Peterson vs. Salter Harris physeal fracture classification systems: How consistent are they?

Abstract

The Salter-Harris (S-H) classification establishes five types of physeal fractures and is the most widely used to describe them. However, the Peterson classification includes fracture patterns not covered by the S-H classification, such as fractures crossing the metaphysis and extending to the physis and open injuries with loss of physis. Nonetheless, it is necessary to validate and compare its reliability with that of the S-H to enhance diagnostic possibilities and clinical interpretation. This study aimed to compare the consistency of intra and interobserver agreement of the S-H and Peterson classifications. A cross-sectional study was conducted to evaluate the reliability of 100 radiographs in the anteroposterior and lateral views. Five pediatric orthopedic surgeons and five third- and fourth-year orthopedic residents were asked to classify both fracture types on two occasions, with an eight-week interval between the two occasions. The intraobserver agreement for S-H in the expert and resident groups was κ = 0.66 and 0.61, respectively; for Peterson, it was κ = 0.67 and 0.61, respectively. The mean total agreement percentages for S-H and Peterson were 84.5% vs. 83.9%, respectively (t = 0.254; P = 0.8; Cohen's d = 0.20). The overall interobserver agreement for S-H in the first and second observations was κ = 0.57 and 0.63, respectively, and for Peterson, κ = 0.58 and 0.62, respectively. The S-H and Peterson classifications may have comparable reliability and moderate-to-substantial interobserver agreement, with no significant differences in overall agreement percentages.