Intrinsic Risk Factors for First-Time Noncontact ACL Injury: A Prospective Study of College and High School Athletes

Abstract

Background:

A comprehensive understanding of the intrinsic risk factors for anterior cruciate ligament (ACL) disruption is important for identifying individuals at increased risk for suffering this trauma and developing interventions to mitigate risk.

Hypothesis:

A variety of risk factors predispose athletes to first-time, noncontact ACL injury and some of these differ between male and female athletes.

Study Design:

Prospective cohort study with nested case-control sampling.

Level of Evidence:

Level 2.

Methods:

Sport teams at 28 high schools and 8 colleges were monitored prospectively over 4 years, and 109 of 130 athletes who suffered their first noncontact ACL injury participated in the study. At the time of injury, matched control subjects were randomly selected from among the case’s teammates and a total of 227 athletes participated. Demographic characteristics, joint laxity, lower extremity alignment, strength, and personality characteristics were measured. The association of each risk factor with injury risk was assessed by conditional logistic regression.

Results:

The risk factors that were associated with ACL injury in both male and female athletes included having a parent with prior ACL injury and increases of the following variables: body weight, anterior displacement of the tibia relative to the femur, genu recurvatum, and generalized joint laxity. Risk factors that are unique to female athletes included increased body mass index, increased trunk flexion strength, and prior non-ACL knee injury. The risk factors specific to male athletes were decreased standing quadriceps angle, decreased hip adduction strength, and chronic disease.

Conclusion:

A diverse set of risk factors predispose both male and female athletes to ACL injury, whereas others appear to be sex-specific.

Clinical Relevance:

Different approaches for assessing risk and preventing ACL injury are needed for male and female athletes. In addition, personalized prevention strategies may be needed to target the specific characteristics that place an individual at increased risk of suffering this trauma.

Young athletes are at substantial risk of suffering severe knee injury such as an anterior cruciate ligament (ACL) disruption, which creates altered biomechanics of the articular surfaces, and is frequently associated with the onset and progression of posttraumatic osteoarthritis about the knee.^{1,2,4,18,23,24} With a current lack of effective therapies to prevent and treat posttraumatic osteoarthritis in young, active individuals, prevention of ACL injury is of paramount importance. ¹⁸ A thorough understanding of the risk factors for ACL injury is essential to assess susceptibility to this trauma and inform the development of injury prevention strategies.

There has been considerable effort focused on identifying both the extrinsic and intrinsic factors that are associated with increased risk of sustaining an ACL injury. Systematic literature reviews of prognostic studies for ACL injury have revealed that extrinsic characteristics such as type of sport, ⁷ level of play, ⁷ playing surface, and weather conditions influence the likelihood of injury.^22,29 Intrinsic risk factors that have been identified include the athlete’s sex,^7,16,25 family history of an ACL injury, ¹⁴ and a number of neuromuscular and anatomic measures.^{22,28,30,31,36} Many prior studies of intrinsic risk factors had retrospective, unmatched case-control designs that may not have adequately controlled for athletic exposure or the potentially confounding effects of extrinsic risk factors such as an athletes’ sex, sport, and level of play. ⁷ A few prospective risk factor studies have been conducted but many had limited statistical power because of the small numbers of injuries that were observed.

Recently, a study was conducted to assess intrinsic risk factors for first-time, noncontact ACL injury by monitoring high school and college sport teams to prospectively identify ACL injuries with potential risk factors measured on each case before surgery using data from the uninjured contralateral leg as a surrogate for pretrauma measurements on the injured leg. ³⁶ Comparable measurements were obtained from randomly selected control subjects on the same team who had not sustained an ACL injury on or before the date of injury for the case. This nested case-control design has the advantage of closely controlling for age and sex, as well as athletic exposure, sport, and other extrinsic variables, while not requiring preseason measurement of risk factors for all athletes being monitored for injury. In addition, under the assumption that the contralateral leg is a valid surrogate for the injured leg before trauma, the results are comparable to those from a fully prospective study. ⁹ The primary aim of the study was to develop multivariate risk models for predicting ACL injury and these results have been published previously.^30,31,34 A secondary aim was to examine the individual associations between ACL injury risk and a wide range of potential risk factors, as they are efficient to use in most clinical settings and cost-efficient to apply in large-scale, field-based screenings, or easily implemented with preseason physical evaluations. The results for measures of knee joint geometry that were found to influence the risk of ACL injury (decreased femoral notch size, decreased ACL size, and increased posterior-inferior directed slope of the lateral tibia plateau articular cartilage) have been reported elsewhere.^5,8,30,31,36 Although important mechanistically, assessment of these risk factors requires 3-dimensional imaging of the knee, that is generally not feasible for screening athletes in school settings.

