A strength-oriented exercise session required more recovery time than a power-oriented exercise session with equal work

Study design

The present study was a randomized cross-over study where each participant completed a heavy-load, strength-oriented session and a moderate-load, power-oriented session, in randomized order (applying the Research Randomizer; Urbaniak & Plous, 2013). To achieve a counterbalanced order, the participants were paired, so that one started with the strength-oriented session and one with the power-oriented session. One to four weeks of rest were allowed between sessions (16 ± 10 days (mean ± standard deviation)).

A test battery of physical performance tests and evaluation of perceived effort and recovery status was applied before, immediately after, and 24 and 48 h after the exercise sessions (Fig. 1). The concentric work (J) performed in the first session was recorded and replicated in the second session, ensuring equal work in both sessions. The exercises were the same for both sessions, but somewhat adapted to serve the purpose of the sessions (Table 1). The primary aim of the study was to compare the recovery rates between sessions when all factors were equal except the external load (50% lower in the power-oriented session than the strength-oriented session).

Three to seven days before the first exercise session, a familiarization session was conducted. The participants were familiarized with all tests and exercises (see details below) and instructed not to conduct any strenuous exercise 48 h prior to the test days. The participants were also instructed to standardize their breakfast before and their meals after the exercise sessions (for 48 h). All supplements and medications were prohibited during the study period.

During the exercise sessions the participants were given a protein bar and a protein drink (both supplements containing approximately 20 g protein, 30 g carbohydrates, and a total of ∼1,000 kJ (Yt, Tine, Oslo, Norway), and an energy drink (30 g carbohydrates; 510 kJ; Yt, Tine, Oslo, Norway) to ensure sufficient protein and energy intake (in total: 40 g protein and 90 g carbohydrates; ∼1,500 kJ). Water was allowed ad libitum.

Participants

Nineteen young, resistance-trained individuals were recruited to this study. Sixteen participants, eight males and eight females, completed all tests and both exercise sessions (24 ± 3 years, 74 ± 12 kg, 1.75 ± 0.11 m; Table 2). Two participants dropped out due to muscle pains (hamstrings and groin) during testing or the exercise sessions; and one was excluded due to technical problems with the test equipment.

The participants were familiar with heavy-load strength training and had been training upper and lower body strength exercises on a weekly basis during the last year (≥2 sessions/week). Of the 16 participants, three were competing at a national elite level (two volleyball and one beach volleyball player), one was an international-level bike trial athlete, and the remaining 12 participants were students at the Norwegian School of Sport Sciences (Oslo, Norway) and engaged in strength training at a recreational level.

The study was reviewed by the Norwegian Regional Ethical Committee of Medical and Health Research (2016/1120). The participants gave written informed consent to take part in the study, following the Declaration of Helsinki (World Medical Association).

Testing and exercises

The familiarization session consisted of all the tests (see below) and 1–3 sets of five repetitions of all the exercises (for both sessions): squat, front squat, trap bar squat, bench press, narrow bench press and push-ups (Fig. 1). The loads were adjusted to get close to a 5-repetition maximum (RM) during the last set. For the power exercises the loads were 50% of the estimated 5RM loads. In both sessions, the exercises were executed with maximal effort in the concentric phase in all repetitions. In the strength-oriented session, the eccentric phase was conducted with a controlled, slow movement (>1 s). In contrast, in the power-oriented session the eccentric phase was faster (<1 s) in order to maximize the power output in the concentric phase. The movement velocity was measured using a linear encoder (see description below).

On the days of the exercise sessions, the participants rated their perceived recovery status (PRS scale; 0–10; Laurent et al., 2011) prior to a warm-up. The warm-up consisted of a 10-minute easy run with increasing velocity (not more than moderate effort) preceding 2 min of individually selected dynamic stretching of both upper and lower body muscles. Thus, each participant followed a warm-up procedure that they were accustomed to, and the procedure was similar before each session (including the familiarization session).

The tests were then conducted in the following order: CMJ, SJ, 10 consecutive multiple jumps (MJ), 20-meter sprint running, maximal push-up force, and power profiles and estimated 1RMs in bench press and squat. Tests were performed before and immediately after the sessions, and again after 24 and 48 h. The power profile tests and 1RM estimation in the bench press and squat were, however, not conducted immediately after the sessions in order to prevent additional fatigue. Finally, 30 min after the sessions the participants rated the perceived exertion (session RPE; 0–10; Foster et al., 2001). The participants were introduced to the ratings and descriptors of both the RPE and the PRS scales at the familiarization session.

