Progressive overload without progressing load? The effects of load or repetition progression on muscular adaptations

Participants

We recruited a convenience sample of 43 resistance-trained volunteers (27 men, 16 women) from a university population (height = 169.5 ± 10.5 cm; body mass = 77.2 ± 16.7 kg; body fat = 23.6% ± 9.5%; age = 23.1 ± 5.3 years; training experience = 3.8 ± 4.0 years). As previously described (Schoenfeld et al., 2019), this sample size was justified by an a priori precision analysis for the minimum detectable change at the 68% level (MDC_68%; i.e., 1SD, which is conservative in that it requires a larger sample to produce a narrow interval) for mid-thigh thickness (i.e., $\mathrm{S}\mathrm{E}\mathrm{M}\times \sqrt{2}=2.93\mathrm{m}\mathrm{m}$), such that the compatibility interval (CI) of the between-group effect would be approximately ±MDC_68%. Based on data from previous research (Schoenfeld et al., 2019), along with their sampling distributions, Monte Carlo simulation was used to generate 90% CI widths for 5,000 random samples of each sample size. To ensure a conservative estimate, as literature values may not be extrapolatable, the sum of each simulated sample size’s 90% CI’s mean and standard deviation was used, and the smallest sample that exceeded MDC_68% was chosen; that is, 18 participants per group (1:1 allocation ratio). Additional participants were recruited to account for the possibility of dropout.

To qualify for inclusion in the study, participants were required to be: (a) between the ages of 18–35 years; (b) free from existing cardiorespiratory or musculoskeletal disorders; (c) self-reported as free from consumption of anabolic steroids or any other legal or illegal agents known to increase muscle size currently and for the previous year; and, (d) considered as resistance-trained, defined as consistently lifting weights at least three times per week (on most weeks) for at least 1 year and regularly working the lower body muscles at least once per week. Participants were asked to refrain from the use of alleged muscle-building supplements throughout the course of the study period.

Participants were randomly assigned to one of two experimental, parallel groups: A group that aimed to increase load while keeping repetitions constant (LOAD: n = 22; 13 men, 9 women) or a group that aimed to increase repetitions while keeping load constant (REPS: n = 21; 14 men, seven women). Randomization into groups was carried out using block randomization, with two or four participants per block (randomized for each block), in R software (R Development Core Team, 2019). Approval for the study was obtained from the Lehman College Institutional Review Board (#2021–2132). Written informed consent was obtained from all participants prior to beginning the study. The methods for this study were preregistered prior to recruitment (https://osf.io/yvhcs).

Resistance training procedures

The RT protocol targeted the lower body musculature and consisted of four sets of the free-weight back squat, leg extension, straight-leg calf raise, and seated calf raise. Participants were prescribed the same upper body RT program with a traditional progression model to follow on alternate training days (without supervision from the researchers) and were instructed to refrain from performing any additional lower body RT for the duration of the study.

Prior to training, participants underwent 10RM testing to determine individual initial training loads for each exercise. The RM testing was consistent with recognized guidelines as established by the National Strength and Conditioning Association (Baechle & Earle, 2008). Training for both routines consisted of two weekly sessions performed on non-consecutive days for 8 weeks. The initial training routines (Session 1) for both groups attempted to maintain an 8–12 repetition maximum (RM) per set per exercise. In subsequent sessions, the LOAD group aimed to increase load while maintaining this target repetition range, whereas the REP group aimed to increase the number of repetitions performed per set while maintaining the initial load. As previously described (Schoenfeld et al., 2016), to help standardize the effort of the training protocols, we verbally encouraged participants to perform all sets to the point of momentary concentric muscular failure, herein defined as the inability to perform another concentric repetition while maintaining proper form. Participants were instructed to perform repetitions in a controlled fashion, with a concentric action of approximately 1 s and an eccentric action of approximately 2 s. Participants were afforded 2 min rest between sets. All routines were directly supervised by the research team to monitor proper performance of the respective routines and ensure participant safety.

