December 6 — 7, 2007 Advisory Committee Meeting Minutes

All-Cause Mortality Subcommittee Report

I-Min Lee, M.D., Sc.D. presented the report on All-Cause Mortality. Dr. Lee acknowledged the participation of William Haskell, Ph.D. from the Committee and Bill Kohl, Ph.D. from CDC on the All-Cause Mortality Subcommittee. Additionally, Dr. Steven Blair served as a consultant. The subcommittee formulated the following five questions to organize their work: 1) What data do we have on the topic? 2) Can we confirm there is an association between physical activity and all-cause mortality and what is the magnitude of the association? 3) What is the minimum amount of physical activity associated with lower rates of mortalities in the studies? 4) What is the shape of the dose-response curve for all-cause mortality? 5) How is the relation between physical activity and all-cause mortality influenced by adiposity levels?

Question 1. What data do we have on the topic?
In extracting data from the CDC database the group attempted to be as inclusive as possible. The subcommittee selected all age groups, all the study designs that were there, all activity types and obviously the health outcome all-cause mortality. This search resulted in 83 studies of which 10 were excluded. The studies were excluded based on the following rationale:

1. Three excluded because of a relationship to adverse events which will be covered in another chapter by the Committee.

2. Three physical fitness studies were excluded because they did not provide direct information on the amounts and kinds of activity, etc, which were relevant to our report.

3. Two cancer studies were excluded as that subject is covered by another subcommittee.

4. One study on the relationship between BMI and mortality stratified by physical activity levels was excluded as there was no data directly related to physical activity.

The group attempted to be as evidence-based as possible and looked for particular information related to volume, intensity, frequency and duration. Of particular interest were special populations such as persons with disabilities. Almost all of the studies that were reviewed were perspective cohort studies. The net result was a large database of more than 300,000 observations in men and nearly 700,000 observations in women with more than 140,000 deaths. The disparity in observations between men and women was primarily due to several large cohort studies that included many women. The majority of the participants in the studies were 40 years of age or older, Caucasian and for the most part healthy.

Question 2. Can we confirm there is an association between physical activity and all-cause mortality and what is the magnitude of the association?
The studies consistently pointed out lower rates of mortality. There was also a significant inverse association with at least with one sex or with one domain of physical activity in 67 of the 73 studies. As all of the studies were observational studies the subcommittee looked for alternate explanations for the associations. Sources of bias such as sick people participating in the studies, follow-up rates and confounding were interpreted to not have a significant impact, if any, on the studies.

Question 3. What is the minimum amount of physical activity associated with lower rates of mortalities in the studies?
There are complications with interpreting data from the studies to find out minimum amounts of physical activity associated with lower rates of mortality because the studies assess physical activity in different ways. For example, if a questionnaire is used the questions asked may differ. Even if the same questions were asked there may still be differences across studies because of the differences in types of activities such as leisure, occupational, household and commuting activity.

Additionally, analyzing data in different ways leads to problems interpreting the science. Typically, studies will have different groupings from low to high with different categories such as inactive, moderately active, highly active, etc. However, one may not know what each of these categories correspond to.

A common way to organize categories are by energy expended, duration of physical activity or frequency of physical activity. Within the studies 12 were done by energy, 9 by duration and 4 by frequency. Studies analyzed by energy indicated that about 1,000 kilocalories a week primarily in leisure time activity was associated with significantly lower levels of mortality. 1,000 kilocalories a week roughly corresponds to 2.5 hours of moderate to vigorous intensity physical activity, which is consistent with current recommendations. Studies classified by duration indicated about 2 hours a week as the minimum amount of activity associated with lower mortality. Studies by frequency, some of which may not have duration built in, showed that activities as infrequently as once a month to one to three times a week of duration, perhaps 30 minutes or longer, was associated with lower risk.

Walking was cited as a good example of physical activity in the 1995 CDC/ACSM recommendations on moderate physical activity. From the studies in the CDC database it appears 2 hours a week corresponded with lower mortality. In terms of distance, 1 – 2 miles a day was associated with lower risk. Pace of walking was very consistently associated with lower risk as well (i.e., the faster your pace the lower your risk). Three studies looked at walking/cycling combined. Although some only looked at active commuting and/or leisure activity combined, these studies generally showed 15 – 20 minutes of walking/cycling per day showed benefit.

Question 4. What is the shape of the dose-response curve for all-cause mortality?
While the minimum amounts of physical activity found to be a benefit in the studies correspond to current health recommendations the relationship is most likely not an all or nothing proposition. Rather, the benefit is more of a gradient response. As such the group wanted to look at the shape of the dose-response curve. As before it can be problematic analyzing different studies due to the different ways physical activity can be classified so each category was reviewed. In plotting the studies a curvilinear relationship was found. While there was a lot of benefit at early levels the benefit tapers off and much more activity is required to gain similar benefits.

The group also questioned the impact of intensity of physical activity, specifically, do all activities count equally? When looking at the results of the studies on the surface the higher the intensity level of the activity the more benefit one receives. In most cases though volume of total energy expended is not accounted for. For example, if 2 individuals engage in the same intensity activity but one has a larger body the person with the larger body expends more energy.

Question 5. How is the relation between physical activity and all-cause mortality influenced by adiposity levels?
The last question the subcommittee looked at is the physical activity mortality association and whether body weight was an influence or does the association exist merely because people who exercise are leaner? Sixty percent of the studies adjusted their findings for body mass index or some measured adiposity and all consistently observed significant inverse associations. A handful of studies looked at individuals of different body mass indices and their results suggested that no matter what your level of adiposity, if you are physically active, at about the levels recommended, you will have significantly lower risks of mortality.