To our knowledge, this is the first study to derive overall and fitness-specific OATsstandard and OATs60+, following the %VO2max and the VTs physical intensity categories, and to compare them with the CTs. Briefly, this work highlights the importance of using METs intensity thresholds adapted to both the assumed 1 MET value and the fitness status of older individuals. In this regard, OATsstandard were found to be 21–24% lower compared to OATs60+, and 58–76% higher in "good/superior" fitness individuals than in "very poor/fair" fitness individuals. When compared with the CTs, higher OATsstandard and OATs60+ were obtained for MPA and HPA than the conventional MPA threshold (3.0 METs), except in the “very poor/fair” fitness individuals. However, lower, similar, or even higher OATsstandard and OATs60+ were obtained for VPA, and SPA compared to the conventional VPA threshold (6.0 METs). As a result, this study revealed appreciable differences between the CTs and the derived OATs, even when calculated following the CTs paradigm (standard METs and %VO2max intensity categories). Therefore, a potential for CTs to misclassify LPA, MPA and VPA in older adults can be inferred, primarily depending on the cardiorespiratory fitness and the METs derivation approach. Alternatively, this study provides resources for the adaptation of the METs system in this population, reporting OATsstandard and OATs60+ according to the fitness status, and following habitual (%VO2max) and alternative (VTs) physical intensity categories.

Looking at the OATsstandard following the VO2max intensity categories, equal or higher MPA thresholds and lower, similar, or higher VPA thresholds were reported according to the fitness subgroup, compared to those of the CTs (Fig. 1A, Fig. 2A). These results partially refute the first proposed hypotheses that both the MPA and VPA thresholds would be lower than those from the CTs, suggesting that the main factor in the correct categorization of physical intensity lies in the level of fitness. In this sense, the derived OATsstandard for MPA were higher in the overall sample (3.7 METs, Fig. 1A) and the “good/superior” fitness subgroup (5.3 METs, Fig. 2C), but nearly identical in the “very poor/fair” fitness individuals (3.1 METs, Fig. 2A) compared to the conventional one (3.0 METs). On the other hand, the OATsstandard for VPA were substantially higher in the “good/superior” fitness subgroup (7.4 METs, Fig. 2A), but notably lower in the overall sample (5.1 METs, Fig. 1A) and the “very poor/fair” fitness subgroup (4.4 METs, Fig. 2A). Indeed, older adults in this “very poor/fair” fitness subgroup would need to exceed 88% of their VO2max to achieve the 6.0 METs of the conventional VPA threshold, which is 13% higher than the VT2 (SPA: 5.1 METs, Fig. 2A) in this subgroup. This contrast with the usual assumption that older adults are physiologically less able to achieve absolute VPA, and explains why moderate-to-vigorous physical activity is often used as a pooled domain of intense physical activity [31]. Similar METs intensity thresholds to those from the "very poor/fair" fitness subgroup (OATsstandard, %VO2max, MPA: 3.1 METs, VPA: 4.4 METs) has been previously reported by the ACSM, suggesting ACSMTs of 3.2 and 4.8 METs for MPA and VPA, respectively (Supplementary Fig. 1). In contrast, 68–71% higher OATsstandard following the VO2max physical activity intensity categories, for MPA (5.3 METs, Fig. 2C) and VPA (7.4 METs, Fig. 2C) were respectively derived in the "good/superior" fitness individuals, compared to those in the “very poor/fair” fitness subgroup (MPA: 3.1 METs, VPA: 4.4 METs, Fig. 2A). This difference was notably greater that that shown by Mendes [32] in a sample from 20 to 60 years old, obtaining 36% and 22% higher MPA and VPA thresholds in the high fitness compared to the low fitness individuals, respectively. These results directly refute the second hypothesis of the present study, which expected similar OATsstandard, following the %VO2max physical intensity categories, in the higher fitness subgroup compared to the CTs. Therefore, although older age is associated with lower cardiorespiratory fitness [12, 33], a homogeneous criterion should not be followed, and the use of non-fitness-specific METs intensity thresholds may increase the risk of inaccurately classifying physical activity intensity. Furthermore, based on the disparate prevalence of sedentary and inactive lifestyle in older adults [34], caution should be taken if the CTs are used in this population.

As for the METs derivation approaches used, 21–24% lower OATsstandard compared to the OATs60+ were obtained. The reason for this difference is the 1 MET value 23% lower than the standard assumed when deriving the OATs60+. These results align with the third hypothesis, expecting higher OATs when using METs60+ than standard METs. Thus, since the CTs and the OATsstandard are based on standard METs, it is reasonable to consider that they should not be applied when assuming a 1 MET value far from 3.5 mL O2·kg−1·min−1. Therefore, the OATs60+ might be the preferred option when categorizing for the intensity of those METs60+ equivalencies from the Older Adult Compendium of Physical Activities, [13] but also when using estimated or measured RMR values close to 2.7 mL O2·kg−1·min−1. However, no study has compared the use of alternative METs system strategies, including variations in the 1 MET assumption or the METs intensity thresholds applied, with the “classical” one in older adults. Therefore, future studies should evaluate the differences between these two approaches, and whether significant improvements in the reliability of the conclusions are observed from an epidemiological and clinical perspective.