This paper focuses on demographic characteristics, joint laxity, lower extremity alignment, strength, and personality characteristics that can be measured when screening athletes in school settings. The findings provide useful information for validating the risk factors identified in prior studies with less-rigorous designs and for assessing risk in settings that preclude measurement of all the risk factors in our multivariate models.^30,31,34 They also provide insight into the differences in ACL injury risk between male and female athletes.

Methods

This study is part of a prospective cohort study with nested case-control sampling. ³⁴ With this study design, a cohort of college and high school athletes was followed during each sport season and whenever an athlete experienced an ACL injury, a sample of matched control subjects was selected from among cohort members who had not sustained an ACL injury by that time. The University of Vermont’s Committee on Human Research in the Medical Sciences Review Board approved this investigation. Each participant, and a parent, or legal guardian if the participant was younger than 18 years of age, provided signed informed consent before participation.

Complete details of the prospective cohort study design, participant recruitment, and data acquisition were provided in a previous publication. ³⁴ Briefly, the study included all lacrosse, basketball, soccer, field hockey (female athletes only), football (male athletes only), rugby, and volleyball (female athletes only) athletes at 28 high schools and 8 colleges throughout the state that had a licensed athletic trainer available to identify subjects who suffered a severe knee injury thought to involve the ACL. When an injury occurred, the athletic trainer notified the study coordinator who then contacted the potential participant and invited eligible individuals to participate.

Subjects were eligible for the study if they had not previously suffered an ACL injury to either knee, and their current injury occurred while they were engaged in sport activity at the participating schools and was produced by a noncontact mechanism. At the time of enrollment of an injured subject, up to 4 control subjects were randomly selected from the injured subject’s teammates and invited to take part in the study if they had not previously suffered an ACL injury. This approach ensured that the injured athletes and their matched controls would be of the same sex, similar age, and have comparable exposure to the sport activity associated with the injury (eg, similar training approach, number of practice and game exposure, and similar extrinsic factors such as playing surface). Two of the 109 study participants with ACL injury had 4 control subjects, 37 had 3 control subjects, 38 had 2 control subjects, and 32 had 1 control subject, recruited into the study, for a total of 227 control subjects.

Following injury, but before undergoing ACL reconstruction, the injured subjects and matched control subjects visited our laboratory for measurement of 5 categories of potential risk factors. These included demographic characteristics, joint laxity (knee, ankle, and generalized), lower extremity alignment, strength (trunk, hip, knee, and ankle), and personality characteristics. Potential risk factors that were altered by the index trauma and hence could not be measured on the injured knee (joint laxity, the strength of the knee extensors and flexors, and genu recurvatum) as they were measured in the case’s uninjured contralateral knee as a surrogate for the injured knee before trauma. Measurements were made on the corresponding knee of the case’s matched control subject to account for potential differences between right and left legs. A total of 53 potential risk factors were assessed and are listed in Tables 1 to 5. The methods that were used to obtain these measurements have been described in a previous publication, ³⁴ as well as in the Appendix (available in the online version of this article), and the intratester and intertester reliability associated with acquisition of these measurements have been reported.^27,32 The same investigator interviewed all study participants to obtain demographic data. Similarly, measurements of joint laxity, lower extremity alignment, and strength were made by the same investigator to maximize comparability between cases and control subjects. Personality characteristics were evaluated with the Temperament and Character Inventory (TCI), ¹⁰ which was completed by the study participants and used to classify the participant’s temperament in 4 dimensions (Novelty Seeking, Harm Avoidance, Reward Dependence, and Persistence) and character in 3 dimensions (Self-directedness, Cooperativeness, and Self-Transcendence).

Table 1.

