Introduction & Objective: Poor glycemic control (GC) increases the risk of various pregnancy complications. Both A1C and GA/Fruc are convenient as they require only one measurement and do not impose fasting. However, GA/Fruc is possibly a better predictor of complications than A1C during pregnancy when GC quickly changes. This meta-analysis compared the predictive ability of pregnancy complications between A1C and GA/Fruc.
Methods: We comprehensively searched for studies of prediction of maternal or neonatal adverse outcomes using both A1C and GA/Fruc and for their best cut-off values in each study presenting 2 x 2 data (i.e., true-positive, false-negative, true-negative, and false-positive cases).
Results: Of 9 eligible studies, 7 predicted macrosomia and could be meta-analyzed using a hierarchical summary receiver-operating characteristic (HSROC) model. Other complications were impossible to be analyzed because of an insufficient number of data. Pooled specificity (95% confidence interval [CI]) was significantly higher (P=0.02) for GA/Fruc (0.83 [0.70-0.91]) compared with A1C (0.57 [0.35-0.77]) while pooled sensitivity (95% CI) was 0.44 (0.26-0.63) for GA/Fruc and 0.67 (0.50-0.81) for A1C (P for difference, 0.17).
Conclusion: Compared with A1C, GA/Fruc is useful for specifying individuals having a macrosomic baby.
S. Kodama: None. T. Yamada: None. N. Yagyuda: None. K. Fujihara: None. L. Khin: None. M. Kitazawa: None. M. Yamamoto: None. Y. Matsubayashi: None. K. Kato: None. H. Sone: Research Support; Novo Nordisk, Astellas Pharma Inc., Kowa Company, Ltd., Kyowa Kirin Co., Ltd., Eisai Inc., Sumitomo Dainippon Pharma Co., Ltd.
