CLINICAL OBSERVATION AND ANALYSIS OF MOVEMENT QUALITY DURING PERFORMANCE ON THE STAR EXCURSION BALANCE TEST

Abstract

Study Design:

Observational

Background:

The Star Excursion Balance Test (SEBT) is used to evaluate dynamic postural control and screen for injury risk. No prior studies have investigated whether the quality of movement during the SEBT has clinical value and can adequately predict injury.

Purpose:

To develop a visual assessment tool and evaluate the relationship between movement quality and SEBT outcomes.

Methods:

One hundred healthy subjects were included. Baseline demographic, limb length, and individual SEBT performance data were collected. SEBT outcomes were obtained and used to classify individuals as at‐risk or not at‐risk. At‐risk individuals demonstrated anterior right/left reach distance difference greater than 4 cm, and/or normalized composite reach distance less than 89.6% for males or 94% for females. Three independent reviewers, blinded to SEBT outcomes, assessed the anterior reach test on videotape. Reviewers underwent training on a scoring system to assess movement quality at the trunk, pelvis, and knee. The total score of movement faults was used to determine interrater reliability and calculate sensitivity and specificity, in addition to positive and negative predictive values of SEBT outcome.

Results:

Seventy‐one subjects were classified as at risk. Interrater reliability of movement scoring was poor‐moderate for the trunk and pelvis (κ=0.18‐0.43), and moderate for the knee (κ=0.5‐0.6). Rater agreement for total movement score was fair‐moderate (W=0.64‐0.73). Rater assessment of aberrant movement was not predictive of SEBT performance. However, subjects deemed at risk had fewer movement faults per rater assessment. Raters displayed moderately strong specificity (0.59‐0.82) and poor sensitivity (0.14‐0.39) in knee assessment to detect at risk performance on the SEBT.

Conclusion:

Clinical observation of knee movement demonstrated acceptable interrater reliability and moderately strong specificity to detect at‐risk SEBT outcome. Total movement score across all regions demonstrated fair‐moderate agreement. Subjects who were at risk tended to have fewer movement faults.

Level of evidence:

3

INTRODUCTION

The Star Excursion Balance Test (SEBT) is an accepted, low‐cost, and reliable^1,2 instrument used in physical therapy practice as a means of evaluating dynamic postural control. Differences in reach distances between lower limbs on the SEBT have been found to be sensitive in identifying individuals with chronic ankle instability,^3–5 patellofemoral pain syndrome,⁶ anterior cruciate ligament deficiency,⁷ and movement discrepancy in those who have undergone anterior cruciate ligament reconstruction (ACL‐R).⁸ Additionally, the results of the SEBT have been identified as a predictor of injury in high school basketball players⁹ and collegiate football athletes;¹⁰ however, it is encouraging for rehabilitation professionals and athletes that SEBT performance can be improved through an appropriate intervention program.¹¹

Dynamic postural control is one of numerous intrinsic risk factors for lower extremity injury reported in the literature.¹² Intrinsic risk factors for injury include, but are not limited to: age,^13,14 sex,¹⁵ flexibility,¹⁶ strength,¹⁷ neuromuscular control deficiencies,¹⁸ muscle imbalances,¹⁹ and injury history.^16,20,21 Of these risk factors, history of previous injury is one of the most predictive of future injury.²² These findings demonstrate the need for appropriate screening measures to identify modifiable injury risk factors in order to implement preventative strategies prior to initial injury, as subsequent injury can often be more devastating than the initial injury.²³

Neuromuscular control deficiencies including greater knee dynamic valgus and high hip abduction loads during specific standardized tests have identified athletes who may be at risk for future injury.¹⁸ Body region–specific movement assessments, such as kinematic analysis to quantify frontal plane knee motion have been examined in the laboratory setting during various athletic movements including running, single leg squat, and landing from a jump.^24–28 Kinematic differences during the SEBT anterior (ANT) reach between those with ACL‐R and a control group have been demonstrated, including increased hip adduction and decreased hip/knee flexion.⁸ Despite the differences in kinematic profiles, Delahunt et al. found no significant quantifiable reach differences between these two groups.⁸ These findings highlight the importance of observation of movement patterns during rehabilitation in the presence of clinically acceptable quantitative SEBT reach performance. While the previous studies are useful for appreciating movement deficiencies, clinical application of laboratory‐based research is often challenging. Rehabilitation professionals rarely have access to the sophisticated instrumentation and software required for three‐dimensional biomechanical movement analysis, thus creating the need for visual assessment tools to evaluate dynamic postural control in an accessible and financially attainable manner.

