Cognitive impairment is an important public health concern, since the number of elderly people with age-related health problems, such as cognitive decline and dementia, is growing considerably. Research on potentially effective interventions for preventing the progression of cognitive impairment is therefore warranted, and physical exercise and vitamin B supplementation are promising and relatively inexpensive interventions.

Regular exercise is associated with better cognitive function and delayed onset of Alzheimer’s disease,1^–4 and has been shown to improve cognitive function in cognitively healthy and demented older adults.5 6 The specific mechanisms of exercise beneficially affecting cognitive performance in subjects in different stages of cognitive decline are not yet known. A recent study of 59 sedentary adults aged 60–79 years has shown that six months of aerobic training increased brain volume in regions of the brain associated with age-related decline in cognition, compared with non-aerobic training.7 Improvement of vascularisation and increased blood flow to the brain are other possible pathways, which may be promoted by improving aerobic fitness.8 9

Vitamin B supplementation may also affect cognitive function through its effect on homocysteine,10 which is associated with impaired cognition.11^–16 There have however been few methodologically sound trials on the effect of vitamin B supplementation on cognition.17^–19 Vitamin B supplementation may affect cognitive performance via two mechanisms.20 Firstly, vitamin B is needed to transform homocysteine into methionine, which is an essential amino acid for the central nervous system. Lack of methionine results in disorders in neurological and psychological status. Secondly, vitamin B may beneficially influence homocysteine concentrations, resulting in structural vascular changes in the brain.

This paper describes the effects of two interventions on cognitive performance in a sample of community-dwelling older adults with mild cognitive impairment (MCI). MCI is a potential transitional stage between normal cognitive function and Alzheimer’s disease, in which people experience memory loss to a greater extent than is expected for age and education, but do not meet criteria for Alzheimer’s disease.21 The interventions were: (1) one year of a moderate intensity walking programme designed to improve aerobic fitness; and (2) one year of daily supplementation with folic acid, vitamins B-12 and B-6. It was hypothesised that both interventions would beneficially influence cognitive performance.

METHODS

Study design

The study was a double-blind randomised controlled trial (RCT). A two by two factorial design was used to compare the effects of: (1) a walking programme with a placebo activity programme; and (2) vitamin B supplementation with placebo supplementation in a single sample. Thus, half of the subjects in the walking programme got vitamin B supplements and the other half got placebo supplements. This was also the case within the placebo activity programme. This design allows two hypotheses to be tested at the same time, by randomising subjects to two interventions and adjusting the results from each intervention for the effects of the other.22 23 The medical ethics committee approved the study protocol.24 All participants gave written informed consent.

Participants

All community-dwelling inhabitants aged 70–80 years in a Dutch town (n = 5491) were sent an invitation letter in September 2003. Among those willing to participate, inclusion criteria were checked using a two-step screening consisting of a postal questionnaire and a telephone interview.25 The operational criteria for MCI and additional inclusion criteria for the trial are described in table 1.

Randomisation and blinding

After the baseline interview, participants were randomised using the option “random sample of cases” in SPSS. Randomisation was stratified for physical activity level assessed by the LASA physical activity questionnaire.32 Participants were randomised to: (1) walking programme or placebo activity programme; then within each activity programme subjects were randomised to (2) vitamin B supplementation or placebo supplementation. Participants and exercise instructors were blinded to group allocation by being left unaware of which exercise programme was supposed to be effective. The pills were coded as A or B by the manufacturer. The key was decoded after data analysis. All cognitive outcome measures were assessed by trained examiners, who were also blinded to group allocation.

Interventions

Walking programme (WP) and placebo activity programme (PAP)

Both the WP and the PAP were group-based and lasted one year (June 2004–June 2005). Trained instructors supervised exercise sessions twice weekly for one hour. Adherence was defined as the percentage of attended sessions.

The WP was based on an existing aerobic walking programme designed to improve aerobic fitness,33 and intensity was moderate (>three metabolic equivalents). Each session consisted of a warm-up, moderate-intensity walking exercises and cool-down. The instructors monitored exercise intensity during all sessions subjectively—for example, participants were still able to talk, but also showed signs of moderate intensity activity, such as increased breathing frequency. Walking pace and distance increased gradually during the programme. Sessions took place outdoors in municipal parks.

