31 May 2015, Amsterdam
The Mediating Effects of Physical Activity on
the Effectiveness of Cognitive Training
An investigation into the mediating effects of physical activity on the effects of cognitive
training on cognitive functioning in healthy elderly and people who suffered from a
stroke
By: Jochem Jansen (6151612)
The aim of this study is to investigate whether cognitive training is effective as an intervention for cognitive decline and whether physical activity during training is a mediating factor of this effect. This study focuses on the difference in this effect between healthy elderly (n=35) and people who have suffered from a stroke (n=31). Subjects from these two target groups performed a training program on a computer at home for 12 weeks adding up to 58 times half an hour of training. The participants would either do an experimental training designed to enhance cognitive functioning, or the control training designed to not enhance cognitive functioning. Cognitive functioning was measured with tests focusing on verbal fluency, short-term memory, logical thinking, executive functioning and working memory. All subjects improved on a composite measure of cognitive functioning after having completed the training program. No difference in the amount of improvement was observed between the experimental training in comparison to the control training. Furthermore, there was no difference present between the amount of improvement between healthy elderly and people who had suffered from a stroke. Finally, the levels of physical activity during the training program did not influence the amount of improvement. These results indicate that the experimental cognitive training was not more effective than
non-cognitive training as an intervention for non-cognitive decline and levels of physical activity during training did not mediate this effect.
Introduction
Age-cognition relations are well-established both in cross-sectional comparisons, as well as in longitudinal studies. In general, there is a positive relationship between increasing age and cognitive decline (Bishop, Duncan, Brett, & Lawrence, 2004; Kramer & Erickson, 2007; Laurin, Verreault, Lindsay, MacPherson, & Rockwood, 2001; Lustig, Shah, Seidler, & Reuter-Lorenz, 2009; Persson et al., 2006), although there is high inter-individual variability (Buitenweg, Murre, & Ridderinkhof, 2012).
Furthermore, age related cognitive decline are characterized by negative effects such as decline in memory, decision making and cognitive control (Buitenweg et al., 2012; Stephen & Richard, 2009). The Netherlands has an increasing aging population, with a steady increase of people over the age of 65 since 1950 from 7.7.% up to 17.4% in 2014 (Centraal Bureau voor de Statistiek). In addition to the general cognitive decline symptoms in healthy aging, the aging population is also increasingly at risk of having a stroke with an overall prevalence of 4,84% for people up to 65 years of age, and 7,06% for
2
people older than 65 (Di Carlo et al., 2000). For stroke patients it is common to experience additional cognitive decline (Poulin, Korner-Bitensky, Dawson, & Bherer, 2012). With both an ever-increasing large aging population in the Netherlands, as well as additional cognitive decline due to the increased
prevalence of stroke patients, this study highlights the need to effectively manage negative effects associated with older age, including the development of targeted therapeutic interventions. It is of great importance to investigate ways to prevent, slow down or even reverse cognitive decline. This study therefore focuses on two methods that may help mediate cognitive decline: cognitive training and physical activity.
First, research into the effectiveness of using cognitive training to help elderly reduce cognitive decline originates from the 1970's (Lustig et al., 2009). Studies have shown that cognitive training may have a positive effect on the performance of the cognitive system that has been targeted by the training. This holds for both old and young populations (Buitenweg et al., 2012; Lustig et al., 2009). However, because of the natural decline of cognitive functioning due to healthy aging (Damoiseaux et al., 2008; O’Brien et al., 2010; Persson et al., 2006), beneficial effects of cognitive training might be of more importance to the elderly. Furthermore, it has been shown that cognitive training is also effective for people who have suffered from a stroke (Poulin et al., 2012). This underlines the importance to compare the effects of cognitive training on cognitive functioning in the elderly and people who have suffered from a stroke.
