Beneficial Associations of Physical Activity With 2-h but Not Fasting Blood Glucose in Australian Adults

RESEARCH DESIGN AND METHODS—

The 1999–2000 national, population-based cross-sectional AusDiab study was undertaken to estimate the prevalence of diabetes and its precursors in Australian adults aged ≥25 years. The methods, response rates, indications of the representativeness of the sample, and major findings of the AusDiab survey have been reported elsewhere in detail (1,15,22). The sample of 11,247 adults represented 55% of those completing the initial household interview (22). The present analysis uses data from 9,214 adults (4,108 men and 5,106 women) who took part in the biomedical examination, were not pregnant, and were without known diabetes. Participants who reported that their health limited them from performing moderate-intensity physical activities, as well as those who self-reported a history of angina, stroke, or myocardial infarction, were excluded from the analyses. The ethics committee of the International Diabetes Institute approved the AusDiab study design, and all participants provided written informed consent to participate.

After an overnight fast (minimum of 8 h), participants attended a local survey center where an oral glucose tolerance test was performed using World Health Organization specifications (23). Blood specimens were centrifuged and transported daily to the central laboratory where plasma glucose levels were determined (Olympus AU600 automated analyzer). Demographic attributes, parental history of diabetes, education (attended highest level of secondary school available yes/no), current cigarette smoking status (yes/no), television viewing time (hours per week), total household income (≥$1,500/week or <$1,500/week), alcohol intake (self-classified into nondrinker, light drinker, and moderate or heavy drinker), and menopausal status in women (gone/going through menopause yes/no) were assessed using an interviewer-administered questionnaire. Ethnicity was determined by place of birth and language spoken at home. Of responders, 87.9% were born in either Australia or the U.K., and 96% spoke English at home. Aboriginal and Torres Strait Islanders accounted for 0.8% of the total AusDiab sample (22).

Waist circumference was measured halfway between the lower border of the ribs and the iliac crest in the horizontal plane. Using a steel measuring tape, the mean of duplicate measures was used in the analyses. Reflecting increasing health risk (24), low, moderate, and high waist circumference categories were respectively defined as <94.0, 94.0–101.9, and ≥102 cm in men and <80.0, 80.0–87.9, and ≥88 cm in women (14).

Physical activity was measured by the Active Australia questionnaire, which asks respondents about their participation in predominantly leisure-time physical activities (including walking for transport) during the previous week (25). Total physical activity time was calculated as the sum of the time spent walking (if continuous and for ≥10 min) or performing moderate-intensity physical activity, plus double the time spent in vigorous-intensity physical activity (26). Frequency of physical activity was calculated by summing the number of sessions of vigorous activity, moderate activity, and walking. Physical activity was categorized to reflect the current Australian public health recommendation for physical activity (27) as active (≥150 min/week across at least five sessions) and inactive (<150 min/week and/or fewer than five sessions). Frequency of physical activity was included in the calculation of the physical activity variable because although a single bout of exercise can enhance glucose tolerance in the short-term (10), activity needs to be regular and sustained, as the beneficial effect on glucose metabolism disappears quickly once activity ceases (28).

Statistical analysis

All analyses were conducted using Stata Statistical Software Release 8.0 (29) survey commands for analyzing complex survey data. Sample weights based on the 1998 estimated residential Australian population were used to account for the clustering and stratification in the survey design and for nonresponse. Univariate analyses were used to compare physical activity groups for men and women. Sex-specific quintiles of physical activity time (minutes) were used to examine dose-response relationships between physical activity and plasma glucose concentration.

Multivariate linear regression analyses were performed to examine the independent and combined associations of physical activity and waist circumference with FPG and 2-h PG. Six separate multiple regression models, with the inactive group as the reference category, were used to assess the extent to which being active was associated with lower blood glucose within each of the three waist circumference categories for men and for women. To examine the association of physical activity with FPG and 2-h PG concentration across each waist circumference category, participants were categorized by both their physical activity (active or inactive) and their waist circumference (low, moderate, or high), with the active low–waist circumference group being the reference category. All multivariate regression analyses were adjusted for age, parental history of diabetes, cigarette smoking, alcohol intake, television time, income, education, and menopausal status (women only). Sex differences were tested for by interactions within the regression model.

RESULTS—

Fifty-five percent of men and 64% of women did not meet the public health guidelines for physical activity. These physical activity levels are similar to those reported in the 1999 National Physical Activity Survey (26). The majority of participants (80% men and 83% women) had blood glucose readings in the normal range, while 3.4% of men and 2.9% of women were newly diagnosed with diabetes. When analyzed by physical activity category (Table 1), mean waist circumference and mean 2-h PG, but not FPG, levels were higher in inactive men and women compared with active men and women.

