When the model for waist circumference was further adjusted for baseline CV risk factors (Table 3, right side), no attenuation of the association was seen.
In our basic models, we fit a univariate linear growth model for waist circumference, investigating the annual change in waist circumference for all subjects, with no other variables in the model.
To further explore the role of adiposity in the association between a family history of diabetes and incident diabetes, we ran various models, adjusting for waist and hip circumference or adolescent BMI and adult weight change (Table 2).
It was also the most significant determinant in the final model for waist-height ratio, a marker of central adiposity, overall neonatal adiposity as determined by SS + TR and sum of skinfolds, independent of maternal BMI, neonatal gender, total gestational age, maternal education or study group assignment.
In full models additionally adjusted for waist circumference (model 3), however, only low adiponectin remained significantly associated with incident diabetes (0.68 [0.51–0.90], P = 0.007).
We also used the detailed maturation stage data to fit models of waist circumference change, adjusting for the four maturational stages.
In models further adjusting for waist circumference, birth weight and CWG birth 48 months were inversely associated with fasting glucose and/or IFG/DM; whereas CWG 24 48 months was not associated.
Inclusion of baseline high BP in the model for large waist did not materially attenuate the association (right side, Table 2), nor did inclusion of large waist at baseline attenuate the association in the model for high BP.
To investigate whether associations were independent of other risk factors, we additionally adjusted the multivariable model for 1) waist circumference (continuous), 2) hypertension (yes versus no), and 3) hyperlipidemia (yes versus no).