Second, physical exercise during lifetime has been shown to contribute to the prevention of cognitive decline (Laurin et al., 2001; Sofi et al., 2011; van Gelder et al., 2004; Yaffe, Barnes, Nevitt, Lui, &
Covinsky, 2001). However, literature from intervention studies supporting or refuting the claim that physical exercise could help slow down cognitive decline or increase cognitive functioning could not be found. Thus, to date, the potential mediating effects of physical activity on cognitive decline remains to be investigated. It seems reasonable for this effect to be present since physical activity has a beneficial effect protecting cognitive decline, which might be translatable to mediating cognitive decline
interventions. The aim of this study is therefore twofold: a) to investigate whether protective effects of physical activity levels on cognitive decline can also enhance the beneficial effects of cognitive training on cognitive functioning, and b) whether this potential effect is more effective in healthy elderly people, or people who have suffered from a stroke. To this aim, we examined whether the hypothesized
mediating effect of physical activity on the beneficial effects of cognitive training is present after a 12 week online cognitive training program performed by both healthy elderly and people who suffered from a stroke or cerebrovascular accident (CVA group).
3
Methods
1.1 participants
For the current study 35 healthy participants were included for analysis. Exclusion criteria were based on the presence of pre-existing conditions that could be potential confounding factors, namely: epilepsy, mental handicap, neurodegenerative defect, severe psychiatric disorders. Furthermore, a functioning computer with internet connection and the ability to use them was required for the training part. The age of the healthy participants included ranged from 49 to 75 years old (mean = 66, sd = 4.3), consisting of 23 females. Healthy participants were randomly assigned to the control mock training condition (n=14) or the experimental training condition (n=21). The conditions are described in the methods section.
In total 31 stroke participants were included in analysis. People that suffered from strokes with an ischemic or hemorrhagic cause (including brain stem strokes) were included. Sub-arachnoid bleeds were excluded. The same exclusion criteria as for healthy subject also applied to subjects in the stroke group. For this group additional exclusion criteria were set to ensure ability to participate in the testing and training during the study. Additional exclusion criteria were: severe auditory or visual handicap, severe neglect and severe aphasia. The inclusion criteria was that stroke patients had cognitive dysfunction and complaints and assessed as such by an experienced clinician. Participants with paresis or paralysis were included as long as the subject was able to use a computer mouse and send emails. The age of the stroke participants ranged from 42 years old to 77 years old (mean = 60, sd = 7.9), consisting of 13 females. Stroke participants were randomly assigned to either the mock training condition (n=18) or the experimental training condition (n=13). The study design had been approved the ethical commission of the VU medical center (METc) and all participants signed an informed consent form.
1.2 Procedure
Before training, participants were requested to come to the University of Amsterdam to perform a battery of tests about their cognitive functioning and get the explanation of the training program and games. Participants were requested to fill out an intake form concerning ability to participate in the study and to identify possible confounding factors. Finally, the participants performed online tests that lasted about 1 hour in which an additional set of tests were performed of which in this study only a selection was used. This was spread out over three days shortly before the start of training (T0) and repeated shortly after the end of the training period (T2). The order in which the tests were performed was counterbalanced. Since this study focused on one part of a larger ongoing research, a selection of these tests was used for this study. The selected tests are described later in the methods section.
During the visit to the University of Amsterdam, participants received explanation about the course and content of the training. This was provided by the use of instruction videos, an information booklet and additional oral information by the researcher. To ensure the participants performed the training tasks correctly, the participants practiced until they fully understood the game. The training period of
4
playing the training games at home lasted 12 weeks. Except one week, all weeks consisted of 5 training sessions per week of half an hour each. In the sixth week only 3 sessions were planned. This amounted up to a total of 58 sessions in total. The participant was able to choose the days and times of the sessions with a maximum of one session per day.
Training
The design of the training is based on the paper by Buitenweg et al. (2012) which described the elements needed for effective cognitive training. According to this theory two elements are required for the cognitive training to be most effective and to let its benefits reach beyond merely improved task specific performance. The first of these is the adaptive nature of the training. Cognitive training is most effective if the level of the training is appropriate for the subject. This means that it is not too easy as to become boring, nor too difficult as to become demotivating. The other aspect important for cognitive training to be effective is flexibility. The training must focus on switching between different aspects of cognition (e.g. memory and executive functioning), to be effective in transferring effects into out-of-task cognitive performance.
