IJCS | Volume 32, Nº4, July/August 2019

346 Almeida et al. Cardiorespiratory fitness classification Int J Cardiovasc Sci. 2019;32(4):343-354 Original Article Study Classification After analysis of CPET results, the 6,568 apparently healthy subjects were separated by sex and age ranges (7-12, 13-19, ten-year intervals from 20 to 70, and > 80 years). After the VO 2peak was measured, individuals were allocated into percentiles and classified into very poor, poor, moderate, good and excellent CRF and then compared. This classification was called the AEMA table. Statistical analysis All data were registered in a database by the same trained, independent investigator. Analysis of these data was performed using the IBM SPSS statistics 23 (IBM Company, USA). Continuous variables were described as mean ± standard deviation and categorical variables in percentage. The Student’s t-test and the chi-square test were used for comparisons between the distributions of continuous and categorical variables, respectively. Correlations of VO 2peak with continuous and categorical variables were made using the Pearson’s test and the Spearman’s test, respectively. Percentage variation was calculated by VO 2peak values of all individuals by sex and age range. Subjects were compared for each table’s (AEMA, AHA, Cooper and UNIFESP) criteria using the Wilcoxon test, Kappa (k) and percentage of agreement (%). An error probability ( α ) < 5% was set as statistically significant. Results The group of patients excluded from the study (n = 4,782) did not show any differences regarding sex, age, anthropometric data as compared with the study population. Demographic data (Table 1) showed a predominantly urban population, of pardo ethnicity for both sexes. Regarding educational attainment, most patients had some high school education, and the family income ranged from 250 to 750 American dollars. Table 2 shows a uniform sex distribution (50.5% of men), with mean age of 40 ± 14 years for men and 43 ± 15 years for women. Overweight was predominant in both sexes, and 53.9% of the individuals were physically inactive (44.9% of men and 63.1% of women). Table 3 describes total blood count, ejection fraction (by doppler color flowmappingwith two-dimensional echocardiography), spirometry and CPET results, which guided the selection of this healthy sample population, and pointed out maximal CPET results (mean R of 1.23 and 1.21 in men and women, respectively). Inaverage,women (49.5%of the sample) showed lower VO 2peak thanmen (24.42 ± 6.7 vs. 33.70 ± 9.0 mL.kg -1 .min -1 , p < 0.001). There was an inverse, moderate correlation between VO 2peak and age in both sexes (R = -0.488, p < 0.001). Correlation of VO 2peak with family income, educational attainment and place of residence was R = 0.236; R = 0.293 and R = - 0.180, respectively. Table 4 shows mean VO 2peak in different age ranges and its percentage variation; a 16.2% increase and 4.0% increase in VO 2peak is observed for men and for women, respectively in the two first age ranges, with a descending trend as age increases in both sexes. Interestingly, such decrease is attenuated in the two last age ranges among women. Table 5 shows the comparison between CRF tables, describing the number and percentage of individuals with lower, similar and higher CRF. As compared with the AEMA table, there was an overestimation of CRF by the AHA, Cooper and UNIFESP classification. We found a significant difference, and low agreement between the CRF tables. Table 6 shows the proposed CRF classification (very poor, poor, moderate, high and very high) of the AEMA table, with VO 2peak intervals distributed by age and sex. Discussion This is an important population-based study reporting the functional capacity evaluated by CPET (VO 2peak ) of Brazilian individuals and its relationship with demographic variables, and that proposes a genuinely national classification of CRF. The findings of the study revealed high discrepancies in CRF classification when AEMA table was compared with AHA, Cooper, and UNIFESP tables. According to these three classifications, individuals were classified as having higher CRF, with disagreement rates of 57%, 62% and 64% when AEMA was compared with the AHA, Cooper and UNIFESP tables, respectively. The AEMA table distinguishes from these three tables, as it includes the age ranges - 7-12, 70-79 and > 80. Most CRF tables were composed with international sample data. For this reason, there may be ethnical and social differences that may affect the classification of the Brazilian population by these tables. We believe that external validity of data collected from foreign