A Comparison of Methods for Estimating Chronic Disease Incidence from Linked Administrative Health Data
View Presentation View Presentation
Mahmoud Azimaee, University of Manitoba  Christopher Bowman, National Research Council  Abba Gumel, University of Manitoba  Janet Hux, Institute for Clinical Evaluative Sciences  William D Leslie, University of Manitoba  *Lisa M Lix, University of Manitoba  Marina Yogendran, Manitoba Centre for Health Policy 

Keywords: record linkage, non-linear regression, case ascertainment, osteoporosis, diabetes

Estimates of chronic disease incidence are an important indicator of the efficacy of prevention programs. Three methods are investigated to estimate chronic disease incidence using multiple years of population-based administrative data. Disease cases were identified in hospital, physician, and pharmacy data from Manitoba, Canada. The methods were investigated for osteoporosis. Two non-linear regression models that correct for over-ascertainment of incident cases due to the finite duration of the observation period were compared to a health state model that estimates incidence from aggregate prevalence and mortality data. Osteoporosis results were validated using clinical data. The health state model produced lower estimates of disease incidence than the regression models for all ages. The models did not consistently show the same trend across multiple years.

Print Friendly
For more information send email: ichps@amstat.orgHPSS 2005