All athletes answered a questionnaire just before participation. Anthropometric data including height, weight and dominant leg were obtained before testing. Prior to testing, participants completed their regular five-minute warm-up of jumping jacks, shuttle run, stretching and agility exercises. Athletes who were injured, undergoing rehabilitation, or those with less than three years of sports experience were excluded from the study. The assessment was performed according to the guidelines without footwear on plain firm ground without rebound property. In case of an invalid trial, the (sub-)test was repeated once.

The study was conducted with the Back-in-Action (BIA) test battery (CoRehab,Trento,Italy). This assessment was originally designed for patients after ACL surgery to guide the return to sport process. Moreover, this series of test offers objective cut-off values for athletes before a return to competition. A detailed description of the test- battery has been published, with an observer independent test-retest reliability that varies between 0.688 and 0.921.^19,21

The test battery determines agility, strength, balance, and speed. Subtests are performed two-legged and one-legged with the dominant (d) and non-dominant (nd) leg in the following order.

The dominant leg was defined as the leg, which an athlete described as his stronger leg in the questionnaire before participation.

The collected data were processed and analyzed using SPSS 26.0 (IBM Statistics, Armonk, NY, USA). Normal distribution was tested using the Kolmogorov-Smirnov-test. Depending on the results analysis of variance (ANOVA), Mann-Whitney-U test (2 groups), or the Kruskal-Wallis test (3 groups) were used for further analysis. For correlation analyses, Pearson correlation was applied. The correlational values (r= Pearson`s r) were interpreted as negligible (0.00- 0.30), low (0.30- 0.50), moderate (0.50- 0.70), strong positive (0.70- 0.90) and very strong positive (0.90- 1.00). Statistical significance was determined at p< 0.05.

Pre-testing anthropometric data is presented in Table 1. Group 1 (G1) athletes were significantly heavier with an elevated BMI compared to G2 and G3 (p< 0.001), whereas G2 and G3 showed comparable body weight and BMI. Significant differences in height were shown between G1 and G2 (p<0.001), but not when compared with G3.

Balance measures of one-legged and two-legged wobble board performances are subsumed in Table 2. The results of the entire group of football athletes demonstrate significantly poorer balance scores for two-leg test (p<0.002) and significantly poorer scores for each leg separately (OL/d: p<0.001; OL/nd: p<0.001) in comparison with the reference population (RP).

For subgroups, G1 showed significantly worse scores in the two-legged and one-legged balance tests compared to the RP (TL; OL/d; OL/nd: p<0.001) as well as to G2 (TL; OL/d; OL/nd: p<0.001) and G3 (TL; OL/d; OL/nd: p<0.001).

G2 and G3 showed no statistically significant differences in the two-legged and one-legged balance tests, equivalent to the results of the RP.

Correlation (Pearson r) of Balance scores to anthropometric data showed a low correlation for body height and wobble board performance (r: TL Bal.: +0.36; OL/d: +0.37; OL/nd: +0.4; p<0.05).

For leg symmetry, 25 of 53 athletes (47%) showed a leg difference of over 10%.

A summary of all CMJ measurements is presented in Table 3 for height measures (cm) and Table 4 for power measures (watt/kg). Absolute jump height was not significantly different between all of the football athletes and the untrained subjects (RP) in the CMJ TL, OL/d and OL/nd (p>0.05). In the TL-CMJ, G3 achieved significantly higher jump heights than G1 (p<0.001), but not in comparison to G2 (p>0.05) or the RP (p>0.05). G1 and G2 showed not statistically different jump heights compared to each other and to the RP (p>0.05).

Jump heights of the OL-CMJ d showed no significant differences between all groups (G1 vs. G2 vs. G3 vs. RP: p>0.05). For the non-dominant limb (OL-CMJ nd) G1 showed significantly lower jump heights than G2, G3 and the RP (p<0.001), which were not significantly different.

