To validate a case definition for diabetes in the pediatric age-group using administrative health data.
Population-based administrative data from Manitoba, Canada for the years 2004-2006 were anonymously linked to a clinical registry to evaluate the validity of algorithms based on a combination of hospital claim, outpatient physician visit, and drug use data over 1-3 years in youth 1-18 years of age. Agreement between data sources, sensitivity, specificity, negative (NPV) and positive predictive value (PPV) were evaluated for each algorithm. In addition, ascertainment rate of each data source, prevalence, and differences between subtypes of diabetes were evaluated.
Agreement between data sources was very good. The diabetes definition including one or more hospitalizations or two or more outpatient claims over 2 years provided a sensitivity of 94.2%, specificity of 99.9%, PPV of 81.6% and NPV of 99.9%. The addition of one or more prescription claims to the same definition over 1 year provided similar results. Case ascertainment rates of both sources were very good to excellent and the ascertainment-corrected prevalence for youth-onset diabetes for the year 2006 was 2.4 per 1,000. It was not possible to distinguish between subtypes of diabetes within the administrative database; however, this limitation could be overcome with an anonymous linkage to the clinical registry.
Administrative data are a valid source for the determination of pediatric diabetes prevalence that can provide important information for health care planning and evaluation.