Validation of a health administrative data algorithm for assessing the epidemiology of diabetes in Canadian children

Pediatr Diabetes. 2010 Mar;11(2):122-8. doi: 10.1111/j.1399-5448.2009.00539.x. Epub 2009 Jun 5.


Objective: To validate a case definition of pediatric diabetes using administrative health data and describe trends in incidence and prevalence over time in Ontario, Canada.

Methods: We sampled hospital records of 700 children from 1994 to 2003 with a prior history of at least one outpatient or hospital record for diabetes mellitus and 300 randomly selected children with no diabetes records. We defined patients as having diabetes based on diagnoses and drug utilization from chart abstraction and compared sensitivity and specificity of a number of combinations of overall health care use using administrative data to develop a highly specific definition. We used Poisson regression to test changes in incidence over time (1994-2003).

Results: Use of four physician claims and no hospital records over a 2-yr period yielded the most specific definition (83% sensitivity, 99% specificity). Using this definition overall age/sex standardized incidence per 100,000 was 32.3 [95% confidence intervals (CI) 30.4, 34.4] and prevalence 241.5 per 100 000 (95% CI 236.2-249.9) in 2003/2004. Overall incidence differs by age, (peaking in 10-14 yr olds) but not significantly by sex. The overall incidence has increased on average by 3.1% per year since 1994 (95% CI 1.02-1.04), with no difference in the rate of increase by age.

Conclusions: Population-based surveillance of diabetes in children is possible using administrative data. This will facilitate further study of trends in incidence but also in use of health services and outcomes. Further work to differentiate type 1 and 2 diabetes will be important.

Publication types

  • Research Support, Non-U.S. Gov't
  • Validation Study

MeSH terms

  • Algorithms*
  • Child
  • Diabetes Mellitus / epidemiology*
  • Female
  • Humans
  • Incidence
  • Male
  • Ontario / epidemiology
  • Prevalence
  • Sensitivity and Specificity