Tests

The countermovement jump (CMJ), squat jump (SJ), and multi-jump (MJ) were conducted on an AMTI force platform (sampling rate: 2000 Hz; OR6-5-1; AMTI, Watertown, MA, USA). Before every test session (for each participant), offset values were acquired and the body weight of the participant was measured and averaged over a 2.3 s period (as recommended by Street et al., 2001). The body weight measured before each jump was confirmed against the initial weigh-in value (less than 5% discrepancy was considered valid). All force data were filtered with a low pass filter (second order Butterworth bi-directional low pass filter; cut-off frequency of 120 Hz).

All jump-tests were performed with the hands fixed on the hips (akimbo). Based on jump height, the average of each individual’s two best attempts of 3–6 jumps was used for subsequent statistical analyses—except for MJ, where only one attempt was made (due to the development of fatigue). The inter-test coefficients of variation (CVs) for the jump tests are listed in Table 2.

All data collected from the AMTI force plate were analysed using a custom-made software (Matlab, The Mathworks, Inc., MA, USA; Biomekanikk AS, Oslo, Norway). From the SJ, duration, concentric peak force, peak and mean power, RFDmax, and jump height were calculated. From the CMJ, duration, eccentric time, peak and mean eccentric and concentric force, RFDmax, and jump depth (lowering of the center of mass (COM)) and jump height were calculated. From the MJ, jump height, vertical stiffness, and reactive strength index (RSI) were calculated.

Jump height was calculated as the squared take-off velocity divided by 2 g for all jumps (SJ, CMJ, and MJ). Take-off velocity was calculated by the impulse–momentum method described by Linthorne (2001) and Street et al. (2001), with the impulse-integral starting from the time point when vertical force exceeded (or fell below for CMJ) 100% of body weight and ending when force fell below 2 N (take-off).

The jump’s phases were calculated as follows: Duration (s) of the SJ was found by backtracking force data from take-off (force <2 N) to the point where the force was 101.5% of body weight. CMJ was divided into an eccentric phase and a concentric phase (Fig. 2), defined by the phase where the COM was descending and ascending, respectively. The initiation of the CMJ (eccentric phase) was found by backtracking the force data from the point of zero velocity, i.e., the deepest position of the COM, to the point where the force was 98.5% of body weight. Eccentric peak force was the highest force measured within the eccentric phase, and eccentric time was defined as the duration of the eccentric phase where the force was greater than that of the body weight (Fig. 2). Peak concentric force was the highest force measured from the point of zero velocity of the COM to the point of take-off.

The maximal rate of force development (RFD) was defined as the largest increase in force over a 5 ms time window during the jump (both for SJ and CMJ; Fig. 2). Specifically, the RFD values (N/s) were calculated from numerical differentiation of the low-pass filtered force measurements using a 4-point method, and the derivative was averaged over 5 ms (10 samples). In the MJ test, the participants were instructed to jump ten consecutive CMJs as high as possible. The vertical stiffness was calculated as the maximal force divided by the downward displacement of the COM, while the reactive strength index (RSI) was calculated as jump-height divided by the ground contact time. All variables are presented as the average of the ten jumps.

Two to three maximal 20-meter sprint runs were performed on a rubberized indoor track (Mondo, Conshohocken, PA, USA) with 3–4 minutes’ rest between trials. The sprints were measured with an electric timing system (Biomekanikk AS, Oslo, Norway) with a timing trigger (single-beamed timing gate 0.6 m after the start line and 0.4 m above ground level) and dual-beamed timing gates placed every 5 m along the sprint track. Participants were instructed to accelerate as fast as possible from a standing start with one foot in front of the other. The inter-test CV for the sprint test is given in Table 2.

After a specific warm-up consisting of ten push-ups with gradually increasing effort and three maximal singles, three single maximal push-ups were assessed on a force platform (sampling rate: 2,000 Hz; OR6-5-1; AMTI, Watertown, MA). One minute of rest was given between the single push-up efforts. The participants were instructed to keep their body “straight” (minimize any movement of the spine and pelvis) and to do a controlled slow eccentric phase to a position where the chest was 2–3 cm above the floor, and then do a push as fast as possible. The hands were allowed to leave the platform if the push-force was large enough to lift the upper body off the ground (the feet were always in contact with the ground). Hand and foot placements were standardized for each participant. If the participants failed to conduct the push as described, the attempt was discarded and repeated (this was, however, a subjective decision by the test leader). The inter-test CV for the push-up test is given in Table 2.