Both the experimental and the control training games were performed on an online platform. However, the content of the games and rules of progression through the levels of difficulty differed per condition. In the control condition only four games were available for the participant to play. These games were all focused on attention. The participants were given a predetermined schedule of which weeks they were allowed to play what level of difficulty. The maximum level of difficulty in the control condition was nine. On the other hand the experimental condition consisted of nine different games and participants could go up to level twenty. The participants were allowed to go to a higher level after reaching at least a 2 out of 3 star score on a level. This ensured that the training was adaptive to the participant’s ability. The games were designed to train executive functioning, memory and logical thinking. More different games were played for shorter time than in the control condition making the training flexible.
5
1.3 measurements
Memory and executive functioning
Letter-Number Sequencing Task
The Letter-Number sequencing task (WAIS-III-NL; Wechsler, 2000) is designed to test memory span and executive functioning. The subject is given a set of numbers and letters and is requested to put them in ascending order starting with the numbers followed by the letters. In total 20 sets were given to the subjects and the score is the amount of correct answers.
Logical thinking
SHIPLEY
The Shipley test (Shipley, 1940) consisted of 20 questions measuring logical thinking. It did so by giving a sequence of numbers, letters or words and asking the participant to fill in a gap in a logical manner (e.g. 1 2 3 4 5 _ ; answer=6). The score would be the amount of correct answers given.
Short-term memory
For measuring short term memory the Ray Auditory Verbal Learning Task (RAVLT) was used (Schmidt, 2004). The subjects were verbally provided with 15 words which they were requested to remember. After all 15 words were provided, the participant was asked to recall as many words as possible. The outcome of the task was the total amount of correct answers over 5 series of 15 words. The RAVLT score was scaled and corrected for age, level of education and gender.
Verbal fluency
The verbal fluency test focuses on executive functioning and association. The subject was asked to verbally provide as many examples of words belonging to a category within one minute. One test session test consisted of two times 1 minute for different categories. The coupled categories used were: City names and girl names, or boy names and supermarket articles. They were counterbalanced over test moments and the scores were scaled and corrected for age, level of education and gender.
Physical activity
To measure the physical activity levels of the participant during the training period, a continuous measure was used. After each training the subjects were asked how many hours they exercised on average per day since their last time of training. This multiple choice question had 4 options: no exercise, less than 15 min, between 15 and 30 minutes, and more than 30 minutes. In the question, exercise was described as an activity that raises the heartbeat. An average of all these answers was taken to provide an overall image of the level of physical activity of the participant during the training period.
6
1.4 data-analysis
To assess cognitive functioning based on data from all the tests, a composite score was created. All tests were weighed equally in the measurement of cognitive functioning. The first step was to transform all scores of the tests into standardized scores to enable comparability between different tests. The z-scores of T0 were calculated using the following formula: z-score= (score – mean) / standard deviation, or, . For measuring the improvement of the participants at T2 compared to T0, the z-scores of T2 were calculated using the distribution parameters of T0. The same formula was used to calculate the T2 z-score, but the values of the mean and standard deviation of the T0 were used. This was done to make sure the same scale was used between tests and the results would not be distorted by any
difference between the scores of T0 and T2. All tests were transformed such that higher scores indicated higher cognitive performance. Finally, taking the average of all z-scores for T0 and T2 separately created a composite score for T0 and T2 of all tests. The analyses were performed on a two by two group structure (see Table 1).
Repeated measure ANOVA’s were performed in all analyses using the composite scores as
dependent variable. All analyses are performed both with and without outliers. The exclusion of outliers never affected the results significantly and thus all results shown are including outliers. The independent variables were the healthy or CVA participant group, and control or experimental training groups. Analyses were performed to check for the influences of the different groups on the improvement on the composite cognitive score as a result of the cognitive training. In addition, a parametric correlation was tested between the measure of daily physical activity and the difference score between T0 and T2 within each subject.
