In their systematic review and meta-analysis, Edwards et al. (1) aimed to ‘directly compare’ the efficacy of isometric exercise and high-intensity interval training (HIIT) for the management of resting blood pressure. They included 38 randomised controlled trials (18 for isometric, 20 for HIIT) in their pairwise meta-analysis and concluded that isometric exercise appears to be superior to HIIT for improving both systolic blood pressure (mean difference between exercise types = 5.29 mmHg, 95% confidence interval 3.97 to 6.61) and diastolic blood pressure (mean difference between exercise types = 3.25 mmHg, 95% confidence interval 2.53 to 3.96). We were interested in these marked differences because they contrast previous findings (2) and, if correct, may necessitate important changes to guidelines. However, in further examining the article, we identified some issues that we believe require attention as they may invalidate the results and are relevant to readers of this journal.

None of the included trials in this review appear to contain both isometric and HIIT interventions; therefore, the authors are unable to ‘directly compare’ the interventions. Instead, by analysing the differences between isometric and HIIT subgroups in the meta-analysis, Edwards et al. (1) are making an inference based on the indirect effect, which assumes that the differences between exercise types can be inferred via a common comparator (in this case, the control group) (3). This is, in effect, the simplest version of a network meta-analysis. However, Edwards et al. (1) only analyse the intervention group’s change from baseline from each study (as shown in Figures 2 & 3 of their article) and report separate meta-analyses of the control change from baseline. This method is not correct. Analysing a single group’s change from baseline in a trial, or a meta-analysis, as opposed to the between-group effect (the change/endpoint in the intervention group minus the change/endpoint in the control group), is misleading because several statistical phenomena like regression to the mean and natural history are not accounted for (4, 5). Regression to the mean in hypertension has been shown to be, at times, as large as the effect of treatment itself (6). In using change from baseline values without comparison to control, the data presented in the meta-analyses are now equivalent to cohort studies and causal inferences are unable to be made (7). By extension, attempting to estimate the indirect effect using change from baseline scores, rather than between-group differences, is prone to bias and is not recommended (3).

We believe that the results and conclusions presented by Edwards et al. (1) may be biased due to the use of change scores, rather than between-group differences, for their comparison between isometric exercise and HIIT. Therefore, we urge caution for anyone who may consider altering their clinical practice until the results are updated.

Correctly re-estimating the indirect effect of isometric exercise versus HIIT via the control group may provide valuable information to clinicians; the network meta-analysis by Naci et al. (2) found similar effects for endurance exercise and isometric exercise, but did not further explore the potential unique effects of HIIT. However, transitivity – the assumption that it is appropriate to learn about isometric exercise versus HIIT via a common control group – requires strong consideration to ensure the validity of the indirect effect (8). It requires that effect modifiers be similarly distributed across the different comparisons. If intervention duration is thought to modify the effect of exercise, it is important to confirm both studies of isometric exercise versus control (8.2 weeks) and studies of HIIT versus control (8.1 weeks) are delivered for a similar duration in the included studies, which appears to be true. Another effect modifier might be the control group intervention/effect, which does not appear to be similarly distributed across the two comparisons. All isometric exercise studies appear to use a minimal/no exercise control, but several HIIT studies use a moderate-intensity continuous training comparator. Given a no exercise control will have a different effect to a moderate-intensity continuous training control, and these are not similarly distributed across the two exercise types, this will threaten the assumption of transitivity and introduce bias in the indirect effect (8). Therefore, to ensure transitivity holds, it may only be appropriate to include HIIT studies that use a no exercise control. Several other clinical and methodological characteristics may also need to be considered, adding to the complexity of indirect effect estimation, especially as the network of interventions increases. This is best done a priori with a study protocol, utilising a team that contains clinical and methodological/statistical expertise.

References
1. Edwards J, De Caux A, Donaldson J, Wiles J, O'Driscoll J. Isometric exercise versus high-intensity interval training for the management of blood pressure: a systematic review and meta-analysis. British Journal of Sports Medicine. 2021:bjsports-2021-104642.
2. Naci H, Salcher-Konrad M, Dias S, Blum MR, Sahoo SA, Nunan D, et al. How does exercise treatment compare with antihypertensive medications? A network meta-analysis of 391 randomised controlled trials assessing exercise and medication effects on systolic blood pressure. British Journal of Sports Medicine. 2019;53(14):859.
3. Bucher HC, Guyatt GH, Griffith LE, Walter SD. The results of direct and indirect treatment comparisons in meta-analysis of randomized controlled trials. Journal of clinical epidemiology. 1997;50(6):683-91.
4. Bland JM, Altman DG. Comparisons against baseline within randomised groups are often used and can be highly misleading. Trials. 2011;12(1):264.
5. Bland JM, Altman DG. Comparisons within randomised groups can be very misleading. BMJ. 2011;342:d561.
6. Salam A, Atkins E, Sundström J, Hirakawa Y, Ettehad D, Emdin C, et al. Effects of blood pressure lowering on cardiovascular events, in the context of regression to the mean: a systematic review of randomized trials. J Hypertens. 2019;37(1):16-23.
7. Tennant PWG, Arnold KF, Ellison GTH, Gilthorpe MS. Analyses of ‘change scores’ do not estimate causal effects in observational data. International Journal of Epidemiology. 2021:dyab050.
8. Salanti G. Indirect and mixed-treatment comparison, network, or multiple-treatments meta-analysis: many names, many benefits, many concerns for the next generation evidence synthesis tool. Research synthesis methods. 2012;3(2):80-97.

