. Effect of a 24-Month Physical Activity Intervention vs Health Education on Cognitive Outcomes in Sedentary Older Adults: The LIFE Randomized Trial. JAMA. 2015 Aug 25;314(8):781-90. PubMed.


Please login to recommend the paper.


  1. It is unfortunate that the trial has the same design that nearly every U.S. trial of dietary supplements to date has had—participants were not selected who had low levels of the nutrients in question. They may all have had very healthy diets that provided sufficient levels of the nutrients of interest. There is also limited presentation of analyses in the publication that considered dietary intake levels so that the reader can judge how likely the trial was to even be able to test whether the supplements are effective in persons with low intake. The epidemiological studies that have found protective associations did not find protective benefits at very high levels of intake obtained through supplements. Rather, the contrast was moderate intake levels from food being protective of cognitive decline versus people with low dietary intake levels. This trial does not test this. The trial findings cannot be generalized to those in the population who have marginal to low nutrient intake of the nutrients in question. For example, based on a number of epidemiological studies, the protective level of benefit of fish is one serving per week and that for green leafy vegetables (rich in lutein) is about one serving per day. If most people in this trial consumed these servings then one would of course not expect to see a positive result.

    View all comments by Martha Clare Morris
  2. I see several issues related to the AREDS2 trial that weaken the generalizability of this type of study's negative conclusions.

    The authors point to cardiovascular disease as one likely cause of cognitive decline in the overall population. For example, they note more cognitive decline in subjects with vascular or heart disease (CVD). It looks like roughly 40 to 45 percent of the study population is on statins. It is not much of a leap to suggest that statin use is likely higher in the CVD population. Protection against CVD is one of the best-established health benefits of fish oil—strong enough to win American Heart Association recommendations to supplement. However, the available data show that omega-3 protected from CVD in earlier studies prior to widespread statin use, but more recent trials have been negative because the majority population now uses statins. That is, omega-3 shows no added benefit over statins in controlling CVD—with the important exception of sudden cardiac death! So, one factor in the negative outcome is the distinct possibility that statin use was sufficiently protective against CVD-related risk of cognitive decline that 1 gram of fish oil had no added benefit. That would be similar to several recent CVD fish oil trials.

    The other big factor likely to reduce cognitive scores in subjects with an average age of about 72 years would be ApoE4-related AD risk, which accounts for nearly 80 percent of the AD in a population under 80 years. Given the epidemiology and ADCS DHA clinical trial data suggesting that DHA is ineffective in ApoE4 carriers (but possibly effective in non-carriers), one might expect a lack of efficacy against early AD-related decline in a population of average age 72. It would be better to look at risk of decline in subjects over 80 where ApoE3 predominates.

    The third issue with this trial is that it treats a general population in which one-third of the subjects in both groups have relatively high dietary omega-3 intake. As Martha Morris has pointed out, there is not much point in testing dietary supplements on those in the upper percentiles of intake that appear to be protective. This study didn't seem to measure the blood levels but one can assume that many randomized subjects had adequate baseline dietary intake.

    Sink et al. show a possible effect of physical activity on executive function, which the authors say is most sensitive to exercise. The group they chose to study is impaired and the level of exercise is necessarily less aerobic than what one might require to achieve a fitness level in many participants. However, the biggest issue is that the ApoE4 carriers are less than 80 percent of the sample, because the authors have not enriched for those at risk for AD. Preclinical data, including unpublished data from our lab, suggest that ApoE4 risk is mitigated by exercise. If you look at the Wisconsin Registry for Alzheimer Prevention data presented by Stepanie Schulz at AAIC 2015, you see that exercise benefited AD biomarkers, such as PiB, in ApoE4 carriers (see Aug 2015 conference news). You can see the biomarker benefits in the whole group, but it is the ApoE4 group that drives it (see Okonkwo et al., 2014). The Lifestyle study has older participants and is 80 percent non-ApoE4 and doesn't see an exercise benefit.    

    I would say flip the study groups to have a better chance to get positive trials. Select for younger people who carry on ApoE4 risk allele in the exercise trials, and select for older non-E4 for testing the omega-3/antioxidant supplements.


    . Physical activity attenuates age-related biomarker alterations in preclinical AD. Neurology. 2014 Nov 4;83(19):1753-60. Epub 2014 Oct 8 PubMed.

    View all comments by Gregory Cole
  3. I am familiar with this trial, as it was coordinated through my institution. I moved to Wake Forest from the University of Washington as the trial was in close-out phase and have never participated in trial conduct, data analysis, interpretation, or manuscript prep (i.e., I have no conflicts of interest to report). There are differences between this study and our PACE study that was recently featured in Alzforum, which likely account for disparate findings:

    Dose of exercise is by far the most likely reason for differences: On average, a “dose” of aerobic exercise was much less than that attained by other studies of structured exercise. The authors acknowledge that the intensity of prescribed exercise may have been insufficient. They used accelerometry data to identify “moderate” levels of physical activity, using >760 counts of activity per minute as a threshold. Others (e.g., Sasaki et al., 2011) have used much higher values to identify the lower limit of what might be considered "moderate" physical activity (e.g., >2,689 counts per minute). Furthermore, the LIFE trial emphasized “physical activity,” not necessarily activity that was aerobic. Our work specifically examines the cognitive and AD-modifying effects of aerobic exercise. It may be that although all types of exercise are beneficial for the body and mind in one way or another, high-intensity aerobic exercise may be essential to improve cognition and slow progression of AD pathology.

