Article
A Comparative Analysis of 150 Thumb Polydactyly Cases from the CoULD Registry Using the Wassel-Flatt, Rotterdam, and Chung Classifications
Search Medline for
Authors
-
Ann Van Heest
- University of Minnesota, Minneapolis, United States
-
Eliza Thompson
- University of Minnesota, Minneapolis, United States
-
Julie Agel
- Harborview Medical Center, Seattle, United States
-
Andrea Bauer
- Boston Children's Hospital, Boston, United States
-
Charles Goldfarb
- Washington University St. Louis, St. Louis, United States
-
Caroline Hu
- University of Minnesota, Minneapolis, United States
-
Amy Moeller
- University of Minnesota, Minneapolis, United States
-
Susan Novotny
- Gillette Children's Specialty Healthcare, St. Paul, United States
-
Deborah Bohn
- University of Minnesota, Minneapolis, United States
| Published: | February 6, 2020 |
|---|
Outline
Text
Objectives/Interrogation: The purpose of this investigation was to compare the inter-rater reliability of the Wassel-Flatt (WF), Rotterdam, and Chung classification systems in patients enrolled in a prospective, multicenter cohort study. Our hypothesis is that the WF, Rotterdam and Chung classifications would have similar inter-observer reliability for classification of 150 radial polydactyly cases.
Methods: Five pediatric hand orthopedic surgeons, one orthopedic resident, and one medical student independently reviewed clinical photographs and radiographs of 150 radial polydactyly cases from the CoULD (Congenital Upper Limb Database) registry. After completion of a training set, each rater classified each case according to WF, Rotterdam, and Chung classifications. The intraclass correlation coefficient (ICC) was calculated using two-way random measures with perfect agreement.
Results and Conclusions: To demonstrate that a full spectrum of cases was available, Table 1 [Tab. 1] exhibits the range of classifications as graded by the 7 raters:
The ICC for inter-rater reliability was 0.93 for the WF classification, 0.98 for Rotterdam types I-VIII, and 0.88 for the Chung classification. On average, the raters felt that 59% of the thumbs were accurately captured in the WF, 90% of thumbs in the Rotterdam, and 48% of thumbs in the Chung classification systems.
While the Chung classification system provides the simplest description of thumb polydactyly (4 types) and was designed to better inform surgical treatment, it had the lowest inter-observer reliability and was considered to capture the clinical presentation in the smallest percentage of cases. The Wassel-Flatt classification provides more detail (8 types), and was intermediate in its inter-observer reliability and accuracy in capturing the clinical presentation. Although the Rotterdam classification is most complex, providing the most level of detail (8 types, with 5 subtypes for each, and 3 location descriptors for each), it had the highest inter-observer reliability when considering only type (not sub-types) and was considered to accurately describe the greatest proportion of cases.
