Background
Good cognitive functioning (CF) is important for wellbeing and an independent life [
1]. Yet, aging is associated with a decline in CF [
2], and especially older adults with chronic diseases are at a higher risk of poorer CF levels than healthy older adults [
3]. Largest declines in CF are seen in the executive functions (working memory, inhibition, shifting) and processing speed [
1,
4]. These functions are necessary to learn, understand and perform complex daily actions [
2], and therefore important for wellbeing and an independent life [
1]. Next to a socially integrated network and cognitive challenging leisure activities, regular physical activity (PA) is the factor with highest potential to slow down the rate of cognitive decline and to prevent dementia [
5‐
7].
In fact, these claims are supported by reviews combining different types of research (e.g., cross-sectional studies, animal studies, intervention studies, etc.) on the protective effect of PA on cognitive decline [
8‐
10]. Both cross-sectional and longitudinal cohort studies consistently show the positive association between PA and CF, suggesting that long-term maintenance of sufficient PA may counteract age-related decline of CF [
4,
11]. Moreover, active older adults engaging in PA during the lifespan are at a lower risk for cognitive decline and impairment than inactive older adults [
8]. In particular, executive functions seem to be the cognitive functions benefitting most from PA [
12,
13]. Furthermore, improvements in processing speed and long-term memory have been demonstrated as well in older adults [
14].
However, evidence from intervention studies on the effect of PA on CF in older adults is inconsistent [
15]. In fact, meta-analytic reviews of randomized controlled trials (RCT) have reported large variations in effects sizes in cognitive outcomes associated with an increase in PA. Some meta-analyses have found moderate cognitive improvements as a result of PA intervention in older adults [
9,
12‐
14], whereas others observed none to limited improvements for delay of cognitive decline in the older adult population even when interventions were successful in increment of PA behavior [
16‐
18]. Although the Cochrane review by Young et al. [
16] did not identify any relationship between PA interventions and CF, they deemed it possible that certain subgroups of older adults, such as those with lower starting levels of fitness, could profit more from PA interventions. This is supported by a meta-analysis and systematic review by Cai et al. [
19] on effects of exercise on CF in chronic disease patients. They found a positive overall effect of exercise interventions on CF. However, 22 out of 35 included studies only involved patients with Mild Cognitive Impairment or Alzheimer’s disease. In addition, the remainder of the included studies in this meta-analysis focused their intervention on only one chronic disease (i.e. cancer, heart failure) while most older adults suffer from multiple chronic diseases [
20].
Furthermore, older adults with chronic diseases have the lowest levels of PA [
21,
22], mostly caused by experienced fatigue and pain [
21,
23]. Although a few interventions exist to improve PA behavior in this specific population, they are often site-situated, which is high demanding, more expensive, and mainly focus on exercise [
24,
25]. Computer-tailoring interventions are a cost-effective solution to improve PA behavior in older adults [
26], and can thus also be so for older adults with chronic diseases.
In this light, the computer-tailored PA stimulating intervention Active Plus was developed and evaluated for people aged over 50 years in 2010 [
27]. Active Plus participants receive three personalized PA advice letters (online or print delivered) in four months. Earlier research in the general population of older adults of 50 years or over demonstrated that the Active Plus group self-reported to be 1.5 h per week more active at moderate-to-vigorous intensity after one year compared to the control group [
28], even in older adults with impaired mobility [
29]. At a later time, the computer-tailored program was fitted to a more elderly (≥65 years) population of single adults who suffered from chronic diseases [
30]. This adapted version of Active Plus was effective in increasing PA behavior three months after baseline, but no effects were found after six months [
31]. However, this concerned an implementation study without a control group, making it impossible to draw definite conclusions on the PA effects of the adapted Active Plus in older adults with chronic diseases.