After adjusting for confounders, including waist circumference, meeting the public health guidelines for physical activity was independently associated with 2-h PG in both men (β −0.26 [95% CI −0.50 to −0.02], P = 0.037) and women (β −0.22 [−0.40 to −0.04], P = 0.016). However, the association between physical activity and FPG was nonsignificant (men P = 0.323, women P = 0.216) and remained nonsignificant after excluding waist circumference from the model (men P = 0.198, women P = 0.541). Having a moderate or high waist circumference was independently associated with a higher FPG and 2-h PG compared with those who had a low waist circumference (P for trend <0.001 for both men and women), with every 1-cm increase in waist circumference resulting in a 0.015- mmol/l (95% CI 0.01–0.08) and 0.012- mmol/l (0.01–0.02) increase in FPG, and a 0.04-mmol/l (0.03–0.05) and 0.03- mmol/l (0.02–0.04) increase in 2-h PG in men and women, respectively (P < 0.001 for all).

For each 30-min/day increase in physical activity time, 2-h PG reduced by 0.08 mmol/l (95% CI −0.13 to −0.02, P = 0.008) in men and 0.07 mmol/l in women (−0.12 to −0.01, P = 0.025), whereas the changes in FPG were nonsignificant. Figure 1 illustrates these dose-response relationships across quintiles of physical activity time. Although men had a higher FPG at each quintile of activity compared with women, there was no significant sex interaction for either FPG or 2-h PG.

Participants were categorized by both their physical activity and their waist circumference. Figure 2 presents the results from the multivariate regression analysis for FPG (Fig. 2A) and 2-h PG (Fig. 2B). When the association between physical activity and blood glucose was assessed within each of the three waist circumference categories, negligible differences in FPG concentration were observed in active men and women compared with inactive men and women, with differences <0.07 mmol/l. By contrast, being physically active was associated with significantly lower 2-h PG in low–waist circumference men (β −0.25 [95% CI −0.49 to −0.02], P = 0.036) and high–waist circumference women (β −0.30 [−0.59 to −0.01], P = 0.044) compared with their inactive counterparts, with nonsignificant negative associations observed in low–waist circumference women (β −0.18 [−0.45 to 0.09], P = 0.180), moderate–waist circumference men (β 0.17 [−0.63 to 0.30], P = 0.472) and women (β −0.17 [−0.51 to 0.17], P = 0.327), and high–waist circumference men (β −0.38 [−0.77 to 0.02], P = 0.061). When the association of physical activity with blood glucose was examined across each waist circumference category, it was observed that FPG values increased in a step-like progression across waist circumference categories with physical activity having negligible association with FPG. By contrast, a more linear relationship was observed for 2-h PG, implying that both physical activity and waist circumference are important for 2-h PG. In both sexes, 2-h PG levels in the active moderate–waist circumference category were not significantly different from 2-h PG levels in active participants with a low waist circumference. The pattern of change across the combined waist circumference and physical activity categories were similar for men and women with no significant sex interaction for either FPG or 2-h PG.

CONCLUSIONS—

In this large, cross-sectional study of Australian adults, meeting the public health guideline for physical activity (≥150 min/week across five or more sessions) was independently associated with lower 2-h PG but not FPG. Similarly, a dose-response relationship was observed between physical activity and 2-h PG, but not FPG. When waist circumference and physical activity were considered simultaneously, being active was associated with significantly lower 2-h PG in high–waist circumference women and low–waist circumference men compared with their inactive counterparts. In addition, mean 2-h PG concentration in those with a moderate waist circumference who were active was not significantly different from those with a low waist circumference. Considering that the analyses were conducted on a large subset of the total AusDiab study sample (22), these findings are likely to have high generalizability to the broader Australian adult population.

The association of physical activity with 2-h PG but not FPG is consistent with previous work in this area (30–32). However, previous observations are restricted to an intervention study of 18 overweight women (30), an intervention on 88 Dutch adults at increased risk for developing diabetes (31), and a cross-sectional study on the unique, high-risk Pima-Indian population (32). This is the first study to have described this association in a large, nationally representative Caucasian population. The finding that moderate levels of physical activity were associated with lower 2-h PG values, even in men and women with moderate or high waist circumference, could have important public health implications. While nondiabetic levels of both FPG and 2-h PG are independently associated with cardiovascular disease (17), compared with FPG, 2-h PG is considered to be more predictive of all-cause and cardiovascular mortality (33). Fasting hyperglycemia and 2-h postload hyperglycemia have related but differing physiological bases. Both are instances of insulin resistance (or reduced sensitivity); however, the relevant sites of insulin resistance differ between these measures. Fasting hyperglycemia predominantly reflects hepatic insulin resistance with normal muscle insulin sensitivity; postload hyperglycemia is characterized by moderate-to-severe skeletal muscle insulin resistance with normal or mildly reduced hepatic insulin sensitivity (34). This might explain the association with 2-h PG but not FPG in our study, since physical activity is directly associated with improved insulin sensitivity (35,36). Furthermore, the absence of an association with FPG concurs with the findings from controlled clinical studies in people with and without type 2 diabetes who have generally observed no changes in FPG following exercise training (37–39).

