score –0.41 (–1.05, 0.22) 0.69 (0.09, 1.29) 0.68 (1, 57)

RAVLT Immediate Recall, score 0.92 (0.75, 1.09) –0.02 (–0.18, 0.14) 0.24 (1, 56)

Verbal memory (delayed), score

RAVLT delayed recall 0.61 (0.50, 0.72) 0.00 (–0.11, 0.10) 0.49 (1, 56)

Fuld delayed recall 0.10 (0.09, 0.11) 0.27 (0.26, 0.29) 0.74 (1, 59)


Figure copy 4.00 (4.39, 3.60) 3.95 (4.34, 3.56) 0.98 (1, 57)

Clock –0.08 (–0.13, –0.03) –0.16 (–0.21, –0.11) 0.84 (1, 59)


Figure delayed recall 0.12 (0.38, –0.14) 1.99 (2.25, 1.74) 0.27 (1, 56)

Motor function, seconds

Pegboard, dominant hand 10.91 (9.38, 12.43) 7.70 (6.23, 9.17) 0.67 (1, 58)

Pegboard, nondominant hand 8.09 (4.83, 11.34) 14.96 (11.88, 18.04) 0.34 (1, 57)

Physiologic outcomes

Weight, kg –0.58 (–0.72, –0.44) 0.12 (–0.01, 0.26) 0.18 (1, 59)

BMI –0.21 (–0.25, –0.16) 0.03 (–0.01, 0.07) 0.24 (1, 59)

Fat mass, kg –1.04 (–0.92, –1.15) –0.77 (–0.66, –0.88) 0.56 (1, 59)

Lean mass, kg 0.45 (0.37, 0.52) 0.67 (0.59, 0.74) 0.56 (1, 59)

Abdominal fat, % –1.80 (–2.06, –1.54) –0.86 (–1.12, –0.61) 0.31 (1, 56)

Leg extension 60s–1 –2.85 (–2.72, –2.98) 12.15 (12.28, 12.02) 0.06 (1, 56)

Leg flex 60 s–1 –2.70 (–3.01, –2.40) 6.80 (6.50, 7.11) 0.08 (1, 56)

Insulin, uU/mL 3.23 (2.75, 3.71) 1.54 (1.08, 2.01) 0.36 (1, 56)

Glucose, mM/L 0.00 (–0.07, 0.07) –0.10 (–0.18, –0.03) 0.54 (1, 54)

Am J Prev Med 2012;42(1) A-12

a Marginal Ms and SDs reported based on repeated measures ANCOVA controlling for age and education. b For ANCOVA, repeated measures, group X time; the first df in parentheses refers to the effect (group X time) and the second refers to the error term. c No significant changes were expected, given prior research literature.

COWAT, controlled oral word association test; LDST, letter digit symbol test; RAVLT, Rey auditory verbal learning test

Am J Prev Med 2012;42(1) A-13

Appendix I

Adverse events and discontinuation of intervention during 3-month exercise intervention

Cybercycle Control bike Total 7 — Knee or sciatica pain while cycling 2a 2 Acute illness (upper respiratory to emergency room) — 1 Other injuries (hurt back lifting, car accident) 1 1 Cancer diagnosis and treatment 2 — Frustrated interacting with bike computer 1 2 Vertigo while cycling 1a —

a These two events occurred in the same participant. For some riders, the sense of motion evoked by virtual scenery may induce vertigo; this was rare in the current study, perhaps due to the use of a recumbent bike, which provides greater distance from the video display.

Am J Prev Med 2012;42(1) A-14

Appendix J

Neuropsychological and physiologic outcomes: three executive function measures and BDNF

Note: Change in executive function and BDNF before and after 3 months of exercise. Cybercyclists are represented by solid lines; control cyclists are represented by dashed lines. (A) Color Trails 2-1; (B) Stroop C; (C) Digit Span Backwards; (D) BDNF. Group X time interactions, controlling for age and education, were significant (p=0.007, 0.05, 0.03, and 0.05, respectively).

BDNF, brain-derived neurotrophic factor

Author's personal copy...ghost rider (last best ride). During Month 3, participants were instructed...

Documents

Transcript of Author's personal copy...ghost rider (last best ride). During Month 3, participants were instructed...