Bench press and squat performance were assessed using a linear encoder (Musclelab Linear Encoder; Ergotest Innovation, Langesund, Norway). The string of the encoder was attached to the bar, with the device measuring displacement (d) and time of the concentric phase (200 Hz sampling rate; 0.019 mm resolution). The start and end of the concentric phase were detected as a 5 ms period of no movement (<0.004 m/s) or immediately by a change in direction (within 5 ms). The calculations of velocity and force from each load were based on the entire concentric phase (i.e., average/mean velocity and force; v = d/t; acceleration [a] = v/t, force [F] = mg + ma).

In both the bench press and squat the participants completed sets of three maximal repetitions at four different loads, with ∼5 s between each lift and 2–4 min between sets. All repetitions were conducted with maximal effort in the concentric phase. The external loads were 25, 50, 75 and 90% of estimated 1RM (estimated during the familiarization session). The attempts with the highest velocity from each load were selected for further analysis. A concentric force-velocity relationship (linear regression) and a power-velocity relationship (parabolic curve) were established and peak power and 1RM were estimated (software from Ergotest Innovation, Langesund, Norway). Peak power was calculated as the apex of the parabolic power-velocity relationship. The 1RM was calculated from the intercept of the load (mg) − velocity relationship and the force (mg + ma) − velocity relationship.

For the squat, the participants were instructed to squat down to a position where the femur was parallel with the floor, in a slow, controlled manner, and then extend as rapidly and powerfully as possible. For the squat we estimated force from the system mass (90% of body mass and the external mass), while for the bench press, only the external mass was used. We used 90% of the body weight for the squat calculations, as suggested by the manufacturer (Ergotest Innovation, Langesund, Norway). This is very close to the 88% of body weight suggested by others using a similar linear encoder device (Cormie, Mcbride & Mccaulley, 2007). The inter-test CVs for the bench press and squat tests are given in Table 2.

Statistics

The data were analysed in spreadsheets that enabled adjustment of one or two predictor variables in the changes within or difference between sessions (Hopkins, 2007). The spreadsheet is fundamentally based on the T-test but gives the opportunity to adjust for baseline to control for the regression to the mean effect. All data were log-transformed, and changes are reported as percentages with their associated 95% confidence interval (CI).

The reliability of the tests was based on the familiarization session and the two pre-tests (before each session). The coefficient of variation (CV) was calculated as described by Hopkins (2000). The smallest worthwhile change was calculated as the baseline between subjects’ standard deviation (SD) multiplied by 0.2 (Hopkins, 2004).

Effects were evaluated using clinical magnitude-based inferences (MBD) (Hopkins et al., 2009; Hopkins, 2019), a method appropriate for small samples. The magnitude of changes within and difference in mean between sessions were assessed by standardization (mean change/difference divided by baseline SD of all subjects), and the resulting standardized effect evaluated with a modification of Cohen’s (1992) scale: <0.2, trivial; 0.2–0.6, small; 0.6–1.2, moderate; >1.2, large (Hopkins et al., 2009). The subjective variables (RPE and PRS) were evaluated with the following scale: <10% trivial, 10–30% small, 30–50% moderate, 50–70% large, 70–90% very large, and 90–100% extremely large (Hopkins, 2010).The initial RPE and PRS values were therefore factored by 10 (0–100).

To make clinical inferences about true values of effects in the population studied, the effects were expressed as probabilities of harm or benefit in relation to the smallest worthwhile change (0.2 of SD; Hopkins et al., 2009). A clear change within or difference between the two exercise modalities corresponds to the case of an effect that is almost certainly not harmful (<0.5% risk of harm) and possibly beneficial (>25% chance of benefit). The effect is shown as the difference or change with the greatest probability, and it is shown qualitatively using the following scale: 25–75%, possibly; 75–95%, likely; 95–99.5%, very likely; >99.5%, most likely (Hopkins et al., 2009). In addition, p-values were included, and effects were considered significant at p < 0.05 if the 95% CI did not overlap zero (p < 0.01 with 99% CI).

Correlations between variables were obtained using Pearson’s r (and 95% CI). We restricted the correlation analyses to testing between objective and subjective variables that showed a difference between sessions, in order to minimize the risk of observing random correlations.

