Validity of the bench press one-repetition maximum test predicted through individualized load-velocity relationship using different repetition criteria and minimal velocity thresholds

1. Introduction

Exercise intensity is a key variable when designing resistance training programs [1, 2]. Traditionally, exercise intensity is prescribed as a percentage of an individual’s one-repetition maximum (1RM) [1]. To correctly program resistance training based upon %1RM, coaches must attain a 1RM value. Briefly, the direct determination of 1RM consists of a gradual load-progression until reaching the maximum weight which an individual can lift only once [3]. Although direct assessment of 1RM provides a reliable indicator of maximal strength [4], its implementation has been frequently discouraged as it is not only time-consuming, but also physically and psychologically demanding [1, 5, 6]. Consequently, a challenge for sport scientists has been to develop more practical and less fatiguing strategies to accurately predict the 1RM [5, 7, 8, 9]. The individualized load-velocity relationship has been shown to be a potential alternative for predicting the 1RM across a variety of exercises [8, 10, 11]. This is owing to the advancement and proliferation of accurate and reliable technologies for measuring barbell velocity [12, 13, 14], as well as the low intra-subject variability observed for the linear inverse relationship between the relative lifted loads (%1RM) and movement velocity [15, 16]. However, a number of important aspects of load-velocity modeling still remain unresolved.

The determination of the individualized load-velocity relationship requires the assessment of movement velocity against two or more loads [7, 17]. Typically, this is done through an incremental loading test in which individuals perform 3 repetitions when lifting light loads ( ⩽ 60% of 1RM), 2 repetitions for medium loads (60–85% of 1RM), and 1 repetition for heavy loads ( ⩾ 85% of 1RM) [17, 18, 19, 20]. The next logical step to create an individual’s load-velocity profile is to select the velocity of each load according to a specific repetition criterion [7, 21]. Some researchers have used as a criterion the average velocity of all repetitions performed at each load to obtain a more robust metric [18, 22, 23], while the most frequent criterion is to use the fastest repetition of each load to represent the maximal performance of the individual [5, 10, 17, 24, 25]. However, the latter may be affected by outliers related to erroneous lifts [21]. However, no study has examined whether the repetition criteria affects the accuracy in the estimation of the 1RM through the individualized load-velocity relationship.

Once the final velocity output of each load has been selected, the next step is to determine the individual relationship between the absolute load and velocity through a linear regression equation [17, 18]. Thereafter, the 1RM can be estimated using the regression equation as the load associated with a minimal velocity threshold (MVT; velocity at 1RM or of the last repetition of a set to failure) [7, 17, 19]. In practice, coaches have two options for selecting the MVT used to predict the 1RM: (I) an individual MVT recorded in a previous session during a 1RM trial [5, 24] or during the last repetition of a set to failure [19] and (II) a general MVT proposed in the scientific literature (e.g., 0.17 m ⋅ s - 1 for the bench press) [17]. The 1RM has been also predicted using the individual MVT recorded during the 1RM assessment performed in the same session in which the load-velocity relationship was modelled [6, 9]. From a practical point of view, this is illogical as there is no need to predict the 1RM in the same session where it has been determined. More important is to determine if a previous attained MVT can accurately predict 1RM in a subsequent session. Moreover, the use of a general MVT simplifies the testing procedure because it would not be necessary to perform a direct 1RM or repetitions-to-failure test at any time point [8], but it has been suggested that using the individual MVT could increase the accuracy in the 1RM prediction compared with using a general MVT [17]. However, the only study that have compared the accuracy in the estimation of the 1RM between general and individual MVTs failed to show significant differences in the deadlift exercise performed with and without lifting straps [9]. It is important to note that, although it does not make sense in practice, this study used the individual MVT recorded during the 1RM trial that was performed in the same session in which the individualized load-velocity relationship was modelled. Due to the low reliability of the MVT [6, 26], it is likely that the accuracy of the 1RM prediction could be worse when using the individual MVT obtained in a preliminary session. However, no study has compared whether the accuracy in the estimation of the 1RM differs when using the individual MVT obtained in the same or a preliminary testing session.

To fill gaps in the scientific literature, the present study was designed to analyze the influence of certain methodological factors (repetition criteria and MVT) on the 1RM prediction capability of the individualized load-velocity relationship. Specifically, the aim of this study was to compare the accuracy in the estimation of the free-weight bench press exercise between six velocity-based 1RM prediction methods based on 2 repetition criteria (fastest and average velocity) and 3 MVTs (general MVT of 0.17 m ⋅ s - 1 , individual MVT of the preliminary session, and individual MVT of the validity session). We hypothesized that the MVT would not significantly affect the accuracy in the prediction of the bench press 1RM [9]. However, no specific hypothesis was formulated regarding the differences between the repetition criteria in the 1RM prediction accuracy due to the lack of similar studies.