Associations between demographic variables and risk of ACL injury

	Male Athletes (N = 112)			Female Athletes (N = 224)
	Mean (SD)	Odds Ratioa [95% CI]	P Value	Mean (SD)	Odds Ratioa [95% CI]	P Value
Height, cm	179.4 (7.8)	1.03 [0.97-1.10]	0.33	167.7 (7.3)	1.00 [0.59-1.55]	0.85
Weight, kg	78.8 (14.5)	1.04 [1.00-1.09]	0.04	64.1 (9.8)	1.04 [1.01-1.08]	0.03
Body mass index	24.4 (3.8)	1.10 [0.96-1.27]	0.16	22.7 (2.7)	1.20 [1.05-1.36]	<0.01
Sports participation, h/week	10.7 (5.2)	1.08 [0.94-1.23]	0.28	9.7 (5.7)	0.96 [0.86-1.08]	0.51
Years played sport	8.5 (4.2)	0.92 [0.79-1.07]	0.28	7.9 (3.8)	0.98 [0.84-1.13]	0.77
	N (%)			N (%)
Parent had prior ACL injury	17 (15.7)	4.43 [1.26-15.61]	0.03	30 (13.7)	4.55 [1.83-11.17]	<0.01
Nonwhite race	13 (11.6)	2.90 [0.90-9.28]	0.07	16 (7.1)	2.70 [0.90-8.08]	0.08
Chronic disease	15 (13.4)	3.82 [1.13-12.95]	0.03	46 (20.5)	0.49 [0.22-1.10]	0.08
Prior knee injury	22 (19.6)	0.86 [0.32-2.32]	0.77	33 (14.7)	3.08 [1.35-7.04]	<0.01
Prior hip/thigh injury	19 (17.0)	1.16 [0.40-3.38]	0.78	15 (6.7)	1.34 [0.44-4.07]	0.60
Prior lower leg injury	5 (4.5)	1.44 [0.24-8.67]	0.69	17 (7.6)	1.53 [0.52-4.56]	0.44
Prior ankle/foot injury	49 (43.8)	1.18 [0.49-2.81]	0.71	97 (43.3)	0.90 [0.49-1.67]	0.75
Prior lower leg surgery	9 (8.0)	2.86 [0.49-6.60]	0.24	11 (4.9)	1.00 [0.24-4.12]	>0.99
Left leg dominance	15 (13.4)	2.99 [0.68-3.03]	0.15	19 (8.5)	0.42 [0.12-1.53]	0.19
Use braces	30 (26.8)	0.94 [0.36-2.42]	0.91	64 (28.6)	1.37 [0.68-2.79]	0.37
Use medications	14 (12.5)	1.70 [0.51-5.68]	0.39	42 (18.8)	0.67 [0.31-1.42]	0.29

^aFor continuous measures, odds ratios correspond to the association with ACL injury per unit increase in the measurement.

P values shown in boldfaced text were statistically significant. ACL, anterior cruciate ligament.

Table 5.

Temperament and Character Inventory: association with ACL injury risk per 5 unit change in scale

	Male Athletes (N = 110)			Female Athletes (N = 218)
	Mean (SD)	Odds Ratio [95% CI]	P Value	Mean (SD)	Odds Ratio [95% CI]	P Value
Novelty seeking	60.5 (9.9)	0.88 [0.72-1.07]	0.21	59.6 (10.2)	1.12 [0.97-1.29]	0.13
Harm avoidance	48.1 (10.6)	1.22 [0.99-1.51]	0.06	53.5 (11.1)	0.88 [0.77-1.01]	0.07
Reward dependence	66.8 (9.6)	0.84 [0.68-1.04]	0.12	73.8 (10.0)	0.83 [0.71-0.96]	0.01
Persistence	73.2 (10.4)	0.93 [0.77-1.13]	0.48	75.1 (10.7)	1.01 [0.87-1.17]	0.91
Self-directedness	75.1 (10.7)	0.98 [0.79-1.21]	0.84	76.1 (10.4)	0.93 [0.80-1.08]	0.37
Cooperativeness	75.3 (8.9)	0.90 [0.73-1.11]	0.33	80.2 (8.8)	0.87 [0.74-1.03]	0.11
Self-transcendence	41.4 (9.7)	0.89 [0.73-1.09]	0.27	41.7 (9.5)	0.99 [0.84-1.16]	0.90

P values shown in boldfaced text were statistically significant. ACL, anterior cruciate ligament.

Statistical Analysis

Conditional logistic regression, in which cases were compared with their matched control subjects, was used to assess the association between each variable and the risk of ACL injury. For data from a prospective study with case-control sampling, this analysis yields odds ratios (ORs) that are comparable to the hazard ratios from a Cox regression based on the entire cohort, with only a slight loss of statistical power. ⁹ By selecting control subjects from among each case’s teammates, the estimated effects of the intrinsic risk factors are independent of age, sex, sport, and level of play. The matching also controlled for extrinsic factors, such as the playing surface, environmental conditions, and exposure to athletic activity. The strength variables were analyzed as peak torque (Nm) and peak torque normalized to body weight (Nm/kg). Results for all analyses were similar, but it is difficult to assess associations with risk using the normalized variables because of the confounding effect of body weight on risk of ACL injury. The normalized results are therefore not presented. All analyses were performed separately for male and female athletes. P values <0.05 were considered statistically significant.