The PAP, developed by experienced exercise instructors, consisted of an introduction, low-intensity (<three metabolic equivalents) non-aerobic group exercises and a closing. Sessions included relaxation, activities of daily living, balance, flexibility and postural exercises. The PAP was carried out in community centres.

Vitamin B supplementation (FA/B12/B6) and placebo supplementation

Participants randomised to FA/B12/B6 supplementation received one pill daily, containing 5 mg vitamin B-11 (folic acid), 0.4 mg vitamin B-12 (cyanocobalamin) and 50 mg vitamin B-6 (pyridoxine hydrochloride) for a year. These vitamin supplements are available on prescription in The Netherlands. For the purpose of this study, the package and vitamin pills could not be identified as an existing supplement. Participants in the control group received an identical-looking placebo pill. Pills were packed in one-week blister packs labelled for each day of the week. Compliance was verified by counting pills in returned blister packs.

Measures

Measurement of baseline variables is described in detail elsewhere.24 Cognitive measures were collected at baseline and after six and 12 months during a standardised interview. The mini mental state examination (MMSE) was used to measure general cognitive function for descriptive purposes. The maximum score is 30, and a score below 24 was considered abnormal.29 Four other tests were administered to assess different aspects of cognition:

• Memory by the auditory verbal learning test (AVLT) (immediate and delayed recall; maximum scores 75 and 15 words, respectively).34

• Executive function by the verbal fluency test (VFT). Participants were asked to name words starting with a particular letter during a one-minute period. In each administration of the test, three letters were given. The score was the total number of named words.35

• Information processing speed by the digit symbol substitution test (DSST). The participant was asked to draw symbols corresponding to nine digits below numbered boxes during a 90-second period. The score was the number of correctly drawn symbols.36

• Attention by the abridged stroop colour word test (SCWT-A). The SCWT-A consists of three tasks; (1) reading eight rows of five written colours; (2) naming the colours of eight rows of five red, green, blue or yellow coloured rectangles; (3) naming the colour of ink for the words red, green, blue or yellow. The score was the time needed to complete each task.37

Statistical analysis

Sample size

Since participants were selected on the basis of memory complaints and impairments, memory was considered the most important cognitive outcome in our study. Therefore, sample size calculations were based on AVLT data.21 The average difference between subjects with MCI and mild Alzheimer’s disease and subjects with MCI and Alzheimer’s disease for immediate recall was five words. On the basis of this difference, it was estimated that 136 participants would be required (α = 0.05 and β = 0.80). To allow for a dropout of 25%, the final sample size was 179 participants.

Data analysis

Differences in baseline characteristics between intervention and control groups (for example, WP versus PAP and FA/B12/B6 versus placebo) were tested using independent t tests, Mann Whitney U tests and χ² tests. For both interventions, data were analysed on a modified intention-to-treat basis, including participants with at least one post-baseline assessment. Longitudinal regression analysis was used to assess changes in each outcome measure. The two follow-up measurements, defined as dependent variable and multi-level analysis with two levels, were (1) time of follow-up measurement (values corresponding with performance after six and 12 months’ intervention); (2) individual. Reported regression coefficients indicate differences between WP and PAP, or FA/B12/B6 supplementation and placebo supplementation. Data were analysed using crude and adjusted models. In the crude models, the independent variables were exercise intervention, vitamin intervention and baseline performance. In the adjusted models, education, baseline activity level, baseline vitamin status, adherence to the exercise programme and compliance with the supplementation were added as covariates. Effect modification by gender was checked. Because gender was an effect modifier in four of seven outcome measures, analyses for all outcomes were stratified by gender.

Since adherence with the WP and the PAP was suboptimal, a post hoc analysis was performed to check for effect modification by session attendance. Finally, a per protocol analysis, including only participants who attended at least 75% of the exercise sessions was performed. This cut-off point is in accordance with previous exercise intervention studies in elderly people.38^–40 Data were analysed using SPSS (release 12.0.2). A significance level of 5% was used for between-group comparisons and 10% for interaction terms.

