ABC | Volume 110, Nº1, January 2018

Original Article Pitanga et al Physical activity and cardiovascular risk factors Arq Bras Cardiol. 2018; 110(1):36-43 Obesity was identified by BMI, being applied the equation BMI =weight(kg)/height(m) 2 and adopted the following cutoff point: obesity = 0 if BMI <30.0 and obesity = 1 if BMI ≥ 30.0. Data analysis procedures Descriptive measures (proportions) were calculated for all categorized variables. Analyzes were stratified by gender at first. The differences between men and women as to variables were identified by the chi-square test. Associations between dependent (different cardiovascular risk scores) and independent variables (LTPA and CPA) were analyzed by logistic regression. The following were considered as potential confounding variables: age, obesity, family income, educational level, and functional status. Variables presenting simultaneous evaluation (tetrameric matrix) of correlation coefficient rho < 0.60 and p ≤ 0.05 upon bivariate analysis were selected as model. Analysis of confounding variables was made by comparing Odds Ratio (OR) of the crude association and adjusted association for possible confounders. The parameter used to identify the difference between associations was 10%. Then logistic regression analysis was performed, starting with the complete model and then removing each of the possible confounding variables that resulted in alteration equal to or greater than 10% in the association between LTPA/CPA and cardiovascular risk scores. 24 The modeling process did not identify effect-modifying variables, and variables age, obesity and educational level were considered confounders for men, while only age and education were identified as confounders for women. Therefore, the best model to analyze the association between LTPA/CPA with cardiovascular risk scores was adjusted for age, obesity, and educational level for males and for age and educational level for females. Dose-response effect was also assessed for the association between LTPA and cardiovascular risk scores. Dummy variables were created for comparison between the reference group (sedentary) and each strata of the PA variable (not very active, active, very active). The Mantel Haenszel test was used to evaluate homogeneity of OR values between variables’ strata, with a significance level set at 0.05. The confidence interval was set at 95% (95%CI), and the statistical software Stata version 12.0 was used. Results A total of 6,222 men (45.3%) and 7,499 women (54.7%) were included in the study. Sample characteristics are shown in Table 1. The former were reported as higher family income, more active in free time and while commuting, with higher values for cardiovascular risk scores analyzed, while the latter were found to be more educated and more frequently obese. There was a higher percentage of retired women and no statistically significant differences between men and women as to age. The association between LTPA/CPA and cardiovascular risk scores in males and females are presented in Tables 2 and 3. LTPA is inversely associated with almost all cardiovascular risk scores analyzed, while CPA is not significantly associated with none of them. Tables 4 and 5 show us the existence of a dose-response effect in association between LTPA and cardiovascular risk scores, especially among men. Discussion This study analyzed the association between LTPA/CPA with different cardiovascular risk scores. LTPA was shown to be inversely associate with risk scores analyzed, while CPA was not. These results, especially regarding LTPA, were similar to those found among 41,053 male and female Finns when moderate or high LTPA levels among both men and women, and daily walking or cycling for work only among women were found to be associated with reduced risk for coronary events. 14 Another study which analyzed healthy behaviors, including PA measured by accelerometry, and showed an inverse dose-response association between healthy positive behaviors and risk for atherosclerotic diseases. 25 In our study, we also found a dose-response effect in the association between LTPA and cardiovascular risk scores, mainly for males. That is, the higher the level of PA, the lower the risk of cardiovascular events. The dose-response effect we found in this study has been reported for a long time. Kohl, 5 has shown, in a vast literature review, the inverse dose-response association between PA and cardiovascular events, especially coronary heart disease, in different longitudinal studies. Important to note that the classification adopted in this study had the amount of LTPA calculated based on both its duration and intensity. In other studies conducted by our research group, 11,12 in which PA was classified by intensity alone, only moderate PA was shown to hold relation with absence of hypertension and diabetes. Thus, one can assume that increasing physical activity levels to achieve greater health benefits should be suggested, bearing in mind both their intensity and duration. Results found in the association between LTPA and cardiovascular risk scores are expected, considering that the main variables composing scores are separately associated with LTPA. Studies have pointed out that LTPA is inversely associated with high BP levels, 7,12 diabetes, 11,26 lipid changes, 27 and risk of coronary heart disease. 28 According to our results, the associations reported in previous studies are more consistent among men than among women. 28,29 Regarding CPA, we could not demonstrate associations with cardiovascular risk scores, although previous studies have found a relationship between this type of activity and diabetes and cardiovascular mortality in individuals with type 2 diabetes. Important to note that these associations, when it comes to mortality by cardiovascular disease, have lost significance after additional adjustments for LTPA and CPA. 30,31 These findings most probably show that the instrument used in our study to assess PA (IPAQ) does not distinguish CPA intensity—walking or cycling, for example. Thus, if subjects’ displacement is done slowly, health benefits may not be significant. In this sense, a recent publication with data from ELSA-Brasil reported that the association between CPA and arterial hypertension was positive in women, but not statistically significant in men, while the association between LTPA and arterial hypertension was inverse 38