Clinical scoring criteria have been developed to qualify other functional tasks including the tuck jump assessment,²⁹ lateral step down test,^30–32 and single leg squat test^26,31,33 with clinically acceptable levels of reliability.^26,29–33 Movement quality of these functional tasks were assessed based on the presence or degree of aberrant movement at the trunk, pelvis, hips, and knees. These aberrant movement patterns, in part are consistent with those associated with lower extremity injury from a previous investigation.³⁴

The SEBT has proven valuable as an instrument for predicting lower extremity injury^9,10 and for pre‐participation athlete screening. Although previous SEBT protocols do not suggest the need for qualitative analysis, the SEBT provides a platform to observe movement quality, which may guide future intervention and assist with the decision‐making process of when to allow an athlete to return to competition. No prior studies have investigated movement quality observed in a clinical setting during the SEBT or determined if a relationship exists between movement quality and quantitative values on the SEBT. The purposes of this study were to (a) rate movement quality as observed during performance of the SEBT ANT reach, (b) determine if a relationship exists between movement quality and quantifiable outcomes of the SEBT, and (c) develop reliable scoring criteria to qualitatively describe movement during the SEBT. The authors’ hypothesis was that those individuals with more observable movement faults as defined by the predetermined scoring criteria would more likely be determined to be at risk on the SEBT, and that visual assessment of the knee would be the most reliable between raters.

METHODS

Subjects

Healthy subjects were recruited from a university population as a sample of convenience. A power analysis revealed that 100 subjects were needed to complete testing in order to attain the study’s objectives based on the proportion of subjects at‐risk from previous studies.^9,10

To be included in this study, subjects were required to be between the ages 18 and 35 years old. Subjects were excluded from participation if they reported (a) previous low back, hip/pelvis, knee, ankle, or foot surgery within the past year; (b) injury to the lower extremities or low back in the previous six months; (c) lower extremity amputation; (d) currently undergoing treatment for inner ear, sinus, head cold, or upper respiratory infection; (e) known balance impairment due to neurological disorder, vestibular disorder, medication use, or other; (f) concussion within the previous three months; or (g) pregnancy. Compensation was not provided to the subjects for their participation. Prior to participation, each subject read and signed an informed consent form approved by the Institutional Review Boards of the University of Wisconsin – La Crosse and Gundersen Health System. Presence of pain during testing was not a specific objective of the current study, as the study population was limited to healthy, non‐injured subjects.

Procedures

Two examiners performed data collection for this study during 2013‐2014. Both examiners were sports physical therapy residents with similar levels of physical therapy clinical experience. Data collection took place in a university classroom setting. Initial data collected for each subject included sex, age, height (cm), body mass (kg), limb dominance (defined as the foot used to kick a soccer ball), and self‐reported activity level using the Tegner Activity Level Scale.³⁵ Prior to data collection, the examiners met to clarify procedures of the tests to be performed. Examiners performed pilot testing on five healthy subjects and made final modifications to the testing procedures.

Lower Limb Length

Limb length measurements were attained as described in a previous study.³⁶ Measurements were performed with the subjects supine on a treatment table. Subject position was standardized by having the participant lie supine with bilateral hips and knees flexed to approximately 45°. The participant performed a bridge maneuver and slowly lowered the pelvis to the treatment table. The examiner passively straightened the lower limbs and provided a gentle distraction force. Limb length was measured in centimeters using a cloth tape measure from the inferior aspect of the anterior superior iliac spine (ASIS) of the pelvis to the ipsilateral distal medial malleolus.

Star Excursion Balance Test Protocol

The SEBT testing protocol was adapted from a previous study.⁸ Participants viewed an instructional video detailing procedures and demonstration of the SEBT prior to testing. During this video, the participant was educated on what constituted an unsuccessful trial. The trial was considered unsuccessful if the subject (a) lost balance or failed to maintain unilateral stance, (b) shifted weight onto the reach foot when touching the measuring tape, (c) failed to perform a controlled return of the reach foot to the starting position prior to reaching another direction, (d) did not maintain heel contact between the stance limb and floor, (e) did not maintain hands on the pelvis, (f) moved or lifted the stance foot, or (g) failed to touch the measuring tape. If the trial was deemed unsuccessful, it was discarded and repeated until a successful trial was performed.

