Clinical Implications of Landing Distance on Landing Error Scoring System Scores

The sample-size calculation for this study was based on data from 2 previous studies of the LESS, one according to the original protocol (50% body height)¹⁰  and the second using a modified protocol (25% body height).³  Both investigations involved similar cohorts (29 and 20 physically active males, age = 18.5 ± 2.5 and 20 ± 2.0 years, respectively). We applied standard 2-tailed hypothesis equations, 95% power (β = .05), 5% significance level (α = .05), critical values of the t distribution, and data from these studies^3,10  to calculate the required sample size. These equations indicated that 64 participants were needed to identify group differences in mean LESS scores between the jump distances. To account for 10% potential withdrawals and missing data, we recruited 70 participants.

Participants had to be 16 to 30 years old, regularly engaged in physical activity (at least once a week) at any level, and free from injury, pain, or any other concern that would limit physical activity. Both sexes (males and females) were included. Previous injuries were not an exclusion criterion. The study protocol was approved by our institution's health research ethics committee (HREC [Health]#41) and adhered to the Declaration of Helsinki. All participants signed a written informed consent document that explained the potential risks associated with testing.

The testing procedure in both experimental conditions was identical to that described by the developers of the LESS,¹  with the exception of the landing distance from the box. During the jump-landing task, we required participants to jump from a 30-cm-high box under 2 landing conditions: (1) set distance of 50% of their body height (d_50%), and (2) self-selected distance (d_ss) for which landing distance was not set. We instructed participants to immediately jump upward for maximal vertical height upon landing. We emphasized actively jumping (not dropping) off the box with both feet, jumping as high as possible straight up on landing from the box, and completing the task in a fluid motion. We did not provide any feedback on landing technique unless a participant was performing the task incorrectly. After receiving the task instructions and practicing jumps for familiarization (typically 1), each person performed 3 successful jump-landing trials under each landing condition. The order of the 2 landing conditions was randomized. We allowed participants to rest between trials within conditions to limit fatigue until they felt ready to perform the task (typically 1 minute); at least 15 minutes of seated rest was allowed between conditions. All tests were completed in a single experimental session.

Two standard video cameras capturing images at 120 Hz (model RX10 II; Sony Corporation) with an actual focal length of 8.8 to 73.3 mm (35-mm equivalent focal length of 24–200 mm) recorded the jump-landing tasks. We mounted the cameras on tripods placed 3.5 m in front of and to the right side of the landing area with a lens-to-floor distance of 1.3 m. A qualified physiotherapist who had conducted more than 400 LESS evaluations replayed the videos using the Kinovea software (version 0.8.15) and scored all 6 trials using the 17-item LESS scoring sheet.¹  The mean LESS score from the 3 trials completed under each condition was used for statistical analysis. The physiotherapist could not be blinded to the landing condition due to the visibility of the landing distance on the videos; however, the assessor was blinded to the participants' scores under the alternate condition. The physiotherapist also used the Kinovea video-analysis software to measure the length of the jump (distance from the box to the heel closest to the box during landing) for each trial after calibrating the video to a 1-m ruler.

We assessed the effect of landing condition on (1) group mean LESS score, (2) group-level risk categorization (proportion of participants categorized as high [LESS: ≥ 5 errors] and low [LESS: < 5 errors] injury risk), and (3) individual-level risk categorization (consistency of high or low injury risk category). The landing distance (in cm and expressed as a percentage of body height) and the proportions of specific LESS errors in the 2 conditions were also compared.

The influence of the landing condition on group mean LESS score, group-level risk categorization, and landing distance was estimated using a generalized estimating equation (GEE).¹⁴  The GEE approach provides an estimate of the average effect in a population, applying robust standard errors to account for within-individual correlations. We used the GEE model with a Gaussian (normal) distribution to explore the influence of the landing condition (d_50% versus d_ss) on the group mean LESS score and landing distance. We applied a binominal distribution to explore the influence of landing condition on group-level risk categorization to estimate the odds ratio (OR) of being categorized as high injury risk in the d_ss compared with the d_50% condition. Both GEE models applied an exchangeable correlation structure.

To explore the individual-level risk categorization, we assessed the agreement (n and %) in risk categorization between the landing conditions using ORs and the 2-tailed McNemar test. The OR indicates whether a landing condition was more likely to categorize individuals as high injury risk: specifically, the number of participants at high risk exclusively in the d_50% condition divided by the number of participants at high risk exclusively in the d_ss condition. We calculated the McNemar test to compare 2 proportions, ie, whether the proportions of participants at high risk significantly differed between conditions. McNemar tests were also used to compare the proportions of specific LESS movement errors for LESS items 1 to 15 (scored 1 = error present, 0 = error absent) between conditions. Due to different scoring of items 16 and 17 (0 to 2 errors), we used the 2-sided Wilcoxon signed rank test to compare these items between conditions.

We set the significance level at P ≤ .05 for all analyses. The statistics were computed using Microsoft Excel for Office (version 365) and RStudio (version 1.1.463) in R (version 3.5.2). All participants finished the study, and the complete dataset was analyzed.