An order-effect is a potential risk with a crossover design (Woods, Williams & Tavel, 1989). Hence, we tested the session-order effect by including session order as a covariate. The effects were trivial (0.0–0.5%), so we did not further include the order effect to avoid too many covariates with the relatively low sample size (baseline value was already included).

Previous studies

Small to trivial impairments of neuromuscular performance were observed after both the exercise sessions. More specifically, measures of CMJ and SJ heights and sprint times were reduced by 1–8%, which are at the low end compared to previous studies (∼2–20%; Raastad & Hallen, 2000; Howatson, Brandon & Hunter, 2016; Raeder et al., 2016; Davies, Carson & Jakeman, 2018; Hiscock et al., 2018). We believe that this discrepancy is because our participants were well trained, and more importantly, they were familiarized with the exercises and tests.

In line with the existing literature (Linnamo, Hakkinen & Komi, 1998; Brandon et al., 2015; Howatson, Brandon & Hunter, 2016), a heavy-load strength-oriented session attenuated the neuromuscular system more than a low or moderate load power-oriented session. However, in previous studies where the exercise work was controlled for, the differences between strength- and power-oriented sessions were close to eliminated (Mccaulley et al., 2009; Hiscock et al., 2018). Our observations confirm these findings, but add some nuances to this picture, as we observed some clear differences between the strength-oriented session and the power-oriented session, such as for CMJ eccentric peak force. Nevertheless, the differences in recovery rates between resistance exercise sessions of different modes (strength- and power-oriented) with similar exercise work must be expected to be rather subtle in magnitude, but these differences may still be relevant information and important for athlete monitoring and training planning. When small differences are of importance, we must, however, ensure that we have adequate measurement methods.

Methodological issues: reliability and fatigue sensitivity

To discriminate between the recovery rates of closely related exercise modalities such as strength- and power-oriented sessions, highly reliable (day-to-day) tests must be applied. Based on our familiarization session, and two pre-session tests, we observed very high reliability for the sprint test (CV: ∼1%). CMJ and SJ height and estimation of 1RMs had good reliability (CV: 3–5%), while peak power in the squat and bench press and MJ height had acceptable reliability (CV: ∼9–10%). Push-up peak force reached near acceptable reliability (CV: ∼11%). Overall, the reliability of tests applied in this study is in line with those of others (Raastad & Hallen, 2000; Hopkins, Schabort & Hawley, 2001; Byrne & Eston, 2002; Cronin, Hing & Mcnair, 2004; Cormack et al., 2008; Taylor et al., 2010; Gathercole et al., 2015a; Gathercole et al., 2015b). One exception among our tests was the RFDmax gleaned from CMJ and SJ, which demonstrated poor reliability (CV >20%). Previous studies confirm a moderate to poor reliability for RFD measurements in single joint knee-extension (CV = 7–17%) (Buckthorpe et al., 2012), and for CMJ and SJ (CV = 16–18%) (McLellan, Lovell & Gass, 2011; Gathercole et al., 2015a). The low reliability for RFD is probably related to the complexity of the task (Maffiuletti et al., 2016), meaning that it is more difficult to achieve a true maximal RFD than a maximal force. This seems to be reflected in studies showing larger increases in RFD than in maximal force within a session (rehearsal) and as an effect of training (Holtermann et al., 2007). That said, better reliability of RFD measures might be achieved by other types of tests than the jump tests applied here; indeed, the isometric mid-thigh pull test appears to be a preferable choice to assess RFD in the lower body (Haff et al., 2015; Hornsby et al., 2017).

Performance tests may also be evaluated by comparing the “smallest worthwhile change” (SWC) with the typical error (Cormack et al., 2008): If the SWC is larger than the typical error, the test should allegedly be able to (confidently) detect relevant and meaningful changes. Among our tests, jump height and measures of force (concentric and eccentric peak force) demonstrated CVs equal to or lower than the SWCs (see Table 2). Nevertheless, an evaluation of tests must be applied in practice. Gathercole et al. (2015a) used the term “fatigue sensitivity”, which refers to a test’s ability to detect impairments in neuromuscular function after exercise. As the conditions of the neuromuscular system change, due to different forms of central and peripheral fatigue (Enoka et al., 2011), high reliability measured in the rested state is not necessarily valid for the fatigued state. In fact, tests of isolated joints, such as isokinetic knee-extension assessments, appear to demonstrate larger changes than multi-joint tests, such as sprint and jump tests after different multi-joint activities (Byrne & Eston, 2002; Andersson et al., 2008; Howatson, Brandon & Hunter, 2016). To this end, we suggest that tests enabling subtle changes in the movement pattern, such as sprint and CMJ, may be highly reliable, but may lack fatigue sensitivity. Subtle movement/technique compensations that optimize the conditions for the current state of the neuromuscular system may indeed “mask” fatigue if only jump height in a CMJ is considered (Van Ingen Schenau et al., 1995; Gathercole et al., 2015b).