RESULTS

Participant flow and characteristics

After baseline measurement, 179 participants were randomised; 27 of them were excluded from the analysis, because they only provided baseline data (fig 1). Compared with the remaining 152 participants, a higher percentage of these 27 participants were married (71% versus 52%, p = 0.05), and fewer were current smokers (0% versus 14%, p = 0.04). No significant differences existed between measures of cognition. Baseline variables for men and women are shown in table 2. Participants in the WP did not differ significantly from participants in the PAP. In the FA/B12/B6 supplementation group, a higher percentage of men had lower education and a higher percentage of women had intermediate/high education compared with the placebo-supplementation group.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Flow chart. FA/B12/B6, folic acid, vitamins B-12 and B-6 ; MCI, mild cognitive impairment; PAP, placebo activity programme; SO, significant other; T12, follow-up after 12 months; T6, follow-up after 6 months; TI, telephone interview; WP, walking programme. *Excluded from analysis (only baseline data available); †included in analysis.

Adherence and compliance

Thirty participants, 19 men and 11 women, did not attend a single session, but were included in the intention-to-treat analysis. Median adherence (25th–75th percentile) to both exercise programmes, including these participants, was 63% (2%–81%) and did not differ between groups. At baseline, adherent men attending at least 75% of the sessions (n = 33) were more often living with a partner (82% versus 65%, p = 0.03) and reported to be less physically active than non-adherent men (n = 52), (median 64 (25th–75th percentile 32–82) versus 87 (34–139) minutes/day, p = 0.04). Only 18 women attended at least 75% of the sessions. The most frequent reasons for dropout from the exercise programmes after the start were health-related problems. No adverse events of the WP or PAP were reported.

Median compliance (25th–75th percentile) with the (vitamin) supplementation was 100% (99%–100%). Two participants stopped taking vitamin FA/B12/B6 pills after reporting adverse side effects—that is, sleep problems and increased forgetfulness; one participant discontinued the placebo pills after reporting not feeling well.

Intention-to-treat analysis

In the intention-to-treat analysis (table 3), no significant main effect of walking or FA/B12/B6 supplementation was found in men. In women, a beneficial main intervention effect of FA/B12/B6 supplementation was found on the DSST (β 2.9 (95% CI: 0.6 to 5.3) symbols, p = 0.02). Appendix tables A1 and A2 present cognitive performance at baseline and after six and 12 months’ intervention.

Per protocol analysis

In women, an interaction was observed between the exercise programmes and adherence. With each percentage increase in session attendance, in the WP, the SCWT-A combination task performance improved by 0.3 seconds (p = 0.04), and the AVLT delayed recall score improved by 0.04 words (p = 0.06). No interaction was observed in men. However, in men attending at least 75% of the sessions (n = 33), the WP beneficially affected AVLT delayed recall (β 1.5 (95% CI: 0.1 to 3.0) words, p = 0.04). The number of women attending at least 75% of the sessions (n = 18) was insufficient to perform a reliable per protocol analysis.

DISCUSSION

Neither the walking programme nor FA/B12/B6 supplementation improved cognition in these community-dwelling older adults with MCI within one year. The walking programme, however, was efficacious in improving memory in men and memory and attention in women, in those with better attendance.

To our knowledge, this is the first intervention study in people with MCI from the general population. While this is a novel approach,38 39 any benefits of the WP and FA/B6/B12 supplementation may have been overshadowed because of the fluctuating course of MCI in community-based samples.40 Longer interventions and follow-up may have provided better insight into the potential preventive effects of these interventions on cognitive decline. Unfortunately, this was not possible in the present study.

The exact mechanisms by which aerobic exercise and vitamin supplementation may affect cognition are not yet known and some overlap between the hypothesised mechanisms may exist in terms of “vascular changes”. In theory, the factorial design of this trial allowed us to examine possible interaction effects between walking and vitamin supplementation. Unfortunately, because of logistical and financial constraints, it was not possible to recruit a larger sample size in order to have sufficient power for the examination of potential interaction. Therefore, the effect of both interventions was examined independently and results for one intervention were adjusted for the possible influence of the other.

The lack of effect may have also been caused by insufficient power owing to the unanticipated post hoc stratification for gender. Because gender was an effect modifier, reporting the results for both genders together would make no sense. This resulted in a decrease of statistical power to detect differences between FA/B12/B6 supplementation and placebo pills, and the walking programme and the placebo activity programme.