Table 1: Division participants over groups
CVA Healthy Total Mock training 18 13 31
Experimental training
13 21 34
7
RESULTS
When grouping subjects from all groups, an improvement in the composite cognitive score after training in comparison to before training was found (F(1,54) = 9.07, p = .004). When we checked for an interaction effect of the healthy and CVA group on this improvement due to the training received (see Figure 1), no difference in improvement was found (F(1,54) = .557, p = .459). That is to say, overall, the healthy and CVA group both equally benefitted from the training received.
Improvement was found in the cognitive functioning scores as a result of the training when
combining the CVA and healthy groups. This suggests that in general, the training may be effective. The expectation was that the experimental training would be more effective at enhancing cognitive
functioning scores than the mock training. When comparing the two training conditions it was observed that the CVA significantly improved after training (F(1, 26) = 6.92, p = .039) as did the healthy group (F(1,32) = 4.38, p = .044). However, when we looked at the difference between the training conditions within both the healthy and CVA group, no interaction effect was observed. This holds for both the CVA group (F(1,26) = 0.29, p = 0.659), as for the healthy group (F(1,32) = 0.653, p = 0.619)(see Figure 2). This indicates that the experimental condition did not differ significantly from the control condition in improving the cognitive functioning score.
To test whether physical activity has an influence on the magnitude of improvement as a result of the cognitive training, the parametric scale of daily physical activity was used. This should provide a realistic approximation of the amount of actual physical activity of the participants because of the high frequency data acquisition. Even though it was not a significantly normal distribution (SW = 0.965, df= 63, p=0.075) there was not one category over represented indicating that the categories were well
Figure 1: Composite z-score of all CVA group subjects and healthy group subjects before (T0) and after (T2) training
Figure 2: Composite z-score of healthy group comparing Experimental and Mock training subjects before (T0) and after (T2) training
8 defined. The improvement score was
measured as described in the data analysis section. The CVA and healthy group were combined together to increase the power of the results. This could be done because there was no significant difference in effect of the training between these groups. The same holds for the different training groups, thus these were also grouped together in this analysis. This means no distinction was made between the CVA and healthy group, not experimental or control condition. The result was the improvement of subjects as a result of the training regardless of group. No significant correlation was found between the amount of improvement and daily physical activity r(59) = -.221, p = .108. These results even suggest a negative correlation between physical activity and improvement. All groups were checked for correlation of composite improvement scores
and physical activity, however, no significant results were found (CVA-control r(14) = -.152, p = .602; CVA-experimental r(10) = .045, p = .901; healthy-control r(12) = -.285, p = .369; healthy-experimental r(17) = -.382, p = .130). This means there was no correlation between physical activity during training and the improvement of the cognitive functioning score.
On average participants significantly improved their cognitive functioning scores after training regardless of group and type of training. However, besides the fact of having done the training, no factors seemed to influence this improvement. There was no difference between CVA and healthy, nor the experimental training and the control one. Furthermore, physical activity during training did not seem to influence the effectiveness of the training.
Discussion
In this study healthy elderly people and people who suffered from a stroke were subjected to an online training intervention program for 12 weeks. The aim of this research was to see whether specially designed cognitive training is effective in these groups and whether levels of physical activity have any influence on this effectiveness. In short, there seemed to be no difference between the specifically designed cognitive training as compared to the control mock training, however, on average both groups did improve their cognitive functioning score after training. This contradicts the research by Buitenweg et al. (2012) that states that cognitive training that is adaptive and flexible, should be more effective than mock training in improving cognitive functioning. Furthermore, the hypothesis that physical activity might enhance of effectiveness of cognitive training was not confirmed.
