Comparison of Icd-9 to Icd-10 Crosswalks Derived By Physician and Clinical Coder Vs. Automated Methods

OBJECTIVES: Coding algorithms are critical for identifying patient samples, comorbidities, and outcomes in studies of claims or electronic medical record data, but algorithms developed with International Classification of Diseases, Ninth Revision (ICD-9) codes are obsolete in current data. The present study sought to compare ICD-9 to ICD-10 crosswalks from General Equivalence Mappings (GEMs) and compare them to crosswalks derived by a clinician and clinical coder, to evaluate whether automated methods are sufficient for deriving ICD-10 algorithms. METHODS: Ten conditions from various therapeutic areas were selected for comparison. Existing ICD-9 algorithms were entered into GEMs to derive ICD-10 crosswalks, and a physician and clinical coder completed a questionnaire to guide the development of ICD-9 to ICD-10 crosswalks for the same conditions. Differences between the crosswalks were summarized using descriptive statistics and the theoretical impact of the differences were assessed qualitatively. RESULTS: Crosswalks identified by the physician/coder were typically far more inclusive than those from GEMs. Crosswalks from GEMs were missing a mean of 1477 codes (median: 45; range: 19-462) compared to those from the physician/coder, while the physician/coder crosswalks missed far fewer (mean: 11.3; median: 5.5; range: 0-53) compared to GEMs. Crosswalks for diabetes, diabetic neuropathy, hypoglycemia, and peripheral vascular disease had the most discrepancies (>130) while crosswalks for acute myocardial infarction and hypertension had the fewest (