For decades, there has been a consensus amongst scientists and practitioners that one of the ways people become stronger after resistance training is that their muscles become bigger. This increase in muscle size is termed muscle hypertrophy.
Over the past 3 years, however, Jeremy Loenneke’s research group at the University of Mississippi has postulated hypertrophy from resistance training makes little or no contribution to the increase in strength that occurs after training. In fact, Loenneke’s group has said the simultaneous increases in muscle size and strength from training might be “completely separate phenomena” (Buckner et al. 2016; Buckner et al. 2017; Dankel et al. 2018; Mattocks et al. 2017).
THE “COMPLETELY SEPARATE” HYPOTHESIS
Loenneke’s group contends there is insufficient evidence to conclude training-induced increases in muscle size cause improvements in muscle strength. The group provides various rationales for their hypothesis, but the fundamental one is that research on resistance training and hypertrophy is often correlational in nature. As the statistical adage goes: “Correlation is not causation.”
Loenneke’s group has been particularly critical of research published by Jonathan Folland’s group at Loughborough University. Folland’s group reported moderately-sized statistical correlations between changes in muscle size and strength after 12 weeks of resistance training and concluded hypertrophy “contributes to” increased muscle strength (Erskine et al. 2014; Balshaw et al. 2017). In 2017, Loenneke’s group wrote a letter denouncing one of Folland’s studies, stating it was “another study not designed to answer the question” (Buckner et al. 2017). Folland’s team fired back with a reply stating Loenneke’s “completely separate” hypothesis “appears to deny knowledge/appreciation of theory and evidence” and is “untenable” (Balshaw et al. 2017).
Welcome to the debate on muscle hypertrophy.
Many researchers, including me, are skeptical of the “completely separate” hypothesis. Yet, Loenneke’s group has valid points, namely that correlational designs are insufficient for determining causation. The group has proposed the best way to determine if training-induced increases in muscle size cause or contribute to increased muscle strength is to design a study that produces “differential effects on muscle size based on group membership (i.e., one group increases muscle size and one does not) and observe how this impacts muscle strength” (Dankel et al. 2018)). Loenneke’s group recently employed this study design and found that a group of participants who practiced a 1 repetition-maximum strength test for 8 weeks did not experience hypertrophy but they did improve their muscle strength (Mattocks et al. 2017). On the other hand, a second group who performed high-volume resistance training experienced hypertrophy but only increased their strength by the same amount as the other group. Because both groups increased their muscle strength by the same amount, but only the high-volume group experienced hypertrophy, Loenneke’s team concluded hypertrophy likely does not contribute to increased strength from resistance training, thus supporting their “completely separate” hypothesis.
Debate over, right?
In a recent commentary, I, along with NeuRA colleagues Dr. Harrison Finn and Prof. Rob Herbert explain that the design used by Loenneke’s team is also inadequate for resolving the hypertrophy debate, and their conclusion that hypertrophy does not contribute to increased muscle strength was unwarranted.
In the study by Loenneke’s team, the group that practiced the strength test did not experience hypertrophy. Thus, hypertrophy could not have contributed to their strength increase. However, the hypertrophy that occurred in the high-volume group could have contributed to that group’s strength increase. As we state in our paper: “[J]ust because strength can increase without hypertrophy does not mean that hypertrophy, when it occurs, does not contribute to improvements in muscle strength.” In other words, the two groups in Loenneke’s study could have increased muscle strength to the same degree by a differential weighting of mechanisms. The strength increase in the first group could have been due solely to neural adaptations (e.g., increased motoneuron recruitment or firing rates), whereas the strength gain in the high-volume group could have been due to a combination of neural and muscle mechanisms (e.g. muscle size, pennation angle).
CAUSAL MEDIATION ANALYSIS
In the debate on muscle hypertrophy, scientists want to know the degree to which hypertrophy mediates the relationship between a training intervention and an increase in muscle strength. The mediating role of hypertrophy has been tricky to figure out because of the potential for confounding of the mediator-outcome relationship.
Confounders can be thought of as variables that have effects on both the mediator and the outcome. When not accounted for, confounders make it difficult to determine if a proposed mediator causes the observed change in the outcome. In randomised-controlled trials, the randomisation procedure controls for confounding between the intervention (resistance training) and the outcome (strength) and between the intervention and the mediator (muscle size/hypertrophy). However, randomisation does not control for confounding between the mediator and the outcome (i.e., the hypertrophy-strength relationship). Thus, in the study by Loenneke’s team, where the unit of randomisation was training volume rather than hypertrophy, potential for confounding of the hypertrophy-strength relationship was present, just as it was in the studies from Folland’s group.
