wqs_data: Exposure concentrations of 34 PCB (simulated dataset)
In gWQS: Generalized Weighted Quantile Sum Regression

wqs_data

R Documentation

Exposure concentrations of 34 PCB (simulated dataset)

Description

We created the 'wqs_data' dataset to show how to use this function. These data reflect 59 exposure concentrations simulated from a distribution of 34 PCB exposures and 25 phthalate biomarkers measured in subjects participating in the NHANES study (2001-2002). Additionally, 8 outcome measures were simulated applying different distributions and fixed beta coefficients to the predictors. In particular 'y' and 'yLBX' were simulated from a normal distribution, 'ybin' and 'ybinLBX' from a binomial distribution, 'ymultinom' and 'ymultinomLBX' from a multinomial distribution and 'ycount' and 'ycountLBX' from a Poisson distribution. The regression coefficients used to generate the outcomes 'yLBX', 'ybinLBX' and 'ycountLBX' were set to:
LBX105LA = 0.3
LBX138LA = 0.6
LBX157LA = 0.2
LBXD02LA = 0.45
LBXD04LA = 0.15
LBXF06LA = 0.3
LBXF07LA = 0.45
then the following terms were added to generate the variables 'y', 'ybin' and 'ycount':
URXMC1 = 0.15
URXMOH = 0.45
URXP02 = 0.2
URXP10 = 0.3
URXUCR = 0.2
All the remaining coefficients were set to 0.
The coefficients to generate 'ymultinomLBX' were set as below:
level B:
LBX138LA = 0.8
LBXD04LA = 0.2
level C:
LBX105LA = 0.4
LBX157LA = 0.3
LBXD02LA = 0.6
LBXF06LA = 0.4
LBXF07LA = 0.6
and the following terms were added for 'ymultinom':
level B:
URXMC1 = 0.2
URXP02 = 0.3
URXP10 = 0.4
URXUCR = 0.3
level C:
URXMOH = 0.6
The 'sex' variable was also simulated to allow to adjust for a covariate in the model. This dataset can thus be used to test the 'gWQS' package by analyzing the mixed effect of the 59 simulated PCBs on the continuous, binary or count outcomes, with adjustments for covariates.