The SEBT was performed in three reach directions as described previously.⁹ Three tape measures labeled in 0.5 cm increments were secured to the floor. One tape measure was oriented to the apex, which was defined as the ANT reach direction. Two additional tape measures were anchored at 135° to the apex, which were defined as the posteromedial (PM) and posterolateral (PL) reach directions. The stance foot starting position was defined as the anterior border of the second toe at the junction of the three reach direction lines of the SEBT. The starting position for the reach foot was defined as the area immediately adjacent to the stance limb in a non–weight‐bearing position. Participants placed their hands on their iliac crests to standardize starting position while standing barefoot. Participants were required to tightly tuck in their shirts into their waistband, which was aligned to be level with their ASIS. A strip of white athletic tape was used to accent the ASIS landmarks. Shorts were fit to allow for visualization of knee position. One‐inch red stickers were placed over each tibial tuberosity for improved visualization. Reliability of the examiners collecting SEBT performance data was not established for the present study; however, previous investigations have demonstrated high intratester and intertester reliability during the SEBT with recommendations for completion of practice trials to account for learning effects.^1,2

Four practice trials were performed in all directions on each limb prior to administration of the test.³⁷ Participants were given verbal feedback if they were performing the test incorrectly. No additional instructions were given regarding movement strategies or posture. Participants were given a two‐minute rest period prior to formal testing. Formal testing included three test trials for each lower extremity for each of the ANT, PM, and PL reach directions. Each test trial was named and recorded according to stance limb and reach direction with the standardized testing order applied as follows: right ANT, left ANT, right PM, left PM, right PL, and left PL reach directions. The sum of the greatest reach distance for each of the three directions was divided by three times limb length and multiplied by 100 to calculate composite (COMP) reach distance for each leg.⁹

A test trial failure was determined if a successful attempt was not established within six attempts for a particular direction. All test trials were observed and recorded by a principal examiner. The ANT reach trial with the greatest reach distance was indicated on the examiner data sheet for retrospective rater viewing. Each ANT reach test trial was videotaped and transferred to DVD format for retrospective viewing. A video camera was positioned three meters in front of the subject to capture a full frontal plane view. Digital images were stored on a secure hard drive and transferred to DVDs for rater viewing.

SEBT Movement Scoring and Clinician Raters

From the quantitative scores attained from the SEBT, subjects were categorized into two groups: (a) those at‐risk for future injury and (b) those not at‐risk. Individuals were classified as at‐risk if they met any of the following criteria: (a) a difference in reach distance between limbs in the ANT reach directions of greater than 4 cm,⁹ (b) COMP reach distance less than 94.0% of limb length for females,⁹ or (c) COMP reach distance less than 89.6% of limb length for males.¹⁰ All other subjects were categorized as not at‐risk.

The ANT reach test trial with the greatest reach distance for each limb was independently viewed and scored by three physical therapist raters of varying clinical experience (18, 6, and 1.5 years) because this was the reach direction most predictive of lower extremity injury⁹ (Figure 1). Raters were blinded to SEBT quantitative values, as well as the determination of risk status. Three predetermined aberrant movement patterns (Table 1) were selected as a modification of the scoring criteria devised from previous investigations³⁰ of the lateral step down test for the trunk, pelvis, and knee region. In addition, the altered kinematic profiles of hip and knee motion in an ACL‐R population from a previous study were used to define the aberrant movement patterns.⁸ These movements were scored in a dichotomous manner as either present (1) or absent (0) throughout the ANT reach SEBT test trial for each limb. Total movement score was defined as the sum of movement faults for each respective limb, ranging from 0 to 3. Clinician raters were instructed to limit each trial to two real‐time viewings, starting with the first subject. Raters received formal training on the scoring criteria prior to viewing the test trials.

Figure 1.

Star Excursion Balance Test Anterior Reach

Table 1.

