Table 3

Standardized polytomous logistic regression estimates for ORs of each glucose status with β-cell dysfunction and insulin resistance considering categories of BMI at the baseline survey and before pregnancy

ORs (95% CIs)Ptrenda
Normal glucose 
(n = 779)Prediabetes (IGT and/or IFG) (n = 401)Diabetes 
(n = 83)
BMI at baseline survey (kg/m2)
 β-Cell dysfunctionb
  <241.003.23 (2.22–4.68)13.6 (4.06–45.3)<0.001
  24–27.90.35 (0.21–0.58)0.90 (0.56–1.44)5.30 (2.42–11.6)<0.001
  ≥280.09 (0.02–0.37)0.31 (0.15–0.65)3.02 (1.42–6.44)<0.001
  Ptrendc<0.001<0.0010.031
 Insulin resistanceb
  <241.002.75 (1.58–4.77)7.78 (2.61–23.2)<0.001
  24–27.95.44 (3.39–8.75)13.6 (8.19–22.6)26.0 (11.2–60.4)<0.001
  ≥2817.4 (9.47–31.9)31.1 (17.5–55.4)45.8 (18.5–113)0.035
  Ptrendc<0.001<0.0010.010
BMI before pregnancy (kg/m2)
 β-Cell dysfunctionb
  <241.002.42 (1.73–3.40)8.81 (3.83–20.3)<0.001
  24–27.90.43 (0.26–0.73)0.90 (0.54–1.48)5.30 (2.44–11.5)<0.001
  ≥280.41 (0.14–1.19)0.85 (0.39–1.85)4.23 (1.70–10.5)0.006
  Ptrendc0.003<0.0010.078
 Insulin resistanceb
  <241.002.70 (1.80–4.04)7.07 (3.18–15.7)<0.001
  24–27.92.86 (1.86–4.39)11.6 (7.29–18.3)19.3 (8.30–44.8)<0.001
  ≥287.50 (3.69–15.2)10.7 (5.52–20.8)21.6 (7.86–59.5)0.103
  Ptrendc<0.001<0.0010.067
  • The number of subjects in each group is presented in Table 1.

  • Adjusted for age, time postpartum, sitting time, dietary fiber, and monounsaturated fat, polyunsaturated fat, and saturated fat intake (continuous variables for above all variables) and education, family history of diabetes, smoking, passive smoking, drinking alcohol, and leisure-time physical activity (categorical variables for above all variables).

  • aTesting trend for differences from normal glucose to prediabetes to diabetes in each BMI category.

  • bβ-Cell dysfunction is defined as the lower quartile of HOMA-%β, and insulin resistance is defined as the upper quartile of HOMA-IR.

  • cTesting trend for differences from BMI <24 to 24–27.9 to ≥28 kg/m2 in each glucose category.