The football players overall generated a significant higher power output in all jump tests (CMJ TL: p<0.001; OL/d: p<0.001; OL/nd: p<0.001) than the RP.

CMJ jump height symmetry over 90% was achieved by 18 of 53 athletes (34%).

The power as measured during the TL-CMJ showed no significant differences in G1, G2, G3 (p>0.05). However, G2 and G3 showed significantly better power output values compared to the RP (p<0.001), while G1 did not (p>0.05). Regarding the power output in the single-leg CMJ, no differences were found between the groups OL/d and OL/nd, but athletes were superior to the RP (G1 vs. RP: p<0.001, G2 vs. RP: p<0.001, G3 vs. RP: p<0.001).

53% (28 of 53) of the tested subjects achieved a symmetry index of >90% for power development in the CMJ.

Time measure results of the Parkour and Quick feet task are presented in Table 5. In the Parkour jumps, American football athletes overall and each group individually were slower than the RP (p<0.001).

G1 performed significantly slower compared to G2 and G3 with the dominant (G2 vs G1: p<0.001; G1 vs. G3: p<0.001) and non-dominant legs (G2 vs. G1: p<0.001; G3 vs. G1: p<0.001).

Regarding the Quick-Feet test completion times between football athlete and RP groups (p>0.05) were not significantly different. However, G3 was significantly faster than G1 (p<0.05), while no other significant differences between the groups occurred (G1 vs. G2: p>0.05; G2 vs. G3: p>0.05).

For correlation measures, a low correlation was found for on legged balance scores with increased time in the parkour test. (OL/d Bal-OL/d Park.: r = +0.46, p<0.05; OL/nd Bal - OL/nd Park.: r = +0.43, p<0.05). A Correlation for power measures of the CMJ and parkour times of the same limb was not found.

Average symmetry index measures are given in Table 6. Mean measures of the differences in performance between the two legs (symmetry index) of the American football athletes for the one-legged balance and one-legged CMJ jump height were over 10% (mean under 90%) for all three position groups. There were no significant differences between the position groups (p>0.05).

For counter-movement jumps, football players generated about 46.86W/kg (TL), 36.94 W/kg (OL/d) and 37.36W/kg (OL/nd), which significantly greater than the reference population. For American Football, explosive power in movement is an important performance parameter, as it is a discontinuous sport with a series of recurrent intense anaerobic power peaks.^33,34 Vertical jumps, broad jumps, the 40m Dash, and CMJ are part of the NFL Combine and therefore a fundamental part of the training routine used by football players, which likely affects the development of power.^33,35 The fact that jump heights of 36.3cm (TL), 22.21cm (OL/d) and 22.8cm (OL/nd) were similar to untrained individuals, is probably due to the elevated body weight of players.

Although the power measures of the American football athletes were superior to the untrained reference group, they are worse when compared to values seen in prior studies. Leutzinger et al. evaluated high school athletes (15-17y) using the NFL Scouting Combine tests, including the vertical jump. Their athletes achieved a vertical jump height of 62.9 cm.^35,36 This may be due to the fact that the vertical jump test of the NFL combine (as investigated by Leutzinger et al.) is executed with help of the arms, which can lead to about 25% increase in jump height of CMJs according to Sayers.^37,38 Still, after multiplication by 1.25, the current results add up to 45.28 cm, which is still diminished, but comparable to 12-year-old players.³⁹ This may be due to the methods used in the current study: the digital Myotest accelometer versus Leutzinger`s analogue Vertec jump trainer are different and accuracy might be impaired.³² Unfortunately, only a limited number of studies depict power development in Watt/kg of CMJ in 16-17 years olf American football players. McKay and Leutzinger reported approx. 6008 watts, 64.99 W/kg body weight.^35,36 When applying Sayers formula, peak anaerobic power output averaged at approximately 4806 watts, about 53 W/kg in this investigated population of American football players.³⁷ Once again, different measurement devices may partly account for these differences or even hide greater differences, as calculation of watt/kg with the Myotest sensor remains unanalyzed and Vertec measurements in an analogue manner rely on half inch increments only.³⁰ Power development itself is the most useful performance parameter in football: it is associated with the career longevity, a predictor for injury risk and the RTS capability.⁴⁰ As the BIA test battery was originally designed to discriminate athletes recovering from an ACL injury regarding their readiness to return to sport, this aspect may be particularly interesting. Regarding ACL injury, the quadriceps to hamstring ratio is an important indicator of function in ACL deficient knees.⁴¹ Additionally, quadriceps strength predicts the long term function after ACL reconstruction.^42,43 Clearly, isometric measures of strength were not undertaken in this field test investigation and therefore comparison is insufficient. Worth mentioning, power development ratio in counter movement jumps measured by a force platform was most recently correlated to ACL injury in the future by Pontillo.⁴⁴

