Table 6.4 summarizes the published prospective epidemiological studies of physical activity and development of type 2 diabetes. In each case, the adjustment variables are noted, but less detail is given about individual studies than in Table 6.3 since they provide somewhat weaker evidence for prevention than RCTs. This is primarily true because any one factor cannot be varied independently of others which may also be associated with altered risk.
In the case of physical activity, people who engage in greater amounts of activity may themselves select other healthy behaviors which are incompletely measured and adjusted for (e.g. less cigarette smoking or altered diet). There may also be a genetic component to physical activity, either based on spontaneous activity100 or clustering of other genetic factors which facilitate activity101. Analyses of twin studies have shown contradictory results concerning asso ciations between physical activity and glucose levels, with one showing a strong genetic effect101, while the other did not102. Regardless of this concern, prospective studies still provide some of the strongest data suggesting that higher levels of physical activity may protect against type 2 diabetes.
One of the earliest studies to explore the role of physical activity was published by Medalie and colleagues from the Israeli Ischemic Heart Disease Project103. They did not show detailed results, but noted no association of five-year clinical diabetes incidence with reported measures of physical activity.
In contrast, most other studies have found beneficial effects. For example, the Nurses Health Study followed over 87000 nurses in the USA for eight years. The first report104 found that at least weekly vigorous activity was associated with a lower risk of self-reported diabetes (RR = 0.83, CI 0.74-0.93; ARR = 8.8/10000 person-years) . A recent report utilized interim measures of reported activity, rather than a single baseline measure, and explored the type and amount of activity more thoroughly105. There was a graded decrease in diabetes incidence over eight years with increasing metabolic equivalent (MET) levels (1 MET is equivalent to 3.5 ml of oxygen utilization per kilogram body weight per minute), which was consistent among nurses whose only form of physical activity was walking. This provides important evidence that feasible levels of increased activity may result in lower diabetes incidence rates.
Fifteen-year follow-up of University of Pennsylvania male alumni was reported by Helmrich et al.106. In a widely quoted result, each 500 Kcal increase in leisure time activity was associated with a RR for diabetes of 0.94 (CI 0.90-0.98; ARR = 1.6/10000 person-years), with the largest reductions in people doing vigorous activity at baseline (RR = 0.69; ARR = 7.5/10000 person-years). Whether the leisure time recall instrument is capable of such fine distinctions remains an open question, since other studies have not found that such small increments result in reductions in diabetes risk107.
In a two-year follow-up study of residents of Malta, Schranz et al.108 reported that moderate to high total physical activity (not just leisure time) reduced the risk of diabetes among people with normal OGTTs (RR = 0.53; ARR = 2.4/100 person-years) but not among those with IGT. Similar effects were seen when stratified by obesity and positive family history.
Self-reported diabetes was studied in a nested case-control design among a cohort of Iowa women by Kaye and co-workers109. They found that over a two-year period the age-adjusted odds ratio for medium versus low reported physical activity was 0.7 (CI 0.5-0.9) and for women reporting high levels of activity it was 0.5 (CI 0.4-0.7) . The overall incidence of diabetes in this cohort was about 1% at two years. BMI, waist-hip ratio and education were also independently associated with diabetes. Once obesity measures were adjusted for, there was no longer an independent association of physical activity or smoking.
Table 6.4. Prospective studies of physical activity and type 2 diabetes mellitus incidence.
Population (no. of subjects)
Follow-up (years) Age
Medalie et al. 1974R Mansonei al. 1991lc Hu et al. 1999105
Israel Ischemic Heart 5
Disease Study (10059) US Nurses Health Study 8
US Nurses Health Study 8
40-65 Weekly vigorous vs. not 40-65 Q2 vs. Q1
No association 0.83 (8.8/10000 PY) 0.77
RR not stated
Was this article helpful?