Some decades ago, Tom Beauchamp and James Childress proposed four principles for biomedical ethics (i.e., respect for autonomy, non-maleficence, beneficence, and justice). They postulated that such an approach, called principlism, could be applied universally. 1
The relationship between regular physical activity and the prevention of some diseases has been disseminated widely in scientific literature. 2 Pugh et al. 3 highlighted the importance of broadening the debate on this relationship and not relying solely on the principle of beneficence. It would also be necessary for the authors to acknowledge practically the principle of non-maleficence. Within this perspective, Pugh et al. 3 commented on the risk of damage, possibly even death, from vigorous physical exercise for the practitioners (whom they called patients).
It is worth noting that the principles of non-maleficence and beneficence have played a central role in the history of biomedical ethics. However, respect for autonomy and justice seem to be often neglected. 1 Even though we may agree on some points with Pugh et al. 3, it is imperative to bring other bioethical principles to the debate.
Thus, we would like to contribute, although briefly, to the debate on the topic addressed by Pugh et al. 3 and suggest that the focus on non-maleficence should be broadened. In addition, we highlight the indispensable focus on the principle of justice and autonomy.
Regarding the expansion of the non-maleficence principle, it has been shown that occupational physical activity can be harmful to one’s health, extending the physical activity paradox beyond the harmful effects of intensity. 4 This finding contradicts the idea that physical activity in its various dimensions is beneficial.
Another point that deserves wide attention concerns inequalities and how they impact various aspects of individuals’ relationships with physical exercise. Various studies have indicated that the vulnerability of certain social groups—such as black people, people of low socioeconomic status, and women—interferes with the performance of leisure-time physical activities or even other behaviours considered as healthy 5–6. We view this as an affront to the principle of justice. Moreover, some researchers have disregarded such inequalities and reinforced blaming the individual for his or her ‘failure’ to engage in behaviour considered appropriate, even arguing that it leads to enormous economic costs to society. 7 This position clearly aligns with neoliberal precepts since it places the responsibility for taking care of oneself on the individual, thus removing it from the State. Why, in general, are women less involved with leisure-time physical activities than men? Why is the same true for black people? Why is physical inactivity lower among black women? How do violence, mobility, and urban planning impact physical activity? Are more impoverished people equally likely to have access to leisure-time physical activity? The possibility of exercising does not seem equal for everyone.
Although we recognise that the focus of Pugh et al. 3 was correct medically, we consider that inequities interfere decisively in the possibility that any isolated measure effectively produces some benefit and/or avoids some harm. The principles of beneficence and non-maleficence may not be guaranteed if effective measures to combat inequities are not adopted. Regarding the context of peripheral countries, it is essential to consider adequate bioethical theories, rather than principlism. 8
In particular, socially excluded people—because of their unequal position in society—may sometimes be unable to make autonomous decisions. The capacity to make decisions is in itself incomplete, relative, non-global, and determined by a series of changing contextual aspects.
We also consider it necessary to question the supposed universality of bioethical principles since they are often presented within a colonialist perspective. This posture is reflected in individuals’ concepts of autonomy, which we believe should be decolonised. Decolonisation stimulates the debate about the knowledge and experiences of the community to reveal their richness. This process is in line with a vision of social justice: achieving health equity considering individual and collective empowerment. Therefore, taking advantage of the knowledge and perspectives of social groups to contribute to the development of specific interventions seems essential for yielding effective results.
Finally, we highlight the insistence on considering the relationship between physical exercise and health as a problem that is only biomedical. The resulting discourses reinforce the idea of the medicalisation of exercise. Thus, the practitioner becomes a patient and the pleasure for the practice is always neglected.

References
1. Beauchamp TL, Childress JF. Principles of biomedical ethics. 8 ed. New York: Oxford University Press, 2019.
2. Booth FW, Roberts CK, Laye MJ. Lack of Exercise Is a Major Cause of Chronic Diseases. Compr Physiol 2012; 2:1143-1211.
3. Pugh J, Pugh C, Savulescu, J. Exercise prescription and the doctor’s duty of non-maleficence. Br J Sports Med 2017; 51:1555-1556.
4. Holtermann A, Schnohr P, Nordestgaard BG, Jacob Louis Marott JL. The physical activity paradox in cardiovascular disease and all-cause mortality: the contemporary Copenhagen General Population Study with 104 046 adults. Eur Heart J 2021; 42:1499-1511.
5. Brodersen NH, Steptoe A, Boniface DR, Wardle J. Trends in physical activity and sedentary behaviour in adolescence: ethnic and socioeconomic differences. Br J Sports Med 2007; 41:140-144.
6. Abichahine H, Veenstra G. Inter-categorical intersectionality and leisure-based physical activity in Canada. Health Promot Int 2017; 32:691-701.
7. Allender S, Foster C, Scarborough P, Rayner M. The burden of physical activity-related ill health in the UK. J Epidemiol Community Health 2007; 61:344-348.
8. Garrafa V, Porto D. Intervention bioethics: a proposal for peripheral countries in a context of power and Injustice. Bioethics 2003; 17: 399-416.