Table 2. Pooled estimates (meta-analysis) at various cut-offs for diagnostic accuracy of HbA1c (%) and FPG (mg/dL) for diabetes.

	Threshold value used for diabetes	Number of studies	Number of cases (true positives & false negatives) & participants	Sensitivity (95% CI)	Specificity (95% CI)	Positive Likelihood ratio (95% CI)	Negative Likelihood ratio (95% CI)
HbA1c (%)	5.7	7	2506/29076	0.888 (0.830–0.927)	0.657 (0.531–0.765)	2.588 (1.878–3.566)	0.171 (0.119–0.246)
	5.8	8	3127/36863	0.818 (0.749–0.871)	0.781 (0.680–0.857)	3.738 (2.587–5.401)	0.233 (0.175–0.310)
	5.9	7	2958/34866	0.770 (0.6874–0.837)	0.834 (0.742–0.898)	4.644 (3.080–7.0022)	0.276 (0.209–0.363)
	6.0	10	3381/39115	0.757 (0.681–0.819)	0.893 (0.843–0.929)	7.084 (4.896–10.254-)	0.272 (0.208–0.356)
	6.1	7	2543/27679	0.726 (0.596–0.826)	0.932 (0.873–0.964)	10.605 (6.166–18.240)	0.294 (0.199–0.436)
	6.2	4	2118/23217	0.655 (0.538–0.7554)	0.935 (0.872–0.968)	10.042 (5.672–17.781)	0.370 (0.279–0.490)
	6.3	6	1710/17151	0.654 (0.574–0.727)	0.945 (0.902–0.970)	11.960 (6.940–20.610)	0.366 (0.297–0.450)
	6.4	5	2059/21670	0.624 (0.527–0.712)	0.950 (0.904–0.975)	12.589 (7.079–22.387)	0.396 (0.317–0.494)
	6.5	17	5132/64928	0.502 (0.417–0.588)	0.973 (0.953–0.984)	18.328 (11.067–30.353)	0.512(0.432–0.605)
FPG (mg/dL)	126	10	3438/45917	0.594 (0.466–0.710)	0.988 (0.965–0.996)	47.825 (19.104–119.729)	0.411 (0.305–0.555)

* Estimates are rounded off to nearest number or three decimal places.