Dietary adherence

Data were collected similar to as previously described (Schoenfeld et al., 2015). Specifically, to avoid potential dietary confounding of results, participants were advised to maintain their customary nutritional regimens. Dietary adherence was assessed by self-reported 5-day food records (including at least 1 weekend day) using MyFitnessPal.com (http://www.myfitnesspal.com), which has good relative validity for tracking energy and macronutrient intake (Teixeira et al., 2018). Nutritional data were collected twice during the study: 1 week before the first training session (i.e., baseline) and during the final week of the training protocol. Participants were instructed on how to properly record all food items and their respective portion sizes consumed for the designated period of interest. Each item of food was individually entered into the program, and the program provided relevant information as to total energy consumption, as well as the amount of energy derived from proteins, fats, and carbohydrates for each time-period analyzed.

Measurements

The following measurements were conducted pre- and post-study in a separate resting session. Participants reported to the lab having refrained from any exercise other than activities of daily living for at least 48 h prior to baseline testing and at least 48 h prior to testing at the conclusion of the study. Anthropometric and muscle thickness assessments were performed first in the session, followed by measures of muscle strength. Each strength assessment was separated by a half-hour recovery interval to ensure restoration of resources. Subjects were allowed to consume food ad libitum after anthropometric testing.

Blinding

To minimize the potential for bias, we incorporated two levels of blinding into the design and analysis of this study. First, the researcher who obtained the ultrasound measurements was blinded to group allocation. Second, the statistician performed blinded analyses; only after the analyses were completed did the research assistant unveil the correct dataset. We were not able to blind the strength-related tests, and thus cannot completely rule out the potential for bias in these measures.

Statistical analyses

Data were analyzed in R (version 4.2.0) (R Development Core Team, 2019). Neither baseline nor within-group inferential statistics were calculated, as baseline significance testing is inconsequential (Senn, 1994) and within-group outcomes are not the subject of our research question (Bland & Altman, 2011), although we descriptively present within-group changes to help contextualize our findings. The effect of group (LOAD vs REP) on each outcome variable was estimated using linear regression with pre-intervention score included as a nuisance parameter (Vickers & Altman, 2001). In addition, we included sex as a covariate since we stratified by sex. All outcomes were modeled using ordinary least squares, except for muscle endurance, which was modeled using Poisson regression with a log link function since the data are counts. Importantly, the log link function exponentiates the linear predictors such that the estimated effects are multiplicative (e.g., group A performed 1.5-times more repetitions than group B) rather than the additive (e.g., group A performed 10 more repetitions than group B). As such, the results estimated using the Poisson model are presented multiplicatively. Model residuals were qualitatively examined for structure and heteroscedasticity. We computed 90% CIs of the adjusted effects using the bias-corrected and accelerated bootstrap with 5,000 replicates. Rather than relying on traditional null hypothesis significance testing, which has been criticized for its use in the biomedical and social sciences (Amrhein, Greenland & McShane, 2019; McShane et al., 2017), we drew inferences via an estimation approach (Gardner & Altman, 1986). That is, we did not wish to binarize the presence of an effect or no effect; rather, we sought to draw inferences about the magnitude and uncertainty of the effects, whether they were close to zero or otherwise.

Secondary analyses were performed on nutrition data, which were analyzed similarly to the MT and strength data; that is, using multiple regression with group dummy-coded and pre-intervention nutrition scores and sex as covariates of no interest. The results of these secondary analyses are presented using mean adjusted effects and their standard errors.

Finally, we performed leave-one-out sensitivity analyses to assess the potential undue influence of any single participant. To do so, we removed each participant, one at a time, and re-estimated the intervention effect and its bootstrapped CIs without the removed participant. This was repeated for each participant in the sample. Participants with undue influence may bias the point estimate (e.g., if they inflate the effect, the point estimate will decrease when they are removed) and increase the variance (i.e., the effect estimate becomes more precise when they are removed).

Hypertrophy

The effect of REPS relative to LOAD on MT was negligible across all muscles except the RF, and with tight CIs. When summing the sites of the RF, REPS had an adjusted effect of 2.8 mm, and the data were compatible with values ranging from −0.5 to 5.8 mm (Fig. 2).

Strength

1RM_SQUAT’s point estimate slightly favored LOAD as compared to REPS, with an adjusted effect of 2.0 kg. However, the data were compatible with a wide spread of effects, ranging from 7.8 kg in favor of LOAD to 2.4 kg in favor of REPS (Fig. 3A).

Muscle endurance

REPS could perform an estimated 2% more repetitions in the leg extension exercise following the intervention as compared to LOAD. The data were compatible with 7% more repetitions for LOAD to 14% more repetitions for REPS (see Fig. 3B).