2. Methods

3. Results

Trivial non-significant differences were observed for the 1RM ( P = 0.850; ES = 0.01) and MVT ( P = 0.482; ES = 0.27) between both testing sessions. The between-session reliability was high for the 1RM (CV = 2.5%; ICC = 0.99) and unacceptable for MVT (CV = 19.2%; ICC = 0.01).

The ANOVA did not reveal a significant main effect for repetition criterion ( F ( 1 , 15 ) = 0.74; P = 0.402) or MVT ( F ( 2 , 30 ) = 1.86; P = 0.173). Furthermore, the interaction repetition criterion × MVT did not reach statistical significance ( F ( 2 , 30 ) = 1.08; P = 0.354). The errors were categorized as low (range = 2.7–3.7%) for all 1RM prediction methods (Fig. 1).

Trivial non-significant differences ( P ⩾ 0.556, ES ⩽ 0.02) and practically perfect correlations ( r ⩾ 0.99) were observed between the actual and all predicted 1RMs (Table 2). Bland-Altman plots showed low and comparable systematic bias (range: - 1.0 to 0.0 kg) and random errors (range: 2.7 to 3.4 kg) for all 1RM prediction methods (Fig. 2). No heteroscedasticity of the errors ( r 2 ⩽ 0.05) were observed for any 1RM prediction method with respect to the actual 1RM.

OPEN IN VIEWER

This study was designed to explore the influence of repetition criterion and MVT on the 1RM prediction capability of the individualized load-velocity relationship during the free-weight bench press exercise. The main finding of this study was that all prediction methods provided an accurate estimation of the bench press 1RM, being the accuracy comparable for the 2 repetition criteria (fastest velocity and average velocity) and 3 MVTs (general MVT of 0.17 m ⋅ s - 1 , individual MVT of the preliminary session, and individual MVT of the validity session). Therefore, in principle practitioners are free to choose the different repetition criteria and MVT as the latter do not seem to affect the accuracy in the estimation of the free-weight bench press 1RM. However, a general MVT could be recommended to simplify the testing procedure because a direct 1RM assessment would not be required at any time.

There is compelling evidence that the individualized load-velocity relationship modelled using various submaximal loads (from 2 to 5) can provide an accurate estimation of the 1RM during the bench press exercise performed in a Smith machine [11, 32, 33]. For example, the individualized load-velocity relationship modelled considering only 2 loads has been shown to provide an accurate estimation of the 1RM during different variants of the bench press exercise regardless of the grip width with absolute errors lower than 5% [32]. By contrast, previous studies have warned about the lack of precision (absolute errors $>$ 5%) of the individualized load-velocity relationship to predict the 1RM during free-weight exercises such as back squat [6] or deadlift [9, 24], raising the question of whether its 1RM prediction capability is influenced by the execution mode (machine-guided vs. free-weight). This question has been recently addressed by Hughes and colleagues [34] who found an acceptable and comparable magnitude ( $P>$ 0.05) and reliability (CV ⩽ 2.6% and ICC ⩾ 0.98) for predicted 1RM between Smith machine and free-weight variants of the bench press and prone row exercises. Instead, the 1RM predictions performed for back squat and overhead press exercises were more comparable ( P = 0.099–0.961 vs. 0.001–0.006) and reliable using a Smith machine (CV = 2.6–3.8%; ICC = 0.97–0.98) compared to free-weights (CV = 3.9–7.7%; ICC = 0.78–0.98). Those findings highlight that the accuracy of the predicted 1RM from the individualized load-velocity-relationship can be affected by the training equipment for some exercises but not for others. In line with the study of Hughes et al. [34], the results of this study also confirm that the free-weight bench press 1RM can be obtained with a high precision from the individualized load-velocity relationship. This is an important finding since most athletes commonly use the free-weight bench press exercise for training [35].

The load-velocity relationship has been indistinctly constructed using either the fastest or average velocity of the different repetitions (usually from 3 to 1) recorded against a range of loads [5, 22, 23, 25]. To our knowledge, this is the first study to examine whether the accuracy of individualized load-velocity relationship to predict the 1RM is affected by the repetition criteria. Our findings show that the accuracy of the predicted 1RM is not affected by the repetition criterion selected for the development of an individual’s load-velocity profile in the free-weight bench press exercise. It is important to note that the within-subject variability in the velocity recordings was low for the different loads included in the modeling (CV = 4.3, 3.7, 4.9, 6.5 and 6.6% for ≈ 25, 40, 55, 70 and 85% of 1RM). These results are in line with those reported by Carroll and colleagues [23] in the back-squat exercise (CV = 6.4, 8.6, 6.4 and 6.4% for ≈ 60, 70, 80 and 90% of 1RM). However, while the repetition criterion based on the average velocity provides a more robust measure, the repetition criterion established according to the fastest velocity may be related to an erroneous execution of the lift [21]. For example, in this study, the absolute differences between the measured and predicted 1RM for a representative participant were reduced by 3–5% using the average velocity compared to the fastest velocity due to the large amount of variation observed for the velocity measurements (CV range = 2.5 to 13.7%). Therefore, caution should be taken when more than a 10% difference is observed between trials [7] particularly if the fastest velocity is used as repetition criterion to developed the load-velocity relationship of the individual. To solve this problem, an extra repetition should be performed and the most extreme value discarded from the analysis [7].

