Comparison of diagnostic accuracy for diabetes diagnosis: A systematic review and network meta-analysis
Issued Date
2023-01-24
Resource Type
eISSN
2296858X
Scopus ID
2-s2.0-85147453802
Journal Title
Frontiers in Medicine
Volume
10
Rights Holder(s)
SCOPUS
Bibliographic Citation
Frontiers in Medicine Vol.10 (2023)
Suggested Citation
Duong K.N.C., Tan C.J., Rattanasiri S., Thakkinstian A., Anothaisintawee T., Chaiyakunapruk N. Comparison of diagnostic accuracy for diabetes diagnosis: A systematic review and network meta-analysis. Frontiers in Medicine Vol.10 (2023). doi:10.3389/fmed.2023.1016381 Retrieved from: https://repository.li.mahidol.ac.th/handle/20.500.14594/82455
Title
Comparison of diagnostic accuracy for diabetes diagnosis: A systematic review and network meta-analysis
Author's Affiliation
Other Contributor(s)
Abstract
Aim: Fasting Plasma Glucose (FPG) and Hemoglobin A1c (HbA1c) are used as diagnostic tests for diagnosing diabetes mellitus, but it is unclear which test has the best diagnostic accuracy. This systematic review and network meta-analysis aimed to estimate the diagnostic accuracy of HbA1c ≥ 6.5%, FPG ≥ 126 mg/dl, and the combination of HbA1c ≥ 6.5% or FPG ≥ 126 mg/dl (HbA1c| FPG), compared with Oral Glucose Tolerance Test (OGTT) ≥ 200 mg/dl for diagnosis diabetes. Materials and methods: We performed a comprehensive search in PubMed, Embase, Cochrane Library, and Scopus from inception to September 24th, 2021. Inclusion criteria were any study design comparing HbA1c ≥ 6.5%, FPG ≥ 126 mg/dl, and HbA1c ≥ 6.5% or FPG ≥ 126 mg/dl with OGTT ≥ 200 mg/dl as the reference test. Data were independently extracted, risk of bias was assessed using QUADAS-2 by two reviewers. Network meta-analysis was done using a bivariate regression model using the Bayesian framework. The relative ranking of all tests was also assessed. Results: Out of 5,026 studies, 73 were included. The sensitivities of HbA1c, FPG, and HbA1c| FPG were 0.51 [95% Credible Interval (CrI): 0.43, 0.58], 0.49 (95% CrI: 0.43, 0.55), and 0.64 (95% CrI: 0.51, 0.75), while the specificities were 0.96 (95% CrI: 0.94, 0.97), 0.98 (95% CrI: 0.97, 0.98), and 0.95 (95% CrI: 0.88, 0.98), respectively. The corresponding positive likelihood ratios (LR) were 13.36 (95% CrI: 8.91, 20.72), 21.94 (95% CrI: 15.04, 31.88), and 11.78 (95% CrI: 5.48, 26.56). HbA1c| FPG is superior based on sensitivity, whereas FPG is ranked best based on specificity and positive LR. Conclusion: Our findings suggest that FPG ≥ 126 mg/dl should be recommended as the best diagnostic test for diabetes. Systematic review registration: https://www.crd.york.ac.uk/prospero/, identifier CRD42021282856.