Diagnosis of Gestational Diabetes Mellitus with Point-of-Care Methods for Glucose versus Hospital Laboratory Method Using Isotope Dilution Gas Chromatography-Mass Spectrometry as Reference

Table 3

Summary of modified Bland-Altman comparison.

	Valid samples ()	Bias (%)	MARD (%)^a	CV (%)^b	95% limits of agreement
					Lower^c	Upper^d
HC201+
Fasting	135	4.3	6.3	6.5	–7.6	16.5
1 h	50	4.9	5.8	5.2	–3.7	14.6
2 h	132	3.5	5.1	5.6	–6.7	13.4
Overall	317	4.2	5.8	6.0	−7.6	15.9
HC201RT
Fasting	135	–1.8	5.1	6.1	–12.2	9.4
1 h	50	0.3	3.4	4.3	–6.5	8.3
2 h	132	–2.7	4.8	5.2	–12.1	6.7
Overall	317	−1.8	4.7	5.6	−12.8	9.2
CL
Fasting	134	5.9	6.0	3.8	0.0	13.0
1 h	51	8.1	8.2	4.3	0.7	14.8
2 h	132	5.7	6.0	4.3	–0.9	15.1
Overall	317	6.1	6.3	4.2	−2.1	14.3

Modified Bland-Altman analysis comparing the difference (glucose value obtained by each method–reference value) with the reference value. The bias is the mean relative difference as a percentage of the reference value. A bias of 5% means that the value is on average 5% higher than the reference value. ^aMean absolute relative difference as a percentage of the reference value. ^bStandard deviation of the bias. ^cLower 95% limits of agreement define bias −1.96 CV. ^dUpper 95% limits of agreement define bias +1.96 CV. CL: central laboratory; CV: coefficient of variation; HC201+: HemoCue 201+; HC201RT: HemoCue 201 RT; MARD: mean absolute relative difference.