The 1RM is calculated from an individualized load-velocity profile regression equation as the load associated with an exercise-specific MVT [7, 17]. This MVT can be selected from a reference value for all individuals from the literature (“general MVT”) [8, 10, 11, 22] as well as from an actual (“individual validity MVT”) or previous (“individual preliminary MVT”) maximal lift for each participant [5, 6, 9, 24]. Our results indicate that these three MVTs approaches provides an accurate estimation of the 1RM during the free-weight bench press exercise (absolute and random errors ⩽ 3.7% and 3.4 kg, respectively). These findings are in line with previous studies that have used a general MVT of 0.17 m ⋅ s - 1 [11, 32, 33] or an individual preliminary MVT [34] to predict the bench press 1RM using different variants (concentric-only and eccentric-concentric) or execution modes (machine-guided and free-weight). An even more important and novel findings is that the accuracy in the estimation of the bench press 1RM did not improve when individual MVTs were considered instead of a general MVT. These results confirm the previous findings of Jukic and colleagues [9] who found that the accuracy of the 1RM was similar using either a general MVT (absolute errors from 7.3 to 15.4 kg) or an individual MVT recorded in the validity session (absolute errors from 6.6 to 16.3 kg) during the deadlift exercise performed with and without lifting straps. Furthermore, it is important to note that the MVT in this study demonstrated a poor consistency (CV = 19.2%; ICC = 0.01) and, therefore, it would not be appropriate to use the values obtained in a preliminary session to estimate the 1RM on subsequent occasions [5, 24]. Similarly, it would be illogical and impractical to use the MVT of the actual 1RM or MV of last repetition to failure to predict the 1RM within the same session [6, 9]. Collectively, these results suggest that practitioners can estimate the bench press 1RM using a general MVT. This will provide a simplified and less time-consuming assessment of bench press 1RM.

Finally, since the accuracy of the predicted 1RM could be affected by the sex (lower errors for women compared with men) [10] or resistance training experience (lower MVT for highly resistance-trained athletes than less-trained athletes) [36], the main limitation of our study is that the current findings are applicable to men with ∼ 1 year experience in resistance training (bench press 1RM = 1.0 ± 0.2 kg ⋅ body mass - 1 ). However, the errors in the current study were homoscedastic (i.e., uniform errors) which indicates that those with lower or higher bench press 1RMs than our sample ( ∼ 72 kg) can still attain an accurate 1RM prediction from the individualized load-velocity relationship. On the other hand, it should be noted that a single linear velocity transducer was used to measure the velocity of barbell despite the fact that the velocity may be different on each side of the barbell [37]. This decision was made to increase the ecological validity of our findings because athletes rarely use two linear position transducers simultaneously.

5. Conclusions

This study confirms that the individualized load-velocity relationship is a highly valid approach for predicting the 1RM during the free-weight bench press exercise. Furthermore, our results demonstrate that the accuracy of the bench press 1RM prediction is not meaningfully affected by the repetition criteria or MVT. This indicates that the load-velocity relationship can be constructed using either the fastest or average velocity of the different repetitions recorded against a range of external loads. However, since the fastest velocity may be an outlier related to an erroneous lift [21], caution should be taken when more than a 10% in velocity is observed between trials [7], particularly if the fastest velocity is used as repetition criterion to develop the load-velocity relationship of the individual. However, it is also plausible that the outlier could be produced by a very slow velocity and in this case the precision could be compromised by using the average velocity of several trials. The accuracy in the estimation of the bench press 1RM did not improve when individual MVTs were considered instead of a general MVT. Practically, this is important because the general MVT can be recommended to simplify the testing procedure and avoid a maximal lift at any time point of the training cycle [8, 9].

Author contributions

CONCEPTION: Alejandro Pérez-Castilla, John F. T. Fernandes, and Amador García-Ramos.

PERFORMANCE OF WORK: Alejandro Pérez-Castilla.

INTERPRETATION OR ANALYSIS OF DATA: Alejandro Pérez-Castilla, John F. T. Fernandes and Amador García-Ramos.

PREPARATION OF THE MANUSCRIPT: Alejandro Pérez-Castilla.

REVISION FOR IMPORTANT INTELLECTUAL CONTENT: Alejandro Pérez-Castilla, John F. T. Fernandes, and Amador García-Ramos.

SUPERVISION: Amador García-Ramos.

Ethical considerations

All participants were informed of the procedures to be used and signed a written informed consent form before initiating the study. The study protocol adhered to the tenets of the Declaration of Helsinki and was approved by the Institutional Review Board of the University of Granada (IRB approval: 988/CEIH/2019).

Funding

This study was funded by the Centro Mixto Universidad de Granada – Mando de Adiestramiento y Doctrina (5/4/20 TR-COMBATE).