Professional Documents
Culture Documents
Pedro Gamito, PhD,1,2 Jorge Oliveira, PhD,1,2 Diogo Morais, MSc,1,2 Cátia Coelho, MSc,3 Nuno Santos, MSc,3
Catarina Alves, BA,3 Ana Galamba, BA,3 Miguel Soeiro, MSc,4 Madhurrima Yerra, BA,5 Hannan French, BA,5
Lily Talmers, BA,5 Tiago Gomes1 and Rodrigo Brito, PhD1,2
Downloaded by Australian National University from www.liebertpub.com at 07/24/18. For personal use only.
Abstract
As the demographic structure in western societies ages, the prevalence and impact of cognitive decline rises.
Thus, new solutions to tackle this problem are required. The use of Information and Communication Tech-
nologies (ICT)-based cognitive exercises has emerged in the last few decades, though with inconsistent results.
Hence, we conducted a pre-post treatment study to further investigate this approach. We designed a set of
virtual reality exercises that mimic activities of daily living by which the patient can train different cognitive
domains. Twenty-five participants, ages 65–85, underwent 12 training sessions between the pre-treatment and
post-treatment assessments. Significant increases were seen between the two assessments for some of the
neuropsychological measures: visual memory, attention, and cognitive flexibility. Results also suggest that
participants with lower baseline cognitive performance levels improved most after these sessions.
Introduction tive domain and design choice were the primary factors
behind the efficacy of the specific technique used.6 A set of
1
School of Psychology and Life Sciences, Lusophone University of Humanities and Technologies, Lisbon, Portugal.
2
HEI-Lab: Digital Human-Environment Interaction Lab, Lisbon, Portugal.
3
Junta de Freguesia de Benfica, Gabinete de Psicologia do Pelouro dos Direitos Sociais e Saúde, Lisbon, Portugal.
4
Department of Informatics, Universidade de Lisboa Instituto Superior Tecnico.
5
University of Michigan College of Literature Science and the Arts, Ann Arbor, Michigan, Lisbon, Portugal.
1
2 GAMITO ET AL.
computer-based approaches6 also found interesting results to and 74 (M = 74; standard deviation [SD] = 5.27; ranging from
support these solutions, but they warned against using these 65 to 85 years) were recruited at the center and participated
solutions outside of a broader rehabilitation program super- in this study. Their average education level was 6 years of
vised by specialists. A more positive outlook is presented in a school (SD = 2.42; ranging from 4 to 12 years). Exclusion
review in which the authors conclude that the computerized criteria were as follows: (a) having been clinically diagnosed
training is at least as effective as traditional solutions.11 In that with a psychiatric or neurological disorder and (b) a history
same review, the authors found that most of the issues raised of alcohol or substance abuse. Two participants were ex-
when addressing computer-based solutions are not confirmed cluded because of depression, and one participant, because of
by the results. Furthermore, these solutions were found to transient ischemic attack.
often provide swifter, less labor-intensive, and more effective
alternatives to traditional approaches. Measures
It is now common to find serious games (SG) allied to
The neuropsychological measures used in this study are
virtual reality (VR). SGs have all the properties of their less
all well-established measures with good psychometric
serious cousins. However, these games are designed not for the
properties: the Montreal Cognitive Assessment (MoCA),20
entertainment of players, but to produce a positive impact on
used for cognitive screening; the Frontal Assessment Bat-
the player’s mental and physical health.12 VR/SG apps sim-
tery (FAB),21 a set of six subtests that assess executive
ulate real-life environments. Unlike traditional paper-and-
functioning; the Wisconsin Card Sorting Test (WCST),22
Downloaded by Australian National University from www.liebertpub.com at 07/24/18. For personal use only.
