Does the JK Classification Have Good Reliability for Making Treatment Choices and Prognosticating Outcomes in Radial Polydactyly?

Background: The Wassel-Flatt classification has been widely used to categorize radial polydactyly, but it has limitations in describing variations such as the hypoplastic and triphalangeal types. To address these limitations, the modified Wassel-Flatt classification was introduced. However, this classification also has drawbacks in categorizing of Type VII and fails to include angular-type hypertrophied epiphysis of the distal phalanx. Therefore, the JK classification, a revised version of the modified Wassel-Flatt classification, was recently proposed. However, its reliability and usefulness in prognosticating for postoperative outcomes, when compared with other classifications, have not been thoroughly investigated.

Questions/purposes: (1) What are the interobserver and intraobserver reliability values for the JK classification, and how do these compare with the Wassel-Flatt and the modified Wassel-Flatt classifications? (2) Which classification types were associated with higher risk of reoperation? (3) Does the JK classification offer prognostically useful information, as measured by Japanese Society for Surgery of the Hand (JSSH) scores?

Methods: Between October 2017 and March 2020, a total of 129 patients with radial polydactyly underwent surgery by a single surgeon in our institution, and these patients were considered eligible candidates for this study. Of those, we excluded patients older than 5 years of age at the time of surgery (one patient), those with < 24 months of postoperative follow-up (16 patients), and those with incomplete medical records (two patients). Ultimately, 85% (110) of patients were included in our evaluation of interobserver and intraobserver reliability as well as in our assessment of reoperation risk and JSSH scores. The mean ± SD age at surgery was 8.8 ± 4.8 months, and the mean follow-up period was 30 ± 4.2 months. Fifty-eight percent (64 of 110) were boys and 42% (46) were girls. Patients were categorized using the Wassel-Flatt, modified Wassel-Flatt, and JK classification systems. Interobserver and intraobserver reliability were assessed using Cohen and Fleiss kappa coefficients. Reoperation risk was compared across the different classifications. Postoperative clinical outcomes were evaluated using the JSSH scoring system with patient categorization based on the JK classification.

Results: The JK classification demonstrated almost perfect interobserver (κ = 0.86 [95% confidence interval (CI) 0.81 to 0.89]) and intraobserver (κ = 0.94 [95% CI 0.88 to 0.99]) reliability. The modified Wassel-Flatt classification also exhibited almost perfect interobserver (κ = 0.82 [95% CI 0.78 to 0.86]) and intraobserver (κ = 0.92 [95% CI 0.85 to 0.98]) reliability. The Wassel-Flatt classification showed substantial interobserver reliability (κ = 0.68 [95% CI 0.62 to 0.74]) and almost perfect intraobserver reliability (κ = 0.85 [95% CI 0.77 to 0.92]). The interobserver reliability of the JK classification was higher than that of Wassel-Flatt classification (p < 0.001). However, there was no difference in the interobserver reliability of the JK and modified Wassel-Flatt classification (p = 0.73). There were no differences in intraobserver reliability between the JK classification and either the modified Wassel-Flatt classification (p = 0.99) or the Wassel-Flatt classification (p = 0.23). Of the three classifications, only the JK showed an association between a specific type (Type VIIB) and risk of reoperation (4 of 6; p < 0.001), whereas the other types had lower reoperation risk. Among the nonhypoplastic types according to the JK classification, post hoc analysis revealed that Type VIIA (median [IQR] 16 [13.5 to 18.5]) had lower JSSH scores than Type II (19 [19 to 20]) and Type IV thumbs (19 [18 to 20]; p < 0.01); the differences in these scores are clinically important. When dichotomized into hypoplastic and nonhypoplastic types based on JK classification, the median (IQR) JSSH scores of hypoplastic types (20 [20 to 20]) were higher than those of nonhypoplastic types (19 [18 to 20], difference of medians 1; p < 0.001).

Conclusion: The JK classification demonstrates an almost perfect interobserver and intraobserver reliability for classifying radial polydactyly. Among the three classification systems evaluated, only the JK classification identified an association between Type VIIB and a higher risk of reoperation. Additionally, according to the JK classification, patients with nonhypoplastic types exhibited lower JSSH scores compared with those with hypoplastic types. Within the nonhypoplastic types, Type VIIA had lower JSSH scores than Type II and Type IV. These findings may aid surgeons during preoperative planning and patient counseling.

Level of evidence: Level II, prognostic study.