The design of the study

The study compared three groups performing either 1 set (1-SET), 3 sets (3-SET) or 5 sets (5-SET) of the same exercises, three times per week (full body routine). The characteristics of the training protocol were:

• Seven exercises per session: flat barbell bench press, barbell military press, wide grip lateral pulldown, seated cable row, barbell back squat, machine leg press, and unilateral machine leg extension.
• Each set was performed for 8–12 repetitions to momentary concentric failure, with 90 seconds rest between sets.

Furthermore:

“The load was adjusted for each exercise as needed on successive sets to ensure that subjects achieved momentary failure in the target repetition range. Thus, if a subject completed more than 12 repetitions to momentary failure in a given set, the load was increased based on the supervising researcher‟s assessment of what would be required to reach momentary failure in the desired loading range; if less than 8 repetitions were accomplished, the load was similarly decreased.”

Rest periods

Probably one of the most important characteristics of the training used was the short rest between sets. According to the author, this was due to practical limitations of using a very high volume routine with longer rest periods as it would take a long time to complete (which begs the question, if its of limited pratical value then why such design?).

Schoenfeld himself has shown that longer rest periods between sets (~3 minutes) appears to be better for muscle hypertrophy and strength than short periods (~1 minute). Short inter-set rests (1 minute) seem to blunt resistance exercise-induced myofibrillar muscle protein synthesis (MPS) (up to 4h after the session) and anabolic intracellular signalling, compared to longer (5 minutes) rests when training at 75% of 1RM. However, there appears to be no effect of short (30 seconds) or long (2.5 minutes) rest periods when training with low loads (40% 1RM) in untrained subjects. If the same is observed for trained subjects, the apparent discrepancy might be explained by the fact that short rests do not allow for enough recovery between heavy sets, so either the load/weight (if the repetitions are held constant) or the number of repetitions (if the weight is held constant) has to be reduced. In either case, the stimulus for muscle growth (load) is reduced.

Taking this into consideration, by design, the training program would favor the 5-SET condition for the simple reason that you can compensate less effective sets with more work than on lower volume conditions. In other words, when performing a suboptimal training program, you might need more sets to compensate.

Exercise selection, how sets are counted and comparison with Ostrowski et al. 1997

Another important point is the exercise selection. MT values were measured, in addition to quadriceps muscles, for the biceps and triceps. However, there were no “direct” exercises targeting the triceps or biceps; triceps were hit by the pushing movements and biceps by the pulling movements. Because of this reason (and likely because their involvement in these exercises might increase as the number of sets increase due to fatigue) more work is needed to see any effect. This agrees with the fact that the group performing 1 set saw virtually no growth in both biceps and triceps: they were effectively doing no work for either muscle.

This fact is related to another common criticism: the way in which sets were counted. According to their previous meta-analysis, the authors counted as “1 set” those exercises in which biceps and triceps were hit, but not as a direct exercise (ie. pushing movements as a tricep exercise, pulling movements as a bicep exercise). Whereas I personally (and most) don’t “count” these compound movements as an exercise for biceps or triceps, some people count them as half a set. Thus, using this more logical criteria, the 5-SET group performed 15 sets for triceps and biceps, whereas the 3-SET performed 9 sets and the 1-SET, 3 sets.

There is only one other study that can be directly used for comparing the results of this study, namely, Ostrowski et al., 1997. Using the same rationale for counting the number of sets (half for pressing movements except close-grip bench press, one per direct triceps exercise), subjects in the high volume group in Ostrowski et al. performed 20 sets for triceps. It is importat to note that this comparison assumes that one set of a push compound movement that uses the triceps promotes the same stimulus as one set that directly targets the triceps, which seems to be unlikely. Accordingly, there is a plateau in the magnitude of change in Ostrowski at 10 sets: doubling the amount of volume did not induce any significant gains (0.1%). Therefore, a threshold appears to occur already at moderate volume (10 or 14 sets, depending on how you count) when performing direct tricep work. If you don’t perform any tricep work, then you have to perform much higher volumes to get any significant stimulus. As the number of sets increase and the primary muscles become more fatigued, the last sets might also provide a more direct work to both triceps and biceps. So this means that more volume is not inherently better for these muscle groups; rather, that under these circumstances, their involvement increases as the sets progress. Thus, more sets increase the work performed by the triceps in the 5-SET compared to the 3- and 1-SET groups.

Perhaps a more direct comparison would be that using the values for the quadriceps: in Schoenfeld et al., the 5-SET group performed 45 sets of direct leg training, whereas the high volume group in Ostrowski et al. performed 12 sets. Increases in MT (Schoenfeld) and CSA (Ostrowski) were 12–13% (depending on which quad site) and 13.1%, respectively. Thus, subjects in Schoenfeld et al had to perform almost 4 times more sets to promote a similar response as the subjects in Ostrowski et al. While the authors only mention that “the reason for this discrepancy remains unclear”, the answer might lie in the different training regimens.

