qgcompint: quantile g-computation with effect measure modification

installation

official release

install.packages("qgcomp")

developmental release (not always guaranteed to be stable)

devtools::install_github("alexpkeil1/qgcompint", build_vignettes=TRUE)

for help

vignette("qgcompint-vignette", package="qgcompint")

Quick start

library(qgcomp)
library(qgcompint)
  set.seed(40)
dat <- data.frame(y=runif (50),
                  x1=runif (50),
                  x2=runif (50),
                  z=rbinom(50, 1, 0.5),
                  r=rbinom(50, 1, 0.5))
 
 # quantile g-computation without effect measure modification
 qfit <- qgcomp.glm.noboot(f=y ~ z + x1 + x2,
           expnms = c('x1', 'x2'),
           data=dat, q=2,
           family=gaussian())
 # no output given here          
 
 # with effect measure modification by Z
 (qfitemm <- qgcomp.emm.glm.noboot(f=y ~ z + x1 + x2,
           emmvar="z",
           expnms = c('x1', 'x2'),
           data=dat, q=2,
           family=gaussian()))



> ## Qgcomp weights/partial effects at z = 0
> Scaled effect size (positive direction, sum of positive coefficients = 0)
> None

> Scaled effect size (negative direction, sum of negative coefficients = -0.278)
>    x2    x1 
> 0.662 0.338 

> ## Qgcomp weights/partial effects at z = 1
> Scaled effect size (positive direction, sum of positive effects = 0.0028)
> x1 
>  1 

> Scaled effect size (negative direction, sum of negative effects = -0.0128)
> x2 
>  1 

> Mixture slope parameters (delta method CI):

> 			  Estimate Std. Error Lower CI Upper CI t value  Pr(>|t|)
> (Intercept)  0.58062    0.11142  0.36224  0.79900  5.2112 4.787e-06
> psi1        -0.27807    0.20757 -0.68490  0.12876 -1.3397    0.1872
> z           -0.10410    0.15683 -0.41148  0.20329 -0.6637    0.5103
> z:mixture    0.26811    0.26854 -0.25822  0.79444  0.9984    0.3235

> Estimate (CI), z=1: 
> -0.0099575 (-0.34389, 0.32398)

Current package capabilities/limitations

• Single modifiers only (e.g. interaction terms between a modifier and the mixture can only be estimated for a single modifier at a time). This also implies that no interaction terms between the modifier and other covariates can be considered.
• Cox model: linear only specifications (i.e. linear effects of a mixture, but not covariates)
• Features missing relative to qgcomp package: zero inflated and hurdle models, multinomial models, tobit regression

Interpretation

• coefficients
  • psi1 coefficient: effect of the mixture in the referent category of the effect measure modifier (z in this case)
  • z coefficient: main effect of the effect measure modifier (will change based on name of effect measure modifier)
  • z:mixture coefficient: interaction term between the effect measure modifier and the entire mixture
• weights/partial effects
  • first set of weights/sum of negative/positive coefficients: - weights: proportion of positive/negative partial effect in the reference stratum effect measure modifier (here, z=0)
    • partial effect: sum of main term coefficients for exposure in a given direction (these will sum to psi1, the effect estimate in the reference stratum effect measure modifier (here, z=0))
  • If the modifier is binary, a second set is printed:
    • weights: proportion of positive/negative partial effect in the index stratum effect measure modifier (here, z=1)
    • partial effect: sum of main term coefficients + interaction term coefficients for exposure in a given direction (these will sum to the mixture effect estimate in the index stratum effect measure modifier (here, z=1))
• Additional estimates (only given if effect measure modifier is binary): The effect of the mixture in the index category of the effect measure modifier (here, z=1)