Abstract
Background:
Lateral ankle sprain (LAS) is the most common injury in basketball, and identifying at-risk players is of high importance.
Purpose:
To evaluate the ability of the Ankle-GO score to identify elite basketball players who will suffer LAS during a competitive season as well as other potential predictive factors.
Study Design:
Cross-sectional study; Level of evidence, 2.
Methods:
A total of 48 elite basketball players (22 male, 26 female; mean ± SD age, 17.9 ± 3.7 years) performed the Ankle-GO score on both limbs during a preseason session. This score is a cluster of functional tests and self-reported questionnaires evaluating ankle function. LAS incidence was recorded throughout the competitive season for each ankle of the players. Potential predictive variables (Ankle-GO, body mass index, history of previous LAS, and sex) associated with LAS were compared between injured and uninjured athletes. The areas under the receiver operating characteristic curve (AUCs) and multivariable logistic regression models with odds ratio (OR) and 95% CI were used to identify potential factors of LAS.
Results:
A total of 17 occurrences of LAS occurred during the season. Ankle-GO score was lower among injured ankles (16.9 ± 3.6 points vs 19.6 ± 3.2; P = .004; d = 0.8) and was associated with probability to sustain an LAS (AUC, 0.71; 95% CI, 0.57-0.84; P = .008). Athletes with an Ankle-GO score <19 points and those who suffered an LAS during the previous years were more likely to experience a new LAS on the same ankle (OR, 5.5; 95% CI, 1.6-19.2; P = .007, and OR, 3.5; 95% CI, 1.1-10.9; P = .03, respectively).
Conclusion:
The Ankle-GO score can help to identify elite basketball players who will suffer LAS during a competitive season.
Lateral ankle sprain (LAS) accounts for the majority of ankle injuries in basketball, comprising 80.2% of all ankle sprains.13,21,31 These injuries frequently occur during high-risk activities such as rebounding, defending, and shooting, and general player contact, often resulting from stepping on another player’s foot.9,21,31,32 Younger athletes, particularly those <26 years old, experience a higher incidence of LAS. 13 Beyond its high occurrence, LAS poses considerable challenges in recovery, often leading to recurrent sprains and chronic ankle instability, severely affecting athletic performance and career longevity.2,21
The prevention of LAS remains a critical focus in sports medicine, as its recurrent nature and associated long-term complications demand effective risk identification and management strategies.7,21,35 A recent meta-analysis identified risk factors for LAS in basketball players, including higher incidence rates in male compared with female athletes (0.90 vs 0.82 per 1000 athlete exposures), increased risk at the elite level (1.87 per 1,000 athlete-exposures) compared with intermediate (1.12) and amateur (0.54) levels, and greater likelihood of injury during matches (2.51 in competition vs 0.80 in training). 33 Other factors include higher body mass index (BMI); neuromuscular stability deficit; and a history of lower extremity injuries, particularly to the hip, hamstring, quadriceps, foot, ankle, or knee.1,6,29,35,38
Predicting athletes who are at risk for sprains has been the subject of increasing interest, especially in professional sports, where preventive measures such as strapping or focused physical rehabilitation can target areas of weakness. In the early 2000s, the single-leg stance (SLS) test was found to predict the likelihood of ankle sprains in high school and intercollegiate athletes, with athletes exhibiting poorer performance showing significantly higher risk.12,34 More recently, the Star Excursion Balance Test (SEBT) and the single-leg hop test have been identified as valuable preseason screening tools in soccer, football, and basketball players.11,17 The multitude of functional performance tests associated with injury risk highlights the necessity of a composite score to standardize assessments and guide preventive strategies. The Ankle-GO score addresses this need by providing a cluster of performance tests and self-reported questionnaires to evaluate ankle function comprehensively.24,25,28 More specifically, it has been shown to be predictive of reinjury ≤2 years after an LAS 24 and could help to identify patients who will fully recover 1 year after injury. 25 The Ankle-GO score was specifically developed to identify deficits and guide return to sport (RTS) after LAS into a composite score by assessing static and balance, hopping and plyometric performance, and self-reported function.27,28 However, its ability to predict LAS risk specifically in basketball remains unexplored.
The primary aim of this study was to evaluate the ability of the Ankle-GO score to predict the risk of LAS injury among elite basketball players during a competitive season. We hypothesized that players exhibiting low Ankle-GO scores would be more likely to suffer an LAS.
In addition, because injury incidence is complex and multifactorial,4,7 the secondary aim was to identify other contextual predictors of LAS injury. Based on existing evidence, it was hypothesized that injury history, sex, and BMI would be significant contributors of LAS injury.