The program used by Ostrowski et al. involved (all sets to failure):

• 12 reps/set for the first 4 weeks.
• 7 reps/set for weeks 5 to 7.
• 9 reps/set for the final 3 weeks.
• Rest was 3 minutes between sets.

(Interestingly, they mention that “the high volume group comprised the training for the active control group, since questionnaire responses had established that this volume was typically used by most subjects prior to the study”. Unfortunately, the usual training for the subjects in Schoenfeld et al., as well as the baseline characteristics per group, are unknown.)

As shown above, one of the main differences is that the number of repetitions were held constant, and there were three weeks in which repetitions were used with a greater load (used in the following discussion as the weight in the bar) (7 reps), than in the Schoenfeld et al. study, in which a repetition range was used. Therefore, for ~1/3rd of the study period, subjects trained with heavier weights. This further argues that the training program performed by subjects in Schoenfeld et al. was suboptimal, as implied by the design.

A more subjective, albeit real-world speculation, similarly supports the idea that subjects in Schoenfeld et al. were not training to “true” failure. Anyone who has trained to failure would agree that performing 5 sets of squats to “true” failure, with only 90 seconds rest in between would necessitate a big reduction in load after the first set (or to not reach true failure in those sets). On top of this, they also had to perform 5 sets of leg press and 5 sets of leg extension. This in addition to the other exercises, three times per week. It is very hard for me to believe that someone can endure such a training, to real failure, with 90 seconds rest and do it in ~68 minutes. If anything, it shows that such set up might be suboptimal as nearly 4 times of work is needed compared to elicit a similar hypertrophic response as the subjects in Ostrowski et al., which performed only 12 sets. Further, the difference in the quality of the sets (and training intervention) for each study is exemplified by the fact that 9 sets in the Schoenfeld study only increased rectus femoris (RF) MT by 3.3%, whereas 3 sets were sufficient in Ostrowski for inducing a 6.7% increase; 3 sets in Ostrowski et al. promoted a greater increase in RF MT than 27 sets in Schoenfeld et al. (5.3%)! All of this, of course, assuming that by the time MT was measured, acute swelling was already negligible.

Is training volume load the answer?

There is another variable that is not mentioned in either study: training volume load (reps x weight). I’m not sure why, given that some people argue that it is not necessarily the number of sets that define volume as the driver of hypertrophy, but the total volume load (or total workload). Thus, the number of sets, by increasing the training volume load, could then promote more hypertrophy than lower volume training.

However, there is evidence, in trained subjects (minimum of 2 years of training), that training with a higher load (“intensity”) leads to greater hypertrophy despite achieving a much lower training volume load (Mangine et al., 2015). In this study, subjects were randomized to two groups:

• 4 sets of 10–12 repetitions with ~70% of 1RM, with 1-min rest intervals (VOL)
• 4 sets of 3–5 repetitions with ~90% of 1RM, with 3-min rest intervals (INT)

Importantly, in this study, the authors standardized the groups with a pre-intervention training protocol, so all subjects started from the same position training-wise.

A big limitation of this study is that the VOL group used a 1 minute rest between sets, which has been shown to be suboptimal. Nevertheless, this study contradicts the thesis that volume (or volume load) is the primary driver of hypertrophy, as better hypertrophy results (by DXA-measured lean mass and ultrasound, albeit not all statistically significant) were observed for the group that performed the lower volume load:

Even if you argue that results are not different (due to lack of statistical significance), this still means that equal muscular gains can be achieved with approximately half the amount of total volume load when other training variables are manipulated. Interestingly, of all the muscles analyzed, the greatest difference was observed for the chest, which is not measured in any of the Schoenfeld studies that I am aware of.

Other study from Schoenfeld et al., which found that a higher intensity (as defined by higher percentages of 1RM) routine is worse than higher volume, lower intensity routine, used a load that is arguably too low in repetitions to promote sufficient hypertrophic adaptations (2–4 reps). Moreover, in this study, rest was equal in both conditions (2 minutes) and target repetitions fixed. As 2 minutes might be too short to recuperate from a ~3RM set to failure (compared to 8–12), and repetitions had to be kept at 2–4, subsequent sets after the first 3RM set would provide very little stimulus, as the weight would have to be reduced without an increase in repetitions. And again, as with the current study, biceps and triceps were measured as a proxy for overall muscle growth despite no direct exercises being performed for these muscles.