Figure 3: Correlation between physical activity measure of subjects and their improvement on the cognitive functioning scores
Com p o sit e imp ro ve m en t score
9
The lack in difference of effectiveness between the two training conditions could have multiple reasons. The first of which is the idea that the design of the cognitive training does not matter for the results it will yield. In other words, simply performing a training session, control or experimental, may result in improved cognitive functioning in the target groups. However, it seems unlikely that training itself has an effect, yet the design of the training is not of influence. It is suggested that there was an effect of cognitive training as compared to mock training but that the current study was underpowered to detect such an effect. A reason for this could be the large variation within the population and sample. This broad variation within the sample and population would require either a very large sample, or a very large effect of the intervention for it to result in a significant effect. Secondly, the lack of difference between the training groups with a general improvement after training might be due to the retest effect. If both trainings would be equally ineffective in improving cognitive measures used in this study, the act of doing the same test for the second time might explain the average improvement of these scores. Furthermore, it could be the case that the measures chosen to test cognitive functioning levels are not measuring the improvement on cognition that the training actually accomplished Since the scale of the measurements is all transformed to a z-score in the same manner, a slight improvement on one scale might weigh the same in the composite score as a large improvement on another. This could not be prevented since most of the measures could not be corrected due to lack of information on the tests. However, analyzing the different components of the composite score separately could identify whether this is the case. The lack of difference between the CVA and healthy group could be explained by the variation within the groups begin too big to find a significant result. Alternatively, it could be the case that that both healthy elderly and CVA patients might benefit equally from the training.
Another finding in this study is that there does not seem to be a positive mediating effect of physical activity on the effects of the training. This could be explained by the lack of effectiveness of the training or the measurements not testing for the right cognitive functioning. Another explanation would be that in fact physical activity during training does not influence the effectiveness of the training. This indicates that physical activity merely has a protective function for cognitive decline as mentioned in the
literature (Laurin et al., 2001; Sofi et al., 2011; van Gelder et al., 2004; Yaffe et al., 2001) rather than a mediating effect on cognitive improvement in the case of intervention studies.
Even though this study seems to suggest cognitive training in elderly with increased levels of physical activity do not seem to make a difference in CVA patients or healthy elderly, this study is not conclusive. To re-evaluate these findings in future research, a couple of aspects can be taken into account. First of all the sample size could be larger to find smaller differences in effect and help diminish the effect of large variation within the groups. Secondly, the full control group could be implemented that does not receive any training whatsoever. This group would be able to distinguish whether the improvement of all groups in this study is due to the training, or to retesting. And finally, the measure for cognitive functioning could be changed. This change would prevent differences between cognitive improvement reached by the subjects and the measures used in this study. Furthermore, a more balanced measure can then be used whereas in this study all tests got the same weight making the data prone to being skewed. A standardized balanced broad scale test for cognitive performance would circumvent this problem. Furthermore, I might be worthwhile to test for effect on individual measures of cognitive functioning (e.g. verbal fluency) for there might be an effect on some aspects of cognitive functioning but not on all.
10
This study did not find any beneficial effects of cognitive versus non-cognitive in healthy elderly or those who have suffered from CVA, however, this does not necessarily mean they could never benefit from this. Even a small benefit from cognitive training could be of great value for these populations because it is an easy to use intervention to aid in improving quality of life (Buitenweg et al., 2012). Therefore more research is needed to either replicate our findings or show which form of training would provide better results. Furthermore, even though in this study no effects of physical activity on the effectiveness of cognitive training were found, it is worth revisiting since it helps protect against cognitive decline and other factors in this study might be responsible for the lack of effect.
Bibliography
Bishop, S., Duncan, J., Brett, M., & Lawrence, A. D. (2004). Prefrontal cortical function and anxiety: controlling attention to threat-related stimuli. Nature Neuroscience, 7(2), 184–8.
doi:10.1038/nn1173
Buitenweg, J. I. V, Murre, J. M. J., & Ridderinkhof, K. R. (2012). Brain training in progress: a review of trainability in healthy seniors. Frontiers in Human Neuroscience, 6(June), 183.