In our paper, we suggest causal mediation analysis is the approach best-suited to resolve the debate on muscle hypertrophy, as it can help control for confounding of the hypertrophy-strength relationship. Causal mediation analysis advances earlier approaches to mediation, and readers interested in learning the history of this technique, its philosophical underpinnings, and why it is an advancement over previous approaches, are encouraged to read Judea Pearl’s recent book, “The Book of Why: The New Science of Cause and Effect.”
One of the merits of causal mediation analysis is that it employs causal maps to help researchers control for confounding. These maps are called directed acyclic graphs (DAGs), and they illustrate proposed causal pathways between a training intervention and an outcome. DAGs are developed using rigorous rules (Shrier and Platt 2008). These rules force researchers to first consider all potential confounders and mediators in a study’s design and then make the assumptions of the entire causal model explicit in graph format.
Figure 1 above is a rudimentary DAG associated with the debate on muscle hypertrophy. In the figure, the line with the question mark is the effect that both Folland’s and Loenneke’s groups have been debating. In our opinion, causal mediation analysis is the best way to determine the size of this effect. In our paper, we recommend a study in which (a) participants are randomised to a training or control group; (b) muscle size and strength are measured before and after the intervention; (c) all potential confounders of the hypertrophy-strength relationship are measured; and (d) software for causal mediation analysis (see our paper for examples) is used to estimate the size of the mediating effect of hypertrophy on strength.
The degree to which muscle hypertrophy from resistance training contributes to increased muscle strength is still unclear, but both Folland’s and Loenneke’s groups have made positive contributions to our knowledge in this area. The moderate-sized correlations between changes in muscle size and strength after resistance training in studies from Folland’s group indicate muscle size should be included as a potential mediator in causal models that seek to explain how people become stronger with resistance training (Erskine et al. 2014; Balshaw et al. 2017). Loenneke’s group has pointed out that more rigorous methodologies are needed to determine causality, and results from their study indicate that muscle size should not be the only mediator tested in causal models (Mattocks et al. 2017). In our paper, we acknowledge these contributions but suggest causal mediation analysis is the best way to determine if training-induced increases in muscle sizes cause increased muscle strength (Nuzzo et al. In Press).
Moving forward, resistance training researchers will likely need to collaborate with others to resolve the debate on hypertrophy. Free software is available to create DAGs and assist with causal mediation analysis (http://www.dagitty.net), but resistance training researchers would benefit from collaborations with those who are skilled in mediation analysis and DAG development. Moreover, sample sizes in resistance training studies are often small and mediation analysis might require more data points. International collaborations between different research centers would help overcome this issue, but I do not expect there to be any phone calls between Loughborough University and the University of Mississippi any time soon.
Nuzzo JL, Finn HT, Herbert RT. Causal mediation analysis could resolve whether training-induced increases in muscle strength are mediated by muscle hypertrophy. Sports Med. In Press.If you cannot access the paper, please click here to request a copy.
Balshaw TG, Massey GJ, Maden-Wilkinson TM, Morales-Artacho AJ, McKeown A, Appleby CL, et al. Changes in agonist neural drive, hypertrophy and pre-training strength all contribute to the individual strength gains after resistance training. Eur J Appl Physiol. 2017;117(4):631–40.
Balshaw TG, Massey CD, Maden-Wilkinson TM, Folland JP. Muscle size and strength: debunking the “completely separate phenomena” suggestion. Eur J Appl Physiol. 2017;117(6):1275–6.
Buckner SL, Dankel SJ, Mattocks KT, Jessee MB, Mouser JG, Counts BR, et al. The problem of muscle hypertrophy: revisited. Muscle Nerve. 2016;54(6):1012–4.
Buckner SL, Dankel SJ, Mattocks KT, Jessee MB, Mouser JG, Loenneke JP. Muscle size and strength: another study not designed to answer the question. Eur J Appl Physiol. 2017;117(6):1273–4.
Dankel SJ, Buckner SL, Jessee MB, Mouser JG, Mattocks KT, Abe T, et al. Correlations do not show cause and effect: not even for changes in muscle size and strength. Sports Med. 2018;48(1):1–6.
Erskine RM, Fletcher G, Folland JP. The contribution of muscle hypertrophy to strength changes following resistance training. Eur J Appl Physiol. 2014;114(6):1239–49.
Mattocks KT, Buckner SL, Jessee MB, Dankel SJ, Mouser JG, Loenneke JP. Practicing the test produces strength equivalent to higher volume training. Med Sci Sports Exerc. 2017;49(9):1945–54.
Pearl J, Mackenzie D. The book of why: the new science of cause and effect. New York: Basic Books; 2018.
Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Med Res Methodol. 2008;8:70.
Dr. Jim Nuzzo (@JamesLNuzzo) is a Postdoctoral Fellow at NeuRA. He studies how strength training alters the neural connection between the brain and muscles. To read Jim’s other blogs, click here.