IJCS | Volume 32, Nº4, July/August 2019

351 Almeida et al. Cardiorespiratory fitness classification Int J Cardiovasc Sci. 2019;32(4):343-354 Original Article Considering the CRF classification per se, each classification system has its own particularities. Cooper classification system included individuals aged older than 13 years, regardless of the physical activity level. 15 The AHA table was composed of subjects aged between 20 and 69 years, physically active or not. 16 Finally, the UNIFESP table selected only physically inactive individuals aged between 20 and 59 years. 17 In the AEMA table, 6,011 individuals were allocated to different age ranges and comparedwith the AHA´s table. A discrepancy of 56.7% (higher or lower CRF) was found, in addition to a low agreement (Kappa 0.291). When the AEMA classification was compared with the Cooper classification, 6,269 subjects were allocated, with a disagreement of 62.4% and low agreement (kappa 0.220). Finally, in the UNIFESP classification, 5,460 individuals were allocated, with disagreement of 63.9% and kappa of 0.201. It is of note the high percentage of disagreement in the CRF level of the three tables compared with the AEMA classification (56.7 - 63.9%, p < 0.001). These findings seem to be explained by the difference between study populations and by the presence of genetic factors. 11,34-38 Regarding ethnicity, our sample was characterized by a mixed population, representing white, black and Indian ethnicities, comparably to IBGE (Brazilian Institute of Geography and Statistics) data. 39 No difference between the sexes was found regarding educational attainment, family income, place of residence, physical activity level. In addition, different methods to estimate VO 2peak were used between the CRF classification systems. While VO 2peak was estimated by exercise duration in Cooper and AHA tables, in our study, this parameter was measured during CPET, and was not estimated by formulas. Previous data published by our group showed that oxygen consumption equations, such as Wasserman’s and Jones’ equations, may overestimate the oxygen consumption by 11.3% and 31.4%, respectively. UNIFESP classification included physically inactive individuals aged between 20 and 59 years. Although this table was also composed using data from the Brazilian population, comparison between this table with the AEMA table revealed the greatest revealed the greatest discrepancy (63.9%). Discrepancy (63.9%). This may be explained by the fact that, in the age ranges of 40-49 and 50-59 years, the values of oxygen consumption were the same for the classification scale, affecting the agreement between the CRF levels. 40 Therefore, considering the distribution of our study population by CRF levels, there was disagreement in CRF classification by the AEMA table compared with the AHA, Cooper and UNIFESP classifications. It is worth mentioning that the treadmill or cycle- ergometer test evaluates clinical, hemodynamic, autonomic, electrical andmetabolic responses to exercise. Information about CRF guides themedical staff to inform patients and family members about aerobic fitness of the subjects, prescribe exercise and evaluate their prognosis. 10 Since the present study showed a great discrepancy in CRF classification between the AHA, Cooper and UNIFESP tables compared with the AEMA table, our findings may be relevant for clinical practice in different ergometric laboratories in Brazil. Our proposed table provides a more accurate classification of CRF compared with other tables derived from foreign populations, since it was developed with Brazilians’ data, thereby eliminating possible biases of international tables. Limitations Since data collection was not performed in all federated states of Brazil, the possibility that our findings may not have external validity throughout the country cannot be ruled out. However, the sample was composed of individuals coming from all the country (84% from Paraiba State and 16% from other states), of different ethnicities andmultiracial background, whichwe known as a mixed-race, national sample. It is worth pointing out that comparison of our data with data matched by IBGE age groups, we did not find any difference (p = 0.401), including a similar distribution by sex. 39 Our sample showed a higher prevalence of overweight subjects, which is in accordance with data reported by 2012 Vigitel 41 (i.e., 51% of the Brazilian population). Also, the prevalence of individuals that practice regular physical exercise or exercise on the way to work is 47.7%, 41 which is similar to our population. The lownumber of individuals aged between 7 and 12 years may be explained by the low frequency of clinical indication of CPET at this age range. On the other hand, at the age of 80’s, there were few people who were healthy andmet all inclusion and exclusion criteria of the present study. Nevertheless, despite this limitation, we believe that it is important for the clinician to have initial reference values for this age group. Finally, all CPETs were performed using a treadmill, and the applicability of our findings to a cycle ergometer should be tested.