To Örebro University

Background and aims: The Changing Diagnostic Criteria for Gestational Diabetes (GDM) study (CDC4G) is a stepped wedged randomized controlled trial in Sweden on the effects of introducing the 2013 WHO criteriaf or diagnosing GDM. Almost all Swedish pregnancies are registered in the Swedish Pregnancy Register (SPR). The CDC4G study provides a unique opportunity to validate the GDM diagnosis in the SPR. We aim to 1) validate the diagnosis of GDM in the SPR using the laboratory values from the oral glucose tolerance tests (OGTT) in the CDC4G study as the gold standard; 2) explore effects of change in diagnostic criteria on validity and prevalence of the diagnosis of GDM. Secondary aim is to investigate whether incident GDM diagnoses during pregnancy are recorded by the midwife when entering the follow-up postpartum registration in the SPR.

Materials and methods: Data from the SPR were compared with data from the CDC4G eCRF (gold standard measurements: venous OGTT values fasting, 1-h and 2-h) among 6080 screened individuals in 2018. We also investigated if the GDM diagnosis, set at the maternity ward was registered by the midwives at the postpartum follow-up (SPR tickbox). We present the sensitivity, specificity, positive (PPV) and negative (NPV) predictive value for each question. The study was approved by the Uppsala-Örebro regional Ethical Review board (2016/487), and by the Swedish Ethical Review Authority (2019/02148, 2020/02856, 2021/02055).

Results: Validating the ICD-code GDM (O24.4) in the Swedish Pregnancy Register resulted in 84.7% sensitivity, 96.7% specificity, PPV of 91.8%, and NPV of 93.5%. The prevalence of the GDM diagnosis more than tripled using the new criteria (Table 1). Both the sensitivity and specificity of the follow-up postpartum registration of GDM were considerably lower than for the GDM ICD-code, 76.6% and 87.6%, respectively. There were some minor differences in the accuracy of the registration before and after the switch to the new criteria for GDM, see table 1.

Conclusion: The coding of GDM in clinical practice, that is transferred to the SPR needs to be improved. We recommend researchers to use data based on ICD coding, instead of manually entered SPR data, until the quality of the variable has improved.