As indicated above, there were trivial changes in CMJ height and peak power 24 and 48 h after both sessions, but clear changes in CMJ eccentric time and CMJ eccentric peak force. Similar findings have recently been reported by others (Gathercole et al., 2015a). These observations indicate that the participants’ ability to use the eccentric phase was impaired in the recovery phase, but some compensations in the execution of the jump apparently minimized the reductions in jump height and power production. After the strength-oriented session the reduction in eccentric peak force seemed related to a slower eccentric phase during the CMJ; i.e., increased eccentric time, since the lowering the of centre of mass was not changed. On the contrary, the participants appeared to lower their centre of mass more after the power-oriented session than at pre-test, especially at 48 h. Future studies should investigate changes in the kinetics and kinematics (movement strategies) of a CMJ in the recovery phase compared to the rested state. However, we suggest the eccentric peak force is a more sensitive marker of fatigue and neuromuscular impairments than jump height and maximal power.

We found no clear meaningful differences between sessions or in the recovery rates between sessions for CMJ and SJ heights. This contrasts with observations by Byrne & Eston (2002), who reported that SJ height was reduced more and recovered slower than CMJ (and drop jump) height after a squat exercise session (10 × 10 repetitions at 70% of body weight). The discrepancy between findings may be related to more muscle damage in the study by Byrne & Eston (2002) than the present study—as indicated by a larger drop in performance (Paulsen et al., 2012). Moreover, studies have investigated various measures of RFD and observed that the impairment and recovery of RFD differ from maximal force (Penailillo et al., 2015; Farup et al., 2016). In our study, we extracted RFDmax from CMJ and SJ, and despite low reliability, we report small unclear and possibly clear differences between sessions at 24 and 48 h—in accordance with previous observations (Gathercole et al., 2015a). Thus, we recognize RFDmax values from jump tests as possibly fatigue sensitive, but we warn about high day-to-day test variability (as discussed above). Moreover, the reader should be aware that sampling frequency and the methods used for (concentric/eccentric) phase identification may affect the outcomes of SJ and CMJ analyses (Owen et al., 2014; Eagles et al., 2015). Hence, comparisons across studies must be made with caution.

From the force-velocity tests in bench press and squat we calculated peak power and estimated 1RM. The 1RM values had allegedly good reliability (CV<5% and CV<SWC), but contrary to the peak power, the 1RM values showed trivial changes after both exercise sessions. Although it has been suggested to be worth using (Jovanovic & Flanagan, 2014; Scott et al., 2016), force-velocity estimated 1RM appears to have limited value for monitoring small changes in recovery status; i.e., estimated (or predicted) 1RM tests appear to have low fatigue sensitivity. We applied ∼90% of 1RM as the heaviest load, which may have been too low to get an accurate estimation of 1RM in the squat, as observed by some (Banyard, Nosaka & Haff, 2017). For the bench press, however, ∼90% of 1RM should be adequate for precise 1RM estimations—at least in an unfatigued state (Jidovtseff et al., 2011; Garcia-Ramos et al., 2018a).

