Velocity-based resistance training: do women need greater velocity loss to maximize adaptations?

Twenty-four healthy young men (26.4 ± 3.9 years, 181.1 ± 5.4 cm, 81.9 ± 11.3 kg) and twenty-five women (25.5 ± 3.8 years, 166.6 ± 7.2 cm, 60.7 ± 6.1 kg) were pair-matched to four different training groups: men training until 20% velocity loss (VL20M, n = 12) or 40% velocity loss (VL40M, n = 12), and women training until 20% velocity loss (VL20W, n = 13) or 40% velocity loss (VL40W, n = 12) within each set. Subjects were allocated based upon their sum of 1-RM in back squat and bench press following a reverse counterbalancing (i.e., ABBA) sequence, after separation of men and women. All subjects were motivated, free of any illness or injuries and had at least one-year experience of systematic resistance training with the back squat and bench press exercises commonly used as part of their recreational training program. Before inclusion, the subject underwent a medical evaluation process, which included resting ECG scan and medical history questionnaire that were examined by a physician. All included subjects signed informed (written and orally) consent forms, and the study was approved by the local ethics committee.

During the intervention, 3 women (two from VL20W and one from VL40W) dropped out because of injuries (not obtained during the study) or personal reasons. In addition, one man (in VL40M) was not able to complete the squat training program due to injury sustained during the study’s training intervention, although he completed bench press training and testing. The final number of subjects in each group was: VL20M: n = 12, VL40M: n = 11, VL20W: n = 11, VL40W: n = 11 and characteristics upon entering the study are presented in Table 1. Subjects were requested to not perform any other type of strenuous physical activity during the study period.

The present study consisted of an 8-week VBRT intervention in the squat and bench press exercises. Two groups, VL20M and VL20W, performed their sets during the training program until they reached a velocity loss of 20%. The other two groups, VL40M and VL40W, performed their sets until they achieved a velocity loss of 40%. All groups completed a total of 15 training sessions over the 8-week training period (Table 2). The subjects trained twice per week in weeks 1–4 and 6–8, with one training session held in week 9 prior to post-tests. Mid-training tests were performed in week 5 and no training sessions took place. The subjects were tested on four occasions during the study: Control-tests (Week 2, Control), before training (Week 0, PRE), after 8 training sessions (Week 5, MID) and after 15 training sessions (Week 10, POST). The measurements were conducted on two different sessions separated by < 48 h. The first testing session consisted of body mass and muscle size measurements. The second testing session consisted of CMJ and progressive loading tests in the Smith-machine full back squat (SQ) and the Smith-machine bench press (BP) exercises, in that order. Both testing and training sessions were performed in a research laboratory under the direct supervision of investigators, at the same time of the day (± 1 h) and under the same environmental conditions.

Smith-machine full back squat incremental loading test: The subjects were instructed to perform the eccentric phase of the exercise descending at moderate velocity (approx. 2 s decent) until the thighs were bellow parallel, while the concentric phase was performed with maximal velocity. The feet needed to stay in contact with the ground and the bar on the shoulders during the whole exercise. MPV of each testing load was recorded using a linear velocity transducer and software (T-Force System, Ergotech, Murcia, Spain). The initial load was 41 kg in men and 31 kg in women. The load was increased with 10‒15 kg steps in each set until MPV of the trial was lower than 0.5 m s⁻¹. Thereafter, the load was increased by 5 or 2.5 kg in each set. Subjects were allowed to perform from one (in the heaviest sets) to six (in the lightest sets) repetitions per measured load. Recovery periods between sets were 2 min until the MPV was 0.5 m s⁻¹ and were 3 min thereafter. Strong verbal encouragement was used by a researcher during the concentric phase of the repetition. The measurement ended when the subject reached the load that led to failure. The same absolute incremental loads were tested in Control, PRE, MID and POST to be able to compare the MPV change between the tests. If the subject successfully achieved their previous 1-RM, the test was continued until failure.

