In this population-based case-control study we found an inverse relationship between leisure-time physical activity and the risk of acute myocardial infarction in the groups of lean, normal-weight and overweight individuals, but not in the group of obese persons. Overweight individuals (BMI 25.0–29.9) who where physically active twice or more per week during their leisure time had an equal or decreased risk of myocardial infarction compared with normal-weight but sedentary persons. In contrast, an increased risk was noted among the physically active obese persons (BMI ≥ 30).
The overall results, where increased risk is associated with overweight/obesity, and decreased risk is associated with leisure-time physical activity, as observed in this study, are in accordance with several other studies [2–6, 11–13].
The joint effect of BMI and leisure-time physical activity, with equal or reduced risk of myocardial infarction among the overweight active persons compared with normal-weight sedentary persons corresponds to the results presented by Blair and Brodney [16]. Somewhat unexpectedly, we did not observe any reduced risk among the obese persons who were active, which is in contrast to others [16, 20]. Instead, an increased risk of myocardial infarction was observed in this group.
However, there are studies where lack of an inverse relationship between physical activity and coronary heart disease in the highest BMI strata have been noted [21, 22]. In some other studies, the active obese persons still had a higher cardiovascular or total mortality compared with inactive but normal-weight persons, irrespective of an inverse association between physical activity and mortality within the obese group [2, 23].
It is known that obesity is associated with several adverse cardiovascular conditions, such as high blood pressure, and left ventricular hypertrophy [24, 25]. Furthermore, some small studies have indicated that high BMI is associated with less efficient cardiac performance and higher blood pressure response during exercise, a higher level of oxidative stress after an exercise session, as well as a lower heart rate reserve [26–30]. It has also been shown that overweight subjects, even if they are regularly engaged in vigorous sport activities during leisure time, still have equal or lower heart rate variability at rest compared with sedentary lean subjects [31]. It may be hypothesized that intense exercise puts a very high strain on the cardiovascular system in obese subjects, which could result in an increased risk of acute myocardial infarction. Our results suggest that advising middle-aged or older individuals who are obese regarding physical training should be done with care, and that obese subjects should be encouraged to attain an initial weight reduction before taking up vigorous exercise.
Study limitations and strengths
One advantage of the SHEEP study is that we have exposure information regarding recent leisure-time physical activity at the time prior to inclusion, as well as BMI at the time of inclusion in the study; this might explain some of the differences in our results compared with other studies. People might change both their BMI and physical activity level several times during life due to different reasons, and it is still unclear how different trajectories of the combination of BMI and physical activity relate to the risk of myocardial infarction.
In the analyses, we used the information regarding physical activity in the age interval to which the subject belonged at the time of the case occurrence. We also carried out analyses using information about physical activity in the age interval prior to inclusion in the study. The results were, however, unchanged.
We used data on height and weight from the clinical examination as the primary source of information regarding the calculation of BMI, and self-report of height and weight if data from the clinical examination were missing. For those where we had information from both the clinical examination and self-report, the BMI calculated from self-reported height and weight was closely correlated to the data from the clinical examination (BMIclinical data mean (std) 26.2 (3.9); BMIself-report mean (std) 25.9 (3.8); r = 0.92). Furthermore, when we restricted the analysis to the subjects for whom we had information on BMI from the clinical examination, the main results regarding the effect of leisure-time physical activity within the obese group did not change (data not shown).
Regarding the information about leisure-time physical activity/exercise we had to rely on self-report. This may lead to misclassification of the level of leisure-time physical activity. If the misclassification was non-differential between cases and controls, this would lead to estimated odds ratios closer to the null value than the actual true value, at least for the most active group when compared to the sedentary group. However, if the reporting of physical activity depended on disease status, the estimated odds ratios may be over- or underestimated. In general it might be expected that people tend to overestimate their physical activity level, due to social desirability. However, if the cases were less prone to do this and instead reported less activity compared with healthy controls, our estimated odds ratios regarding physical activity would be biased away from the null value. Since our overall results regarding leisure-time physical activity are in accordance with previous studies, including cohort studies where the problem of differential misclassification is unlikely to occur, we do not think that differential misclassification explains the major part of the findings regarding leisure-time physical activity in this study.
However, it has been shown that people in higher BMI strata are less precise in their reporting of leisure-time physical activity level than normal-weight people [32]. This might be part of the explanation for the weaker association between leisure-time physical activity and myocardial infarction among overweight and obese persons compared with normal-weight persons, observed in this study. This potential (non-differential) misclassification of physical activity is, however, unlikely to explain the opposite direction of the estimated relative risk among the most active obese persons when compared with the normal-weight or obese sedentary persons. Furthermore, we find it unlikely that obese cases should be more prone to overestimate previous physical activity level, compared with obese control subjects.
A possible explanation for the differences in the results when all cases vs. non-fatal cases only were included (apart from different quality in the proxy information for the fatal cases) could be a higher survival rate after a myocardial infarction for physically active persons, as has been reported by Wannamethee, et al [33].
The potential role of selection bias must be considered. In general, we had quite high response rates for both cases and controls (78 % for cases, 72 % for controls), which should prevent serious distortion of the results due to selection bias. However, the response rate was higher for non-fatal than for fatal cases (84 % and 62 %, respectively, with response for fatal cases provided by close relatives). Under the assumption that non-response was related to physical activity level (i.e. that low physical activity level was more common among non-respondents, and the proportion of non-respondents was higher among controls than non-fatal cases), we would have expected that the results from the analyses including non-fatal cases only would have yielded odds ratios at least equal to, or further away from the null value, compared with the odds ratios obtained from the analyses including all cases, if a substantial selection bias would have been present. This was not the case in our study. Furthermore, the main results for obesity and leisure-time physical activity are in agreement with several cohort studies, where selection bias is less likely to occur.
In some groups, e.g. in the group of lean subjects, the number of cases and controls were small. This is also the case when analyzing the combined effect of BMI and leisure-time physical activity in separate strata of men and women. The small number of cases and controls in these sub group analyses results in rather wide confidence intervals, which prevent any firm conclusions to be made from these analyses. However, the main results of the combined effect of BMI and leisure-time physical activity were similar in both men and women, and also in groups of low- and high socioeconomic status, in smokers and non-smokers, as well as in those below and above 60 years of age.