Category(s):
Health
Status: Active
Principal: Chris Folkins
Project Number: P0112
Year Approved: 2023
Canadian Chronic Disease Surveillance System (CCDSS) data provides an estimate of chronic disease prevalence in New Brunswick based on diagnosis information from hospital discharge abstracts or inferred from physician notes in NB Physician Billing data. CCDSS is frequently used for research purposes as a means of disease case identification for administrative data studies. Other possible approaches for case identification in administrative data include the use of diagnostic lab values, or the incidence of Medicare billing for physician services that are unique to a particular disease.
Previous NB-IRDT studies have found a high rate of discordance between CCDSS diagnosis of COPD and lab results (spirometry) that reflect a COPD diagnosis, i.e. many patients with spirometry results reflecting COPD do not appear in CCDSS COPD, and many COPD cases identified in CCDSS COPD have spirometry results that do not reflect COPD (or lack spirometry results altogether). Furthermore, preliminary data shows a similar discordance between CCDSS COPD and incidence of billing for the Chronic Disease Management (CDM) – COPD service code in NB Physician Billing data, i.e. not all patients for whom a CDM-COPD code is billed are identified in CCDSS COPD as a COPD case. These findings suggest a lack of concordance across various administrative data approaches that may be used to estimate chronic disease prevalence and identify cases for use in data linkage and administrative data research. This lack of concordance has implications for the accuracy and validity of research conducted using NB administrative data.
This project aims to examine whether similar discordance exists across three different approaches that may be used to identify diabetes cases in NB administrative data, namely: CCDSS diabetes, Medicare billing for the CDM-diabetes service code, and results of hemoglobin A1c (HbA1c) blood tests that reflect a diagnosis of diabetes. The purpose of this work is to provide an understanding of the strengths and limitations associated with each of these methods of diabetes case identification, and to more generally illustrate and characterize the potential limitations that can exist within NB administrative data. Our findings will improve understanding of administrative data products hosted by NB-IRDT, and support better-informed decision making and data interpretation by researchers undertaking studies of chronic disease in NB using administrative data.