MPV performances were retrospectively classified as “low” or “high” velocity according to our previously published sex-specific values (Pareja-Blanco et al. 2020c). The cut-off was set at 70% 1-RM for both exercises, which was the median relative load within the range used during training (65–75%1-RM, Table 2). In SQ, cut-offs of 0.73 and 0.65 m s⁻¹ were used for men and women, respectively. The highest MPV score from each absolute load was taken as the best performance, and then mean MPV from all low and all high velocities were calculated for each subject. The coefficient of variation (CV) for Control to PRE was 2.8% for 1-RM, 5.6% for MPV with low velocities and 14.0% for MPV with high velocities.

Smith-machine bench press incremental loading test: MPV of each repetition during the test was recorded using the same measurement system as in the SQ incremental loading test (T-Force System, Ergotech, Murcia, Spain). The initial load was 36 kg in men and 10–20 kg in women depending on the estimated 1-RM of the individual subject. The load was increased in 2.5‒15 kg steps in each set until MPV of the trial slowed to 0.4 m s⁻¹. Thereafter, the load was increased by 5 or 2.5 kg in men and by 2.5 or 1 kg in women. The subjects were given instructions to perform from one (in the heaviest sets) to six (in the lightest sets) repetitions per set in this exercise, as in SQ. The recovery periods between sets were 2 min when the highest MPV in the last set was over 0.4 m s⁻¹ and 3 min when MPV in the last set was under 0.4 m s⁻¹. The subjects were advised to perform the eccentric phase of the exercise using moderate velocity and stop the bar on the chest. After a ~ one-second pause, a verbal signal was given by a researcher and the subjects performed the concentric phase of the movement as fast as possible. The stop on the chest was used to minimize bouncing on the chest and, thus, to standardize the technique (Pallarés et al. 2014). The bar needed to stay in the hands of the subject during the whole exercise. Strong verbal encouragement was used by a researcher during the concentric phase of the repetition. The measurement ended when the subject reached the load that led to failure. If the subject achieved their previous 1-RM, the test continued until failure. The cut-off to distinguish between “low” and “high” velocity, corresponding to 70% 1-RM, were 0.58 (men) and 0.54 (women) m s⁻¹ (Pareja-Blanco et al. 2020c). The CV for Control to PRE was 2.7% for 1-RM, 4.5% for MPV with low velocities and 10.6% for MPV with high velocities.

Countermovement jumps (CMJ) were performed on a force platform (Faculty of Sport and Health Sciences at the University of Jyväskylä, Finland) with Signal software version 4.14 (Cambridge Electronic Design Ltd., Cambridge, United Kingdom) used to record (sampling rate 2000 Hz) and analyze data. Vertical displacement of a subject’s center of mass was calculated using the following equation: h = I²/(2 gm²), where h = vertical displacement of a subject’s center of mass, I = vertical impulse of force, g = acceleration due to gravity, m = mass of the subject. The subjects performed the measurement by lowering their body rapidly and then pushing against the ground with an explosive change of direction. Hands were kept on the hips throughout the whole movement and the subjects were encouraged to jump as high as they can in every measurement. Subjects performed 3 jumps separated by 1 min of rest, and the highest of the 3 jumps was taken into further analyses. The CV for Control to PRE was 3.8%.

Surface electromyography (EMG) was recorded from the vastus lateralis and medialis of the right leg during SQ and CMJ. Bipolar silver-silver chloride surface electrodes (Ambu BlueSensor N, Copenhagen, Denmark) with an inter-electrode distance of 20 mm were attached to muscle-specific locations following SENIAM guidelines (Hermens et al. 2000). Small indelible ink tattoos (diameter < 1.0 mm) were placed during Control to ensure that the electrodes were replaced exactly to the same spots in each measurement (Häkkinen and Komi 1983). The telemetric recording system (Noraxon Inc. Scottsdale, Arizona, USA) with a sampling frequency of 3000 Hz was used for data collection. The EMG signal was down-sampled to 2000 Hz and transmitted via A/D converter (Cambridge Electronic Design Ltd., Cambridge, United Kingdom) to a personal computer where Signal 4.14 software (Cambridge Electronic Design Ltd., Cambridge, United Kingdom) was used to data recording and analyzing. The data were band-pass filtered over 20–350 Hz and root-mean squared for both muscles separately, and the average of vastus lateralis and medialis EMG activity (lateralis + medialis/2) were analyzed over the entire concentric phase of the CMJ measurement and back squat 1-RM trial.

