143
USE OF PROCEDURE CODES IN A PHYSICIAN CLAIMS DATABASE: A VALIDATION STUDY
J Wyse1, L Joseph2, A Barkun1, M Sewitch1
1McGill University, Department of Medicine; 2McGill University, Department of Epidemiology and Biostatistics, Montreal, Quebec
Aims: Procedure codes submitted by physicians as claims for services are often used in population level database research. Accuracy of this information, however, has not been rigorously studied nor validated in Canada. Information on polypectomies from these databases may be useful in examining the impact of screening colonoscopies and monitoring colorectal screening programs.
The objective of this study was to determine the level and predictors of agreement between a physician claims database for polypectomy and polypectomy documentation in the endoscopy report.
Methods: A retrospective cohort study was undertaken of patients aged 50 to 80 years who underwent a colonoscopy in 7 Montreal hospitals. Indication for colonoscopy was obtained from the physician. Physician claims records for the procedure code 0749 (polypectomy) were obtained from the Régie de l'Assurance Maladie du Québec (RAMQ), the agency responsible for administering the provincial health insurance plan in Quebec. Medical charts were reviewed for each subject, and information regarding the performance of a polypectomy was abstracted directly from the endoscopy report. Accuracy of the physician claims (RAMQ) database was then assessed using the endoscopy report as the gold standard. Differences in the accuracy measures between different specialties (gastroenterologists and non-gastroenterologists) and between different hospital sites were also assessed.
Results: 665 patients underwent a colonoscopy by a total of 37 different endoscopists. Polypectomy was recorded in the endoscopy report in 185 (27.8%) patients. Screening was the indication for 351 (49.6%) of the colonoscopies. Procedure code in the physician claims (RAMQ) database had a sensitivity of 85.9% (95% CI [79.9-90.4]), a specificity of 99.2% (95% CI [97.7-99.7]), a positive predictive value of 97.5% (95% CI [93.4-99.2]), a negative predictive value of 94.8% (95% CI [92.4-96.5]), and a concordance of 95.5% (95% CI [93.5-96.9]). Kappa was 0.883 (95% CI [0.843-0.924]). Gastroenterologists were 5.1% (95% CI [2.5-7.6]) more likely than non-gastroenterologists to submit procedure codes that were discordant from the endoscopy report. Between-institution accuracy levels varied by as much as 17.1% (95% CI [1.7-32.1]). However, given the high rates of overall concordance, these disparities may not represent meaningful differences.
Conclusions: This study supports the use of physician claims databases as a good source of Canadian data for identifying patients undergoing polypectomies. Nonetheless, given the sensitivity of 85.9%, a 14.1% underestimate in the RAMQ database for the number of polypectomies performed may still significantly impact population level projections and underscores the importance of validation studies.