Single leg mini squat: an inter-tester reproducibility study of children in the age of 9–10 and 12–14 years presented by various methods of kappa calculation

BACKGROUND Multiple studies suggest that reduced postural orientation is a possible risk factor for both patello-femoral joint pain (PFP) and rupture of the anterior cruciate ligament (ACL). In order to prevent PFP and ACL injuries in adolescent athletes, it is necessary to develop simple and predictive screening tests to identify those at high risk. Single Leg Mini Squat (SLMS) is a functional and dynamic real-time screening test, which has shown good validity and reproducibility in evaluation of postural orientation of the knee in an adult population. The aim of this study was to determine the inter-tester reproducibility of SLMS in the age group of 9-10 and 12-14 years by evaluating postural orientation of the ankle, knee, hip and trunk. Further on, this study exemplify the divergence of kappa values when using different methods of calculating kappa for the same dataset. METHODS A total of 72 non-injured children were included in the study. Postural orientation of the ankle, knee, hip and trunk for both legs was determined by two testers using a four-point scale (ordinal, 0-3). Prevalence, overall agreement as well as four different methods for calculating kappa were evaluated: linear weighted kappa in comparison with un-weighted kappa, prevalence-adjusted bias-adjusted kappa (PABAK) and quadratic weighted kappa. RESULTS The linear weighted kappa values ranged between 0.54-0.86 (overall agreement 0.86-0.97), reflecting a moderate to almost perfect agreement. When calculating un-weighted kappa (with and without PABAK) and quadratic weighted kappa, the results spread between 0.46-0.88, 0.50-0.94, and 0.76-0.95, reflecting the various results when using different methods of kappa calculation. CONCLUSIONS The Single Leg Mini Squat test has moderate to almost perfect reproducibility in children aged 9-10 and 12-14 years when evaluating postural orientation of the ankles, knees, hips and trunk, based on the excellent strength of agreement as presented by linear weighted kappa. The inconsistency in results when using different methods of kappa calculation demonstrated the linear weighted kappa being generally 15% lower than the quadratic weighted values. On average, prevalence-adjusted bias-adjusted kappa increased the un-weighted kappa values by 7% and 12% by children aged 9-10 and 12-14, respectively.