Methods
Study Design
A 1-year prospective cohort study was conducted from August 2023 to June 2024 among elite (international-level) basketball players. The study was performed in accordance with the Declaration of Helsinki. All athletes provided informed consent, and the study received institutional ethics approval.
Inclusion and Exclusion Criteria
Male and female players from the national training center (INSEP, France) as well as national teams were recruited. Exclusion criteria were (1) history of previous surgery to the musculoskeletal structures (ie, bones, joint structures, and nerves) in either lower extremity, (2) history of fracture in either lower extremity requiring realignment, and (3) acute injury to the musculoskeletal structures of other joints of the lower extremity in the previous 3 months.
Patient and Public Involvement
Patients were not engaged in the development, conduct, or oversight of the study.
Ankle-GO Score
The primary independent variable of the study was the preseason Ankle-GO score. 28 This is a validated 0-to-25 scale (higher = better function) used to assess functional capability in sporting patients with LAS during the RTS continuum. It clusters 6 components targeting LAS-related deficits: 4 functional tests (SLS, 30 the modified SEBT [mSEBT], 10 the side hop test [SHT], 8 and the Figure-of-8 Test [F8T] 5 ), plus 2 patient-reported outcome questionnaires—the Foot and Ankle Ability Measure (FAAM), which includes Activities Daily Living (ADL) and Sports subscales, 19 and the Ankle Ligament Reconstruction–Return to Sport after Injury (ALR-RSI) scale assessing psychological readiness 26 (Table 1, Figure 1). All athletes completed the Ankle-GO once in the preseason (August 2023) under a single experienced evaluator (B.P.).
List of Tests and Questionnaires Used for the Construction of the Ankle-GO Score and System to Determine the Points for Each Component a
ADL, Activities of Daily Living; ALR-RSI, Ankle Ligament Reconstruction–Return to Sport after Injury; F8T, Figure-of-8 Test; FAAM, Foot and Ankle Ability Measure; mSEBT, modified Star Excursion Balance Test; SHT, side hop test; SLS, single-leg stance test.
Tests included in the Ankle-GO score.
The SLS test involved standing on 1 leg (knee ~10° of flexion, eyes closed, hands on hips) for 20 seconds. Each time an error occurred—such as removing hands from hips or stepping—the participant was penalized, and fewer errors indicated better static balance. The mSEBT required the participant to stand on 1 foot at the center of a “Y” and reach the opposite foot maximally in 3 directions (anterior, posteromedial, posterolateral), with scores reflecting total reach distance normalized to leg length. The SHT consisted of hopping side to side between 2 lines (30 cm apart) for 10 total hops as quickly as possible, and shorter times signified better performance. In the F8T, the participant completed 2 figure-of-8 loops (5 m apart) at maximal speed, with lower times denoting superior agility. The FAAM rated ADL and Sports tasks on a 5-point scale, then results were summed and converted to percentages, whereas the ALR-RSI scale used 12 items scored 0 to 10 to evaluate psychological readiness, then results were summed and converted to a percentage.
Outcome Measures
The primary outcome measure of the study was the rate of LAS during 1 season. Ankles of the athletes were dichotomized according to their injury status (injured vs uninjured). LAS was defined according to the recommendations of the International Ankle Consortium as “an acute traumatic injury to the lateral ligament complex of the ankle joint resulting from an excessive and sudden inversion mechanism of the rear foot, possibly combined with an adduction or plantar flexion of the foot, precluding participation in sports.”7,11 The diagnosis was made clinically by the team physician, who assessed pain, swelling, and ligamentous laxity via the talar tilt and anterior drawer tests, 22 and it was confirmed by imaging when necessary. The study primarily examined how the preseason Ankle-GO score predicted LAS risk during the season.
Secondary outcome measures included evaluating the influence of injury history, sex, and BMI on LAS risk.
Data Collection and Follow-up
LAS injury incidence was prospectively collected by the medical staff of the teams during the competitive season (from September 2023 to June 2024). Other data related to demographics such as age, sex, BMI, and injury history were collected from medical files by a blinded researcher (M.K.M.) upon enrollment in the study.