The other two recent studies

Very closely to the publication of Schoenfeld et al., 2018, two more studies addressing the effects of volume on muscle hypertrophy in trained subjects were published. One was the previously mentioned Haun et al., 2018. This study showed that with increasing volumes up to 32 sets per week, most of the gains were achieved at ~20 sets per week. As mentioned above, MT and CSA values increased only marginally, and lean body mass increases (assessed by DXA) at the higher volume portion of the study were largely driven by extracellular water accumulation (leaving only 200g of “gains” when increasing volume from 20 to 32 sets).

Of note, training was performed at 60% 1RM, with RIR of ~4 in the final week (so very far from failure). Total training volume load increased 3.2 times from week 1 to week 6. When comparing mid (3 weeks, 20 sets) to the final week (6 weeks, 32 sets), almost doubling the training volume load (1.64 times) produced 200g grams of lean body mass gains, as well as inconsistent differences in VL MT (+5%), biceps MT (-2.6%), but a 12.5% increase in average fiber CSA. Interestingly, for several of the outcomes, there was a regression from increasing sets from 10 per week to 20 per week.

It is important to mention that, similar to Schoenfeld et al., 2018, there was no direct bicep work in the training protocol. There was also no group performing a different training set up as the study wanted to address differences between supplementation regimens under very high volume conditions, so there is no training control.

Overall, these results show something similar to what Schoenfeld et al., 2018 suggests: if training with a suboptimal set up, more sets are needed to elicit any hypertrophic response, if any.

In this study, trained subjects performed either 9 (LOW), 18 (MID) or 27 (HIGH) sets of biceps exercises comprising one direct (seated supine biceps curl) and two compound (supine bent over row and supine grip pulldown) exercises. In contrast to Haun et al. and Schoenfeld et al., the compound movements involved more biceps work as they were performed with a supine grip. Using the counting rationale as previously, these would correspond to 6 sets for LOW, 12 sets for MID, and 15 sets for HIGH. However, supination in the compounds makes more reasonable to count these compounds as 1 set instead of half a set.

Training was performed at ~75% of 1RM, 2 RIR (close to failure), with a slower eccentric tempo than the other two studies (3 seconds) and resting for 3 minutes between sets. The big advantage of this study is that it trained the biceps directly and measured its response upon different volumes.

Despite a much greater total volume load performed by the HIGH group, MT increased the most in the MID group, albeit the difference was not statistically significant between groups. However, the effect size was much larger for MID compared to the other groups. Thus, the “dose-response” was observed when going from 9 to 18 sets, which regressed when performing 27 sets.

The main limitation of this study is that participants were allowed to train outside of the study (but were not allowed to perform any exercise involving the elbow flexors). Moreover, the MID and HIGH trained two times per week versus only one in the LOW group, which means that both the volume (sets) and frequency was increased in the MID and HIGH groups.

Nevertheless, from the three recent studies, this is the only one in which bicep MT was assessed after performing a training protocol that included direct bicep work, thus evaluating directly the effect of volume on muscle growth.

Bottomline

There are several caveats to the interpretation that there is a graded dose-response of resistance training volume and muscle hypertrophy. When comparing studies, several conclusions arise:

• There might still be residual acute muscle swelling after a high volume routine at 48–72h, which could confound the results of measurements done in this time frame and the true magnitude of the response to high volume training protocols.
• Schoenfeld et al. always use a full body training program that does not include any direct exercise for biceps and triceps. However, these are the only upper body muscle groups measured and used as a proxy for upper body skeletal muscle hypertrophy. There is virtually no data that I am aware of measuring other upper body muscle groups and their relationship with training volume.
• Studies that include direct work for biceps or triceps show that there is a plateau in the apparent dose-response relationship between sets at a moderate intensity (as % of 1RM) and muscle growth between~10–20 sets. This cannot be extrapolated to other muscles.
• There is no consistent dose-response relationship between training volume load and muscle hypertrophy.
• The last Schoenfeld study has limited practical applicability (nearly all hypertrophy programs include exercises that train biceps and triceps directly) and shows that more work is needed when the routines utilized are suboptimal for muscle growth.
• Comparison of the changes observed in Ostrowski et al., 1997 and Schoenfeld et al., 2018 show that the latter used a very inefficient training protocol, which needs higher volumes to elicit a hypertrophic response.
• The above shows that the quality of the training protocol is more important than the quantity. When the latter is lower, the amount of sets needed is higher to get any response.
• Different training variables (tempo, time under tension, intensity/% of 1RM, etc.) could modify the volume-hypertrophy response.

In conclusion, based on the available evidence, a direct, dose-response relationship between volume and muscle hypertrophy is unwarranted. The number of sets required to elicit an optimal hypertrophic response depends on the characteristics of the training program; increasingly greater gains are not necessarily achieved with higher training volumes.