Movement Scoring Criteria

Body region	Movement pattern	Score
Trunk	Trunk alignment leaning or rotating in any direction	1
Pelvis	Loss of horizontal plane greater than 10° (pelvic drop)	1
Knee	Tibial tuberosity medial to the 2nd toe	1

*Scored independently for each lower extremity throughout the anterior reach

Statistical Methods

The primary intent of this study was to examine the relationship between movement quality as determined by an experienced physical therapist and quantifiable SEBT outcomes. The degree to which rater assessments are able to predict risk status derived from the SEBT was assessed across the three raters for each side of the body (left vs right) and for the three body regions (trunk, pelvis, and knee) of interest, for a total of six comparisons. Calculations used the Cochran‐Mantel‐Haenszel test and resulting odds ratios, and homogeneity of the odds ratios was assessed via the Breslow‐Day test. Interrater agreement for categorical responses was assessed via Cohen kappa with modifications for more than two raters, while agreement for ordinal responses was assessed via Kendall coefficient of concordance (W). Both calculations used the MAGREE macro. Groups were compared using Pearson χ² tests for categorical data and the Wilcoxon rank sum tests for ordinal data. All calculations were performed with SAS/STAT 9.3 (Cary, NC), and a p‐value of less than 0.05 was considered significant for all comparisons.

RESULTS

Participants

One hundred seven subjects volunteered for this study, seven of whom were excluded due to recent history of injury to the lower extremities or low back (n=4), sinus or upper respiratory infection (n=2), and recent history of concussion (n=1). One hundred subjects completed the testing protocol and were included in data analysis (Table 2). None of the subjects reported experiencing any personal harm during testing. Subjects were not significantly different in age, body mass index (BMI), or Tegner activity level with respect to SEBT risk classification. Of the subjects included in data analysis, females were significantly more likely than males to be classified as at‐risk on at least one limb, per COMP risk criteria (Table 3).

Table 2:

Subject demographics

Characteristic	Value
Age, mean y ± SD	20.97 ± 2.55
BMI, mean kg/m2 ± SD	23.37 ± 2.96
Sex	64 females, 36 males
Limb dominance	95 right, 5 left

SD= standard deviation; BMI= body mass index.

Table 3.

Participant Classification of Risk

Participant classification	Females n = 64	Males n = 36	Total n = 100
COMP or ANT reach at‐risk	54 (84.4%)	17 (47.2%)	71 (71%)
COMP and ANT reach at‐risk	4 (6.3%)	2 (5.6%)	6 (6%)
COMP reach at‐risk	51 (79.7%)	15 (41.7%)	66 (66%)
ANT reach at‐risk	7 (10.9%)	4 (11.1%)	11 (11%)

COMP=composite; ANT=anterior.

Rater Agreement

In general, rater agreement with respect to the presence or absence of movement faults in the trunk and pelvis regions was poor to moderate, with kappa values ranging from 0.18 to 0.43 (Table 4). Agreement in knee assessment was consistently better when compared with that of the other regions (κ=0.5‐0.6). Rater agreement on the total number of movement faults noted for each side of a subject was fair to moderate, with better agreement on the right side (W=0.732) than on the left (W=0.644).

Table 4:

Interrater agreement, reported as Kappa coefficients and percent agreement.

Body region	Kappa		% Agreement
	Right	Left	Right	Left
Trunk	0.21	0.18	45	42
Pelvis	0.43	0.36	58	55
Knee	0.60	0.50	80	69

Rater Prediction of Risk

Collectively, rater finding of aberrant movement patterns was not predictive of an at‐risk performance on the SEBT (Appendix 1). The only significant association between collective rater assessment and the COMP reach was a negative association between aberrant trunk movement patterns and the results of the SEBT on both the left and right sides; however interrater agreement in these regions was poor. With the exception of the left pelvic region, homogeneity of the odds ratios via the Breslow‐Day test was acceptable. Additionally, subjects deemed at risk tended to have fewer movement faults per rater assessment (Appendix 2).

Examined individually, the sensitivity of rater assessment to detect at‐risk performance per the SEBT was poor and varied significantly between raters and across body regions (Table 5). All three raters displayed moderately strong specificity in knee assessment—generally around 70%; however, specificity was notably lower and displayed larger variation between raters for the pelvis and trunk. When two or more raters agreed, collective rater knee specificity was acceptable for knee assessment (Appendix 3).

Table 5:

Sensitivity, Specificity, Negative and Positive predictive values.

