Introduction and Objective: Data are limited as to whether continuous glucose monitoring (CGM) can be used to accurately diagnose gestational diabetes mellitus (GDM). We aimed to develop GDM diagnostic criteria based on CGM metrics at 24-28 weeks’ gestation.Methods: Planned analysis of a longitudinal, multi-site prospective cohort of gravidas without pregestational diabetes who wore a blinded CGM for 10 days at 24-28 week’s gestation and had an OGTT on the day of CGM placement. The primary outcome was GDM diagnosed using Carpenter-Coustan criteria. A2GDM (treated with medication) was a secondary outcome. In a training dataset, a range of 24-hr CGM metrics were considered as candidate diagnostic variables, including time above glycemic thresholds, measures of glycemic variability (e.g. mean amplitude of glycemic excursions [MAGE]) and central tendency. Continuous metrics maximizing area under the ROC curve (AUC) were selected for further analysis. For these top metrics, optimal binary diagnostic cutoffs were then identified via grid search to maximize specificity while maintaining sensitivity ≥75%. These final criteria to most accurately diagnose GDM were validated by estimating sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) in a separate dataset.Results: Of 2178 participants, 81.3% had valid CGM data; 93.3% completed OGTTs; 15.4% were diagnosed with GDM (7.1% A2GDM). The best continuous metrics (AUC, 95% CI) for GDM diagnosis included MAGE (0.73, 0.68-0.78), time above 150 mg/dL (0.71, 0.66-0.76), and time above 140 mg/dL (0.7, 0.65-0.75). The optimal MAGE threshold was ≥43 mg/dL (specificity 59.7%, sensitivity 75.4%, PPV 14.5%, NPV 96.4%). Time ≥140 and 150 mg/dL also performed poorly, with thresholds ≥2% and ≥1%, respectively. Findings held true in the validation dataset and were similar for A2GDM.Conclusion: Detection of GDM at 24-28 weeks with CGM data was limited by modest performance, including low specificity. CGM metrics do not predict GDM well enough to replace OGTT for diagnosis.
A. Merriam: None. M. Banker: None. P. Catalano: None. F.L. Facco: None. M. Feghali: None. W. Grobman: None. E. LeBlanc: None. W. Lowe: None. M. Mourad: None. C. Oshiro: None. C. Powe: Research Support; Current; Dexcom, Inc. Other – Associate Editor of Diabetes Care, Honoraria for Educational Materials; Current; American Diabetes Association. Other – Royalties for Up To Date chapters; Current; Wolters Kluwer (Up To Date). Other – Speaker; Ended; Medscape. U. Reddy: None. D. Rouse: None. D. Scholtens: None. J. Sherr: Other – research support, consultant, advisory board member; Current; Abbott Diabetes. Other – advisory board member, consultant; Current; Vertex Pharmaceuticals Incorporated. Consultant; Current; Ypsomed AG. Research Support; Current; Dexcom, Inc., JDRF, Provention Bio, Inc., National Institutes of Health. Other – research support, consultant, advisory board member; Current; Insulet Corporation, Medtronic. Research Support; Current; Sanofi. Advisory Panel; Current; sequel med tech. A.C. Spadola: None. K. Vesco: None. E. Werner: None. N. Zork: None. L.M. Yee: None.
The GO MOMs study is supported by grants from the National Institute of Diabetes and Digestive and Kidney Diseases. U01DK123795 to Massachusetts General Hospital; U01DK123791 to Kaiser Permanente; U01DK123759 and U01DK123745 to Northwestern University; U01DK123799 to Yale University; U01DK123783 to Women & Infants Hospital of Rhode Island. Dexcom provided the CGM systems used in the study free of charge.
Source link
Leave a Reply