Statistical Analysis
A priori power analysis for the primary aim determined that a minimum sample size of 46 players was necessary to achieve a statistical power of 0.80 and a "2-sided" alpha level of 5% based on an expected area under the curve (AUC) of the receiver operating characteristic (ROC) of 0.75.23,24 The proportion of injured players during the season was expected to be 30%.13,35
The analysis and presentation of data were consistent with the Checklist for Statistical Assessment of Medical Papers. 18
For the primary aim of this study, the predictive validity of the Ankle-GO score to identify patients whose ankles would suffer an LAS was evaluated using an ROC curve. The AUC was determined with a precision score considered to be null (AUC, 0.5), low (0.5 < AUC < 0.7), fair to good (0.7 ≤ AUC < 0.9), high (0.9 ≤ AUC < 1), or perfect (AUC, 1). 14 The optimal cutoff score was calculated using the Youden index (J = sensitivity + specificity − 1). Ankle-GO scores were then dichotomized as either above (positive) or below (negative) the cutoff point to simplify the interpretation of risk and the related odds ratios (ORs). In addition, discriminant validity of the Ankle-GO score was assessed using independent t tests between patients who suffered an LAS and those who did not.
For the secondary aim of this study, 4 potential predictors of LAS injury were studied (Ankle-GO score, sex, LAS history on the same ankle, BMI). 7 Data were checked for normality and homogeneity of variance using Shapiro-Wilk and Levene tests. Relationships between potential predictors variables and injury status were assessed using different statistical tests based on the nature of the data: χ2 or Fisher exact tests for categorical variables, independent t tests for normally distributed data, and Mann-Whitney U tests for skewed measures. Only variables with a significance level of P < .20 between the 2 groups (injured and uninjured) were entered into a logistic regression model. In addition, all potential predictive variables were tested using bivariate Pearson r correlations. If any combinations with P < .20 returned a correlation r > 0.8, only 1 of the variables was included in further analyses. Variance inflation factors (VIFs) were used to assess multicollinearity and outliers were searched using standard residual values and Cook distance. Linearity for quantitative predictors was assessed using the Box-Tidwell procedure. After verifying that these statistical assumptions of regression were met, a multivariable logistic regression (stepwise) was conducted to determine whether the remaining potential predictive variables were associated with injury status (dependent variable) during the season. ORs and 95% CIs were reported for the variables associated with the probability of suffering an LAS. There were no missing data at the end of the follow-up period (N = 48 athletes; n = 96 ankles).
To investigate whether a history of previous LAS influenced the preseason Ankle-GO score, an independent-samples t test was conducted. Additionally, the independence between the variable “Ankle-GO < 19” and “previous LAS” was evaluated using multicollinearity diagnostics (tolerance and VIF).
The statistical analysis was performed by a blinded assessor using JASP (Version 0.19.1.0; University of Amsterdam), and SPSS Version 12.0 (IBM Corp). The alpha level adopted for the significance of the regression models was .05.
Results
Among the 48 elite basketball players included in this study (22 men and 26 women; mean ± SD age, 17.9 ± 3.7 years), 17 occurrences of LAS were recorded during the competitive season. More specifically, 16 players (33%) suffered an LAS, including 1 female player who suffered an LAS on each ankle (Table 2).
Baseline Characteristics and Ankle-GO Scores for Injured Versus Uninjured Players a
Data are presented as mean ± SD unless otherwise indicated. Bold P values indicate statistical significance. LAS, lateral ankle sprain.
Ankle-GO scores did not significantly differ between ankles with or without a history of LAS (18.86 ± 3.75 vs 19.25 ± 3.29; P = .60). Moreover, multicollinearity diagnostics showed high tolerance (0.993) and low VIF values (1.007) for both variables, indicating no collinearity between prior LAS and the dichotomized Ankle-GO score (<19) in the regression model.
The mean time between the measurement session and the occurrence of an LAS was 4.2 months. In terms of injury history, 28 players suffered an LAS during the previous competitive season, 7 of whom suffered an LAS on both ankles. Table 3 shows the scores obtained in the several components of the Ankle-GO score among injured and uninjured players during the preseason testing session.
Scores Obtained From the Different Components of the Ankle-GO Score in the Injured and Uninjured Players a
Data are presented as mean ± SD. ADL, Activities of Daily Living; ALR-RSI, Ankle Ligament Reconstruction–Return to Sport after Injury; F8T, Figure-of-8 Test; FAAM, Foot and Ankle Ability Measure; mSEBT, modified Star Excursion Balance Test; SHT, side hop test; SLS, single-leg stance test.
Primary Aim
The mean Ankle-GO score was significantly lower among injured ankles (16.9 ± 3.6 points vs 19.6 ± 3.2; P= .004; d = 0.8) (Table 2, Figure 2A). The estimated probability of sustaining an LAS decreased as Ankle-GO scores increased (Figure 2B).
