Background Identifying individuals with a high risk for ACL injury is important for injury prevention. Current screening methods exist but rely on expensive and time consuming biomechanical analysis, limiting large scale application.

Objective The aim of this experiment was to investigate observational ACL injury risk assessment performance.

Design Individual raters viewed 25 frontal plane video clips of female high school athletes performing a drop vertical jump. Five video clips were repeated. The decision task was to rate the athletes risk for an ACL injury on a 10 point scale (1=very low, 10=very high). Concurrent 3D peak knee abduction moment (KAM) was calculated for the video-taped athletes and served as the criterion for ACL injury risk level.

Setting Raters viewed the video clips and responded using a computer.

Participants A convenience sample of 40 allied health professionals/students served as raters.

Risk factor assessment In order to perform signal detection analysis on the judgments, ratings greater than 5 were considered high risk.

Main outcome measurements To assess judgment performance, Spearman rho (ρ), sensitivity (A') and decision bias (B_D) were calculated. Intra-rater consistency was assessed by calculating typical error between the 5 repeated video clips. Rater performance was also compared to the ACL Nomogram (KAM prediction model).

Results Correlation between continuous KAM and subjective ratings ranged from −0.16 to 0.39 (Nomogram ρ=0.64). However, the best performer (M.S. Physiotherapist) displayed a sensitivity value of 0.85 (B_D=0.19) which was similar to the Nomogram measure of 0.97 (B_D=−0.48). In general, intra-rater consistency was sufficient with 83% of raters displaying typical error values less than 1.

Conclusions Superior performance during visual assessment of ACL injury risk appears to be difficult but achievable. Research describing expertise mechanisms is warranted.