R/tab_univariate.R
tab_univariate.RdProduce odds ratios, risk ratios or incidence rate ratios
tab_univariate(
x,
outcome,
...,
perstime = NULL,
strata = NULL,
measure = "OR",
extend_output = TRUE,
digits = 3,
mergeCI = FALSE,
woolf_test = FALSE
)A data frame
Name of A TRUE/FALSE variable as your outcome of interest (e.g. illness)
Names of TRUE/FALSE variables as exposures of interest (e.g. risk factors)
A numeric variable containing the observation time for each individual
Name of a TRUE/FALSE variable to be used for stratifying
results. Note that this results in a different output table - giving you a
table of crude measure, measures for each strata and the mantel-haeszel
adjusted measure for each exposure variable listed in ...
Specify what you would like to calculated, options are "OR", "RR" or "IRR" default is "OR"
TRUE/FALSE to specify whether would like all columns in the outputs (default is TRUE) Non-extended output drops group odds or risk calculations as well as p-values
Specify number of decimal places (default is 3)
Whether or not to put the confidence intervals in one column (default is FALSE)
Only if strata specified and measure is "RR" or "OR". TRUE/FALSE to specify whether to include woolf test for homogeneity p-value. Tests whether there is a significant difference in the estimates between strata.
Inspired by Daniel Gardiner, see github repo Real data set for example from http://sphweb.bumc.bu.edu/otlt/mph-modules/bs/bs704-ep713_confounding-em/BS704-EP713_Confounding-EM7.html
# set up data set, first as 2x2x2 table
arr <- array(
data = c(10, 35, 90, 465, 36, 25, 164, 175),
dim = c(2 , 2 , 2),
dimnames = list(
risk = c(TRUE , FALSE),
outcome = c(TRUE , FALSE),
old = c(FALSE, TRUE)
)
)
arr
#> , , old = FALSE
#>
#> outcome
#> risk TRUE FALSE
#> TRUE 10 90
#> FALSE 35 465
#>
#> , , old = TRUE
#>
#> outcome
#> risk TRUE FALSE
#> TRUE 36 164
#> FALSE 25 175
#>
# Create data frame from 2x2x2 table
library("tidyr")
#>
#> Attaching package: ‘tidyr’
#> The following objects are masked from ‘package:Matrix’:
#>
#> expand, pack, unpack
a <- arr %>%
as.data.frame.table() %>%
tidyr::uncount(weights = Freq) %>%
dplyr::mutate_all(as.logical) %>%
tibble::as_tibble()
# get the results from tab_univariate function
tab_univariate(a, outcome, risk, strata = old, digits = 6, measure = "OR")
#> variable est_type exp_cases unexp_cases cases_odds exp_controls
#> crude risk crude 46 60 0.7666667 254
#> old: TRUE risk old: TRUE 36 25 1.4400000 164
#> old: FALSE risk old: FALSE 10 35 0.2857143 90
#> MH risk MH NA NA NA NA
#> unexp_controls controls_odds ratio lower upper p.value
#> crude 640 0.3968750 1.931759 1.2811369 2.912796 0.001456825
#> old: TRUE 175 0.9371429 1.536585 0.8839320 2.671127 0.126045958
#> old: FALSE 465 0.1935484 1.476190 0.7056245 3.088241 0.298455837
#> MH NA NA 1.516129 0.9739216 2.360198 NA
tab_univariate(a, outcome, risk, strata = old, digits = 6, measure = "RR")
#> variable est_type exp_cases exp_total exp_risk unexp_cases
#> crude risk crude 46 300 15.33333 60
#> old: TRUE risk old: TRUE 36 200 18.00000 25
#> old: FALSE risk old: FALSE 10 100 10.00000 35
#> MH risk MH NA NA NA NA
#> unexp_total unexp_risk ratio lower upper p.value
#> crude 700 8.571429 1.788889 1.2486872 2.562790 0.001456825
#> old: TRUE 200 12.500000 1.440000 0.8989988 2.306566 0.126045958
#> old: FALSE 500 7.000000 1.428571 0.7316106 2.789484 0.298455837
#> MH NA NA 1.436364 0.9769870 2.111738 NA