to bake a cake), working memory tasks (e.g., shopping at a stability in the calculation of a standardized change between
grocery store), auditory memory tasks (e.g., listening to and test scores.28
remembering news on TV), executive functions (e.g., se- Test-retest scores were then calculated for each neu-
lecting the appropriate clothes to wear, arranging shoes in a ropsychological variable to estimate the stability over time of
shoe closet), as shown in Figure 1—from top-left clockwise, each measure used in the calculation of the RCI. This ap-
the bathroom, living room in the apartment, virtual kitchen proach was developed as an alternative to the limitations of
test, shoe closet test, bedroom, outdoor environment, phar- comparisons between subjects and of situations in which
macy, and art gallery test. The level of difficulty for each cases cannot easily be matched with controls in their indi-
type of task was gradually increased throughout the sessions, vidual characteristics.29
so that, for instance, during the first session the participants The resulting Z-score in RCI measures the performance of
had to complete simple tasks, and during the last session they the participant during post-treatment assessment relative to
had to complete more complex tasks (e.g. to buy several items their baseline for each of the measures. Accordingly, a
in the grocery store). The detailed description of the inter- change exceeding Z – 1.96 ( p < 0.05) reflects a reliable
vention protocol is presented in Supplementary Table S1; neuropsychological change between test scores for a 95
Supplementary Data are available online at www.liebertpub percent confidence interval (CI) (2.5 percent in a positive
.com/cyber). direction and 2.5 percent in a negative direction for a two-
tailed test - [a/2 at 95 percent CI]).
Downloaded by Australian National University from www.liebertpub.com at 07/24/18. For personal use only.
FIG. 1. Screenshots
depicting some of the tasks
used in VR/SG cognitive
stimulation. VR/SG, virtual
reality/serious games.
4 GAMITO ET AL.
where Sdiff reflects the standard error of difference for test pre-treatment assessment to a normative score (i.e., above 15
scores: points) at the time of post-treatment assessment. The dif-
ferences in all the mean scores were tested with a paired
qffiffiffiffiffiffiffiffiffiffiffiffiffi
sample t-test, which showed significant differences between
Sdiff ¼ 2ðSE Þ2 pre-treatment and post-treatment assessments for the number
of perseverative errors of WCST (t[21] = 2.493; p = 0.021),
which accounts for the standard error of measurement (SE): the score in the memory trial of RCF (t[22] = -3.991;
pffiffiffiffiffiffiffiffiffiffiffiffiffi p = 0.001), error percentage (t[16] = 3.463; p = 0.003), and
SE ¼ S1 1 rxx concentration index of d2 (t[16] = -2.926; p = 0.010).
where the S1 is the SD at pre-treatment, whereas rxx repre- Test-retest reliability scores
sents the test-retest reliability.
The RCIs were calculated individually for each neu- Participants in this sample were retested 3 months after
ropsychological measure. On the basis of 1,000 bootstrap their initial tests. The same neuropsychological instruments
samples, the correlations between the RCI and variables such were used in both assessments. These instruments, which are
as age, education and global cognitive ability at pre- well established, have been extensively studied in the geri-
treatment were estimated with 95 percent CIs, a level chosen atric population. Table 1 depicts the test-retest scores (rxx)
for our sample. These scores were below acceptable (<0.50)
Downloaded by Australian National University from www.liebertpub.com at 07/24/18. For personal use only.
Table 1. Descriptive Statistics and Test-Retest Scores for the Neuropsychological Variables
Pre-treatment Post-treatment
M1 S1 M2 S2 t rxx
MoCA 22.88 3.05 23.00 4.00 -0.244 0.79
FAB 14.68 2.28 15.48 1.27 -1.528 0.33
WCST trials 125.42 7.70 119.83 13.83 1.687 0.36
WCST errors 57.21 20.19 47.00 21.82 1.762 0.18
WCST P errors 32.21 20.15 23.87 13.26 2.493* 0.51
WCST categories 2.96 2.01 3.87 1.76 -1.622 0.35
RCF copy score 28.88 5.42 28.74 5.16 0.763 0.67
RCF copy time 5.71 3.07 5.96 3.50 0.775 0.57
RCF memory score 11.98 5.24 14.98 5.74 -3.991** 0.80
RCF memory time 3.01 1.77 3.35 1.97 -0.609 0.49
d2 targets processed 337.95 77.13 316.70 79.63 1.106 0.18
d2 errors, % 14.57 14.42 4.23 5.20 3.463** 0.60
d2 global performance 286.32 55.72 303.91 66.32 -1.087 0.41
d2 concentration index 86.84 52.74 118.52 23.67 -2.926* 0.57
ECQ 35.20 5.16 33.56 3.35 1.459 0.42
BDI II 10.28 6.31 10.78 8.31 0.048 0.79
M1, mean values at pre-treatment; S1, SD at pre-treatment; M2, mean values at post-treatment; S2, SD at post-treatment; rxx, test-retest
reliability.
