An algorithm to differentiate diabetic respondents in the Canadian Community Health Survey
Edward Ng1, Kaberi Dasgupta, Jeffrey A Johnson
1Health Information and Research Division at Statistics Canada in Ottawa, Ontario K1A 0T6. Edward.Ng@statcan.ca
Insights
A new algorithm accurately classifies diabetes types in Canadian adults. The Ng-Dasgupta-Johnson algorithm distinguishes between type 1, type 2, and gestational diabetes using survey data.
Area of Science:
- Public Health
- Epidemiology
- Biostatistics
Background:
- Diabetes mellitus is a significant public health concern in Canada.
- Accurate classification of diabetes types is crucial for effective management and research.
- Previous classification methods may not fully capture the nuances of diabetes in the Canadian population.
Purpose of the Study:
- To introduce and validate the Ng-Dasgupta-Johnson algorithm for classifying diabetes types.
- To apply this algorithm to a large Canadian population dataset.
Main Methods:
- Utilized data from the Canadian Community Health Survey (CCHS) cycle 1.1 (2000/2001).
- Included 6,361 respondents diagnosed with diabetes.
- Employed the Ng-Dasgupta-Johnson algorithm, analyzing responses on diabetes during pregnancy, medication use, insulin timing, and age at diagnosis.
Main Results:
- The Ng-Dasgupta-Johnson algorithm resulted in a 5% type 1 and 95% type 2 diabetes classification.
- This contrasts with an earlier algorithm's 10%-90% split for type 1 and type 2 diabetes.
- The observed distribution aligns with the rising prevalence of obesity, a key type 2 diabetes risk factor.
Conclusions:
- The Ng-Dasgupta-Johnson algorithm provides a refined method for diabetes type classification in population-based surveys.
- This improved classification is vital for understanding diabetes epidemiology in Canada.
- The findings underscore the need for continued monitoring and intervention strategies for type 2 diabetes.
Objectives:
This article describes an algorithm to classify respondents to cycle 1.1 (2000/2001) of the Canadian Community Health Survey (CCHS) according to whether they have type 1, type 2 or gestational diabetes.
Data Source:
The data are from the chronic disease module and the drug module of cycle 1.1 of the CCHS.
Analytical Techniques:
A total of 6,361 respondents to cycle 1.1 of the CCHS reported that a health care professional had diagnosed them as having diabetes. The Ng-Dasgupta-Johnson algorithm classifies this group according to whether they have type 1, type 2 or gestational diabetes, based on their answers to CCHS questions about diabetes during pregnancy, use of oral medications to control diabetes, use of insulin, timing of initiation of insulin treatment, and age at diagnosis.
Main Results:
Application of an earlier algorithm to CCHS cycle 1.1 results in a 10%-90% split for type 1 and type 2 diabetes. By contrast, the Ng-Dasgupta-Johnson algorithm yields a 5%-95% split. This is not unreasonable, given the rapid rise in obesity, a major risk factor for type 2 diabetes, in Canada.