This study also reported OATs following the VTs intensity categories (HPA and SPA), showing, as initially hypothesized, consistently higher METs intensity thresholds than the same OATs following the %VO2max intensity categories (MPA and VPA). For instance, the overall sample’s OATsstandard, following the VTs physical activity intensity categories, showed an HPA threshold of 4.2 METs (53%VO2max) and a SPA threshold of 6.1 METs (76%VO2max), which are respectively higher compared to those obtained at 46% (MPA: 3.7 METs, Fig. 1A) and 64%VO2max (VPA: 5.1 METs, Fig. 1A) using the %VO2max intensity categories. However, when these same overall OATsstandard were compared to the CTs, a similar SPA threshold (6.1 METs, 75% of VO2max, Fig. 1A) to the conventional VPA threshold (6.0 METs) was obtained. This issue was previously studied by Iannetta [16], who found a high risk of physical intensity misclassification when using "fixed universal METs" thresholds (LPA, MPA, VPA, maximal) compared to using VTs intensity categories (MPA, HPA, and SPA). These authors showed an increase in the HPA range as VO2max increased, in contrast to the fixed MPA range in the CTs (≥ 3.0 METs to < 6.0 METs). Similar results were found in the present study, reporting HPA ranges in the OATsstandard, following the VTs physical activity intensity categories, of 1.4 METs in the "very poor/fair" fitness subgroup and 3.1 METs in the "good/superior" fitness subgroup. Considering the importance of VTs from a physiological perspective, their use should be encouraged, providing more far-reaching evidence, including more certainty about metabolic pathways, the use of energy substrates, or metabolic stress during a given intensity category.

This study suggests that a non-adapted use of the METs system in older adults may inadvertently undermine the validity of physical activity recommendations and prescription from both public health and clinical perspectives. Among the concerns this may entail, the unfair application of the conventional VPA threshold (6 METs) is one of the most significant, requiring older individuals with low fitness to achieve near-maximal effort (6.8 METs). This may explain the low levels of VPA frequently detected in older adults, many of whom lead sedentary and inactive lifestyles, and the widespread use of moderate-to-vigorous physical activity in this population [35]. Furthermore, if this low “absolute” VPA accumulation is then combined with MPA, the distinct effects these activity intensities may have separately is obscured, thereby limiting the depth of research findings. This poses new methodological challenges focused on the proper assessment of both fitness and physical activity in older adults. As there is now a global trend towards precision medicine, accurate estimation of physical activity intensity is essential [36, 37]. This aligns with recommendations for utilizing intensity categories based on the VTs, which better capture the metabolic stimulus and can improve comparability across studies. Otherwise, the validity of research delving into the dose amount and intensity of the physical activity needed to improve health will be compromised, reducing the efficacy, efficiency, and safety of physical activity interventions [38]. This is particularly critical for clinical subgroups, highlighting those with cardiovascular diseases [20, 39]. Therefore, a one-size-fits-all approach in the older population may not be appropriate, and personalized exercise programs considering individual fitness are essential for optimizing health outcomes. To this end, new approaches should also be explored by deriving individualized METs intensity thresholds based on additional determinants such as measured or estimated RMR, sex, body composition, physical performance, or clinical status [12, 33, 40].

This work presents various strengths and limitations. To our knowledge, this is the first study to provide overall and fitness-specific OATsstandard and OATs60+ based on GXT protocols in a relatively large older adult sample, and using two different approaches to categorize physical intensity. However, this study included data from two separate studies, FenotipAGING and PRO-Training. Since neither of them was originally designed to address our specific research question, we did not perform a priori power and sample size calculations. Therefore, we should acknowledge that these OATs might not apply satisfactorily in a different sample. Nevertheless, post-hoc power computation analyses were conducted afterwards, revealing a large effect size and achieving a statistical power (1-β probability error) above 0.90 for all primary outcomes. It should also be noted that, according to the methodologies in the FenotipAGING and the PRO-Training studies, subjects followed different GXT protocols, and distinct metabolic devices were used, which could limit to some extent the comparability of VO2max among subjects. However, regarding the GXT protocols, no differences in RER were detected among subgroups, suggesting that cardiorespiratory fitness was consistently assessed across individuals. Furthermore, meticulous attention was paid when calibrating both metabolic devices, and standardizing the measurement conditions. Besides the OATsstandard, this study also reported OATs60+ that could be useful to categorize the intensity of the METs60+ based equivalences from the Older Adult Compendium of Physical Activities, or when assuming a measured or estimated individual RMR closer to 2.7 mL O2·kg−1·min−1. Furthermore, OATsstandard and OATs60+ following two different strategies for physical activity intensity categorization were developed, highlighting the OATs based on VTs intensity categories. This will allow for more physiologically meaningful findings. Therefore, this study creates a new scenario by providing alternative strategies for adapting the METs system for improved use in adults ≥ 60 years old. Nevertheless, future studies should test whether using OATs, compared to the CTs, can make a real difference from a clinical and epidemiological perspective.