Muscle size was measured by anatomical cross-sectional area (CSA) in the axial plane of vastus lateralis and by muscle thickness of the triceps brachii, both from the right limbs, using a B-mode ultrasound device (SSD-a10; Aloka, Tokyo, Japan). A 10 MHz linear-array probe (60 mm) coated with water-soluble transmission gel and housed in a custom-made convex support was used. CSA measurements were performed using the extended-field-of-view function as previously described (Walker and Häkkinen 2014; Walker et al. 2020), which has been shown to be a valid and repeatable method when assessing muscle CSA changes over time (Ahtiainen et al. 2010). Triceps brachii measurements were taken at the mid-point between the medial epicondyle and the acromion. The probe was oriented in perpendicular to the skin and positioned with minimal contact to avoid tissue deformation. Three images were taken during each measurement and the average of the three values was taken forward for further analyses. All measurements were taken and analyzed (ImageJ software version 1.44, National Institutes of Health, Bethesda, Maryland, USA) by the same experienced researcher. The CV for Control to PRE was 3.5% for vastus lateralis CSA and 4.0% for triceps brachii thickness.

The descriptive characteristics of the training program are presented in Table 2. The training program included two main exercises, SQ and BP, which were always performed in that order. In addition, three sets of supplementary trunk exercises (8–12 reps with 1 min inter-set recovery), for abdominals and lower back, were performed at the end of each training session using low loads and moderate velocities (approx. 2 s concentric and 2 s eccentric duration). The technique for SQ and BP was as described in the Measurements section, i.e., concentric phase performed with maximum voluntary velocity, SQ performed without a pause from eccentric to concentric transition, the bar stopped on the chest ~ 1 s before the concentric phase in the BP exercise etc. The MPV of each repetition in every training session was measured with a linear velocity transducer (T-Force Measurement System, Ergotech, Murcia, Spain). In SQ and BP, the training intensity was determined by using the velocity-based estimation of 1-RM (Sánchez-Medina et al. 2017; González-Badillo and Sánchez-Medina 2010). The load of the bar was adjusted if needed after the first repetition of the first set in every training session to match with the velocity (± 0.03 m s⁻¹) that corresponds the desired load (% of predicted 1-RM). The training loads were progressively increased from 65 to 75% predicted 1-RM during the training program (Table 2). The inter-set rest interval was 3 min in all sessions. Strong verbal encouragement was used in every training session to perform the concentric phases with the maximum possible velocity. Each subject performed an individual number of repetitions in each training session based on their velocity-loss allocation and their ability to withstand fatigue.

Based on a previous study (Pareja-Blanco et al. 2020a), sample size estimations suggested that 12 subjects per group would satisfy alpha (0.05) and power (0.95) assumptions. Test–retest (Control to PRE) reliability was measured by the standard error of measurement (SEM), which was expressed in relative terms through the CV. SEM was calculated as the root mean square of the intra-subject total mean square. Mean values and standard deviations of each group were calculated according to standard procedures. All data were checked for normality using the Shapiro–Wilk test prior to performing statistical analyses. Independent T-tests assessed potential between-group differences between 20 and 40% velocity-loss groups (within sex) at baseline (PRE). Training session data, i.e., repetition number and velocity, were assessed by two-way ANOVA (2 sex × 2 velocity loss). Repeated measures ANCOVA (2 sex × 2 velocity loss × 4 time-points), using pre-training values as covariate, was used to evaluate the effects of the training program between sexes and velocity-loss protocols. Where appropriate, Bonferroni post hoc tests were applied for pairwise comparisons. Possible relationships between training-induced changes in tested variables was explored by Pearson’s product-moment correlation test. To examine the magnitude of differences between velocity-loss protocols (within sex), Hedge’s g effect sizes were calculated from between-group differences in the relative changes (Δ%) from PRE to POST. The cut-offs for small, medium and large effect were < 0.3, 0.3‒0.8, and > 0.8, respectively. All statistical analyses were performed using IBM SPSS Statistics software (SPSS version 26, Chicago, Illinois, USA).