    LIFE participants were physically frail, by definition, and so were very different than people we enrolled in our exercise studies. Physical frailty likely impacted the adults’ ability to exercise at higher intensity, which brings us back to dose. On the other hand, Sink et al. found that the adults who scored more poorly at enrollment of tests of physical function (more frail) improved more on tests of executive function (Fig. 2c). It is possible that for these folks who have more difficulties mobilizing, any exercise was challenging and therefore their “dose” may have been a bit higher, relatively, than for those who are able to move about with more ease. Although this statement appears to contradict what I said earlier, the overall statement remains true: Reaching exercise doses that confer cognitive benefits would likely be more difficult for these individuals in the LIFE cohort.

    The LIFE study tells us that low-dose exercise can be beneficial for physical function and help reduce physical frailty, which is very important since so many older adults are physically compromised. The LIFE study also tells us, though, that in order to see cognitive benefits, the “dial” controlling intensity of aerobic exercise has to be turned up.

    Finally, as the authors pointed out, the LIFE cohort was predominantly cognitively normal (12-13 percent MCI/dementia after 24 months), whereas our previous studies included only adults at high risk for cognitive decline.


    . Validation and comparison of ActiGraph activity monitors. J Sci Med Sport. 2011 Sep;14(5):411-6. Epub 2011 May 25 PubMed.

    View all comments by Laura Baker
  4. Dr. Baker points out several good reasons why the results from the LIFE study differ from the PACE study that were recently featured on Alzforum. There are two other reasons why the LIFE study is unique from most other studies of physical activity on cognition, including PACE. The first is that our participants were much older than those typically enrolled in physical activity/exercise interventions. Our mean age was 80, compared to 65 in PACE. This 15-year difference in age is important and could help explain why our participants may have still achieved “moderate” levels of physical activity in spite of our seemingly low accelerometer definition of moderate activity of >760 counts/min. Dr. Baker notes that moderate activity has been defined as accelerometer counts that are greater than 2,690/min. But the cited study (Sasaki et al., 2011) based this definition on young (mean age 27 years), healthy participants and thus it may not be an appropriate definition of moderate activity for older adults. Indeed, a recent article by Zisko et al. highlights the importance of taking age and cardiorespiratory fitness into account when defining activity levels, especially for the elderly. Due to declining VO2 max, many elderly cannot reach the moderate/vigorous activity levels set for younger adults. Using objectively measured VO2 max, they determined that moderate intensity activity for a population of 70- to 77-year-olds corresponded to 669-4048 counts/min depending on gender and fitness level. Our definition of >760 counts/min for moderate activity is consistent with Zisko‘s results given the age and frailty of our population.

    The second important factor to remember when interpreting the results of the LIFE study in the context of other physical activity interventions is that, unlike many studies, we did not use a “stretching control.”  Because cognitive outcomes were not the primary outcome of LIFE (mobility disability was), the control group consisted of cognitively and socially engaging health-education seminars.  Both cognitively and socially engaging activities have been shown to be beneficial to cognition.  Thus, in LIFE, we may have diminished our ability to show that physical activity is superior to a control condition in cognitive outcomes because our control condition likely also benefited cognition.


    . New relative intensity ambulatory accelerometer thresholds for elderly men and women: the Generation 100 study. BMC Geriatr. 2015 Aug 4;15:97. PubMed.

    . Validation and comparison of ActiGraph activity monitors. J Sci Med Sport. 2011 Sep;14(5):411-6. Epub 2011 May 25 PubMed.

    View all comments by Kaycee Sink
  5. I strongly endorse Martha Clare Morris' comments. It is about time that trials of nutrients do not get funded (or published) unless the baseline level of the nutrient concerned (or, second-best, its intake) is measured. In addition, of course, the final level of the nutrient should be reported. In this way, we will know if the population under study included a significant proportion with poor nutrient status and if the treatment corrected this. The treatment outcomes can then be assessed by stratification of the population for baseline nutrient level. In the Chew trial, as the authors stated "the study was conducted in a select population of well-nourished and highly educated persons." The omega-3 fatty acid levels were not measured in this trial, so we have no idea if any of them had insufficient levels to protect their brain.

    A second problem is that in this healthy elderly group, those taking placebo showed an extremely small degree of cognitive decline in the period studied. How can it be expected that any treatment would slow down what is already a clinically insignificant degree of cognitive decline?

    A third problem with the study is that it was done on people with macular degeneration, which is well known to be associated with elevated levels of plasma homocysteine (Huang et al., 2015). The authors would not have been aware of a recent study that showed that omega-3 fatty acids do not benefit the brain in people with high homocysteine (Jerneren et al., 2015). In the latter report it was shown that a high omega-3 fatty acid status only slows brain atrophy in people with a good B vitamin status, i.e., with low homocysteine.

    I conclude that the Chew trial cannot be used to claim that omega-3 fatty acids are of no benefit for slowing cognitive decline. 


    . Homocysteine and the risk of age-related macular degeneration: a systematic review and meta-analysis. Sci Rep. 2015 Jul 21;5:10585. PubMed.

    . Brain atrophy in cognitively impaired elderly: the importance of long-chain ω-3 fatty acids and B vitamin status in a randomized controlled trial. Am J Clin Nutr. 2015 Jul;102(1):215-21. Epub 2015 Apr 15 PubMed.

    View all comments by A. David Smith

Make a Comment

To make a comment you must login or register.

This paper appears in the following:


  1. Neither Exercise Nor Supplements Boost Cognition in Two Large Studies