Vitamin supplementation

A benefit of FA/B12/B6 supplementation was observed only on attention in women. This is in line with the conclusions of recent meta-analyses, that published trials do not yet provide adequate evidence for an effect of supplementation with folic acid, vitamins B-12 and B-6 , alone or in combination, on tests for cognitive function in people with normal or impaired cognitive function.17–19 41 A possible explanation for the lack of effect in our study may be the absence of screening for baseline vitamin deficiencies or hyperhomocysteinaemia. However, there was no main intervention effect on cognition in post hoc analyses of participants with baseline folate deficiency and/or hyperhomocysteinaemia (n = 85, data not shown). A higher dose of vitamin B supplementation also appears to be unwarranted, since the FA/B12/B6 supplementation significantly improved homocysteine and vitamin B concentrations within six months (The effects of exercise and folic acid, vitamins B-12 and B-6 on homocysteine concentrations in adults with mild cognitive impairment, paper submitted). These changes did not result in improved cognitive function. A longer duration of vitamin B supplementation may, however, be necessary for the consolidation of neurological and vascular changes in order to promote cognitive function. In contrast to short-term interventions, Durga et al recently observed a beneficial effect on cognitive performance after three years of supplementation with 800 μg of folic acid in subjects with raised homocysteine and normal vitamin B-12 concentrations at baseline.42

Walking programme

The lack of a main effect of exercise may have been caused by the moderate adherence to the exercise programmes. It is important to mention that the correlation between general cognition and adherence was only small and non-significant. Moreover, MMSE scores at baseline and after six and 12 months did not differ significantly between adherent (⩾75% of the sessions) and non-adherent (<75% of the sessions) participants (data not shown). While many previous studies have included data only from participants who completed the programme, a strength of the present study is that data were analysed according to the intention-to-treat principle, including all randomised participants with available data, irrespective of exercise adherence. As a result, data from many participants (30 out of 152) who did not attend a single exercise session were included in the intention-to-treat analysis. The inclusion of data from non-adherent participants may have underestimated the actual intervention effect in our study. The importance of adherence is underlined by the fact that the effect of the walking programme on some of the cognitive outcomes increased with increasing adherence.

In our study, baseline physical activity levels were relatively high. Around 80% of the women and 90% of the men reported being at least moderately active for 30 minutes or more per day. In comparison, in the general Dutch population aged 65–75 years, only 60% of women and 70% of men make this claim.43 Despite the high self-reported baseline activity levels, the WP was efficacious in improving aerobic fitness compared with the PAP (Feasibility and effectiveness of two group-based exercise programmes for community-dwelling older adults with mild cognitive impairment, paper submitted). Since larger effects of exercise on cognition are expected in sedentary subjects,5 it is possible that a threshold physical activity level exists, above which additional moderate-intensity physical activity does not result in further cognitive benefit.

It is also possible that the contrast between the programmes was not large enough to induce between-group differences. However, according to heart rate recordings, there were clear differences between the moderate intensity of the WP and the low intensity of the PAP (Feasibility and effectiveness of two group-based exercise programmes for community-dwelling older adults with mild cognitive impairment, paper submitted). Indeed, the use of a low-intensity placebo-activity programme with non-aerobic exercise in the present study could be seen as a strength, as it offered the opportunity to blind both participants and instructors to the exercise intervention, and to exclude the Hawthorne effect of attention.

Although the rigorous design of our study (including the intention-to-treat analysis, blinding of participants, exercise instructors and outcome assessors, and a range of valid measurements to assess cognition) reduces the likelihood of bias, it may also explain the absence of beneficial main effects. Etnier et al44 performed a thorough meta-analysis on the effect of exercise on cognition, and found that the effect size decreased as experimental rigour increased. It must be noted that this meta-analysis was published in 1997 and did not include more recent methodologically sound positive trials.

Unfortunately, our results do not correspond with two recent meta-analyses on the effects of exercise on cognitive function in cognitively healthy and demented subjects, which found significant positive effects.5 6 The conclusions of these meta-analyses may have shifted to a more positive view as a result of publication bias.45 Publication bias may arise from the tendency of journals to reject studies without a positive outcome, and from the fact that authors are less likely to submit statistically non-significant results for publication.46 Dissemination of both “positive” and “negative” findings is of great importance for science development as it provides better insight into which interventions are effective and which are not.

In conclusion, our results do not support the role of vitamin B supplementation in improving cognitive function in people with MCI. However, our findings suggest that regular participation in moderate intensity walking may improve aspects of cognition, especially memory, in community-dwelling people with MCI. Regular participation in moderate intensity physical activity should therefore be encouraged in this population group. Future larger trials are needed to confirm these findings. Another direction for future research is to examine ways to increase exercise frequency among this population.