Adherence to training was 98 ± 3% in VL20M, 97 ± 5% in VL40M, 95 ± 6% in VL20W, and 95 ± 4% in VL40W, with no statistical differences between-groups. The total number of repetitions completed by the 20% velocity-loss groups was ~ 78% and ~ 76% in SQ in men and women, respectively, and was ~ 73% and ~ 62% in BP in men and women, respectively, compared to their respective 40% velocity-loss groups (P < 0.001, Fig. 1). There was a significantly greater number of low-velocity repetitions in both 40% velocity-loss groups (P < 0.001). In SQ, all women (i.e., VL20W + VL40W) performed more repetitions with high velocity than all men (i.e., VL20M + VL40M) (P = 0.003), whereas men performed more low-velocity repetitions than women (P = 0.02). In BP, VL20M completed more repetitions at high velocity than VL40M (P = 0.03), but this difference was not observed between women (VL20W vs. VL40W). Also, all men (i.e., VL20M + VL40M) performed more repetitions than all women (i.e., VL20M + VL40M) (P = 0.03).

At PRE, SQ 1-RM did not differ (P > 0.05) between VL20M (113.7 ± 28.0 kg) and VLM40 (111.7 ± 13.7 kg) nor between VL20W (68.0 ± 12.6 kg) and VL40W (67.1 ± 19.9 kg). All groups of men and women significantly increased their SQ 1-RM from PRE to MID and from PRE to POST (Fig. 2A), without significant between-group differences. Effect sizes for between-group differences in relative changes PRE to POST showed a medium effect in men (g = 0.31) and women (g = 0.56).

Surface EMG during the SQ 1-RM did not change from PRE to POST in any of the groups. Averaged EMG activity was 0.452 ± 0.179 mV (VL20M), 0.564 ± 0.144 mV (VL40M), 0.532 ± 0.135 mV (VL20W), and 0.503 ± 0.230 mV (VL40W) at PRE for the four groups.

At Pre, SQ MPV did not differ (P > 0.05) either in high velocities or low velocities between VL20W (0.81 ± 0.08 and 0.46 ± 0.07 m s⁻¹) and VL40W (0.82 ± 0.07 and 0.45 ± 0.06 m s⁻¹). In men, there was a statistically significant between-group difference for low velocities (0.52 ± 0.05 vs. 0.46 ± 0.04 m s⁻¹, P = 0.007) but not when comparing high velocities (0.97 ± 0.08 vs. 0.95 ± 0.06 m s⁻¹). All groups showed statistically significant increases in MPV of both high and low velocities from PRE to POST (Fig. 2B and C), without significant between-group differences.

Effect sizes for between-group differences in relative changes PRE to POST showed a small (g = 0.14) and medium effect (g = 0.51) in men and a medium (g = 0.50) and medium effect (g = 0.38) in women for high and low velocities, respectively. Further, the training-induced changes in SQ 1-RM and MPV in high (r = 0.467, P = 0.033, n = 21) and low velocities (r = 0.471, P = 0.027, n = 22) were significantly related in women. In men, training-induced changes in SQ 1-RM and MPV at low velocities were significantly related (r = 0.479, P = 0.021, n = 23), but changes in 1-RM and MPV at high velocities were not (r = 0.194, P = 0.381, n = 23).

At PRE, BP 1-RM did not differ (P > 0.05) between VL20M (81.2 ± 16.8 kg) and VLM40 (80.5 ± 13.5 kg) nor between VL20W (40.2 ± 9.9 kg) and VL40W (39.3 ± 9.9 kg). All groups showed statistically significant increases from PRE to POST (Fig. 2D), without significant between-group differences. Effect sizes for between-group differences in relative changes PRE to POST showed a small effect (g = 0.04) in men but a medium effect (g = 0.60) in women.

At PRE, BP MPV did not differ (P > 0.05) either in high velocities or low velocities between VL20M (0.83 ± 0.09 and 0.35 ± 0.07 m s⁻¹) and VL40M (0.79 ± 0.09 and 0.35 ± 0.05 m s⁻¹). In women, there was a statistically significant between-group difference for low velocities (0.35 ± 0.03 m s⁻¹ vs. 0.30 ± 0.06 m s⁻¹, P = 0.039) but not for high velocities (0.71 ± 0.10 vs. 0.65 ± 0.05 m s⁻¹). Both low-velocity and high-velocity MPV increased in all groups, but there were no statistically significant between-group differences in training-induced increases in MPV for low or high velocities (Fig. 2E and F).