(A) Preseason Ankle-GO scores among injured vs uninjured ankles. (B) Estimated probability of sustaining a lateral ankle sprain (LAS) based on preseason Ankle-GO scores, with shaded 95% CI.
The predictive ability of the Ankle-GO score was acceptable, with an AUC of 0.71 (95% CI, 0.57-0.84; P = .008) (Figure 3). A cutoff of 19 points yielded a sensitivity of 62% and specificity of 77% in identifying players who would suffer an injury during the season (Youden index = 0.39). The Ankle-GO scores were recoded as being either above or below this cutoff point and entered in the regression model.
Receiver operating characteristic (ROC) curve showing the predictive ability of the Ankle-GO score for a lateral ankle sprain.
Secondary Aim
From the 4 potential predictive factors, only 3 met the initial screening criterion of P < .20. The full model containing Ankle-GO score, injury history, and sex score was statistically significant (Δχ2 [93] = 4.75; P = .03; Nagelkerke R2 = 0.21; Cox and Snell R2 = 0.13), indicating that the model was able to identify athletes who would suffer an LAS during the competitive season. The assumption of multicollinearity was met (VIF = 1.019). An inspection of standardized residual values (>3) and Cook distance revealed no outliers. Goodness of fit was confirmed with the Hosmer-Lemeshow test (P = .45). The model correctly classified 82% of cases (95% CI, 75.3%-93.2%).
Only 2 independent variables made a statistically significant contribution to the model. The Ankle-GO score <19 points (OR, 5.5; 95% CI, 1.6-19.2; P = .007) and injury history (OR, 3.5; 95% CI, 1.1-10.9; P = .03). That is, athletes scoring <19 points at the preseason testing session have 5.5 times higher odds of sustaining an LAS. Similarly, players who suffered an LAS in the previous season had 3.5 times higher odds of sustaining a new LAS on the same ankle.
Discussion
The results of the present study revealed that elite basketball players exhibiting lower preseason Ankle-GO score were at higher odds of sustaining an LAS during the competitive season.
This finding can be attributed to the fact that the Ankle-GO score incorporates balance-oriented tests that have previously shown predictive value for LAS risk. For instance, the SLS has been linked to higher LAS incidence in several populations.12,34 In a cohort of 106 professional football and basketball players, Halabchi et al 12 reported that poor single-leg balance performance and limited plantarflexion was associated with both acute and recurrent LAS. 12 Similarly, Trojian and McKeag 34 demonstrated that athletes (American football, soccer, and volleyball players) unable to maintain stable SLS were 2.54 times more likely to sprain their ankles, a risk that increased further (Relative Risk, 8.82) without prophylactic taping. On the other hand, the mSEBT has also been recognized for its predictive value.11,17 Gribble et al 11 showed that lower mSEBT performance and higher BMI significantly increased LAS risk in high school and collegiate football players, whereas Ko et al 17 found that weaker posteromedial and posterolateral mSEBT performance predicted LAS in adolescent soccer players. Results from the current study confirmed poorer performances on the composite score of the mSEBT in the injured group at baseline (82 ± 6.8 vs 93.2 ± 15.2). The other performance tests of the Ankle-GO score have not been previously studied as predictors for LAS in basketball players. However, their focus on balance metrics underscores their potential relevance; McGuine et al 20 evaluated postural sway—the mean degrees of sway per second—during single-leg balance tests with eyes closed in 210 high school basketball players (16 years old, on average). Athletes who sustained ankle sprains had higher preseason sway scores (2.01 ± 0.32) than those who did not (1.74 ± 0.31) (P = .001). Players with poor balance were nearly 7 times more likely to suffer ankle sprains (P = .0002). While individual components such as the SLS test have shown predictive value, our findings suggest that the combined Ankle-GO score offers stronger predictive performance than any single test alone. This likely reflects the multifactorial nature of LAS risk, which spans static and dynamic balance, functional movement, and psychological readiness. Although administering the full score requires more time than a single test, its superior predictive performance may justify its use, particularly in high-performance environments. Importantly, LAS is among the few musculoskeletal injuries for which effective, evidence-based prevention strategies—such as neuromuscular training, taping, and bracing—are well-established.15,16,29,37,39 Therefore, a screening tool such as the Ankle-GO score is not only valuable for risk identification, but also actionable: it enables targeted implementation of preventive interventions that could reduce injury incidence, recurrence, and long-term complications in at-risk basketball players.