Countermovement jump

CMJ showed negligible changes in both LOAD and REPS. The data were compatible with a relatively small range of effects, ranging from 1.5 cm favoring LOAD to 1.7 cm favoring REPS (see Fig. 3C).

Body composition

Body fat showed small changes across the study period, with minimal between-group effects. LSLM estimates largely corroborated the MT measures, with a small point estimate (0.1 kg advantage to REPS) and inconsequential CI (0.1 kg in favor of LOAD to 0.3 kg in favor of REPS) (Fig. 4).

Dietary changes

Dietary changes were negligible across both LOAD and REPS, with minimal between-group effects (Table 2).

Leave-one-out sensitivity analyses

We performed leave-one-out sensitivity analyses for all outcomes to assess whether any single participant strongly influenced the estimated effects. While some individuals were slightly influential in some analyses (e.g., MG muscle thickness), none were sufficiently influential to shift our conclusions (Fig. S1).

Hypertrophy

Both groups gained appreciable muscle mass over the study period, with pooled mean increases ranging from 6.7% to 12.9% across measurement sites; similar increases were observed between conditions for a majority of MT measurements including the soleus, gastrocnemius, and all 3 VL sites (Table 1; Fig. 2). Overall, these results suggest that, from a hypertrophy standpoint, progressions can be made with load, repetitions, or conceivably a combination of the two over the course of an 8-week training block. The results are generally consistent with the body of literature, which shows similar hypertrophy across a wide spectrum of loading ranges (Schoenfeld et al., 2017).

The similar hypertrophic outcomes observed in our study are in contrast to previous work by Nóbrega et al. (2022), who performed a retrospective analysis using groups from two different studies (Barcelos et al., 2018; Nobrega et al., 2018). Contrary to our findings, their results showed that adjusting load elicited substantially greater increases in muscle cross-sectional area of the VL compared to the group that adjusted repetitions (16.0 ± 4.0% vs 7.9 ± 4.0%, respectively; ES = 2.03 [95% CI: 1.04–3.02]). Several differences between the studies may account for the discordant findings, with perhaps the most important being that Nóbrega et al. (2022) did not employ randomization since it was a retrospective analysis, hindering the ability to draw causal inferences.

Intriguingly, REPS showed a modest superiority for increases in summed MT of the RF (point estimate = 2.8 mm) with CIs ranging from negligible negative effects (−0.5 mm) to relatively large positive effects (5.8 mm); the effects were fairly consistent across proximal, mid and distal sites and were not sensitive to leaving any subject out (Fig. S1). Although the reasons for this finding are not entirely clear, it is possible that higher repetition squat training potentiated greater recruitment of the RF due to heightened accumulated fatigue in the vastii musculature, which henceforth would require greater contribution from the RF toward the end of a set. In contrast, it would likely not be as beneficial for the RF to contribute when squat loads are greater since it would counteract the hip extensors. This hypothesis is purely speculative as we currently lack evidentiary insights into the details of recruitment patterns and fatigue dynamics between the specific contexts. Alternatively, it is possible that the observation was simply due to random chance, especially since the other muscles seemed to have similar growth between conditions. Given the relatively modest magnitude of difference between conditions and that only the RF appeared to benefit from REPS relative to LOAD, this should be considered a preliminary finding that requires replication.

Strength

Increases in 1RM_SQUAT were ~20 kg on average across groups, but slightly favored LOAD, with a point estimate of 2 kg, or about a 10% greater increase in LOAD compared to REPS (Table 1; Fig. 3A). However, the CI encapsulated effects ranging from relatively modest negative effects to appreciable positive effects for LOAD (−2.4 and 7.8 kg, respectively), calling into question the meaningfulness of differences. The overall lack of consistent, appreciable differences between conditions is somewhat surprising given that the literature generally indicates a dose-response relationship between the magnitude of load and gains in dynamic muscular strength (Schoenfeld et al., 2017). Although speculative, it is possible that the relatively null findings between conditions can be explained by the fact that 1RM testing was conducted on a Smith machine while training was performed using the free-weight back squat. Consistent with the principle of specificity, there may be less overall carryover between a free-weight squat and a Smith machine squat, particularly given that both groups trained relatively far from their 1RM in this exercise. Hence, neither group conceivably would have developed the specific coordination and skill required to optimize 1RM squat performance on the Smith machine. To avoid inferential ambiguity and provide clarity to the matter, future investigations may benefit from incorporating multiple measures of strength (Buckner et al., 2017). From these data alone, it seems REPS may provide lifters with another option to increase their maximal strength.

