Precision or Pitfall? Evaluating the Accuracy of ICD-10 Coding for Cemented Total Hip Arthroplasty

Abstract

Background: The International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Procedure Coding System (ICD-10-PCS) was adopted in the U.S. in 2015. Proponents of the ICD-10-PCS have stated that its granularity allows for a more accurate representation of the types of procedures performed by including laterality, joint designation, and more detailed procedural data. However, other researchers have expressed concern that the increased number of codes adds further complexity that leads to inaccurate and inconsistent coding, rendering registry and research data based on ICD-10-PCS codes invalid and inaccurate. We aimed to determine the accuracy of the ICD-10-PCS for identifying cemented fixation in primary total hip arthroplasty (THA). Methods: We retrospectively reviewed all cemented primary THAs performed at 4 geographically diverse, academic medical centers between October 2015 and October 2020. Cemented fixation was identified from the ICD-10-PCS coding for each procedure. The accuracy of an ICD-10-PCS code relative to the surgical record was determined by postoperative radiograph and chart review, and cross-referencing with institution-level coding published by the American Joint Replacement Registry (AJRR) was also performed. Results: A total of 552 cemented THA cases were identified within the study period, of which 452 (81.9%) were correctly coded as cemented with the ICD-10-PCS. The proportion of cases that were correctly coded was 187 of 260 (72%) at Institution A, 158 of 185 (85%) at Institution B, 35 of 35 (100%) at Institution C, and 72 of 72 (100%) at Institution D. Of the 480 identified cemented THA cases at 3 of the 4 institutions, 403 (84%) were correctly reported as cemented to the AJRR (Institution A, 185 of 260 cases [71%]; Institution B, 185 of 185 [100%]; and Institution C, 33 of 35 [94%]). Lastly, of these 480 identified cemented THA cases, 317 (66%) were both correctly coded with the ICD-10-PCS and correctly reported as cemented to the AJRR. Conclusions: Our findings revealed existing discrepancies within multiple institutional data sets, which may lead to inaccurate reporting by the AJRR and other registries that rely on ICD-10-PCS coding. Caution should be exercised when utilizing ICD-10 procedural data to evaluate specific details from administrative claims databases as these inaccuracies present inherent challenges to data validity and interpretation.