In conclusion, to our knowledge, there is a lack of cost-effective and easily accessible PA interventions for an elderly population which suffers from a broad range of one or more chronic diseases, and the effects of PA on CF in this population have not yet been tested. Based on these previous studies, we assumed that the computer-tailored intervention Active Plus might improve PA behavior in older adults with chronic diseases and, as a result, could lead to beneficial effects on CF. Our recent RCT showed that Active Plus was only to a limited extend able to improve self-reported PA behavior in chronically diseased older adults six and 12 months after baseline measurements [
32]. We did not find any significant intervention effects in objectively measured PA. In addition, subgroup analyses showed that more vulnerable participants (e.g., with a higher degree of impairment, age, or body mass index) benefitted more from the intervention on especially the lower intensity PA behaviors.
In this paper, our main research goal was to test the cognitive effects of Active Plus in older adults with chronic diseases. Though the intervention was individually tailored, it might have been that not all subgroups of participants responded similarly to the Active Plus intervention, as we found in the paper on PA effects [
32]. Therefore, we explored whether the cognitive effects differed for subgroups based on degree of impairment, adhering to the PA guidelines (≥ 150 min of moderate-to-vigorous PA), age, gender, body mass index, educational level, and marital status [
33].
Discussion
To the best of our knowledge, this is the first study to assess the cognitive effects of a computer-tailored PA intervention in older adults suffering from a broad range of chronic diseases. Additionally, this study explored the effectiveness of the intervention on cognitive functions in relevant subgroups based on demographics, PA behavior, and level of impairment. Although both the intervention group and the control group significantly improved most of their CF test scores over time, there were no effects of the Active Plus intervention on the assessed domains of CF (verbal memory, shifting, inhibition, and processing speed), nor did we find any cognitive effects of the intervention in subgroups of older adults with chronic diseases.
The finding that both groups improved significantly on all verbal memory outcomes after six months, and on all CF outcomes except inhibition after 12 months, can be explained in multiple ways that might have occurred next to each other. First, the increased CF scores for both groups may have been caused by a learning effect. In CF literature, various reasons have been discussed to explain the improved scores originated by practice, such as reduced anxiety or increased familiarity with the testing environment and procedural learning [
63]. Because almost all participants had no earlier experience with partaking in a study and because they were not confident what the CF tests were all about, it is plausible that this could have elicited stress. Second, the house visits necessary to conduct the CF tests can have caused an improved CF in both groups, because of the personal attention participants received from the researcher or student, as previous research has suggested the existence of the Hawthorne effect [
64]. Third, the assessment of the PA outcome measures themselves could have led to better CF. As both the intervention group and the control group received the same assessments, it is plausible that wearing the accelerometer and filling in the questionnaire (which focused on PA behavior, motivation for PA and intention to PA) led to higher awareness about their own current PA behavior and as a result led to an increase in PA behavior. However, in this RCT, the control group did not improve on either of the eight PA outcomes assessed [
32]. Therefore, it is less likely that the PA measurements themselves have caused an improved CF.
However, the most likely explanation for the contradictory findings between earlier studies on the relationship between PA and CF, and this RCT, is the minimal effectiveness found of Active Plus on PA behavior in older adults with chronic diseases. The Active Plus intervention aims to stimulate PA behavior in older adults with chronic diseases and has already been proven to increase PA behavior effectively in the general older adult population [
28]. However, in this RCT, the intervention was only able to improve self-reported PA to a limited extent and did not improve objectively measured PA in older adults with chronic diseases [
32]. The small significant improvements (walking, cycling, gardening) were found on the PA behaviors performed at a lower intensity than necessary, for example, for sports. Even though there still is a lot unknown about the PA characteristics that lead to optimal results for CF [
13], it is possible that the intensity of PA may be important [
16], and some research suggests that a moderate-to-vigorous intensity is needed [
9,
65]. Therefore, the limited effects we found on self-reported PA may have been too weak to improve CF. The mechanisms that may have led to the poor effects on PA behavior might be due to the target population, the relatively high baseline amounts of MPVA (less room for improvement), the nature of the intervention Active Plus and the design of the RCT itself. However, these explanations are beyond the scope of this paper and have been discussed thoroughly in our paper on the effects of the Active Plus intervention on PA [
32].
