Exercise dose and individual response of healthy adults: is it time to re‐evaluate exercise responsiveness and training recommendations?

I read with great interest the article by Montero and Lundby (2017) addressing the effects of endurance training (ET) dose in cardiorespiratory fitness (CRF) responsiveness in healthy young males, recently published in The Journal of Physiology. While I fully acknowledge the role of CRF as a predictor of all-cause mortality and cardiovascular disease (Kodama et al. 2009) and, consequently, the importance of ensuring sufficient CRF improvements in response to ET in all individuals, there are several aspects of the article regarding individual response that I would like to discuss. Montero and Lundby report that the prevalence of CRF non-response to ET decreased with higher exercise dose and it was absent in groups trained with the highest doses (240 and 300 min per week) after the first 6 weeks, suggesting that the lower levels of the current standard exercise guidelines (150 min of moderate endurance exercise per week; Garber et al. 2011) may not provide a sufficient stimulus to evoke positive adaptations in all individuals. However, the authors define ‘non-responders’ as any individual who failed to demonstrate a change in maximal power output (Wmax) greater than the typical error of the measurement (TE) without including a control group in their study. Several review articles (Hopkins 2000, 2015; Hopkins et al. 2009; Atkinson & Batterham 2015; Williamson et al. 2017) highlight the importance of a relevant comparator arm when examining inter-individual differences in response to training (SDR), emphasising that the results of parallel studies are meaningful only when the standard deviation (SD) of the change score(s) in the intervention group(s) (SDI) is substantially larger than the SD of the change score in the control group (SDC), according to the equation: SDR = √(SDI2 – SDC2). The absence of a non-exercising comparator arm in the study of Montero and Lundby represents a strong limitation in this sense since it impedes the separation of the true inter-individual variability in CRF response to ET from the technical error of measurement and random intra-individual biological variation. A re-analysis of the results provided by the authors, by extrapolating SDC from the test–retest reliability data of the same study (SDC = TE × √2) assuming a similar TE for Wmax (±3.96%) over the first 6-week ET intervention (Hopkins et al. 2001), reveals that most of the inter-individual variability in CRF response observed in the groups exercising ≤180 min per week was indeed due to this random error component. In fact, the SDR for CRF response (free from the influence of technical error and biological variation) and its 95% confidence interval (derived using statistical first principles under the assumption that the sampling variance is normally distributed) for each of these groups are −8.2 W (–14.2 to 8.3 W) (60 min), 6.4 W (–14.1 to 16.7 W) (120 min) and −11.2 W (–16.5 to 4.5 W) (180 min), with negative values that indicate a greater variation in the control group (estimated from the reliability study) than in the intervention group. When these data are interpreted in relation to the baseline between-subject SD as recommended by Hopkins (2015) to assess the magnitude of individual responses by standardization according to given thresholds (0.1, 0.3, 0.6, 1.0 and 2.0 for small, moderate, large, very large and extremely large), the SDR for CRF response in the three lower-dose groups is either negative or borderline trivial–small, but considering the uncertainty, the true inter-individual variability in CRF response could have been negative, trivial or, at most, small [−0.19 units (–0.32 to 0.19 units), 0.10 units (–0.21 to 0.25 units) and −0.27 units (–0.39 to 0.11 units) for 60, 120 and 180 min, respectively]. Of note, the width of the 95% confidence interval for SDR in these groups indicates a reduced precision for the estimate of individual responses. No parallel studies to date have been properly powered to assess inter-individual response variability, as a very large sample size is required to define this parameter precisely (Hopkins 2018). Notwithstanding the uncertainty surrounding individual response estimates, the proportions of ‘responders’ and ‘non-responders’ to ET intervention can be obtained using a spreadsheet (Hopkins 2018) and with knowledge of the smallest important change (e.g. 0.2 × baseline SD; Hopkins 2000). Assuming no mean change in the (estimated) control group over the first 6-week period in the study of Montero and Lundby, high percentages of ‘non-responders’ emerge in the groups training 60 min [100% (100 to 76%)] and 120 min [50% (100 to 50%)] per week; however, CRF non-response is completely absent in individuals exercising 180 min per week [0% (0 to 0%)]. When interpreting these results, it is important to notice that only the last group satisfied the current exercise guidelines and recommendations in terms of training volume (Garber et al. 2011). Rather than being related to inter-individual variability per se, the high prevalence of CRF non-response in the first two lower-dose groups at 6 weeks can be explained by an inadequate exercise dose at the group level. Indeed, when a training stimulus sufficiently powerful to produce meaningful improvements in the mean CRF is applied to the third group, ‘non-responders’ are no longer detected despite a similar (little to no) individual variability between the groups. Importantly, the lack of ‘non-responders’ in the group training 180 min per week supports the adequacy of the lower levels of the current exercise recommendations for developing CRF in all individuals. That being said, it is worth mentioning that Montero and Lundby included high intensity exercise as a part of their training programmes, which may have increased the overall training effect on CRF response compared to moderate intensity protocols (Garber et al. 2011). Montero and Lundby stated that the prevalence of CRF non-response to ET was completely eliminated after the second 6-week ET period, characterized by the addition of 2 × 60 min sessions per week in the participants who fail to demonstrate any improvement in CRF after the first 6 weeks of training. The authors concluded that CRF improvements may be elicited in ‘non-responders’ by using higher training stimuli. Nevertheless, the lack of a relevant control group and/or estimated individual differences in the true CRF response at 6 weeks limits the power of data interpretation and weakens the authors’ conclusions about exercise dose and CRF responsiveness. Furthermore, the choice of increasing training dose in all ‘non-responders’ without at least a comparator arm exercising at the same dose for the entire duration of the ET intervention makes it difficult to isolate the contribution of the training dose to the CRF response from the contribution of the time component. In conclusion, the findings from Montero and Lundby regarding the effects of exercise dose in CRF responsiveness should be interpreted with caution due to the absence of a comparator arm, which limits the control over the sources of variation that may affect the study results. When surrogate control data are used to verify the authors’ findings, little to no individual differences in the true CRF response are observed. Although the present commentary does not deny inter-individual variability in CRF response to ET, until solid evidence about ‘non-responders’ is provided by a robust design approach, a re-evaluation of the current exercise dose guidelines appears unwarranted. The author declares that the submitted work was carried out in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The author would like to express his profound gratitude to Prof. Greg Atkinson, Prof. Alan M. Batterham and Philip J. Williamson for their extraordinary support in the submission of the present article.