Results

The characteristics of the study subjects and the conditions under which their injuries occurred have been presented previously. ³⁴ In brief, the injured athletes (n = 109) and matched control subjects (n = 227) ranged between 14 and 23 years of age, and 35.8% were male. Thirty-five percent of the injured subjects competed at the collegiate level at the time of injury and 65% participated at the high school level. The majority (76%) of injuries occurred during games and about half of these (56.9%) happened at home games.

Demographic Characteristics

Athletes with a parent who had an injured ACL were at significantly increased risk for suffering a first-time noncontact ACL injury (Table 1), and the ORs were similar for male athletes [OR = 4.43; 95% CI = 1.26-15.61] and females [OR = 4.55; 95% CI = 1.83-11.17]. Likewise, body weight was related to ACL injury, with risk increasing 4% for each kilogram increase in body weight for both males [OR = 1.04; 95% CI = 1.0-1.09] and females [OR = 1.04; 95% CI = 1.01-1.08]. Height, leg dominance, use of braces or medications, prior leg surgery, and prior injury to the hip, thigh, lower leg, ankle, and foot were not associated with risk of ACL injury in either male or female athletes(Table 1). In addition, neither the amount of time per week nor the number of years spent participating in sport were associated with risk of injury in male and female athletes. A unit increase in body mass index (BMI) was associated with a 20% increased risk of injury in females [OR = 1.20; 95% CI = 1.05-1.36], but the increase was smaller for males [OR = 1.10; 95% CI = 0.96-1.27] and was not statistically significant (Table 1). There were more notable differences between males and females in the ORs for 2 other demographic variables. Male athletes who reported having a chronic disease (73% of the injured male athletes listed asthma as their chronic disease) were at increased risk of suffering ACL injury [OR = 3.82; 95% CI = 1.13-12.95]. In contrast, the result for the females indicates that those with chronic disease may be at decreased risk of injury [OR = 0.49; 95% CI = 0.22-1.10; P = 0.08], even though the proportion of injured female subjects with a chronic condition who listed asthma as their chronic disease (76%) was similar to males. Having a previous knee injury was highly associated with increased risk of ACL injury in female athletes [OR = 3.08; 95% CI = 1.35-7.04], but not in males [OR = 0.86; 95% CI = 0.32-2.32].

Knee, Ankle, and Generalized Laxity

An increase in generalized joint laxity, as measured by the Beighton score ³ which quantifies the extent of hypermobility of a subjects’ joints, was significantly associated with increased risk of ACL injury in both males and females (Table 2); however, the increase in risk per unit increase in the Beighton score was more pronounced for male athletes [OR = 1.43; 95% CI = 1.14-1.78] than for females [OR = 1.15; 95% CI = 1.03-1.29]. The KT-2000 was used to measure the entire anterior-posterior load-displacement relationship of the tibiofemoral joint. These data were postprocessed to quantify joint laxity [anterior-posterior displacement of the tibia relative to the femur between anterior (130 N) and posterior (90 N) directed loads and the stiffness of the tibiofemoral joint, ie, the slope of the load-displacement relationship at the linear anterior and posterior regions] as described previously. ¹¹ There were significant increases in the risk of ACL injury for each 1.0 mm increase in anterior-posterior knee laxity, as measured by the KT-2000 arthrometer, ¹¹ and this relationship was similar for male athletes [OR = 1.20; 95% CI = 1.03-1.41] and females [OR = 1.27; 95% CI = 1.12-1.45] (Table 2). Among female athletes, an increase in posterior knee stiffness was significantly related to a decrease in risk of ACL injury [OR = 0.93; 95% CI = 0.86-1.0]. Thus, each unit decrease in posterior stiffness of the knee (also considered as a more compliant knee) increased risk of ACL injury by 7.5%. Posterior knee stiffness was not significantly associated with risk of ACL injury in male athletes, and anterior knee stiffness was not significantly related to risk in either male or female athletes (Table 2). A positive talar tilt test of the ankle was not significantly associated with an increased risk of ACL injury for either sex.

Table 2.

