Skip to main content
. 2010 Mar 24;91(5):1185–1194. doi: 10.3945/ajcn.2009.29036

TABLE 5.

Association of serum markers with BMI1

Serum marker2 Estimated value at a BMI (in kg/m2) of 21.8 (≈25th percentile in data set) and 95% CI Estimated percentage change per one-unit increase in BMI 95% CI for percentage change per one-unit increase in BMI P value
% %
Adiponectin (ng/mL) 13.71 (11.38, 16.51) −2.7 (−4.9, −0.5) 0.0189
Leptin (ng/mL) 7.20 (6.21, 8.35) 8.3 (6.2, 10.4) <0.0001
C-peptide (ng/mL) 2.11 (1.84, 2.41) 4.8 (1.03, 1.066) <0.0001
Insulin (μIU/mL) 4.29 (3.48, 5.30) 5.3 (1.025, 1.082) 0.0002
Glucose (μg/dL) 73.58 (69.71, 77.68) 1.1 (0.4, 1.8) 0.0035
IGF-I (ng/mL) 201.78 (180.66, 225.38) −0.6 (−2.1, 0.8) 0.4036
IGFBP3 (ng/mL) 4729.15 (4510.67, 4857.71) 0.30 (0.3, 1.0) 0.2392
C-reactive protein (mg/dL) 0.63 (0.47, 0.83) 9.4 (5.4, 13.7) <0.0001
IL-6 (pg/mL) 2.75 (2.09, 3.60) 4.6 (0.8, 8.4) 0.0161
TNF-α (pg/mL) 1.71 (1.52, 1.93) 0.8 (0.7, 2.3) 0.3082
Estradiol (pg/mL) 10.31 (9.28, 11.44) 2.6 (1.2, 4.1) 0.0004
SHBG (nmol/L) 13.71 (11.38, 16.51) −4.2 (−5.8, −2.5) <0.0001
1

The results are secondary to the main analysis and represent the findings of exploratory model building. Outcomes were first natural-log transformed due to skewness, and then linear mixed-effects models were applied to the data, with BMI, time, dose, and menopausal status as potential predictors. From the models, we estimated the percentage change in outcome per one-unit change in BMI and 95% CIs. To give a sense of baseline outcome values in relation to BMI, we also used the model to estimate the outcome value at the 25th percentile of BMI (as calculated from our data set) and the 95% CI. We chose the 25th percentile rather than the minimum to avoid providing an estimate at what could potentially be an outlier value of BMI. IGF-I, insulin-like growth factor I; IGFB3, insulin-like growth factor binding protein 3; SHBG, sex hormone–binding globulin; IL-6, interleukin-6; TNF-α, tumor necrosis factor-α.

2

n = 48 subjects for all serum markers, except for estradiol, which was assessed only in postmenopausal subjects (n = 23).