*p < 0.05.
**p < 0.01.
BDI-II, Beck Depression Inventory-II; ECQ, Everyday Competence Questionnaire; FAB, Frontal Assessment Battery; MoCA, Montreal
Cognitive Assessment; RCF, Rey complex figure; RCF memory score, Rey complex figure memory trial total score; SD, standard deviation;
WCST, Wisconsin Card Sorting Test (values are percentages, except the number of categories which is the raw score); WCST P errors,
perseverative errors in the WCST in percentage.
COGNITIVE STIMULATION WITH VR IN THE ELDERLY 5
FIG. 2. Percentages of
cases for positive and nega-
tive change (RCI). RCI,
Reliable Change Index.
Downloaded by Australian National University from www.liebertpub.com at 07/24/18. For personal use only.
higher proportion of positive change across the outcomes of Since the elderly are generally vulnerable to cognitive de-
the study. The proportion of positive and negative changes cline, we tested whether a 3-month cognitive stimulation
that were unlikely to occur by chance at a 95-percent CI is program would help improve cognitive functioning in this
depicted in Table 2. Changes that were positive (Z > 1.96) or population. Participants showed statistically significant im-
negative (Z < 1.96) in more than 2.5 percent (a/2 for two- provements in attention (d2), in visual memory (Rey Com-
tailed predictions) would probably be due to a treatment plex Figure), and on two indicators of cognitive flexibility
effect, because only the extreme outcomes were considered (WCST), as shown by group analysis.
significant changes. Table 2 shows that improvements A control group would have allowed us to discern the effects
(Z > 1.96) were observed in 9 percent of the sample for of confounding variables (e.g., practice effects) on treatment
WCST (perseverative errors), 22 percent for the RCF (score outcomes. Because we lacked a control group, we adopted a
in the memory trial), 12 percent in error percentage, and 13 more conservative statistical approach based on RCI. This
percent in the concentration index of d2. None of the cases approach has been used in a variety of clinical research studies
showed a significant negative change in neuropsychological because of its ability to detect how much each individual has
outcomes. changed between the pre-treatment and post-treatment as-
sessments while controlling for the standard error of difference
Correlates of cognitive performance between test and retest scores.28,29 This change results in a
We also explored whether these improvements were re- standardized score, in which scores above or below Z = 1.96 are
lated to individual characteristics such as age, education, or less likely to occur by chance (i.e., only in 5 percent of the
sample for a 95 percent CI). A higher percentage than 2.5
general cognitive ability at baseline. Bivariate correlations
were used with bootstrap sampling. The results show a sig- percent (a/2) of improved cases was observed on cognitive
nificant correlation between the improvements in the error flexibility (9 percent), attention/concentration (12 percent to 13
percent) and visual memory (22 percent). These results show
percentage of d2 and the total score of the MoCA ( r = -0.44;
bootstrapped 95 percent CI [-0.23 to -0.74]; SE = 0.13), that there were strong improvements in main domains involved
which suggests that the magnitude of improvements is in the VR/SG program.
The tasks that were used in the cognitive stimulation
highest for individuals with lower levels of cognitive func-
tioning at baseline. program are general tasks that mimic daily life activities but
are not specific, focused tasks for any of these domains.
Discussion However, the specific functions in which significant differ-
ences were observed, that is, attention, memory, and cogni-
Our study was carried out with elderly participants who tive flexibility, are required in daily activities, such as
had not been clinically diagnosed with cognitive deficits. shopping at the supermarket, preparing food, and choosing
the right clothes to wear; in most sessions, these activities
Table 2. Reliable Change Index were used to train participants. Conversely, no significant
differences were observed among the results of the screening
(+) Change (-) Change tests (MoCA and FAB), which assess general functioning;
Z > 1.96 Z < 1.96 this finding suggests that these improvements were restricted
to specific domains of cognitive functioning and are con-
WCST P errors 9 0 sistent with the results of recent reviews showing improve-
RCF memory score 22 0
ments in attention and memory as specific effects of ICT-
d2 errors 12 0
d2 concentration index 13 0 based interventions.16
This finding contrasts with a previous study, in which a
Values are in percentages. similar VR/SG approach was used but in which we used a
6 GAMITO ET AL.
clinical sample diagnosed with cognitive deficits. In that 5. Ballesteros S, Kraft E, Santana S, et al. Maintaining older
study, we found significant improvements in executive func- brain functionality: A targeted review. Neuroscience and
tioning, which was assessed with the FAB.30 However, in that Biobehavioral Reviews 2015; 55:453–477.