Knee (KT-1000), ankle (clinical exam), and generalized joint laxity (Beighton Score): associations with ACL injury per unit increase in measurement

	Male Athletes			Female Athletes
	Mean (SD)	Odds Ratioa [95% CI]	P Value	Mean (SD)	Odds Ratioa [95% CI]	P Value
Generalized joint laxity (Beighton Score)	1.7 (2.2)	1.43 [1.14-1.78]	<0.01	2.6 (2.7)	1.15 [1.03-1.29]	0.01
KT-1000 anterior-posterior displacement, mm	13.4 (3.0)	1.20 [1.03-1.41]	0.02	13.9 (2.7)	1.27 [1.12-1.45]	0.00
KT-1000 posterior knee stiffness, N/mm	24.0 (4.6)	1.02 [0.93-1.13]	0.64	23.5 (4.6)	0.93 [0.86-1.00]	0.05
KT-1000 anterior knee stiffness, N/mm	21.6 (5.6)	0.94 [0.86-1.04]	0.24	19.8 (4.6)	0.94 [0.88-1.02]	0.13
	N (%)			N (%)
Positive talar tilt test	8 (7.3)	4.68 [0.87-25.1 3]	0.07	40 (17.9)	1.79 [0.78-4.14]	0.68

^aFor continuous measures, odds ratios correspond to the association with ACL injury per unit increase in the measurement.

P values shown in boldfaced text were statistically significant. ACL, anterior cruciate ligament.

Lower Extremity Alignment

The ORs corresponding to a unit change of the lower extremity alignment measures are presented in Table 3. A decrease in passive genu recurvatum (indicating a less-negative angle or decreased hyperextension) was associated with a decrease in risk of noncontact ACL injury in both male athletes [OR = 0.89 per degree; 95% CI = 0.81-0.98] and female athletes [OR = 0.91 per degree; 95% CI = 0.85-0.98]. Hence, each 1 degree increase in passive genu recurvatum (increased hyperextension) increases risk by 12% in male athletes and 10% in female athletes. The corresponding increase in risk for a 1 degree increase (again, corresponding to a more-negative angle) in active genu recurvatum was 10% in males and 13% in female athletes, but the relationship was statistically significant only for female athletes. Similarly, a 1 degree increase in hamstring extensibility (measured as the knee flexion angle with the subject positioned supine and the thigh oriented vertical) increased the risk of ACL injury by 3% in both male and female athletes, but was statistically significant only for female athletes. For the male subjects, a 1 degree decrease in the standing quadriceps angle (Q angle) increased the risk of ACL injury by 14.9% [OR = 0.87; 95% CI = 0.77-0.98], whereas each 1 mm increase in navicular drop produced a 12% [OR = 1.12; 95% CI = 1.02-1.24] increase in risk. For females, the corresponding associations with risk were weaker and not statistically significant. Hip anteversion, pelvic angle, tibiofemoral angle, tibial torsion, and tibial and femoral lengths were not associated with risk of ACL injury for either sex (Table 3).

Table 3.

Lower extremity alignment: associations with ACL injury risk per unit change in the measurement

	Male Athletes			Female Athletes
	Mean (SD)	Odds Ratioa [95% CI]	P Value	Mean (SD)	Odds Ratioa [95% CI]	P Value
Passive genu recurvatum, deg	-8.1 (5.0)	0.89 [0.81-0.98]	0.02	-8.2 (4.5)	0.91 [0.85-0.98]	<0.01
Active genu recurvatum, deg	-7.0 (4.6)	0.91 [0.83-1.01]	0.08	-6.9 (4.1)	0.88 [0.82-0.96]	<0.01
Hamstring extensibility, deg	22.4 (10.3)	0.97 [0.93-1.02]	0.23	17.5 (11.7)	0.97 [0.95-1.00]	0.04
Standing quad angle, deg	8.8 (4.3)	0.87 [0.77-0.98]	0.02	12.4 (4.7)	0.94 [0.88-1.01]	0.08
Navicular drop, mm	10.3 (4.7)	1.12 [1.02-1.24]	0.02	8.5 (4.2)	1.07 [0.99-1.15]	0.09
Tibiofemoral angle, deg	8.2 (3.2)	1.06 [0.91-1.23]	0.47	10.5 (3.2)	0.94 [0.86-1.04]	0.23
Pelvic angle, deg	6.4 (4.0)	1.03 [0.92-1.15]	0.63	6.9 (4.1)	0.95 [0.88-1.03]	0.23
Tibial torsion, deg	12.0 (5.6)	0.98 [0.90-1.07]	0.60	12.9 (5.1)	0.95 [0.89-1.02]	0.17
Hip anteversion, deg	-6.8 (5.6)	0.93 [0.85-1.02]	0.13	-7.5 (5.2)	1.00 [0.94-1.06]	0.99
Tibial length, cm	40.4 (2.6)	1.01 [0.85-1.20]	0.90	37.4 (2.2)	0.96 [0.83-1.13]	0.65
Femur length, cm	45.1 (2.8	1.02 [0.88-1.19]	0.79	43.6 (2.7)	0.97 [0.85-1.11]	0.66

^aFor continuous measures, odds ratios correspond to the association with ACL injury per unit increase in the measurement.

