This prospective cross-sectional study was based on the data obtained from the participants of the Predimed-Plus trial at group allocation. This trial consists of a multicentre, randomized, parallel-group study of primary cardiovascular prevention that compares the effect on cardiovascular morbidity and mortality of an intensive lifestyle intervention based on a hypocaloric Mediterranean diet, increased PA, and behavioral therapy (intervention group) vs a nonintensive Mediterranean diet intervention without calorie restriction (control group). The participants comprised male volunteers aged 55 to 75 years or female volunteers aged 60 to 75 years with an overweight or obese status (body mass index [BMI] ≥ 27 and < 40) and at least 3 of the 5 components of MetS12 but without evidence of cardiovascular disease. The study has been described in greater detail elsewhere.13 The study was approved by the ethics committees of all participating centers and all volunteers signed a written consent form to confirm their voluntary participation. The study is registered in the International Standard Randomized Controlled Trials registry with the number 89898870.

A symptom-limited maximal treadmill test was performed (General Electric model T2100) according to the Bruce ramp protocol and with continuous electrocardiographic monitoring. The participants were asked to exert themselves until they were exhausted; they were allowed to support themselves on the handrails without holding on. Blood pressure and heart rate were recorded every 3minutes and at the end of the test. Expired gas analysis was performed using a system for metabolic tests (MetaSoft CPX testing, GE Medical Systems Information Technologies; Freiburg, Germany) with a gas analyzer (MetaLyzer 3B, Firmware Version 2.0, CORTEX; Leipzig, Germany).

Effort was considered maximal when the respiratory exchange ratio exceeded 1.10.14 When the respiratory exchange ratio was between 1 and 1.10, effort was only considered adequate for inclusion in the present analysis if the heart rate was ≥ 90% of the theoretical maximum (220 bpm – age) and the Borg scale score was ≥ 17.14,15 The value obtained at the final moment of the exercise was considered to be the VO2max for that individual. Based on the Wasserman equations,16 the device software calculated the workload, expressed as metabolic equivalents (1 MET = 3.5mL O2/kg/min) and the predicted VO2max.

Self-reported PA was evaluated using the REGICOR17 and RAPA18 questionnaires.

The REGICOR questionnaire, recently validated in Spain, is based on a population study and includes the version validated for Spain of the Minnesota Leisure-time Physical Activity Questionnaire. It collects information on the type of leisure-time PA, its frequency (d/wk), and its duration (min/d). Intensity was evaluated according to the 2011 Compendium of Physical Activities.19 According to intensity, activities were classified as light (< 4 MET), moderate (4-5.5 MET), and vigorous (> 6 MET).20 These data allowed PA to be dichotomized according to the clinical practice guideline recommendations for leisure-time PA4,7: moderate-vigorous PA ≥ 150min/wk or vigorous PA ≥ 75min/wk or energy expenditure ≥ 500 MET min/wk.

The RAPA questionnaire consists of 7 questions related to aerobic or dynamic PA (RAPA 1) and 2 questions on activities that increase muscle strength and improve balance (RAPA 2). The RAPA 1 also establishes 3 levels of PF (1, 2, and 3) from lowest to highest. Level 3 questions consider the clinical practice guideline recommendations.

Sedentary behavior was evaluated through the Nurses’ Health Study questionnaire validated for Spain21; of the sedentary activities considered, the time spent sitting in hours per day (≥ 7h) and the television viewing time grouped into 3 categories were selected: ≤ 2, 2 to 4, and > 4h/d.20

The Predimed-Plus trial included the “chair test”. This PF test evaluates lower body strength by recording the number of times participants can sit down and get up from a standard-size chair (43-44cm in height) for 30seconds. The reference values for the Spanish population have been published.22 According to the number of repetitions and patients’ age and sex, PF was classified into 3 levels (1, 2, and 3) from lowest to highest. For the present analysis, 15 was selected as the cutoff point to consider an individual to be in good PF, taking into account the mean age of the study population. According to the results of the chair test and RAPA 1, in the Predimed-Plus trial, a final classification was established in 3 levels: lightly active, moderately active, and active.

The study protocol also included accelerometry in a subsample of the participants (20% of the control group and 50% of the intervention group). The GENEActiv GATV04 accelerometer (Activinsights Ltd; Kimbolton, United Kingdom) was used over a 7-day monitoring period. The data were processed in R (R Core Team; Vienna, Austria) using the GGIR package (version 1.2-5). The cutoff points for moderate/vigorous PA were as published.23

Categorical variables are reported as frequencies and percentages and continuous variables as mean ± standard deviation. Comparisons between groups were made by the t test or ANOVA (analysis of variance) for continuous data and by the chi-square test for categorical data.

The Pearson correlation coefficient was used to assess the association between the PF variables obtained in cardiopulmonary exercise testing (VO2max and MET) and the continuous variables of PA and sedentary lifestyle. For variables that were by definition categorical (RAPA and RAPA+Chair), the association was determined by the eta correlation coefficient.

