README.md

In louisahsmith/simpleSens: What the Package Does (One Line, Title Case)

simpleSens is a package designed to make simple sensitivity analysis for unmeasured confounding, selection bias, and misclassification easy.

Installation

library(simpleSens)

Usage

There are two main types of sensitivity analysis that this package allows for:

• calculate an E-value
• calculate a bias factor

# Calculate an E-value for unmeasured confounding
calculate_evalue(RRobs = 4, biases = list(confounding()))
#> 
#> This multiple bias E-value refers simultaneously to 2 parameters:
#> RR_UY and RR_AU
#> (see documentation for definitions)
#> [1] 7.464102

# Calculate an E-value for selection bias and misclassification
calculate_evalue(RRobs = 2.5,
         biases = list(selection("selected"), misclassification("outcome")))
#> 
#> This multiple bias E-value refers simultaneously to 3 parameters:
#> RR_AU|S=1, RR_UY|S=1, and RR_AY*|y
#> (see documentation for definitions)
#> [1] 1.923636

# Calculate an E-value for all three available types of bias
calculate_evalue(RRobs = 1.4234,
         biases = list(selection("general", "S = U"),
                       misclassification("exposure", outcome_rare = TRUE,
                                                     exposure_rare = TRUE),
                       confounding()))
#> 
#> This multiple bias E-value refers simultaneously to 5 parameters:
#> RR_UY|A=0, RR_UY|A=1, RR_YA*|a, RR_UY, and RR_AU
#> (see documentation for definitions)
#> [1] 1.120531

# get sensitivity parameters for a combination of biases
get_param_info(
  biases =
    list(confounding(), 
         misclassification("exposure", exposure_rare = FALSE, outcome_rare = TRUE))
)
#>   necessary.parameter argument.name              bias
#> 1               RR_UY          RRUY       confounding
#> 2               RR_AU          RRAU       confounding
#> 3            RR_YA*|a         RRYAa misclassification
# calculate bound with those parameters
calculate_bound(
  RRUY = 2, RRAU = 1.5, RRYAa = 3,
  biases =
    list(confounding(), 
         misclassification("exposure", exposure_rare = FALSE, outcome_rare = TRUE))
)
#> [1] 3.6

louisahsmith/simpleSens documentation built on March 19, 2020, 12:07 a.m.