P values shown in boldfaced text were statistically significant. ACL, anterior cruciate ligament.

Strength of the Muscles that Span the Hip, Trunk, Knee, and Ankle Joints

For the strength measures, the ORs corresponding to 10 Nm increases in peak torque are presented in Table 4. A decrease in hip adduction strength was associated with an increase in risk of ACL injury in male athletes [OR = 0.79; 95% CI = 0.67-0.93], but not in female athletes [OR = 0.98; 95% CI = 0.85-1.12]. Conversely, a 10 Nm increase in trunk flexion torque was significantly associated with increased risk of injury in females [OR = 1.17; 95% CI = 1.03-1.33], but not males [OR = 0.96; 95% CI = 0.86-1.07]. Adjustment for body weight strengthened the association between hip adduction and ACL injury risk in males, but the relationship between trunk flexion and risk of injury in females was not statistically significant after adjustment for body weight (results not shown). Strength measurements of the hip abductor, flexor, extensor, internal and external rotator muscles, thigh muscles (flexors and extensors), and ankle muscles (dorsi and plantar flexors) were not related to the risk of suffering an ACL injury in either sex.

Table 4.

Hip, knee, trunk, and ankle strength generated as peak torque: association with ACL injury risk per 10 Nm change in peak torque

	Male Athletes			Female Athletes
	Mean (SD)	Odds Ratioa [95% CI]	P Value	Mean (SD)	Odds Ratioa [95% CI]	P Value
Hip adduction, Nm	17.5 (4.1)	0.79 [0.67-0.93]	<0.01	12.1 (2.8)	0.98 [0.85-1.12]	0.72
Hip abduction, Nm	14.9 (3.5)	0.97 [0.85-1.11]	0.67	11.5 (2.6)	1.06 [0.92-1.22]	0.40
Hip flexion, Nm	18.6 (4.8)	0.93 [0.84-1.03]	0.16	12.3 (3.1)	1.09 [0.98-1.21]	0.12
Hip extension, Nm	18.0 (5.7)	0.97 [0.90-1.06]	0.53	11.5 (3.4)	1.05 0.96-1.16]	0.26
Hip internal rotation, Nm	2.7 (1.0)	0.97 [0.64-1.46]	0.87	1.9 (0.5)	1.53 [0.87-2.71]	0.14
Hip external rotation, Nm	3.1 (1.0)	0.83 [0.53-1.28]	0.40	2.1 (0.6)	1.20 [0.69-2.10]	0.51
Knee flexion 15°, Nm	11.6 (3.2)	0.92 [0.77-1.09]	0.33	7.5 (2.0)	1.10 [0.94-1.29]	0.23
Knee flexion 30°, Nm	10.4 (2.9)	1.00 [0.82-1.20]	0.96	6.7 (1.9)	1.10 [0.93-1.30]	0.27
Knee extension 15°, Nm	11.1 (4.1)	1.04 [0.91-1.19]	0.59	8.1 (3.2)	1.09 [0.97-1.24]	0.15
Knee extension 30°, Nm	16.9 (5.3)	0.97 [0.88-1.08]	0.63	12.2 (3.6)	1.06 [0.96-1.16]	0.24
Trunk flexion, Nm	16.9 (4.8)	0.96 0.86-1.07]	0.47	10.1 (3.4)	1.17 [1.03-1.33]	0.02
Trunk extension, Nm	17.9 (5.6)	0.97 0.89-1.06]	0.55	12.7 (3.6)	1.01 [0.93-1.11]	0.80
Ankle dorsiflexion, Nm	4.0 (1.1)	1.03 [0.71-1.49]	0.89	2.6 (1.1)	1.10 [0.84-1.44]	0.48
Ankle plantarflexion, Nm	11.2 (3.3)	0.98 [0.85-1.13]	0.79	8.2 (2.7)	1.11 [0.97-1.25]	0.12

^aFor continuous measures odds ratios correspond to the association with ACL injury per unit increase in the measurement.