A sample of 70 young adults (34 males, 36 females) participated in this study. Age, height, and mass (mean ± SD) for males were 18.9 ± 0.8 years (range = 18–21 years), 180.3 ± 6.7 cm, and 80.1 ± 14.4 kg and for females were 19.6 ± 2.4 years (range = 17–26 years), 169.0 ± 6.9 cm, and 64.5 ± 7.1 kg, respectively. All participants were physically active an average of 3.7 times per week for 6.8 hours per week and for at least 2 years. Their levels of engagement with sport were 51% club level, 20% school level, 19% national level, and 10% recreational.

Participants landed at the prescribed 50% of body height in d_50% and closer to the box in the d_ss condition (P < .001, Table 1). The group LESS score (mean ± SD) was 5.58 ± 1.79 errors (range = 1.67–11.00 errors) for the d_50% and 5.57 ± 1.74 errors (range = 1.30–10.7 errors) for the d_ss condition, with 66% and 64% of participants, respectively, categorized as having a high risk of injury based on the threshold of ≥ 5 errors (Table 1). Given the GEE estimates, the group mean LESS scores (P = .969) and odds of being classified as high injury risk at a group level (OR = 0.94 [95% CI = 0.47, 1.88], P = .859) were similar between conditions.

At an individual level, the risk categorization was inconsistent for 33% (n = 23) of participants between conditions (Figure). Twelve participants were categorized as at high risk of injury exclusively for d_50% (Figure). Their mean difference in LESS scores between conditions was 1.85 errors. In contrast, 11 participants were categorized as at high risk of injury exclusively for d_ss (Figure). Their mean difference in LESS scores between conditions was 1.82 errors. The difference in the proportion of participants at high or low injury risk was not significant between conditions (McNemar test P = 1.000), with slightly greater odds of being at high injury risk (OR = 1.09 [95% CI = 0.48, 2.47]) in the d_50% condition (Figure). The proportion of specific LESS errors was different for ankle plantar flexion at initial contact (d_50% > d_ss, P = .004), knee valgus at initial contact (d_50% < d_ss, P < .001), narrow stance width (d_50% < d_ss, P = .002), toe-out foot position (d_50% > d_ss, P = .008), and knee-valgus displacement (d_50% > d_ss, P = .001; Table 2).

We explored the influence of landing at a distance of 50% of body height (d_50%, as prescribed in the original LESS protocol¹ ) versus a self-selected distance (d_ss, as typically prescribed by strength and conditioning coaches, athletic trainers, and physiotherapists) on mean LESS score, group-level risk categorization, individual-level risk categorization, and the occurrence of specific LESS errors. Landing distance did not influence the mean LESS score or the proportions of participants categorized as at high (LESS ≥ 5 errors) or low (LESS < 5 errors) injury risk; however, the occurrence of specific LESS errors and individual-level risk categorization were inconsistent between the landing conditions. Based on these results, researchers can consider using d_ss to facilitate the testing of large cohorts when only the group mean LESS score or the proportions of participants at high or low injury risk in a given population are of interest, with the caveat that the injury-risk threshold of 5 errors for the LESS has not been validated for d_ss.¹¹  However, in clinical and sport environments, the specific movement errors and injury-risk categorizations are of primary interest and using d_ss during the LESS might lead to different LESS errors and a different risk categorization at an individual level than d_50%. Given that the LESS at d_50% is the protocol that has shown some value in predicting the ACL injury risk at an individual level,¹¹  this protocol should be used in clinical and sport settings as a baseline test until the predictive ability of LESS at d_ss is prospectively examined. In any circumstance, explicit documentation of landing distance is encouraged to ensure the reproducibility of protocols and outcomes.

The original LESS requires individuals to perform a jump-landing task from a 30-cm-high box to d_50%.¹  However, scientists have implemented various landing distances for LESS assessment^3,6  other than the d_50% originally described by Padua et al¹  without knowing how protocol variations influence outcomes. Moreover, anecdotal observations and discussions with clinicians and practitioners in health care and sports indicated that the landing distance was often not set when using the LESS. On average, d_50% equated to a landing distance of 86.19 cm in our study. In contrast, when individuals self-selected their landing distance, they landed at a distance equaling 26.11% of their body height (the equivalent of 45.42 cm), close to the 25% used by DiStefano et al.³  For us, this landing distance was approximately 40 cm closer to the box than d_50%.