Mechanisms for neuromuscular recovery

Exercise-induced impairment of neuromuscular function and the following recovery phase are multifaceted (Lieber & Friden, 2002; Enoka et al., 2011; Paulsen et al., 2012). However, if we consider a particular exercise, such as the squat, and assume a constant range of motion (muscle lengthening/strain) and a given total exercise volume (sets × repetitions), the determining factors would be narrowed down to contraction/lengthening velocity and force. With the criterion of maximal effort (intention to move) in the concentric phase, velocity will be high and force low during light or moderate load power exercises, and vice-versa for heavy load strength exercises (cf. the force-velocity relationship (Huijing, 1998)). Higher concentric forces during the heavy load strength exercises will logically put more mechanical stress on the muscle tissue. However, high-force concentric contractions result in minimal muscle damage and a swift recovery of muscle function within 24 h (Jones, Newham & Torgan, 1989; Lee, Suter & Herzog, 1999; Carson, Riek & Shahbazpour, 2002). Thus, concentric work can probably only explain perturbations in neuromuscular function shortly after exercise (i.e., minutes to a few hours, as a result of metabolic factors; Allen, Lamb & Westerblad, 2008). This led us to suggest that the eccentric phase was probably of greatest importance in the differences in neuromuscular impairment and recovery rates between sessions (Paulsen et al., 2012). In other words, the higher eccentric forces—simply due to higher loads—during the strength-oriented session likely explain the slower recovery compared to the power-oriented session (Faulkner, Opiteck & Brooks, 1992; Black et al., 2008). On the other hand, the between-session differences displayed by the recovery markers were generally small and trivial compared to the significant difference in loads (the loads in the power session were 50% of those in the strength session). Therefore, we propose that the higher eccentric velocity during the power-oriented (compared to the strength-oriented session) caused a substantial mechanical stress on the muscles, despite the moderate loads: Stretch-shortening cycle exercises have, indeed, been shown to induce muscle damage and require days of recovery (Nicol, Avela & Komi, 2006). Future research should investigate this, but we suggest that how the eccentric phase during power-oriented exercise is performed and the utilization of the stretch-shortening cycle could have a major impact on recovery times.

Lower and upper body exercise

In the present study, both upper body and lower body exercises were applied. Studies exploring muscle damage and recovery after eccentric exercise have reported that upper body muscles sustain more damage and require longer recovery times than lower body muscles (Jamurtas et al., 2005; Chen et al., 2011; Chen et al., 2019). However, recovery rates after traditional strength training do not appear to be different between upper and lower body exercises, such as the bench press and squat (Mclester et al., 2003; Korak, Green & O’neal, 2015; Moran-Navarro et al., 2017). In line with these studies, our data demonstrate a similar recovery rate for upper and lower body exercises. Moreover, as for the lower body, the strength-oriented session seemed to induce somewhat more fatigue and longer recovery times than the power-oriented session for the upper body. In contrast to most studies that have investigated recovery after eccentric exercise (as cited above), we recruited well-trained individuals, which points to training status as an important parameter for recovery times—rather than an inherent difference between upper or lower body muscles. Nevertheless, great care should be taken when comparing recovery from different exercises/sessions, because variables such as muscle strain, force and work are very difficult to control for.

Fatigue vs. potentiation and supercompensation

Neuromuscular function can be altered through adaptation to training over weeks and months (Goldspink, 1985), but the neuromuscular system is also history-dependent for shorter time periods. In fact, both fatigue and potentiation are possible outcomes of muscle contractions (Sale, 2002). While heavy loads and large exercise volumes may induce long-lasting neuromuscular fatigue (hours and days), exercises conducted with low volume and high/maximal effort can result in potentiation and enhanced neuromuscular function that lasts for minutes to several hours (Cook et al., 2014; Russell et al., 2016). Interestingly, in the present study the power-oriented session appeared to enhance MJ height at 24 h and squat peak power and push-up peak force 48 h after exercise (note that the push-up peak force at 48 h was trivial and unclear compare to baseline, but clearly different between sessions). This is in line with Tsoukos et al. (2018), who observed increased CMJ height and RFDmax 24 and 48 h after loaded jump squats (40% of 1RM; 5 × 4 repetitions). In contrast to squat peak power, we observed no such “supercompensation” in CMJ, SJ or 20 m sprint (which were all back to baseline at 48 h). Notably, our participants executed a large exercise volume, about three times that of Tsoukos et al. (2018), and fatigue mechanisms may have overshadowed most of the supercompensation effects of power exercises. Moreover, we only followed the participants for 48 h, which means that we do not know whether the supercompensation occurred later after the strength-oriented session (e.g., after 72 h). As final note, potentiation effects (or supercompensation) is indeed relevant for athletes, as it is common practice for “power athletes”, e.g., rugby players, track and field throwers and sprinters, to perform a power-oriented session close to competitions (∼4-48 h; Russell et al., 2016; and own observations from the Norwegian Olympic Center, Oslo, Norway).

