Table 3—

Risk variables for diabetes: results of multiple logistic regression analyses

VariablesβSE of βPOR (95% CI)
Total study population
    Age0.900.04<0.000012.46 (2.27–2.66)
    BMI0.120.060.04491.12 (1.00–1.27)
    Positive family history0.750.09<0.000012.12 (1.77–2.53)
    Waist circumference0.340.05<0.000011.40 (1.27–1.55)
    Higher education0.290.130.02381.34 (1.04–1.72)
    City vs. PUV0.610.10<0.000011.84 (1.51–2.24)
    Town vs. PUV0.320.100.00201.38 (1.13–1.67)
City
    Age0.970.06<0.000012.64 (2.12–2.97)
    Family history0.790.14<0.000012.20 (1.67–2.90)
    Waist circumference0.330.08<0.00011.39 (1.19–1.63)
    BMI0.140.100.1501.15 (0.95–1.40)
    High income0.310.150.03781.36 (1.11–1.83)
    School education−0.450.150.00250.64 (0.48–0.85)
Town
    Age0.980.06<0.000012.67 (2.14–3.00)
    Family history0.550.14<0.00011.73 (1.13–2.28)
    Waist circumference0.340.09<0.000011.40 (1.18–1.68)
    BMI0.050.100.6241.05 (0.86–1.28)
PUVs
    Age0.700.07<0.000012.01 (1.76–2.31)
    Family history1.280.20<0.000013.61 (2.43–5.32)
    Waist circumference0.310.100.0011.36 (1.12–1.66)
    BMI0.200.110.0721.22 (0.98–1.52)
  • The dependent variable was diabetes vs. normal glucose tolerance. Independent variables in all models were age, sex, family history, BMI, waist circumference, income, physical activity, occupation, and education.