LAUSR

BACKGROUND Upper limb orthopedic surgery has no standardized complication classification system. The adaptations in the Clavien-Dindo classification system have only been validated through the creation of scenarios. We hypothesized that the modified Clavien-Dindo-Sink (CDS) classification of hand and upper extremity surgery complications has good intraobserver and interobserver reliability in real-world scenarios. This study assessed the accuracy of the modified CDS classification in real-world scenarios and the absolute incidence of 30-day complications. METHODS The sample size calculation was based on the modified CDS classification. A kappa value of 0.84, accompanied by a 95% confidence interval (CI) of 0.1 and involving 6 observers, indicated the inclusion of 111 events. To achieve the anticipated sample size of 111 complications, 1850 postoperative procedures will be necessary, given the expected 6% incidence of complications. Six evaluators, each with varying levels of expertise in the field, classified the cases according to the modified CDS classification in 2 rounds. RESULTS The incidence of early complications was 7.05% (95% CI = 6%-8%) in a tertiary care university hospital. Intraobserver agreement was moderate to almost perfect; however, interobserver agreement was moderate. The less experienced surgeons presented a better interobserver correlation than the more experienced ones. CONCLUSIONS In a real-world scenario of early postoperative complications of upper extremity surgery, the modified CDS classification presented a lower interobserver correlation than reported in the literature. The surgeon's experience could influence its applicability.