The mean LESS scores and group-level risk categorizations were similar between the 2 landing distances. Changing the landing distance of jumps has been shown to alter landing biomechanics.^15–17  As the jump distance increased from 20% to 80% of body height¹⁶  (35.2–140.7 cm) during the double-legged stop-jump task and from 30% to 90% of maximal jump distance (42–163 cm) during a travelling jump in dancers, anterior tibial shear force, peak forward acceleration of the tibia, peak posterior ground reaction shear force, and vertical ground reaction force (GRF) have been reported to increase.^15,17  These biomechanical variables have been associated with ACL strain and are considered important risk factors for noncontact ACL injury.^18–20  As such, one would expect increased ACL strain and worse LESS scores under d_50% than d_ss due to the increased landing distance, which we did not observe. However, the number of LESS movement errors changed between conditions. Specifically, from the 210 jump-landing tasks (70 participants × 3 tasks) scored, 12 more errors for ankle plantar flexion at initial contact (item 4), 18 more errors for toe-out foot position (item 10), and 21 more errors for knee-valgus displacement (item 15) were scored under the d_50% than the d_ss condition (Table 2). Yet 29 more errors for knee valgus at initial contact (item 5) and 24 more errors for narrow stance width (item 8) were scored under the d_ss than the d_50% condition. Intrarater agreement for these LESS movement errors was 80% to 100%⁶ ; hence, these differences between landing distances would probably not be due to poor intrarater reliability of individual LESS items. The change in the occurrence of specific LESS errors confirmed that altering the jump distance affected gross movement patterns. Nonetheless, the change in movement errors was distributed quasi-equally between the distances (ie, certain errors increased, and others decreased), and the mean LESS scores and group-level risk categorizations were not affected.

Other than landing distance,^21–23  jump-landing biomechanics and neuromuscular control can be influenced by box height,^21–25  footwear,²²  instructions,^21,23  subsequent movement,²⁶  and history of ACL rupture.²⁷  For instance, the vertical GRF was 1.1 times greater barefoot than shod during netball landings,²²  and increasing the heights of boxes from 32 to 72 to 128 cm^22,24  increased the vertical GRF during landing from 3.9 to 6.3 to 11 times body weight, respectively. When vertical GRF increases, individuals are more prone to land with greater knee-flexion displacement to moderate forces and protect the body against high impact loads.²⁵  Huston et al²⁵  showed that the knee angle at initial contact increased from 7° to 12° and maximum knee-flexion angle increased from 88° to 104° when landing height increased from 20 to 60 cm during drop-jump tasks. The “softness” of landing (item 16, Table 2), knee-flexion angle at initial contact (item 1), and knee-flexion displacement (item 12) during the jump-landing task are scored during the LESS.¹  Changes in box height would likely affect LESS scores to a greater extent than changes in landing distance. Some authors have used a 40-cm rather than a 30-cm box²⁸  and tested participants barefoot²⁹  during the LESS, but the clinical implications of these alterations compared with the original protocol are unknown.¹ 

Whereas the odds of being classified as at high injury risk were similar between the d_50% and d_ss conditions at a group level (P = .859) based on the established cutoff score of 5 errors,¹  only a subset of individuals (n = 34, 49%) were categorized as at high risk under both conditions. The difference in landing biomechanics and related difference in the number of specific LESS errors between conditions is the most probable source of inconsistency in risk categorization. The mean differences in LESS scores between conditions for participants who scored at high risk exclusively for d_50% and d_ss were 1.85 and 1.82 errors, respectively. The psychometric properties of the LESS were as follows: standard error of measurement (SEM) for intrarater reliability = 0.19 to 0.52, interrater reliability = 0.71, and test-retest reliability = 0.81 errors.⁵  These SEM values indicate that the magnitude of the difference in LESS scores between participants who were categorized as at high risk exclusively for a given landing condition is clinically meaningful. Therefore, we caution against using d_ss when individual errors and individual-level risk categorization are of interest.

The main limitation of our study was that group-level and individual-level risk categorization were set at 5 errors based on a prospective study from Padua et al.¹¹  Research on other functional movement screens, ie, the Y-Balance Test and Functional Movement Screen, indicated that injury risk thresholds should take sex, sport, and age into account.³⁰  The threshold of 5 errors derives from a population of young (age = 13.9 ± 1.8 years) elite soccer players¹¹  and might not be appropriate for our population of young physically active adults (age = 19.3 ± 1.8 years). However, to date, no other population-specific cutoff score has been established for the LESS. Furthermore, it is important to note that the predictive ability of the LESS for noncontact ACL injury is uncertain based on current evidence.⁵  We also caution that our sample population of 70 active young individuals may limit the generalizability of our findings to younger and older athletes or less active groups.

Group mean LESS scores and the proportions of participants categorized as at high or low risk of injury based on a threshold of 5 errors were similar when landing at a distance of 50% of an individual's height compared with a self-selected distance. The LESS data are commonly averaged and compared between (eg, males versus females, injured versus uninjured participants) or within (eg, preintervention versus postintervention) groups to draw clinical inferences.^2,10,31  In such cases, using d_ss could facilitate the testing of large cohorts by removing the need to individualize landing distances to 50% of body height. However, the change in the occurrence of specific LESS errors confirms that altering the jump distance affected gross movement patterns. Injury risk thresholds for d_ss have not been validated and might provide inconsistent and inaccurate comparisons at an individual level versus LESS findings based on d_50%. In clinical and sport settings, specific movement errors and injury-risk categorizations are of primary interest. Therefore, using the validated protocol of d_50% is recommended until the psychometric properties of LESS at d_ss have been established, given that our data showed differences in occurrence of specific LESS errors between d_ss and d_50% protocols and inconsistent individual-level risk categories in 33% of participants.