Rater	Side	Body region	Sensitivity	Specificity	NPV	PPV
1	Left	Trunk	0.804	0.023	0.083	0.511
		Pelvis	0.518	0.477	0.438	0.558
		Knee	0.393	0.591	0.433	0.55
	Right	Trunk	0.813	0.056	0.143	0.605
		Pelvis	0.531	0.611	0.423	0.708
		Knee	0.297	0.75	0.375	0.679
2	Left	Trunk	0.25	0.568	0.373	0.424
		Pelvis	0.125	0.591	0.347	0.28
		Knee	0.232	0.727	0.427	0.52
	Right	Trunk	0.313	0.556	0.313	0.556
		Pelvis	0.375	0.611	0.355	0.632
		Knee	0.141	0.778	0.337	0.523
3	Left	Trunk	0.518	0.159	0.206	0.439
		Pelvis	0.321	0.591	0.406	0.5
		Knee	0.25	0.818	0.462	0.636
	Right	Trunk	0.547	0.111	0.121	0.522
		Pelvis	0.5	0.528	0.373	0.653
		Knee	0.156	0.778	0.341	0.556

NPV= negative predictive value; PPV= positive predictive value.

DISCUSSION

Movement Quality and SEBT Outcome

Contrary to the authors’ initial hypothesis, those who were deemed at risk on COMP reach tended to display fewer movement faults per rater assessment of total movement score. This inverse movement fault to COMP at‐risk relationship may be due to protective compensations. The subject may be able to maintain adequate control only within a certain range of their “cone of stability.”³⁸ The cone of stability is a mapping within the central nervous system that is used to maintain equilibrium during a movement or task.³⁸ The cone of stability may get distorted in the presence of impairments in proprioception, balance, or neuromuscular control.³⁸ The authors hypothesize that subjects displaying fewer movement faults while still being at risk per SEBT scoring criteria may be unable or unwilling to move toward the limits of their cone of stability, therefore decreasing the use of aberrant movement patterns. It stands to reason that there would be an increase in aberrant movement patterns if the participant reached further toward the limits of their cone of stability. Further investigation is needed to identify whether a clinical relationship exists between movement strategy and SEBT performance.

Rater assessment of aberrant movement with respect to SEBT outcome varied according to each body segment. Lower extremity dynamic postural control is often depicted as a kinematic chain of interworking parts to complete a movement task. It has been demonstrated that aberrant movement of the lower extremity is often characterized by abnormal trunk position, poor pelvis and poor knee control rather than an isolated movement fault at a single location.³⁹ When performing the SEBT, aberrant movement patterns may differ according to a limitation in a specific aspect of the movement. For example, an athlete with a limitation in dorsiflexion range of motion may adopt movement strategies vastly different from those displayed by an athlete with decreased proprioception—and such strategies may be specific to the impairment or the individual. Mobility limitations may alter the movement pattern of the athlete, as ANT reach distance has been shown to be influenced by ankle dorsiflexion range of motion.^40,41 In addition, increased hip strength (i.e. greater proximal stability) has been positively correlated with SEBT performance.^42,43 These various constructs of movement may influence movement quality or strategy according to the limitation placed on the athlete.

The SEBT has demonstrated the ability to predict injury risk by assessing both ANT and COMP reach performance.^9,10 Clinical observation of SEBT movement quality occurred only in the ANT reach direction from a frontal plane perspective, limiting the generalizability of the findings for those with COMP risk outcome. In developing movement scoring criteria, the authors felt it was difficult to define reliable aberrant movement patterns in the PM and PL reach directions. Additionally, the SEBT reach is a three‐dimensional task, which may complicate clinical observation of movement from a single‐plane perspective. The addition of a sagittal plane perspective may have provided further insight into ANT reach movement assessment including relative knee and hip flexion angles. Ideally, movement quality assessment for the ANT reach direction would have been compared with ANT reach outcome; however, the current sample was too small to perform this analysis. Future research may identify movement scoring criteria in all reach directions in order to establish a better comparison with ANT and COMP risk.

Rater Agreement

Confirming the initial hypothesis, interrater reliability of movement assessment for the knee region was the highest of the three body regions. Rater agreement for knee assessment may have been greater due to improved visualization of the tibial tuberosity relative to the foot in the frontal plane. In contrast, contralateral pelvic drop and trunk lean/rotation may be too subtle to visually assess in a reliable fashion without specific anatomical landmarks to score the movement utilizing a single plane video perspective. Overall, agreement for total movement score between raters was fair to moderate, which was similar to that reported for previous investigations of the lateral step down test.^30,32 Examiner reliability for collecting SEBT performance data was not established in the present study, as high intratester and intertester reliability during the SEBT has been demonstrated in previous investigations.^1,2