P values shown in boldfaced text were statistically significant. ACL, anterior cruciate ligament.

Temperament and Character Inventory

Females who scored higher on the Reward Dependence scale of the TCI had lower risk of ACL injury, with a 5 unit increase in the score corresponding to a 17% decrease in risk [OR = 0.83; 95% CI = 0.71-0.96] (Table 5). For males, there was a similar inverse relationship [OR = 0.84; 95% CI = 0.68-1.04] that was not statistically significant (P = 0.12). None of the other temperament or character scales were significantly associated with ACL injury risk in either male or female athletes.

Discussion

This study supports the hypothesis that many different risk factors predispose athletes to ACL injury and that while some of these are similar for both sexes, others differ. The factors that were significantly associated with increased ACL injury risk in both males and females included having a parent who suffered an ACL injury and increases in the following variables: body weight, anterior-posterior knee laxity (anterior-posterior displacement of the tibia relative to the femur), generalized joint laxity, and genu recurvatum of the knee. Female and male athletes had very similar ORs for hamstring extensibility and reward dependence (the temperament portion of the TCI that characterizes an individual’s affinity to respond to signs and signals of reward; specifically, to verbal signals of social support and approval), but the association with injury was not significant for males, likely due to the smaller sample size.

In addition to the shared risk factors, there were a number of sex-specific risk factors, suggesting that different approaches may be needed to evaluate and mitigate ACL injury risk in male and female athletes. Although the effect of increasing body weight on ACL injury risk was the same for male and female athletes, the increased risk associated with BMI in female athletes (20% per unit increase in BMI) was twice as high as the 10% increase in males, which was not statistically significant. BMI can be an inaccurate measure of body fat for athletes because it does not distinguish between fat and muscle mass. If increased body fat is a risk factor for ACL injury, the different results for BMI in males and females may reflect poorer accuracy as a measure of body fat in males due to their greater muscle mass. A more direct measure of body fat may be required to assess body fat in athletes and to understand its relationship to ACL injury. Further, increased navicular drop coupled with decreased standing Q angle, which have important interactive effects on an individuals’ neuromuscular response to a weight-bearing rotational perturbation, were significantly associated with increased risk of ACL injury only in male athletes (see Appendix, available online). Likewise, decreased hip adduction strength, a modifiable risk factor measured as the peak torque, was associated with increased risk of ACL injury only in male athletes.

Fewer male than female athletes reported having a chronic disease (13.4 % vs 20.5%), and for both sexes, asthma was the most prevalent chronic condition (73.4% of males and 76.1% of females reporting a chronic disease). Among those with asthma, the proportions who reported taking medications to treat their condition were also similar in males (54.5%) and females (61.1%). However, chronic disease increased the risk of ACL injury only in males (OR 3.82; P = 0.03) while potentially decreasing the risk in females (OR 0.49; P = 0.08). This may reflect differences in disease severity but may also indicate differences in how each sex perceives and reacts to chronic disease or the medications used to treat it.

Although at higher risk for knee injury, fewer female than male athletes reported suffering a prior non-ACL knee injury (14.7% versus 19.6%) which suggests that female athletes may be less likely to continue participating in sport after injury. It is unclear why prior non-ACL knee injury is only associated with ACL risk in female athletes, but it is possible that among those athletes who continued to participate in sport, the female athletes had more severe knee injuries than the male athletes.

For risk factors that were common to both sexes, comparison of ORs for male and female athletes provide further insight into their differing risk of suffering an ACL injury between the sexes. The ORs estimate the risk of injury associated with a particular risk factor relative to teammates who do not have the risk factor, not relative to all athletes, and therefore depend on the distribution of the risk factor among the matched control subjects. Generalized joint laxity was more strongly associated with risk in men (OR = 1.43) than women (OR = 1.15), but the average Beighton score for women were considerably higher than for men. This suggests that the female control subjects are also at increased risk for ACL injury because of their joint laxity and further increases of the Beighton score do not have as large an effect on risk as in males, for whom increased scores reflect joint laxity more comparable to females. Similarly, decreased hip adduction strength was only associated with increased risk of ACL injury in males, suggesting that the lower average hip adduction strength among females may contribute to their overall higher risk of ACL injury.

