pif.confidence.approximate.loglinear: Confidence Intervals for the Potential Impact Fraction when...

In pifpaf: Potential Impact Fraction and Population Attributable Fraction for Cross-Sectional Data

Description

Confidence intervals for the Population Attributable Fraction for the approximate method where only mean and variance from a previous study is available.For relative risk inyective functions, the pif is inyective, and intervals can be calculated for log(pif), and then transformed to pif CI.

Usage

pif.confidence.approximate.loglinear(Xmean, Xvar, thetahat, thetavar, rr,
  cft = NA, deriv.method.args = list(), deriv.method = c("Richardson",
  "complex"), check_exposure = TRUE, check_rr = TRUE,
  check_integrals = TRUE, nsim = 1000, confidence = 95,
  check_thetas = TRUE, is_paf = FALSE)

Arguments

Xmean	Mean value of exposure levels from a cross-sectional.
Xvar	Variance of the exposure levels.
thetahat	Estimator (vector or matrix) of theta for the Relative Risk function rr
thetavar	Estimator of variance of thetahat
rr	Function for Relative Risk which uses parameter theta. The order of the parameters shound be rr(X, theta). **Optional**
cft	Differentiable function cft(X) for counterfactual. Leave empty for the Population Attributable Fraction paf where counterfactual is 0 exposure.
deriv.method.args	method.args for hessian.
deriv.method	method for hessian. Don't change this unless you know what you are doing.
check_exposure	Check that exposure X is positive and numeric.
check_rr	Check that Relative Risk function rr equals 1 when evaluated at 0.
check_integrals	Check that counterfactual and relative risk's expected values are well defined for this scenario.
nsim	Number of simulations for estimation of variance
confidence	Concidence level (0 to 100) default = 95 %
check_thetas	Checks that theta parameters are correctly inputed
is_paf	Boolean forcing evaluation of paf

Author(s)

Rodrigo Zepeda-Tello [e-mail]

Dalia Camacho-Garc<c3><ad>a-Forment<c3><ad> [e-mail]

Examples

#Example 1: Exponential Relative Risk
#--------------------------------------------
set.seed(46987)
rr      <- function(X,theta){exp(X*theta)}
cft     <- function(X){0.4*X}
Xmean   <- data.frame(3)
Xvar    <- 1
theta   <- 0.4
thetavar <- 0.001
pif.confidence.approximate.loglinear(Xmean, Xvar, theta, thetavar, rr, cft,
nsim = 1000)

#Example 2: Multivariate Relative Risk
#--------------------------------------------
X1       <- rnorm(100,3,.5)
X2       <- rnorm(100,4,1)
X        <- data.frame(cbind(X1,X2))
Xmean    <- t(as.matrix(colMeans(X)))
Xvar     <- cov(X)
thetahat <- c(0.12, 0.17)
thetavar  <- matrix(c(0.001, 0.00001, 0.00001, 0.004), byrow = TRUE, nrow = 2)
rr       <- function(X, theta){exp(theta[1]*X[,1] + theta[2]*X[,2])}
pif.confidence.approximate.loglinear(Xmean, Xvar, thetahat, thetavar, 
rr, cft = function(X){0.8*X}, nsim = 100)

pifpaf documentation built on May 1, 2019, 9:11 p.m.