The possible influence of PA and sedentary lifestyle variables on VO2max was analyzed according to a 3-step process: first, univariable linear regressions were performed that included the dependent variable VO2max and each variable of PA that had shown a significant correlation as an independent variable; second, linear multivariable models were performed that included, apart from the independent variable, age and sex as potentially confounding variables; and finally, a multivariable model was applied that included, in addition to the variables included in the previous step, BMI as a possible confounding variable. In the models using the RAPA scale as an independent variable, “lightly active” was considered the reference category.

All statistical analyses were performed with SPSS 23 software.

Although VO2max is the parameter of choice for evaluating PF or cardiorespiratory functional capacity, its determination is difficult because it requires specialized technology and personnel. An alternative method is to use the duration of the exercise test without expired gas analysis or the METs reached at peak effort estimated from the speed and slope of the treadmill or bicycle watts; the high correlation coefficients of the 2 variables with VO2max justify their use.25 Indeed, METs are the most widely used indicator of PF in the literature.26 For this reason, METs were included in this analysis and their results were comparable to those of VO2max (Table 2 and Table 3), although they overestimate PF vs measured VO2max, as can also be seen from the results obtained. The entire group achieved 10 ± 2.4 METs with a VO2max of 19.6 ± 4.7mL/kg/min, representing an overestimation of 45%.

The REGICOR and RAPA questionnaires provided adequate information on PF and, in particular, on the VO2max and METs reached at maximal effort. Individuals who reported that they adhered to the clinical practice guideline recommendations regarding leisure-time PA had a VO2max 17% to 18% higher than those who did not, representing a difference of 3.1 to 3.2mL/kg/min in absolute numbers, with the same result for comparison of the most and least active groups when the variable had 3 categories. These results are clinically relevant because a 1 MET (3.5mL O2/kg/min) increment in PF is accompanied by 10% to 25% reductions in mortality.1,5 Additionally, in this study, the time spent on leisure-time PA, regardless of its intensity, was significantly related to VO2max, both in the univariable analysis and after adjustment for age, sex, and BMI, with the same findings for total energy expenditure. These results indicate that light PA, not just moderate and/or vigorous activity, may have beneficial effects, as indicated by recent work.27

Analysis of the correlation between the PA recorded in the questionnaires and VO2max revealed that the results agree with those in the literature. The correlation coefficients range between 0, 1, and 0.4 for both measured9,27 and estimated VO2max,3,10 with vigorous PA at the upper end of these values.28 Objective methods do not substantially improve upon the results obtained with the questionnaires. For example, in a study of 1631 young Australians, Schmidt et al.29 obtained correlation coefficients of 0.16 for men and 0.19 for women using pedometers, whereas Dyrstad et al.9 determined correlation coefficients of 0.47 for men and 0.54 for women using accelerometers. Our accelerometry results (r = 0.22) may have been influenced by the small number of patients tested, although the specific results for the 91 individuals who wore the accelerometers are in line with the positive results of the questionnaires in Table 2 (Table of the supplementary data). No explanation was found for the apparent superiority of the questionnaires over accelerometry in the present study, which has robustly compared the usefulness of both methods.

In contrast, the chair test, an easy-to-implement functional test that assesses lower body strength, together with the moderate-vigorous PA of the questionnaires, was the parameter showing the best correlation with VO2max in the univariable analysis and remained significant in the regression analysis. The step test, another method that assesses the lower body, has also shown a positive association with PF.30

Leaving aside the possible influence of the various methods and tests used, these results seem to indicate that the 2 variables being evaluated, although related, are nonetheless markedly different. PA is a multidimensional behavior defined as any movement that increases basal energy expenditure, whereas VO2max is a physiological parameter that reflects the ability of the cardiorespiratory and circulatory system to transport oxygen to cells. The latter has an important genetic component31 that determines individual response to physical exercise.

A surprising finding was the slight or absent relationship between sedentary behavior and VO2max. The only parameter inversely related to PF was the number of hours watching television, but this correlation was slight (r = –0.14) and lost statistical significance in the regression analysis. The number of hours sitting per day was not related to VO2max, as assessed by both questionnaire and accelerometry. Similar results have been published in a study9 that measured the sedentary behavior of 759 Norwegian adults with accelerometers, with no significant differences in hours of sedentary behavior between participants with higher and lower VO2max values. Perhaps the type of sedentary behavior should be taken into account. Indeed, Saidj et al.6 found that the time spent sitting at work was not related to VO2max, whereas leisure-time sedentary behavior had an inverse and significant relationship.

The interpretation of these results was complicated by the small number of participants in some subgroups. In addition, although a significant number of participants had MetS and were overweight, this is a specific population group and the application of the conclusions to other groups would be inappropriate. The present analysis was performed by a single research group, which is usually an inconvenience when it comes to the generalization of results but is a positive aspect in this case because of the reduced variability in the performance of exercise tests with expired gas analysis. Finally, this study was cross-sectional and the level of evidence of the results would have been higher with a longitudinal design.