Table 3

HRs by age-group for the association of prospective gains in BMI with incident diabetes, after adjusting for covariates*

Age-group at baselineP value for interaction
≥30 and <45 years old≥45 and <60 years old
HRP valueHRP value
Primary model
 Blacks
  n (events)1,359 (89)2,297 (391)
  Baseline BMI (5 units)1.29<0.0011.25<0.0010.69
  Log(BMI-years over baseline BMI) (1 unit)1.180.021.020.390.047
 Whites
  n (events)3,266 (110)10,482 (919)
  Baseline BMI (5 units)1.37<0.0011.38<0.0010.92
  Log(BMI-years over baseline BMI) (1 unit)1.35<0.0011.11<0.0010.008
Sensitivity analysis (excluding baseline fasting glucose from the primary model)
 Blacks
  n (events)1,359 (89)2,297 (391)
  Baseline BMI (5 units)1.41<0.0011.25<0.0010.69
  Log(BMI-years over baseline BMI) (1 unit)1.130.080.990.560.065
 Whites
  n (events)3,266 (110)10,482 (919)
  Baseline BMI (5 units)1.60<0.0011.60<0.0010.96
  Log(BMI-years over baseline BMI) (1 unit)1.27<0.0011.060.0010.006
  • *Additional covariates in the primary multivariate analyses include sex, baseline HDL-C (15 mg/dL), log(baseline triglycerides), baseline fasting glucose, baseline prehypertension, baseline hypertension, log(BMI-years below baseline BMI), as well as interaction terms for age-group × baseline covariates, race × baseline covariates, and age-group × race × baseline covariates. The referent group for hypertension and prehypertension is those with normal blood pressure, defined as systolic blood pressure <120 mmHg and diastolic blood pressure <80 mmHg and not using antihypertensive medication. In the sensitivity analyses, baseline fasting glucose was removed as a covariate from the primary multivariable model.

  • †BMI-years is a time-dependent variable calculated using linear interpolation between measured BMI points and the amount of time above or below baseline BMI.

  • P value for interaction is the test of effect modification between age-group at baseline and the risk factor on incident diabetes.