In comparison to an untrained population (RP), the American football players of this study performed significantly worse in multiple subtests. The results in the one-legged balance test (Bal. OL d/nd) were particularly notable, in which American football players performed significantly worse. Previous researchers report ambiguous data on wobble board performances in association with body height, limb length, and the related center of gravity.^45–48 In the current study there was only a low correlation between body height and worse balance scores. During NFL Combine testing balance is not measured, but agility, coordination, and explosiveness are with a three-cone drill, which is similar to the Parkour Jumps used in the BIA test battery. It is debatable whether single-leg balance or the single-leg CMJ results are associated with the results in the single-leg Parkour Jumps. A low correlation between worse one-legged balance scores and higher Parkours times was found in our athletes. However, there is no correlation between power measures in the respective OL CMJs and parkour times, even though the parkour consists of a sequence of repetitive single-leg jumps.

The results of the Quick-Feet test are not significantly different to the results of the reference population. This is surprising, as similar exercises like the Quick-Feet Ladder are an elementary part of football training routine.⁴⁹ A better result for football players was expected in this assessment as was seen in the power results from the CMJ.

In addition to the performance parameters mentioned above, the symmetry values regarding balance, jump height, jump force and parkour times are also relevant for the evaluation of the RTS permission of the BIA test battery. Moreover, this measure has been proposed by many researchers before with varying target values between 85-95%.^28,50 Unexpectedly, the symmetry indices of the BIA test battery were commonly not met by athletes nine months post ACL reconstruction surgery.²⁶ Further research on limb symmetry has shown decrements in physical performance in athletes with inter-limb symmetries as measured by the balance test.⁵¹ In contradiction, regained limb symmetry after ACL injury may overestimate knee function.²⁶ For the BIA test battery, a interlimb difference >10% (symmetry index<90%) is set as cut off for the RTS, which was validated by testing healthy, uninjured athletes.^19,52 Only 53% of the American football athletes in the current study (28 of 53) achieved a symmetry index of >90% for power development in the CMJ, only 18 (34%) in jump height measures, only 25 (47%) in balance scores. This may also be relevant to performance parameters, as Fort-Vanmeerhaeghe et al. demonstrated that increased asymmetry is associated with lower acceleration and consequently slower 30m sprint times, as well as increased risk of injury.⁵¹ This relationship is illustrated in the current results, as G1 players with low symmetry indices and overall worse BIA results have also been considered in the literature to be the most injury-prone playing position.⁷

Research on healthy high-level athletes in Judo and Taekwondo has revealed interlimb differences in a different test, the so called “3-hop test”, of over 10 percent in almost 25% of the athletes and the significance of interlimb differences in uninjured trained athletes remains a matter of debate.^28,50,52,53 Therefore, it remains questionable whether interlimb symmetry is a reliable marker for a RTS decision.^54,55 Also, the ratio between the extremities does not consider absolute performance results. Normalized data for jump and hop tests are necessary.⁵²

