Neuromuscular Intervention Targeted to Mechanisms of ACL Load in Female Athletes

1. Specific Aim 1 Determine the mechanisms by which trunk motion increases knee load in females. Rationale Aim 1 Knee external abduction moment (LOAD) and lateral trunk motion (LTM) are known to be strong predictors of ACL injury risk in female athletes, but it is not known whether these predictors are linked and which athletes are at increased risk. In addition, we do not know the specific neuromuscular mechanisms that predispose these athletes to greater risk of injury. Neuromuscular control of the trunk will be examined relative to GRF position and magnitude, hip torque and knee LOAD. We will determine if increased ΔGRFv and increased relative hip adductor torque (HAdT) underlie increased knee LOAD in females. Central Hypothesis Aim 1 Lateral trunk motion will increase knee LOAD in female athletes by increased ΔGRFv, increased HAdT or by a combination of these two mechanisms during cutting and landing. Hypothesis 1.1. Increased LTM induced by unanticipated lateral cutting will increase knee LOAD by increasing ΔGRFv, adjusting for HAdT, in female athletes. Hypothesis 1.2. Increased LTM induced by single-leg medial drop landing will increase knee LOAD by increasing HAdT, adjusting for ΔGRFv, in female athletes. Hypothesis 1.3. Increased LTM induced by combined drop landing and unanticipated lateral cutting will result in combinatorial increases in knee LOAD by increasing ΔGRFv and HAdT in female athletes. Hypothesis 1.4 Females with low ΔGRFv, low HAdT, and low knee LOAD will form a distinct group or cluster of subjects, those with high ΔGRFv or HAdT and moderate LOAD will form another cluster, and those with high ΔGRFv, high HAdT and high LOAD will form the last distinct group using data collected during DLUC.
2. Specific Aim 2 Determine if NMT that decreases coronal plane trunk motion will decrease knee abduction LOAD in a double-blind cluster (by school and team) randomized controlled trial (RCT).

Rationale Aim 2 This aim will determine how NMT targeted to LTM and its two knee loading mechanisms, ΔGRFv and HAdT, will affect knee LOAD in low, moderate and high LOAD subgroups of female athletes.

Central Hypotheses Aim 2 NMT will increase control of coronal plane trunk motion and decrease knee LOAD by either mechanical (ΔGRFv), neuromuscular (HAdT) or both mechanisms and pre-test low, moderate and high knee LOAD subgroups of female athletes will demonstrate differential effects of NMT.

Hypothesis 2.1 Knee LOAD will be lower in trained than untrained females during landing and cutting.

Hypothesis 2.2 Post-test knee LOAD will not differ in trained high, moderate and low knee LOAD subgroups. Hypothesis 2.3 Post-test ΔGRFv and HAdT will not differ in trained high, moderate and low LOAD subgroups.

Hypothesis 2.3 Post-test knee LOAD, ΔGRFv and HAdT values will not differ from pre-test values in untrained high, moderate and low knee LOAD subgroups.

Program Director/Principal Investigator (Last, First, Middle): Hewett, Timothy, Edwin PHS 398/2590 (Rev. 11/07) Page Continuation Format Page

1. Specific Aims The Specific Aims have not been modified from the original application. Our long-term goals are to determine the mechanisms by which female athletes become more susceptible to ACL injury and to develop interventions that address these mechanisms in order to reduce knee loads and ACL injury risk. The major goal of this proposal is to determine if decreased neuromuscular control of the trunk increases coronal plane knee load in high-risk groups of females. This overall objective of this application is to test the central hypotheses that lateral trunk motion increases knee load and that neuromuscular training that increases control of trunk motion will decrease knee abduction loading in females with moderate and high knee loads. Our rationale for this project is that its successful completion will provide a strong, evidence-based intervention that will effectively decrease ACL injury risk in high-risk female athletes.
2. Studies and Results

We continue to utilize a prospective randomized controlled design for Specific Aims 1 and 2 as outlined in the original funded application. The Boone County Superintendent and the Boone County School Board approved the participation (testing, busing, randomized intervention, etc) of the school systems' girls' athletic programs (Basketball, Soccer and Volleyball) at their monthly school board meeting on Thursday, June 11, 2009 and pre-testing began on September 1, 2009. As outlined, school-sponsored basketball, soccer and volleyball teams from the Boone County, Kentucky school system have been and are almost completely recruited, tested and tracked. Testing will be completed by December 31, 2010. Female subjects from all the county high school and junior high schools are being screened prior to the start of their basketball and soccer and volleyball seasons.

A total of 23 teams yielding 318 basketball players have been pre-screened and trained and post-screened September 1, 2009 through April 30, 2010. A total of 29 teams of approximately 306 soccer and volleyball players (31 of which were not new athletes and were tested in prior basketball season) were pre-screened, trained and post-screened between May 2010 and December 2010. We have faced challenges recruiting soccer players at the junior high school level in the county school system. Some of the junior high schools do not have organized soccer teams. We are addressing this challenge by recruiting volleyball teams within the county in order to fill in all of our randomized blocks. In addition, we will capture those athletes that will go on to play high school soccer, volleyball and basketball within the Fayette county school system (see supplement summary below). Thus far, we have enrolled and pretested (N=31 tested for both seasons) and tested a total of 593 athletes for our randomized controlled trial from 9/1/09 through 12/31/10. The breakdown by sport and school level is summarized in the Enrollment Table. There were 558 athletes made the cuts on their respective teams who were randomized by cluster (team) into the intervention. This number includes all athletes who completed at least 1 intervention training session. 523 athletes had full data sets for analysis.

For the initial biomechanical analyses, we have chosen to focus on Hypothesis 1.4 with the use Latent Profile Analysis (LPA). We will use LPA to examine whether we can cluster subjects with similar characteristics of the GRFv, HAdT, hip moments (minimum and maximum), lateral trunk motion and center of gravity displacement into distinct neuromuscular profiles. Our latent profile analyses will evaluate independent variables across the multiple tasks to form groups of girls with similar neuromuscular patterns for these variables where the heterogeneity of response pattern is minimized within each profile and maximized across profiles. Initially, two profiles are specified for the LPA model, then, in a stepwise fashion, the number of profiles specified is increased by one. At each step, changes in Bayesian information criteria (BIC), adjusted for sample size, is used to assess model fit and identify the number of profiles needed. The point at which the number of profiles is deemed adequate is when no significant drop in BIC is seen when the profile number is increased. In addition, the Lo-Mendell-Rubin adjusted likelihood ratio test is used to determine the optimal number of profiles.(Lo 2001) The analysis was implemented using Mplus 5.(Muthen 2007) Validation of profile group includes examination as an independent predictor of LOAD.

Analysis has begun and profiles have been created for the pre-test movements. Graphs depicting the standardized means along with the 2 generalized profiles are shown in Figure 1. For presentation purposes it is necessary to show the standardized values as the variables are on different scales, with hugely varying mean values. Due to the underlying normality assumption, it was necessary to transform the lateral trunk motion variable to the loge scale. In addition, it is necessary to account for the correlation between some of these variables; in particular maximum hip moment and GRFv and HAdT. We are continuing to examine the potential LPA models at this time and will be checking them against the post intervention data to determine any emergent profiles. Comparison of mean KAM, with and without adjustment for age and pubertal status by profile will also be the focus of future analyses.

Figure 1. Profile analysis of Pre-Test Data Set (N=457) In addition, we have performed this initial assessment of the efficacy of the Randomized controlled intervention and will be presenting these initial results (The effect of Trunk focused Neuromuscular training on Hip Strength) at a national meeting this coming July.