Despite the success of lifestyle interventions (that included increasing physical activity and reducing weight) in improving glucose tolerance and reducing the risk of developing type 2 diabetes in high-risk individuals (32,40,41), few population-based studies have examined the extent to which regular physical activity can attenuate the risk of hyperglycemia that is associated with overweight and obesity. In studies using self-reported physician-diagnosed diabetes as an outcome measure, physical activity typically had a relatively small impact, with the risk of diabetes in the overweight and obese groups significantly greater than in those in the lowest body fat category regardless of activity level (6,11–13). The effect of physical activity is more pronounced, however, when diabetes is determined objectively. In the Aerobic Centre Longitudinal Study, fit obese men had a similar risk of incident diabetes as unfit nonobese men (8), while active Pima Indian women in the second BMI tertile had a similar risk of incident diabetes as inactive women in the lowest BMI tertile (7). We found that in all waist circumference categories, mean 2-h PG values were lower in those who were active compared with those who were inactive. Concentration of 2-h PG did not significantly differ between those who were active and had either a low or moderate waist circumference. These results add to the current “fitness versus fatness” debate in suggesting that physical activity attenuates the risk of having high blood glucose associated with elevated waist circumference. However, the linear relationship between 2-h PG and the joint physical activity and waist circumference categories emphasizes that waist circumference remains an important factor in blood glucose control.

When the association between physical activity and blood glucose was considered within each waist circumference category, the lower 2-h PG values associated with meeting the public health guidelines for physical activity were particularly evident among high–waist circumference women and low–waist circumference men. Previous prospective studies that have explored the relationship between physical activity and blood glucose within categories of body fat have typically reported that the greatest benefit of physical activity was observed in men and women who were overweight or obese, with a nonsignificant effect observed in men and women in the leanest group (11,12,42). The current findings suggest that meeting the public health guideline for physical activity is important not just for those with a moderate or high waist circumference but also for men with a low waist circumference. This has important implications for screening individuals at risk for hyperglycemia. For example, the Australian evidence-based guidelines for the management of type 2 diabetes do not currently include physical activity as a factor to identify who should be tested for undiagnosed type 2 diabetes (43). Inclusion of a physical activity measure in the screening guidelines would help capture the group of men with a low waist circumference who are inactive and who may have a significantly increased risk of higher 2-h PG compared with their active peers.

This is the first study in a large, representative population-based sample to have examined the combined associations of physical activity and waist circumference with blood glucose in men and women. Its major strengths include a large sample size across a wide age range, objective measures of glucose status, detailed anthropometric measurements, and a comprehensive lifestyle assessment. Limitations include the cross-sectional design and the self-report of physical activity. The Active Australia questionnaire and model were used to measure and categorize physical activity based on current public health recommendations concerning duration and frequency of physical activity (27). Although the validity and reliability of the Active Australia questionnaire is acceptable (44,45), in general, questionnaires give only a crude and imprecise estimate of habitual physical activity and misclassification is inevitable. To obtain more accurate data on physical activity, future research should aim to collect objective measures of physical activity, such as accelerometer data (46). Although we found a significant dose-response relationship between physical activity duration and 2-h PG, further work is required to elucidate what particular types, intensities, and settings for physical activity may be most beneficial to improving blood glucose control in men and women with an elevated waist circumference.

Over half of the population that we studied had a moderate or high waist circumference. Although maintaining a healthy weight is important to maintain normoglycemia, weight loss is difficult and time consuming. Meeting the public health guidelines for physical activity may provide a feasible strategy for improving blood glucose control in those who already have an elevated waist circumference. The advantages of using physical activity as an intervention are that it is nonpharmacological and that the direct beneficial effects of physical activity on glucose tolerance and insulin sensitivity are rapid (10). Long-term physical activity can also indirectly decrease 2-h PG and FPG concentrations through weight loss and prevention of weight gain (8,47), which also leads to improved insulin sensitivity (48). While only a prospective study can determine the combined contributions of physical activity and waist circumference to the development of prediabetes and type 2 diabetes, the prospective collection of physical activity data in this population will eventually allow for such an evaluation.

Given the current “diabesity” epidemic, improving the understanding of the association of physical activity with blood glucose in low–, moderate–, and high–waist circumference men and women is both relevant and timely. This large, representative, population-based study highlights that there are important differences between 2-h PG and FPG in relation to physical activity. Our findings demonstrate that 2-h PG appears to be more sensitive than FPG to the beneficial effects of physical activity and that these benefits appear across the waist circumference spectrum.