Muscular endurance

Leg extension endurance increased by ~7 repetitions across both groups and we observed negligible difference between groups, with a CI containing values of no practical significance (Table 1; Fig. 3B). Previous research is mixed as to the effect of the training load on local muscular endurance with some studies showing a benefit to the use of lighter loads and others showing negligible differences across a wide range of loading conditions (Schoenfeld et al., 2021). Notably, studies that base testing on a fixed submaximal load, as was the case in our study, tend to show similar increases in muscular endurance between heavy and lighter loads (Jessee et al., 2018; Buckner et al., 2019), supporting the notion that REPS and LOAD are both viable options to increase muscular endurance.

Countermovement jump

CMJ performance neither improved nor differed between groups (Table 1; Fig. 3C). In athletic populations, the general observation is that as maximal strength increases relative to body mass, indices of explosive performance improve correspondingly (Nuzzo et al., 2008). However, while our population was trained, they were not necessarily athletic. Thus, the combination of a lack of appreciable differences in strength, the lack of specific jump training, and the given population may explain the lack of changes in either group.

It also should be noted that the emphasis of repetitions in both groups was to control the weight, particularly on the eccentric action, but also during the concentric action as well (cf., maximum concentric velocity). Thus, benefits related to highly dynamic strength, such as the stretch-shortening cycle, may not have been as pronounced. Qualitatively, it was also visibly apparent that many participants lacked the specific coordination for efficient performance of the CMJ, perhaps limiting their ability to exploit the effects of the interventions.

Limitations

Our study had several limitations that should be considered when attempting to draw inferences from the data. First, we tried to account for dietary practices via 5-day food diaries at the beginning and end of the study under the guidance of trained nutrition professionals. While food diaries are a well-accepted method for estimating nutritional consumption, evidence indicates widespread discrepancies between what is reported and what is actually consumed (Mertz et al., 1991). It therefore remains possible that despite our attempts to control nutritional intake, between-group differences in energy- and/or macronutrient-related factors may have confounded results. Although possible, body fat estimates via multifrequency BIA indicated similar changes between REPS and LOAD and results were within the standard error of measurement of the modality (Schoenfeld et al., 2018), suggesting a relative group-level maintenance of body fat over the study period; this indicates total energy intake was likely similar between conditions. Second, our sample comprised young resistance-trained men and women; thus, results cannot necessarily be generalized to other populations including adolescents, older individuals, and untrained populations. Third, training and testing were specific to the calves and quadriceps, thus inferences cannot be drawn for other lower body or upper body musculature. Fourth, despite our best efforts to verbally encourage all participants to train to momentary concentric failure, some volitionally stopped short of this directive during training. Participants in REPS appeared to have greater difficulty approaching true failure on average, likely due to greater metabolic acidosis and discomfort. That said, all subjects trained with a high level of effort throughout the study period, which has been shown to be sufficient for maximizing muscular adaptations (Grgic et al., 2021); thus, the degree of effort likely did not influence results between conditions. Future work may wish to obtain ratings of perceived effort and/or repetitions in reserve to directly evaluate subjective estimates of proximity to volitional failure. Fifth, although all subjects had previous RT experience (at least 1 year of consistent lower body RT), their experience varied across the cohort, and as a group, they would not be considered highly trained individuals. Thus, the sample would be more reflective of the average regular gym-goer and results therefore cannot necessarily be generalized to elite athletes and high-level bodybuilders. Moreover, previous squat experience was not a requirement of the study and many of the subjects did not regularly include squats in their training routines. Thus, some of the gains in dynamic strength conceivably can be attributed to initial neuromuscular improvements and may not reflect what would be achieved by those who squat on a regular basis. Finally, our findings are specific to a relatively short training block (8 weeks); it remains questionable as to whether and how results might be influenced by continuing the intervention over a longer timeframe. That said, many individuals plan their training programs in mesocycles lasting several weeks to months, making the results highly practical from a prescription standpoint.