Skip to main content

Table 3 The clinical performance of the REGARDS model for incident type 2 diabetes: Tehran Lipid and glucose study

From: External validation of the American prediction model for incident type 2 diabetes in the Iranian population

Probability threshold (%)

≥ 10

≥ 13

≥ 20

≥ 30

Population aged ≥ 45 years (N = 1835, DM = 281)

High-risk population, %

54.55

39.95

18.00

6.76

Incident diabetes*,%

23.8

29.6

44.24

62.10

Sensitivity, % (95% CI)

84.7 (79.9–88.7)

77.2 (71.9–82.0)

52.0 (45.9–57.9)

27.4 (22.3–33.0)

Specificity, % (95% CI)

50.9 (48.4–53.4)

66.8 (64.4–69.1)

88.2 (86.4–89.7)

97.0 (96.0-97.8)

PPV, % (95% CI)

23.8 (21.2–26.5)

29.6 (26.3–33.1)

44.2 (38.8–49.8)

62.1 (52.9–70.7)

NPV, % (95% CI)

94.8 (93.1–96.2)

94.2 (92.6–95.5)

91.0 (89.5–92.4)

88.1 (86.4–89.6)

LR+ (95% CI)

1.73 (1.61–1.85)

2.33 (2.12–2.56)

4.39 (3.68–5.23)

9.1 (6.5–12.7)

LR- (95% CI)

0.30 (0.23–0.40)

0.34 (0.27–0.42)

0.54 (0.48–0.62)

0.75 (0.70–0.80)

Youden Index, % (95% CI)

35.6 (28.3–42.1)

44.0 (36.3–51.1)

40.2 (32.3–47.6)

24.4 (18.3–30.8)

AUC (95% CI)

0.68 (0.65–0.70)

0.72 (0.69–0.75)

0.70 (0.67–0.73)

0.62 (0.60–0.65)

  1. All analysis was done based on survey data analysis (weighted statistics); PPV: positive predictive value; NPV: negative predictive value; LR: Likelihood ratio.
  2. *Incident diabetes was reported among the high-risk population.