There are some examples of PA interventions that did have a positive effect on CF tested with an RCT [
66‐
69]. The study by Albinet et al. [
66] found an intensive 12 weeks aerobic exercise program effective in increasing executive functioning in older adults as opposed to a stretching program. However, the study had a small sample size (
N = 24) and used only one CF test. The study by Muscari et al. [
67] included 120 healthy older adults. The intervention, which consisted of supervised endurance exercise training three times a week for 12 months, was effective in reducing progression of cognitive decline. This study only assessed the Mini-Mental State Examination as a measure of cognitive function, which is not a reliable test for change scores in this population [
70]. Best et al. [
68] examined the effects of resistance training once or twice weekly as opposed to twice-weekly balance and tone training. Resistance training was beneficial for executive function and the memory domain. However, this study only included women, which affects the generalizability and had a relative low compliance to the program. All of the above interventions were site-situated and therefore quite expensive to execute and demanding for the participants. Liu-Ambrose et al. [
69] tested the effects of a home-based resistance training, balance, and aerobic program on executive functioning. The program significantly improved inhibition in the intervention group versus the control group. However, this study included only participants who had fall incidents and found significant effects on only one of the three executive functioning outcomes (inhibition, updating, set-shifting). Furthermore, this study still relied on the deployment of trained personnel. In conclusion, the studies mentioned above have important limitations, were expensive to execute and demanding for participants. Higher quality studies are needed to clarify the association between PA interventions and cognitive function and to determine which types of PA will have the greatest benefit on specific cognitive domains.
While only the computer-tailored PA stimulating Active Plus intervention is not sufficient to increase PA in the general older adults with chronic diseases population and possibly thereby have an intervention effect on CF, a possible solution to enhance the effect could be a blended approach in which this computer-tailored intervention and face-to-face contact are combined [
71]. Especially in a more elderly population that maybe is less internet and more personal contact-oriented. A blended approach could be a cost-effective solution, as it implies less costly face-to-face contact and improved feeling of self-regulation. For example, the Active Plus intervention, which contains solely personalized advice on how to implement PA in daily life, could be combined with face-to-face contact with a physiotherapist or weekly meetings with a PA group for older adults. Especially as older adults are known to prefer to exercise in groups as opposed to exercise alone [
72]. A blended approach is increasingly being applied in both healthcare and mental healthcare [
71]. There are already some examples of blended approach interventions aimed at promoting PA in older adults [
73,
74]. However, only a few studies exist, and results are mixed. To our knowledge, there are no blended interventions to improve CF or prevent a further cognitive decline through improving PA.
Furthermore, to improve effectiveness of Active Plus, the intervention could be enriched with cognitive training to increase intervention effects on CF. There are already some indications that combining both physical training and cognitive training could lead to better outcomes on CF, since cognitive training appears to improve other CF domains than PA in older adults [
75,
76], and this results in larger effects on CF in older adults than solely physical or cognitive training [
77,
78]. However, not all studies show that cognitive training programs or intellectually demanding activities enhance general CF. At best, they find that such interventions boost one’s performance in tasks similar to the trained task, and do not extend to other cognitive domains. This is an important issue for future research [
79].
Some strengths of the study should be mentioned. First, the current study has a strong research design, as RCT’s are considered the golden standard in effectiveness studies. Second, our study participants were quite mixed and generalizable to a general older adults with chronic diseases population or even the general older adult population. As an example, our research population had almost equal numbers of male and female participants, and the majority of the participants was low educated (e.g., 51%). In addition, our research sample had BMI levels [
80] and a mean number of comorbidities (3.5) [
81] equal to the general older adult population in the Netherlands. However, selective response is probable. Despite these strengths, the study also had some limitations. Although the CF tests are well-validated sensitive tests, sensitivity still might be too low. However, this is an issue in every study that measures CF, and computer-based tasks (e.g. this RCT assessed CF on an iPad) are actually suggested to improve sensitivity of cognitive assessments [
82]. Furthermore, the selective dropout (i.e., older participants, during the intervention period, and those with lower baseline levels of moderate-to-vigorous PA) may have influenced our results, although this is expected to be less harmful because of the reasonably low dropout. In (partly) digital health interventions a dropout rate of 25.1% per cent is thought of as low [
83]. Furthermore, we accounted for these possible confounding variables.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.