study, the sample consisted of patients undergoing community 6. Lampit A, Hallock H, Valenzuela M. Computerized cog-
treatment for alcohol use disorder, which has a different pat- nitive training in cognitively healthy older adults: A sys-
tern of cognitive deficits from the sample in our study. tematic review and meta-analysis of effect modifiers. PLoS
We explored the data set further to determine the rela- Medicine 2014; 11:e1001756.
tionships between the RCIs and individual characteristics 7. Valenzuela M, Sachdev PS. Can cognitive exercise prevent
such as age, education, and general cognitive functioning; the onset of dementia? Systematic review of randomized
the results suggest that the improvements found in attention clinical trials with longitudinal follow-up. The American
Journal of Geriatric Psychiatry 2009; 17:179–187.
(through the decreases in error percentage in d2) were neg-
8. Papp KV, Walsh SJ, Snyder PJ. Immediate and delayed
atively associated with general cognitive functioning at
effects of cognitive interventions in healthy elderly: A re-
baseline. This suggests that if improvements in attention are view of current literature and future directions. Alzheimer’s
the main aim, the most efficient use of this program is to and Dementia 2009; 5:50–60.
target patients showing some signs of cognitive decline. No 9. Gates NJ, Sachdev PS, Fiatarone Singh MF, et al. Cognitive
effects of age or education were found on RCIs within the and memory training in adults at risk of dementia: A sys-
limited range of both in our sample. tematic review. BMC Geriatrics 2011; 11:55.
Downloaded by Australian National University from www.liebertpub.com at 07/24/18. For personal use only.
Our results should be interpreted with some caution, since 10. Johansson B, Tornmalm M. Working memory training for
our study lacked a control group. We sought to minimize this patients with acquired brain injury: effects in daily life.
problem by transforming the outcomes to RCIs, which is a Scandinavian Journal of Occupational Therapy 2011; 19:
more conservative approach to the analysis of pre-post- 176–183.
treatment effects. This does not allow us to discern the effects 11. Kueider A, Parisi J, Gross A, et al. Computerized cognitive
of practice on repeated test scores. We should note, however, training with older adults: A systematic review. PLoS One
that practice effects are more likely to be seen in measures that 2012; 7:e40588.
involve problem-solving abilities;31 problem-solving was not 12. Gamito P, Oliveira J, Morais D, et al. (2011) Serious games for
required in the evaluations of memory and attention, which serious problems: from Ludicus to Therapeuticus. In Kim JJ,
were assessed using the RCF and the d2, respectively. Also, ed. Virtual reality. London, UK: IntechOpen, pp. 527–548.
the 3-month post-treatment interval may have reduced the 13. Gamito P, Oliveira J, Alghazzawi D, et al. The art gallery
practice effects of repeated exposure to these tests. One way of test: A preliminary comparison between traditional neu-
overcoming this limitation would be to use a standard ropsychological and ecological VR-based tests. Frontiers in
regression-based change score approach from an independent Psychology 2017; 8:1911.
14. Larson EB, Feigon M, Gagliardo P, et al. Virtual reality and
normative group drawn from a different study32 that would
cognitive rehabilitation: A review of current outcome re-
account for practice effects; however, there were no data
search. NeuroRehabilitation 2014; 34:759–772.
available for most of the tests used in the current study. 15. Pietrzak E, Pullman S, McGuire A. Using virtual reality
The results of our cognitive flexibility, attention, and and videogames for traumatic brain injury rehabilitation: A
memory tests are consistent with recent reviews17,18 and structured literature review. Games for Health Journal
highlight the advantages of using VR/SG for cognitive in- 2014; 3:202–214.
tervention, even with populations that may be unfamiliar 16. Shin HH, Kim KM. Virtual reality for cognitive rehabili-
with the use of such technology, such as the elderly. As tation after brain injury: A systematic review. Journal of
technology keeps progressing, there is great potential for the Physical Therapy Science 2015; 27:2999–3002.
improvement and increased ecological validity of VR/SG. 17. Coyle H, Traynor V, Solowij N. Computerized and virtual
reality cognitive training for individuals at high risk of
Author Disclosure Statement cognitive decline: Systematic review of the literature. The
American Journal of Geriatric Psychiatry 2015; 23:335–
No competing financial interests exist.
