Reliability of the Walter Reed Classification for Heterotopic Ossification Severity in Amputees

Abstract

At this time, there is no validated tool for describing heterotopic ossification (HO) severity or measuring efficacy of therapies after amputation. This study aims to validate the Walter Reed classification system for reliable characterization of HO burden in residual limbs. Descriptive, Level IV. US Military Trauma Referral Center. Five orthopaedic surgeon raters with varying experience levels assessed orthogonal radiographs of 100 randomly selected residual upper and lower limbs after combat-related amputation. A standardized education session on the Walter Reed HO classification system followed by 2 grading sessions, separated by a 2-week waiting period. Scores were analyzed for interobserver and intraobserver reliability using weighted Cohen's kappa, intraclass correlation coefficient, and Krippendorff's alpha for ordinal variables. After the waiting period, 89% of HO was rated the same by observers with an intraobserver kappa = 0.83 (95% confidence interval [CI] 0.75-0.92). Interobserver reliability was evaluated using group-wise comparison among observers, resulting in an overall Krippendorff's alpha = 0.78 (95% CI 0.76-0.80). Using intraclass correlation coefficient, interrater consistency was 0.94 (95% CI 0.91-0.96). The Walter Reed HO classification system is substantially reliable for use by raters with any level of orthopaedic training. In addition, it is reliable for evaluation of both through bone and disarticulation type amputations in the upper and lower extremities. However, in 11% of cases, a secondary read may vary, suggesting that this system is useful for understanding HO and developing interventions but may benefit from further refinement, including advanced imaging and clinical correlation.