Objective vs subjective measures of recovery

Session RPE (sRPE) for resistance exercise was reviewed by Mcguigan & Foster (2004) and validated for “intensity”; i.e., load in % of 1RM, by Sweet et al. (2004). Later studies have found the sRPE to be related to both volume and work rate during strength training (Scott et al., 2016; Hiscock et al., 2018). The present study ensured equal concentric work, but different loads—i.e., the power-oriented session was performed with 50% of the loads used in the strength-oriented session. Nevertheless, because the power-oriented session lasted ∼12% (∼13 min) longer than the strength-oriented session, the work rate was highest during the strength-oriented session. As the difference in loads (% of 1RM) between sessions was much larger than the difference in work rate, we suggest that the higher loads (% of 1RM) were the dominant factor influencing the sRPE scores (although we acknowledge that this cannot be ascertained with the present study design). Notably, it has been proposed that exercise intensity/load (% of 1RM) influences RPE scores via a positive relationship with the central motor control discharge (Gearhart Jr et al., 2002), cf. the “corollary discharge model” (Pageaux, 2016). However, our participants in both sessions were strongly encouraged to execute every repetition with the intention to move as fast as possible in the concentric phase. Indeed, both the motor-related cortical potentials (MRCP; Slobounov, Hallett & Newell, 2004) and the electromyographic (EMG) amplitude seem independent of load (% of 1RM) if the intention to move is maximal—at least for lower body exercises (Bosco et al., 1982; Hakkinen, Komi & Kauhanen, 1986; Kawamori & Haff, 2004; Mcbride et al., 2010). If we assume that our participants moved maximally in all repetitions, the corollary discharge model seems unable to explain a higher sRPE after the strength-oriented session than the power-oriented session. Consequently, we suggest that the sRPE scores in the present study were influenced by afferent feedback from the muscles; supporting a “combined model” (Pageaux, 2016). The afferent feedback may be a combination of different sensors including tendon organs (“force sensors”) and nociceptor receptors responding to metabolic perturbations. Metabolic perturbations, such as elevated extracellular levels of adenosine, lactate and protons (Allen, Lamb & Westerblad, 2008), stimulate capsaicin fibres (Aδ and C-nerves; Pollak et al., 2014); and accordingly, muscular fatigue may be an important underlying mechanism behind the RPE scores (Hardee et al., 2012; Vasquez et al., 2013). When working at maximal intensity, fatigue will start to develop within seconds (Allen, Lamb & Westerblad, 2008), and probably to a larger degree during the strength-oriented session than the power-oriented session due to more time under tension (i.e., a longer acceleration phase during the lifts and/or less deacceleration). We cannot exclude the possibility that the participants used elastic energy storage and release (the stretch shortening cycle) during the power-oriented session, and thereby had better energy economy during the power-oriented session than the strength-oriented session (Bosco et al., 1982). Higher energy expenditure and more fatigue in combination with the heavier loads could explain the higher sRPE after the strength-oriented session than the power-oriented session. Finally, it is noteworthy that the “contents”/definition of the RPE concept, i.e., effort vs. force, pain and discomfort, and the mechanisms behind RPE, are debatable (Pageaux, 2016). Moreover, the timing of reporting RPE, e.g., during or immediately after an exercise vs. 30 min after a session (i.e., sRPE), may be important for the decisive mechanisms of the RPE scores; thus, more scientific work is needed to better understand the use of sRPE in relation to different modes of resistance exercise.

While sRPE scores are collected after a session, PRS is obtained before an exercise session. PRS is supposed to give an evaluation of the athletes’ readiness and performance status in the upcoming session (Laurent et al., 2011). In the present study, recovery status 24 and 48 h after the strength-oriented session were reported lower compared to the power-oriented session. Indeed, as for sRPE, PRS pointed in the same direction as the objective tests. However, no consistent correlations were found between the PRS and objective variables, such as CMJ eccentric peak force and push-up peak force. Interestingly, the state of recovery was perceived as incomplete both 24 and 48 h after the power-oriented session although performance was back to baseline, or even above (squat peak power and MJ). Recent studies support a partly dissociated time course between objective and subjective recovery status—for both upper and lower body muscles—indicating a slower recovery when assessed subjectively (Zourdos et al., 2016; Ferreira et al., 2017a; Ferreira et al., 2017b; Marshall, Cross & Haynes, 2018). In summary, this advocates for caution in interpreting subjective and objective measures of recovery. In our case (and perhaps most cases), it is conceivable that neither the subjective nor the objective measures revealed the true recovery status. On the objective side we merely measured some properties of the neuromuscular system, leaving the possibility that unassessed properties were not recovered. Interestingly Zourdos et al. (2016) observed a difference in the PRS when assessed before and after warm-up (higher PRS after warm-up). We assessed PRS only before warm-up, leaving the possibility for higher coherence between objective measurements and PRS if evaluated after warm-up.