A potential source of confounding was that previous LAS could influence preseason functional performance, thus artificially lowering Ankle-GO scores. To address this, we compared scores between patients with and without a history of LAS and found no significant difference. Furthermore, multicollinearity diagnostics confirmed that previous LAS and the Ankle-GO <19 variable were statistically independent, strengthening the validity of the predictive model. However, this subgroup comparison may have been underpowered to detect subtle differences, and the findings should be interpreted with caution.
This study demonstrates a 33% risk of LAS in basketball players over 1 season, a rate slightly higher than but consistent with existing literature. For comparison, Herzog et al 13 analyzed 389 NBA players and reported a single-season ankle sprain risk of 25.8% (95% CI, 23.9%-28.0%). The difference may be attributed to the variation in the number of games played per athlete. For instance, Tummala et al 35 reported an incidence rate of 3.71 per 1000 game exposures in a large national cohort of 554 basketball players, highlighting the role of game exposure in determining injury risk.
Our data indicate that athletes with a history of injury in the preceding season have a 3.5-fold higher odds of sustaining another LAS. This finding aligns with the literature, which consistently highlights the strong association between injury history and reinjury risk across various sports.7,13,38 Notably, basketball-specific research supports this relationship: Herzog et al 13 analyzed 796 ankle sprains among 389 NBA players over 4 seasons (2013-2017) and reported that players with a previous ankle sprain in the past year had a 1.41 times higher incidence rate (95% CI, 1.13-1.74) compared with those without a history of ankle sprain (P = .002).
This study does not indicate any significant effect of sex distribution on the risk of LAS. However, this conclusion may be biased due to the relatively small sample size. Although a formal a priori sample size calculation was conducted, it was tailored for the Ankle-GO score, not for sex-based analysis. The literature presents mixed evidence regarding sex differences in LAS risk.3,7,31 Beynnon et al 3 reported that female basketball players were at a significantly higher risk of LAS compared with male players (RR, 4.11; P = .045) in a large cohort of 901 athletes. Conversely, the review by Delahunt and Remus 7 found no consistent evidence of sex as a significant risk factor for first-time inversion ankle sprains, suggesting that other factors such as sport type and competition level might play more critical roles. Moreover, Roos et al 31 analyzed 2429 lateral ligament complex sprains in National Collegiate Athletic Association athletes and observed higher injury rates in men’s basketball (11.96/10,000 athlete exposures) compared with women’s basketball (9.50/10,000 athlete exposures). However, recurrent injuries were more prevalent in women’s basketball (21.1% vs 19.1%).
While BMI has been previously identified as a risk factor for ankle sprains in sports such as football and soccer,11,36,38 the evidence remains weak and underexplored in basketball players. 40 In our elite basketball cohort, we did not observe this association. It is possible that elite basketball players, with their unique physical profiles and taller statures, may not exhibit the same BMI-related risks observed in other athletes.
Last, regarding potential ceiling effects in the Ankle-GO scoring in this elite population, only 5 players (10%) reached the maximal score of 25 points on ≥1 ankle (1 athlete achieved the maximum on both ankles). These findings suggest no ceiling effect within the data set.
Limitations
This study is limited by its relatively small sample size, which may affect the statistical power and generalizability of the findings. Additionally, the findings are specific to elite basketball players, limiting their applicability to other levels of play, other sports, and the general population. Another limitation is the absence subgroup analysis regarding injury severity and mechanisms (contact vs noncontact). Comparing these mechanisms could improve the predictive value of the Ankle-GO score, particularly for noncontact injuries but larger sample size studies are required to perform such analysis.
Conclusion
Ankle-GO score is associated with the probability of sustaining an LAS in elite basketball players during a competitive season. Athletes scoring <19 points had 5.5 times higher odds of suffering an LAS. In addition, players who sustained LAS during the previous season had 3.5 times higher odds of reinjuring the same ankle. Medical and strength and conditioning staff could use the Ankle-GO score to identify at-risk players and should target impairment revealed with the score to prevent the occurrence of LAS. Injury history should also be taken into account to mitigate the risk of LAS in elite basketball teams.
Footnotes
Submitted March 15, 2025; accepted November 13, 2025.
One or more of the authors has declared the following potential conflict of interest or source of funding: A.H. has received consulting fees from Arthrex and Depuy. R.L. is a consultant for Arthrex and a consultant and developer for Serf Extremity and Implant Service Orthopédie. AOSSM checks author disclosures against the Open Payments Database (OPD). AOSSM has not conducted an independent investigation on the OPD and disclaims any liability or responsibility relating thereto.
Data accessibility statement:
Data are available from the authors upon reasonable request.