Considering the original purpose of the BIA-assessment, the results of the current study are particularly interesting. The test battery sets a threshold of a BIA-score of 3.0, a calculated score after all tests, for a RTS permission. This score has a scale from 1 (best) to 5 (worst) in all four qualities/categories: balance, agility strength and speed. Among these qualities, different tests used to form a score for each quality, which is later averaged in the BIA-Score.^19,21 In the hypothetical case of injury or rehabilitation, only 17% (9 out of 53) of the athletes would achieve RTS permission.^19,21 Herbst et al. applied even more restrictive thresholds for high-risk sports (e.g. alpine skiing, soccer, handball or American football), requiring at least a BIA-score of 4.0 to receive RTS permission.¹⁹ Considering these cut-off values, none (0%) of the athletes would reach RTS criteria with their performance, although none of the athletes were recovering from injury, but were participating in routine practice and games. This is consistent with results from Australia and the Netherlands, where the RTS criteria of this test battery were met by only 2.5 % and 17.5% 9 months / 11 months post ACL surgery respectively.^24,27 There are two possible explanations for these unexpected results. It may be that the BIA test battery has too high standards for RTS and therefore may have limited utility. This is consistent with the low passing rates post ACL surgery as described above. Or, in contradiction, the examined American football players just did not perform well at all which may be a warning signal for coaches and staff. Uncovered deficits should be addressed by specific training to minimize the potential risk of injury. In this context, the BIA test battery can be used as a training tool as well for progress measurement.^19,21

Significant differences in performance were noted among the respective position groups in certain tests. Athletes from G2 and G3 had similar anthropometry and achieved equivalent results in the tests, whereas differences in anthropometric characteristica and results were noted with G1, which performed significantly worse in most assessments, but not in power development (CMJ) and the Quick-feet task. It is likely that these assessments are most likely to be part of the training routine and therefore well known to most of the athletes.^36,49

BMI itself appears to correlate with outcomes and injury risk, as has been postulated for the NFL roster status of linebackers and defensive linemen.^40,56 The current results affirm these findings, as low to moderate significant correlations were found between increased BMI and worse results in the assessments As expected, lower BMI (G2 and G3) was associated with better results in balance and agility, which is contrary to the power measurements. When calculating anaerobic peak performance according to Sayers, athletes from G1 showed significantly higher absolute values in anaerobic peak force development (p<0.05), but not with respect to body weight (power per kg). These findings are consistent with results by Leutzinger et. al, where athletes of the offensive line and defensive line (G1) performed worse in the pro-agility drill and L-cone drill, which also require good agility and balance.³⁵ These results are interesting as they present in the early stage of the career, not only in athletes joining the NFL combine. Playing positions that require tackling (G1) had heavier athletes and showed higher power outputs. This is likely due to the requirements of this position which demands explosive power in just a few steps and in one direction. Players, who participate more in the passing and running game, are lighter, show better agility and balance, and similar power when normalized by mass (kg).

Within the context of injury and prevention, the categorization among these three groups is relevant, as well. In youth and professional football, linemen have a threefold increased risk of injury, followed by running backs and wide receivers, possibly due to frequent involvement in play actions and tackling.^2,57–62 The current measurements reveal impaired results for the above listed position Group 1 including linemen (G1), which were alarmingly even below results of an age matched reference population. Comparably, recent studies have found a correlation between injury risk of the lower extremity and BMI >26 kg/m² in high school aged linemen.⁵⁶ In recent studies, high quadriceps strength has been determined as an important predictor of a safe RTS ability.^63,64 In the assessments that determine power (CMJ), G1 achieved high power-output performance values with a high inter-limb symmetry over 93%. Contrary to this, other symmetry values of GI athletes and assessment results are below the thresholds of a safe RTS. Further investigation is needed to determine which predictor (e.g. symmetry of power or balance) has the strongest influence and how performance parameters as well as functional assessment results should relate to each other for a safe RTS. The usefulness of the BIA test battery with the existing cut off values for high school American football players has to be questioned.

The authors report no conflicts of interest.