The results of this investigation confirm a number of risk factors identified in prior studies. The measures of passive restraint of the tibiofemoral joint that were associated with ACL injury risk for both male and female athletes in our study confirm prior studies that found increases in anterior-posterior knee laxity,^20,26,33,37 generalized joint laxity,^20,26,33 and knee genu recurvatum.^19,20,26 were associated with increased risk of ACL injury. Our finding that increased BMI was significantly associated with increased risk in women supports the earlier results from a study of female cadets at the United States Military Academy reported by Uhorchak et al. ³³ Similarly, our finding of a fourfold increase in ACL injury risk for both males and females whose parent had suffered an ACL injury is consistent with previous studies.^25,14,17 This study also revealed that females with increased hamstrings extensibility (ie, increased flexibility) were at increased risk of suffering an ACL injury, consistent with earlier work by Ford who showed individuals with increased hamstrings extensibility may have delayed onset of hamstrings muscle contraction and use their knee in a more extended position during cutting maneuvers. ¹⁵

Our study identified several potential risk factors for ACL injury that had not been previously identified. Although a number of studies have established that prior ACL injury is a major risk factor for subsequent ACL injury to both the ipsilateral and contralateral knee,^11-13,21,35 we believe this is the first study to find that a prior knee injury not involving the ACL is a strong risk factor for a subsequent first-time noncontact ACL injury in female athletes.

To the best of the authors’ knowledge, this is the first prognostic study to investigate the association between an athlete’s temperament and character, and the risk of ACL injury. We found that a low Reward Dependence score was associated with increased risk of injury in female athletes and is a likely risk factor for male athletes. The behaviors associated with low Reward Dependence that might increase the risk of injury are unclear, but, in general, an individual with lower Reward Dependence is less dependent on the responses of others, not as sympathetic in their feelings, and less sensitive to social cues. Future work to understand how temperament and character influence an athlete’s reaction to competitive situations, and how this in turn impacts injury risk, may provide important insights for injury prevention and rehabilitation.

An important feature of this investigation was that its study design (case-control sampling within a prospective cohort) did not require risk factor measurements on all members of all sport teams participating in the study, while producing ORs comparable to the relative risk estimate that would have been obtained from Cox regression of data from all athletes in the cohort. ⁹ However, with this approach, we could not measure some of the potential risk factors on the injured knee (eg, measurements of joint laxity, the strength of the extensors and flexors of the knee, and genu recurvatum) as they were altered by the index trauma. Instead, we measured the uninjured, normal contralateral knee of subjects who suffered an ACL injury and considered it representative of the injured limb before injury, because in prior research, we measured both knees of uninjured subjects with normal knees and found a high correlation between the left and right knees (see Appendix, available online). ⁶ To control for any potential right-left differences, measurements for cases and their matched control subjects were always made on the same leg. Use of the uninjured knee as a surrogate may better reflect the status of the contralateral knee just before injury than a prospective approach that assumes no changes occur in the potential risk factors between the time they are measured (typically in the preseason before exposure to organized sport) and the time of injury. This may not be a valid assumption for potential risk factors such as strength, which are known to change over the course of a sport season.

We made every effort to recruit all athletes suffering an ACL injury and enrolled almost twice as many injured female (70) as male athletes (39) due to their higher incidence of ACL injury ⁷ and their greater willingness to participate. At a significance level of 0.05, the sample sizes in the study provided 80% power to detect ORs of 3.0 and 2.2 in male and female athletes, respectively, for dichotomous variables with moderate (30%) prevalence. The study also had 80% power to detect ORs of 1.6 and 1.4 per each SD of increase in continuous variables for males and females, respectively.

We focused our study on first-time ACL injuries because prior ACL injury is a strong predictor of repeat injury, and the extent of the prior injury, the recovery achieved through surgery and rehabilitation, and the length of time before returning to preinjury activity level are all likely to influence the risk of subsequent ACL injury, making the risk factors very different from those for first-time injury. Similarly, we excluded subjects who were injured by contact with another player or object because the risk factors for contact injuries are likely to differ from those for noncontact ACL injuries.

A major advantage of our study design is that the matching of cases to control subjects from the same sport team enabled more accurate assessment of intrinsic risk factors by closely controlling for exposure to athletic activity, and the potentially confounding effects of sport, playing conditions, and level of play on ACL injury risk. Although extrinsic risk factors are important for determining which teams and conditions pose the greatest risk of ACL injury for their participants, the intrinsic risk factors evaluated in this study can potentially identify which athletes on a team are most likely to sustain an ACL injury, providing an opportunity for personalized counseling or intervention to prevent these debilitating injuries. To accomplish this, the next step should be to conduct a prospective study that evaluates the risk factors we have identified in the current study on a similar cohort of high school and college athletes and validates how well they identify who goes on to suffer ACL trauma and who does not suffer this trauma.