ABC | Volume 110, Nº5, May 2018

Original Article Ramires et al Prevalence of metabolic syndrome in Brazil Arq Bras Cardiol. 2018; 110(5):455-466 Thus, PA variable was initially grouped into three categories, according to World Health Organization recommendation: active (individual that reaches or exceeds moderate physical activity for 150 minutes or vigorous physical activity for 75 minutes per week in at least 10 minutes sessions); inactive (which refuses to practice PA at leisure) and insufficient assets (when performing PA below recommendation). 11 Finally, it was decided to combine inactive and insufficient assets categories, transforming them into dichotomous variable (active/inactive). Variables corresponding to comorbidities analyzed here were previous self-reported cerebral vascular accident medical diagnoses (CVA), chronic renal failure (CRF), depression and other cardiovascular diseases (CVD). For the latter, previous CVD diagnoses reports were considered, such as: infarction, angina, heart failure, among others. Overweight was identified according to body mass index cut-off points (BMI). In individuals aged between 18 and 59 years, values ≥ 25 kg/m² were considered overweight. 11 For those aged 60 years or older, values > 27 kg/m² were considered as overweight. 12 Statistical analysis Statistical analysis were processed with Stata software version 13.0 (Stata Corp., College Station, USA), using survey commands, whose analysis procedures take into account sampling design and weight. 13 The comparison between MS prevalence, for each comorbidity and for disease burdenwas based on their respective confidence intervals of 99% (CI 99%). Prevalence ratios (PR) with their respective CI 99%, were calculated by simple and multiple Poisson regression models. 14 Statistical modeling process was conducted using determinants conceptual model to MS, 10 applying hierarchical approach in analysis and using stepwise forward method for variables introduction, considering as eligible those with p < 0.20 (univariate analysis); variables in which the CI 99% did not include "1" or contributed to model adjustment remained in the model. Associations between MS and potential associated factors were introduced according to sociodemographic, behavioral and comorbidity factors, analyzed by three multiple models. At distal level of analysis (Model 1), the sociodemographic variables age, schooling, skin color, conjugal situation and housing region were considered; for Model 2 composition we used, behavioral variables physical activity and health self‑perception adjusted by Model 1; In model 3, variables referring to proximal determinants (comorbidities) were introduced and their effects being adjusted by model 2. It is emphasized that once variables set was defined in a hierarchically superior model, it did not suffer any alteration in others levels of analysis. Justification for preserving variables in each model was based on result importance for MS occurrence understanding and effect magnitude, as well as its variability, represented here by CI 99%. 14 In addition, analysis were stratified by sex, considering that in descriptive analysis, MS showed to affect in different way male and female population, which may reflect different association factors between the groups. Results Sociodemographic, behavioral and comorbidities characteristics of 59.402 individuals over 18 years are described in Table 1 according to MS absence or presence. Physically inactive individual high frequency was identified (98.1%) and 53.8% who were overweight. Self-perception predominant report of very good or good health was observed (65.9%) and low schooling significant frequency (39.1%) among individuals. Table 2 shows MS prevalence, comorbidities and MS components burden in Brazilian population. Abdominal obesity was the factor with highest prevalence in this study (65.2%, CI 99% 64.4-65.9), followed by high BP (40.7%, CI 99% 39.6-41.7). It was observed that in all comorbidities, women presented the most expressive results, with high BP (46.9%, CI 99% 45.5-48.3) the only condition in which men showed a higher prevalence. In components sum, it is observed that only 1/4 of population did not present any of studied changes (23.8% [CI 99% 22.9-24.7]), while 38.1% (CI 99% 37.2-39.0) of participants already presented at least one MS component and 29.2% (CI 99% 28.3-30.1) coexisted with two considered factors. MS condition was estimated at 8.9% (CI 99% 8.4-9.5) of Brazilian population, with women proportion in this condition (10.3% [CI 99% 9.6‑11.2]) statistically surpassing what was observed in the male population (7.5% [CI 99%, 6.7-8.3]). Table 3 presents MS prevalence according to studied exposure variables. There are greater aggravation prevalence among older individuals (≥ 60 years), lower schooling time (≤ 8 years) and living with a partner. MS was greater among individuals residing in SE/S/CW regions, physically inactive, with overweight and who considered their health precarious. Regarding comorbidities, in general, MS higher prevalence were found among individuals who reported prior CKD diagnosis, stroke and other cardiovascular diseases, in relation to those who said they did not have the disease. In addition, we identified that, regardless of characteristic or condition considered as risk, MS prevalence was always higher among women. Results referring to factors associated with MS in hierarchical modeling process (hypothetical-causal model), different for men and women, are available in tables 4 and 5. In female population final model, we identified that MS probability was higher among individuals in the following situations: age ≥ 60 years (PR 3.20 [CI 99% 2.76-3.72]), education ≤ 8 (PR 1.46 [CI 99% 1, 23-1,74]), living with partner (PR 1.27 [CI 99% 1,11-1,45]), residing in SE/S/CW regions (PR 1.18 [CI 99% 1, 02-1.38]), regular to very bad health self-perception (PR 2.35 [CI 99% 1.99-2.78]), stroke (PR 1, (CI 99% 1.00‑1.86), other CSD (PR 1.29 [CI 99% 1.03-1.62]), overweight (PR 2.09 [CI 99% 1, 79-2.42]) and depression (PR 1, 31 [CI 99% 1.07-1.59]) (Table 4). Regarding male population, final model did not include variables as schooling, skin color, other CVD and CVA, age remaining in ≥60 years (PR 2.60 [CI 99% 2.04-3.31]), living with partner (PR 1.48 [CI 99% 1.17-1.88]), residing in SE/S/CW regions (PR 1.57 [CI 99% 1.28-1.94]), have worse ("regular to very bad") self-referred health (RP 2.59 [IC99% 2.01-3.33]) 457