Abstract
INTRODUCTION: Cardiovascular disease (CVD) surveillance in Quebec and the rest of Canada is carried out using health administrative databases, which in Quebec includes the physician claims database. The presence of billing claims without diagnoses can lead to the number of CVD cases being underestimated. The purpose of this study is to estimate the proportion of CVD diagnoses and CVD cases that may be missing from these databases. METHODS: The study was conducted using a prospective cohort of 8781 participants living in the Québec City area. Access to health administrative databases was granted for the entire 28-year follow-up period. First, we performed frequency analyses to estimate the proportion of missing CVD diagnoses. Then we used validated algorithms to identify CVD cases and estimate the proportion of CVD cases that were potentially not captured over the 28-year period. RESULTS: About one-fifth (22.1%) of the diagnoses in the physician claims database were missing. The proportion of missing CVD cases was estimated at 12.7% for 1991-2018, although this varied with the period covered (1991-1996: 15.5%; 1997-2013: 10.7%; and 2014-2018: 16.3%). CONCLUSION: Approximatively 1 in 10 CVD cases are not identified due to a missing diagnosis. This underestimation of CVD cases is a potential limitation that should be considered when using Quebec health administrative databases to identify CVD cases for surveillance work and epidemiological studies.