Objective: Our aim in this study was to evaluate the accuracy of alternative algorithms for identifying pre-existing type 1 or 2 diabetes (T1DM or T2DM) and gestational diabetes mellitus (GDM) in pregnant women.
Methods: Data from a clinical registry of pregnant women presenting to an Edmonton diabetes clinic between 2002 and 2009 were linked and administrative health records. Three algorithms for identifying women with T1DM, T2DM, and GDM based on International Classification of Diseases---tenth revision (ICD-10) codes were assessed: delivery hospitalization records (Algorithm #1), outpatient clinics during pregnancy (Algorithm #2), and delivery hospitalization plus outpatient clinics during pregnancy (Algorithm #3). In a subset of women with clinic visits between 2005 and 2009, we examined the performance of an additional Algorithm #4 based on Algorithm #3 plus outpatient clinics in the 2 years before pregnancy. Using the diabetes clinical registry as the "gold standard," we calculated true positive rates and agreement levels for the algorithms.
Results: The clinical registry included data on 928 pregnancies, of which 90 were T1DM, 89 were T2DM, and 749 were GDM. Algorithm #3 had the highest true positive rate for the detection of T1DM, T2DM, and GDM of 94%, 72%, and 99.9%, respectively, resulting in an overall agreement of 97% in diagnosis between the administrative databases and the clinical registry. Algorithm #4 did not provide much improvement over Algorithm #3 in overall agreement.
Conclusions: An algorithm based on ICD-10 codes in the delivery hospitalization and outpatient clinic records during pregnancy can be used to accurately identify women with T1DM, T2DM, and GDM.
Keywords: administrative database; base de données administratives; diabète de type 1; diabète de type 2; diabète gestationnel; gestational diabetes; type 1 diabetes; type 2 diabetes; validation studies; étude de validation.
Copyright © 2023 Canadian Diabetes Association. Published by Elsevier Inc. All rights reserved.