Effect sizes for between-group differences in relative changes PRE to POST showed a small effect (g = 0.13) and small effect (g = 0.29) in men for low and high velocity, respectively. However, large between-group effect sizes in women were observed for low (g = 1.42) and high velocities (g = 0.98). As in SQ, the training-induced changes in BP 1-RM and low (r = 0.503, P = 0.024, n = 20) and high velocities (r = 0.518, P = 0.019, n = 20) were significantly related in women. In men, training-induced changes in BP 1-RM and low (r = 0.521, P = 0.011, n = 23) and high velocities (r = 0.744, P < 0.001, n = 23) were also significantly related.

At PRE, CMJ height did not differ (P > 0.05) between VL20M (31.1 ± 4.5 cm) and VLM40 (34.1 ± 3.9 cm) nor between VL20W (24.1 ± 3.5 cm) and VL40W (23.6 ± 5.3 cm). All groups showed significant increases in CMJ from PRE to MID and all but VL40M showed significant increases from PRE to POST (Fig. 3A), but no between-group differences were observed. Effect sizes for between-group differences in relative changes PRE to POST showed a small effect in both men (g = 0.12) and women (g = 0.27).

Averaged EMG during the concentric phase of CMJ significantly increased in three of the groups (VL40M, VL20W and VL40W) at MID compared to PRE (Fig. 3B), and significance remained at POST in VL40M and VL20W. Such a pattern of increase followed by plateau/decrease was observed in both muscles separately and when averaged.

At PRE, vastus lateralis CSA did not differ (P > 0.05) between VL20M (31.4 ± 3.6 cm²) and VL40M (31.0 ± 5.3 cm²) nor between VL20W (22.0 ± 4.3 cm²) and VL40W (21.4 ± 3.3 cm²). All groups increased vastus lateralis CSA, but there were no statistically significant between-group differences in training-induced increases (Fig. 4). Effect sizes for between-group differences in relative changes PRE to POST showed a medium effect (g = 0.59) in men and a small effect (g = 0.23) in women.

One aspect of the study design that may have influenced the ability to determine between-group differences is that the training program lasted 8 weeks. In practice, typical power training mesocycles are 4–6 weeks in duration. It has been shown that power training for 7–12 weeks or longer can lead to a plateau or even reverse some gains in power performance both in men and women depending on the type of the training protocol and the measurements conducted (Häkkinen and Komi 1985; Kyröläinen et al. 1989; Peltonen et al. 2018). Therefore, the possibility that true peaks in performance gains being missed by the testing at weeks 5 and 10 cannot be discounted.

In standardizing the velocity loss for each group (either 20% or 40%), the present training program provides a similar training stimulus by varying the training volume according to the fatigability of the individual. Against our initial prediction, women did not perform a greater total number of repetitions compared to men; men actually performed more repetitions in BP than women. Indeed, the discrepancy of total volume between 20 and 40% velocity loss was larger for BP (~ 73% vs. ~ 62%) than for SQ (~ 78% vs. ~ 76%), and this could potentially be explored in future. Hence, it is possible that VL20W did not attain the required training volume to maximize adaptation in BP.

Despite not including a control group, the present study included a control period, which highlights the variability in the measurements and also the impact of learning. Upon inspecting the variations from Control to PRE, it is clear that the observed changes are a result of our training program and not factors external to the intervention. Therefore, the observations of improvement throughout the training period can be considered as training-induced adaptations. Nevertheless, one weakness of the present study is that three of the groups contained 11 subjects (below the a priori determined requirement of 12) at the end of the intervention; due to drop-outs. This may, at least partly, explain why statistically significant differences in women were not found despite medium and large effect sizes between-groups for bench press data. However, this was the first study conducted in women, and therefore the estimations based on previous men’s data may not be appropriate. Hence, the present study provides a guide for future VBRT studies investigating adaptations in women.