Participants

The percentage of participants in the current study who were found to be at risk per COMP reach criteria was higher than that reported in previous studies.^9,10 One study found a similar risk stratification for both males and females.⁹ Contrary to that finding, in the current study there was a significantly greater number of at‐risk females as compared to males. Athlete gender is only one of the many intrinsic risk factors for lower extremity injury.¹² Increased injury risk among female athletes may be partially attributed to neuromuscular control differences between men and women.^44,45 While the difference between genders exists, dynamic postural control can be improved with an appropriate training program for female athletes.¹¹

SEBT performance is dependent on age, sex, sport, and athletic ability.^9,10,46 Previous studies reported on injury risk in high school athletes⁹ and collegiate football players,¹⁰ while the current study investigated active, college‐aged individuals, with a majority of the subjects classified as recreational athletes. Differences in the total volume of at‐risk subjects may be explained in part by the differences in study population. Utilization of current risk stratifications as a screening tool or for guidance in return to sport decision making may not be as appropriate if applied to a recreational athletic population. Normative SEBT data and injury risk stratifications have yet to be established in individuals similar to those included in the current study. Further prospective investigation is warranted to establish normative data and risk stratification across various populations to expand the clinical utility of the SEBT as a screening tool.

Limitations

The results of the current study are generalizable to a healthy, college‐aged population involved in recreational and competitive athletics. Limitations did not prevent attainment of the study objectives; however, generalizability of this single‐center study was limited in a number of ways. Movement analysis using the SEBT was performed in healthy subjects, making clinical application to those with injuries a challenge. Subjects were not followed prospectively to determine incidence of lower extremity injury. Secondly, application of risk classification from previous investigations^9,10 was limited due to inconsistencies between testing protocols as well as differences in performance between the SEBT and Y Balance Test.⁴⁷ Applications of and clinical utility of the SEBT requires further investigation in order to develop guidelines for appropriate use in the rehabilitation setting.

Conclusion

Clinical observation of knee movement during performance of the ANT reach of the SEBT demonstrated acceptable interrater reliability and moderately strong specificity to detect at‐risk SEBT outcome. Observation of trunk and pelvis movement demonstrated poor to moderate interrater reliability. Total movement score across all body regions demonstrated fair to moderate agreement between raters. Subjects who were at risk tended to have fewer movement faults.

Aberrant movement of the knee during the SEBT ANT reach appears to be the most reliable, and demonstrated the highest specificity for SEBT at‐risk outcome when compared to the trunk and pelvis as identified by clinician raters. Future research should investigate the need for a qualitative assessment of the SEBT.

Appendix 1.

Odds Ratios for Rater Prediction of Risk

Body region	Odds Ratio for prediction of Composite Risk only [95% CI]		Odds Ratio for prediction of Composite Risk or Anterior Risk [95% CI]
	Right	Left	Right	Left
Trunk	0.27 [0.15 – 0.49]	0.22 [0.12 – 0.40]	0.32 [0.17 – 0.58]	0.26 [0.15 – 0.48]
Pelvis	1.21 [0.76 – 1.93]	0.63 [0.39 – 1.01]	1.24 [0.77 – 1.99]	0.58 [0.36 – 0.93]
Knee	0.77 [0.44 – 1.36]	0.81 [0.49 – 1.35]	0.82 [0.46 – 1.45]	1.02 [0.61 – 1.70]

Appendix 2.

Total Movement Score and SEBT Outcome

Total Movement Score	Composite Risk, Right Side		Composite Risk, Left Side		Composite or Anterior Risk, Right Side		Composite or Anterior Risk, Left Side
	No	Yes	No	Yes	No	Yes	No	Yes
0	18	56	19	51	18	56	19	51
1	43	57	56	55	41	59	53	58
2	39	52	42	39	36	55	37	44
3	14	21	24	14	13	22	23	15

Appendix 3.

Rater consensus (2 or more raters agree) Sensitivity, Specificity, Negative Predictive Values (NPV), and Positive Predictive Values (PPV) for SEBT At‐Risk Performance

Total Movement Score	Composite Risk, Right Side		Composite Risk, Left Side		Composite or Anterior Risk, Right Side		Composite or Anterior Risk, Left Side
	No	Yes	No	Yes	No	Yes	No	Yes
0	18	56	19	51	18	56	19	51
1	43	57	56	55	41	59	53	58
2	39	52	42	39	36	55	37	44
3	14	21	24	14	13	22	23	15