Limitations

The present study has limitations. First, we applied a series of tests and we cannot exclude the possibility that the tests themselves induced fatigue that affected the results; e.g., reduced the test reliability. Moreover, we had no control trial in which the participants simply conducted the four test-sessions without participating in an exercise session (see Fig. 1). Consequently, we must be careful interpreting the changes in relation to time after each session (within-session changes); it is possible that the recovery was prolonged due to all the tests.

Second, we calculated the work done based on concentric work; thus, we excluded eccentric work, and we cannot rule out that some differences between sessions could have been explained by this fact.

Third, each participant completed two sessions. Due to the repeated bout effect, a faster recovery must be expected after the second session (Mchugh, 2003). Moreover, since the loads (in % of 1RM) were higher in the strength-oriented session, the adaptative processes may have been better stimulated after the strength- than the power-oriented session (i.e., strengthening of the myofiber cytoskeleton (Paulsen et al., 2009)). If true, this may have created a bias toward faster recovery after the power-oriented session. Furthermore, the time between sessions (the washout period) varied between the participants (1–4 weeks), which means that their training status may have changed slightly. This effect does appear small as the pre-values before each session were very similar, with low to moderate CV for all variables (Table 2). To this end, the order of sessions was randomized, and the impact of session-order was trivial when controlled for. Additionally, we recruited both females and males. The participants had different training backgrounds and we did not control their training in the washout period (except during the 48 h before each session). We did not fully control the diets of the participants. We acknowledge that these factors may have induced biases and variability in our results.

Fourth, we did not include tests that allowed us to distinguish between central and peripheral fatigue, nor did we measure systemic markers of recovery (such as creatine kinase, testosterone and cortisol; Buckthorpe, Pain & Folland, 2014; Hiscock et al., 2018; Tsoukos et al., 2018). This could have given us valuable information about the subtle impairments of neuromuscular performance and recovery between sessions.

Fifth, we acknowledge that the definition of the different variables gleaned from the SJ and CMJ tests are open for debate. Particularly, we want to make the reader aware of the fact that CMJ peak eccentric and concentric force are reached within a very narrow time window in the lowest position of the jump. Thus, collecting only the force in the lowest squat position could yield the necessary information, with the advantage that the point/position is clearly defined (easily reproducible).

Finally, one should be careful about extrapolating the results of this study to other training interventions/programs due the many combinations/possibilities within a strength/power training program that may be important in the recovery process, such as exercises, load, volume, work and interest-rest periods.

Practical applications

Knowledge of recovery from exercise sessions is needed to make qualified assumptions when designing training programs, particularly for elite athletes who must handle large training volumes and avoid overtraining. The present and previous studies have shown that to monitor recovery one must consider a combination of tests and be aware of the error of measurements. In our study, the eccentric peak force during a CMJ and the peak power calculated from a squat force-velocity test were the variables that seemingly best differentiated between a strength-oriented session and a power-oriented session. Further research is warranted to see whether these tests are valid for other modes of resistance exercise and with participants of different performance levels (training status).

In our hands, RFD from CMJ and SJ seem to have too large a day-to-day variability to be recommended for monitoring recovery. Improved standardizations and instructions to the athlete may be worth exploring. Similarly, for the upper body our applied tests were not fully satisfactory in terms of reliability and fatigue sensitivity, implying that more work is needed.

The power-oriented session tended to improve performance in certain tests at 24 and/or 48 h after exercise. Potentiation or a fast supercompensation from power-oriented sessions is highly relevant for athletes preparing for competitions.

Objective and subjective tests of recovery may not correlate. Consequently, both test modalities should be used and interpreted together to ensure a holistic approach (Kiely, 2012). Because the recovery process is so complex, it is important to acknowledge that there is much we do not know or understand; thus, relying on only objective or only subjective measurers could prove inadequate for most athletes.

It appears that the best tests for assessing recovery will differ significantly according to the exercises that have been conducted. Consequently, we cannot expect a “gold standard” test battery. Rather, we need to use a selected number of tests for each specific athlete or group of athletes, and a combination of subjective and objective tests appears advisable.