doi:10.3389/fnhum.2012.00183
Centraal Bureau voor de Statsiek, Den Haag/Heerlen (2015). Bevolking; kerncijfers. Retreived from http://statline.cbs.nl
Damoiseaux, J. S., Beckmann, C. F., Arigita, E. J. S., Barkhof, F., Scheltens, P., Stam, C. J., … Rombouts, S. a R. B. (2008). Reduced resting-state brain activity in the “default network” in normal aging. Cerebral
Cortex (New York, N.Y. : 1991), 18(8), 1856–64. doi:10.1093/cercor/bhm207
Di Carlo, A., Launer, L. J., Breteler, M. M., Fratiglioni, L., Lobo, A., Martinez-Lage, J., … Hofman, A. (2000). Frequency of stroke in Europe: A collaborative study of population-based cohorts. ILSA Working Group and the Neurologic Diseases in the Elderly Research Group. Italian Longitudinal Study on Aging. Neurology, 54(11 Suppl 5), S28–33. Retrieved from
http://europepmc.org/abstract/med/10854359
Kramer, A. F., & Erickson, K. I. (2007). Capitalizing on cortical plasticity: influence of physical activity on cognition and brain function. Trends in Cognitive Sciences, 11(8), 342–8.
doi:10.1016/j.tics.2007.06.009
Laurin, D., Verreault, R., Lindsay, J., MacPherson, K., & Rockwood, K. (2001). Physical activity and risk of cognitive impairment and dementia in elderly persons. Archives of Neurology, 58(3), 498–504. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11255456
11
Lustig, C., Shah, P., Seidler, R., & Reuter-Lorenz, P. a. (2009). Aging, training, and the brain: a review and future directions. Neuropsychology Review, 19(4), 504–22. doi:10.1007/s11065-009-9119-9 O’Brien, J. L., O’Keefe, K. M., LaViolette, P. S., DeLuca, a N., Blacker, D., Dickerson, B. C., & Sperling, R. a.
(2010). Longitudinal fMRI in elderly reveals loss of hippocampal activation with clinical decline.
Neurology, 74(24), 1969–76. doi:10.1212/WNL.0b013e3181e3966e
Persson, J., Nyberg, L., Lind, J., Larsson, A., Nilsson, L.-G., Ingvar, M., & Buckner, R. L. (2006). Structure-function correlates of cognitive decline in aging. Cerebral Cortex (New York, N.Y. : 1991), 16(7), 907–15. doi:10.1093/cercor/bhj036
Poulin, V., Korner-Bitensky, N., Dawson, D. R., & Bherer, L. (2012). Efficacy of executive function interventions after stroke: a systematic review. Topics in Stroke Rehabilitation, 19(2), 158–71. doi:10.1310/tsr1902-158
Schmidt, M. (2004). Rey Auditory Verbal Learning Test: A Handbook (RAVLT).
Shipley, W. C. (1940). A self-administered scale for measuring intellectual impairment and deterioration. Journal of Psychology, 9, 371-377.
Sofi, F., Valecchi, D., Bacci, D., Abbate, R., Gensini, G. F., Casini, a, & Macchi, C. (2011). Physical activity and risk of cognitive decline: a meta-analysis of prospective studies. Journal of Internal Medicine,
269(1), 107–17. doi:10.1111/j.1365-2796.2010.02281.x
Stephen, B. B. R. E., & Richard, R. K. (2009). Aging and Neuroeconomics: Insights from Research on Neuromodulation of Reward-Based Decision Making. Cognitive, Affective, & Behavioral
Neuroscience, 9(4), 365–379. doi:10.3758/CABN.9.4.365
Van Gelder, B. M., Tijhuis, M. a. R., Kalmijn, S., Giampaoli, S., Nissinen, a., & Kromhout, D. (2004). Physical activity in relation to cognitive decline in elderly men: The FINE Study. Neurology, 63(12), 2316–2321. doi:10.1212/01.WNL.0000147474.29994.35
Wechsler, D. (2000). Wechsler Adult Intelligence Scale (WAIS-III) Nederlandstalige bewerking. Technische handleiding. Lisse: Swets & Zeitlinger.
Yaffe, K., Barnes, D., Nevitt, M., Lui, L. Y., & Covinsky, K. (2001). A prospective study of physical activity and cognitive decline in elderly women: women who walk. Archives of Internal Medicine, 161(14), 1703–8. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11485502