359.
18. Bogdanova Y, Yee MK, Ho VT, et al. Computerized
References
cognitive rehabilitation of attention and executive function
1. Giannakouris K. (2010) Regional population projections in acquired brain injury. Journal of Head Trauma Re-
EUROPOP2008: Most EU regions face older population habilitation 2016; 31:419–433.
profile in 2030. Statistics in Focus, 1/2010. Luxembourg: 19. Rosa PJ, Sousa C, Faustino B, et al. (2017) The effect of
Publications Office of the European Union. virtual reality-based serious games in cognitive inter-
2. Hedden T, Gabrieli JD. Insights into the ageing mind: A ventions: A meta-analysis study. In Proceedings of the
view from cognitive neuroscience. Nature Reviews Neu- 4th Workshop on ICTs for improving Patients Re-
roscience 2004; 5:87–96. habilitation Research Techniques. Lisbon, Portugal: ACM,
3. Aguirre E, Woods RT, Spector A, et al. Cognitive stimu- pp. 113–116.
lation for dementia: A systematic review of the evidence of 20. Freitas S, Simo*es MR, Alves L, et al. Montreal Cognitive
effectiveness from randomised controlled trials. Ageing Assessment (MoCA): Normative study for the Portuguese
Research Reviews 2013; 1:253–262. population. Journal of Clinical and Experimental Neu-
4. Elliott M, Parente F. Efficacy of memory rehabilitation ropsychology 2011; 33:989–996.
therapy: A meta-analysis of TBI and stroke cognitive re- 21. Lima C, Meireles L, Fonseca R, et al. The Frontal Assess-
habilitation literature. Brain Injury 2014; 28:1610–1616. ment Battery (FAB) in Parkinson’s disease and correlations
COGNITIVE STIMULATION WITH VR IN THE ELDERLY 7
with formal measures of executive functioning. Journal of 29. Chelune GJ, Naugle RI, Luders H, et al. Individual change
Neurology 2008; 255:1756–1761. after epilepsy surgery: Practice effects and base-rate in-
22. Heaton RK, Chelune GJ, Talley JL, et al. (1993) Wisconsin formation. Neuropsychology 1993; 7:41–52.
card sorting test (WCST)—Manual revised and expanded. 30. Gamito P, Oliveira J, Lopes P, et al. Executive Functioning
Odessa, FL: Psychological Assessment Resources. in Alcoholics Following an mHealth Cognitive Stimulation
23. Osterrieth P. Filetest de copie d’une figure complex: con- Program: Randomized Controlled Trial. Journal of Medical
tribution a l’etude de la perception et de la memoire [The Internet Research 2014, 16:e102.
test of copying a complex figure: a contribution to the study 31. Dikmen SS, Heaton RK, Grant I, et al. Test-retest reliability
of perception and memory]. Archives de Psychologie 1944; and practice effects of Expanded Halstead-Reitan Neu-
30:286–356. ropsychological Test Battery. Journal of the International
24. Brickenkamp R. (2002). The d2 test of attention. Gottingen, Neuropsychological Society. 1999; 5:346–356.
Germany: Hogrefe Verlag GmbH & Co. 32. Cysique LA, Franklin D, Abramson I, et al. Normative data
25. Beck A, Steer R, Brown G. (1996) Manual for the Beck and validation of a regression based summary score for
Depression Inventory II. San Antonio, TX: Psychological assessing meaningful neuropsychological change. Journal
Corporation. of Clinical and Experimental Neuropsychology 2011; 33:
26. Kalisch T, Richter J, Lenz M, et al. Questionnaire-based 505–522.
evaluation of everyday competence in older adults. Clinical
Interventions in Aging 2011; 6:37–46. Address correspondence to:
Downloaded by Australian National University from www.liebertpub.com at 07/24/18. For personal use only.