Using HbA1c to Improve Efficacy of the American Diabetes Association Fasting Plasma Glucose Criterion in Screening for New Type 2 Diabetes in American Indians: The Strong Heart Study

Epidemiology/Health Services/Psychosocial Research

Using HbA_1c to Improve Efficacy of the American Diabetes Association Fasting Plasma Glucose Criterion in Screening for New Type 2 Diabetes in American Indians

¹College of Public Health, University of Oklahoma, Health Sciences Center, Oklahoma City, Oklahoma
²National Heart, Lung and Blood Institute, Bethesda, Maryland
³Aberdeen Area Tribal Chairmen’s Health Board, Aberdeen, South Dakota
⁴Medlantic Research Institute, Washington, DC

Diabetes Care 2002 Aug; 25(8): 1365-1370. https://doi.org/10.2337/diacare.25.8.1365

Download figure
Open in new tab
Download powerpoint
Figure 1—
The ROC curves based on the logistic regression models on FPG, HbA_1c, and HbA_1c and FPG. The areas under the respective ROC curves are denoted as the area for FPG, the area for HbA_1c, and the area for HbA_1c and FPG. They are calculated as 0.640343, 0.7053313, and 0.7237567, respectively. The results from a comparison of these areas are as follows: P = 0.153 for the area for FPG vs. the area for HbA_1c; P = 0.0008 for the area for FPG vs. the area for HbA_1c and FPG; P = 0.463 for the area for HbA_1c vs. the area for HbA_1c and FPG.
Download figure
Open in new tab
Download powerpoint
Figure 2—
The optimal critical lines for R = P/(1 − P) based on the data from the SHS baseline and second exams. Those IFG participants whose FPG and HbA_1c are above or on the line are referred to take an OGTT for possible diabetes (2hPG ≥200 mg/dl).

Table 1—
FPG by 2hPG levels: the SHS
2hPG (mg/dl) FPG (mg/dl)
<110 110–126 >126 Total
Baseline exam
<200 1,489 427 117 2,033
≥200 74 102 180 356
Total 1,563 529 297 2,389
Second exam
<200 1,058 289 57 1,404
≥200 78 86 76 240
Total 1,136 375 133 1,644

Table 2—

The optimal critical line for different cost-to-benefit ratio R

R	FPG at the intercept of OCL and HbA_1c = 0 (mg/dl)	HbA_1c at the intercept of OCL and FPG = 110 (%)	HbA_1c at the intercept of OCL and FPG = 126 (%)	Z*	k*	Specificity (%)	Sensitivity (%)
SHS baseline examination (OCL: 0.89 × HbA_1c + 0.11 × FPG = Z*)
P /(1 − P)	162.93	6.54	4.56	17.924	0.222	76.81	58.82
0.05	153.26	5.35	3.37	16.861	0.090	18.03	97.06
0.10	154.22	5.47	3.49	16.966	0.099	23.89	95.10
0.15	157.74	5.90	3.92	17.353	0.139	44.96	83.33
0.20	160.50	6.24	4.26	17.657	0.180	61.83	71.57
0.25	162.93	6.54	4.56	17.924	0.222	76.81	58.82
0.30	164.31	6.71	4.74	18.076	0.250	81.97	52.94
0.35	164.31	6.71	4.74	18.076	0.250	81.97	52.94
0.40	164.31	6.71	4.74	18.076	0.250	81.97	52.94
0.45	164.31	6.71	4.74	18.076	0.250	81.97	52.94
0.50	169.16	7.31	5.34	18.610	0.362	95.32	26.47
1.00	170.78	7.51	5.54	18.788	0.404	97.42	19.61
SHS second examination (OCL: 0.106 × HbA_1c + 0.078 × FPG = Z*)
P /(1 − P)	120.31	7.61	−4.20	9.433	0.189	36.33	83.72
0.05	117.31	5.40	−6.41	9.197	0.155	9.34	98.84
0.10	117.31	5.40	−6.41	9.197	0.155	9.34	98.84
0.15	117.31	5.40	−6.41	9.197	0.155	9.34	98.84
0.20	120.31	7.61	−4.20	9.433	0.189	36.33	83.72
0.25	120.31	7.61	−4.20	9.433	0.189	36.33	83.72
0.30	120.31	7.61	−4.20	9.433	0.189	36.33	83.72
0.35	127.31	12.78	0.97	9.981	0.287	83.74	30.23
0.40	127.31	12.78	0.97	9.981	0.287	83.74	30.23
0.45	127.31	12.78	0.97	9.981	0.287	83.74	30.23
0.50	128.26	13.48	1.67	10.056	0.303	87.54	24.42
1.00	132.80	16.84	5.02	10.412	0.382	98.96	3.49

R = the costs-to-benefit ratio; P = the prevalence of diabetes in IFG participants, which is estimated as 0.193 and 0.229 based on the data from the baseline and second exam, respectively; k* = the optimal cutoff point; Z* = log[k* /(1 − k*)]− a, where a is the estimated intercept in the fitted logistic regression model on FPG and HbA_1c, and a = −19.176 and −10.891 based on the data from the baseline and second exam, respectively. All variables are defined in research design and methods (statistical analysis). OCL, optimal critical line.

August 2002, 25(8)