Zachary McCaw
Updated: 2020-10-20
Given stratified binary event data for two arms, this package calculates summary statistics comparing two arms with respect to the marginal event rate. Marginal event rates are calculated as the stratum-size weighted-average of the per-stratum event rates, then compared via the risk difference, risk ratio, and odds ratio. Also see:
- StratRMST for comparing restricted mean survival times.
devtools::install_github(repo = 'zrmacc/MargRates')Consider the following 28-day unadjusted mortality data from a recent COVID-19 clinical trial, stratified by respiratory support at randomization
# Event counts.
y0 <- c(283, 682, 145)
n0 <- c(683, 2604, 1034)
y1 <- c(95, 298, 89)
n1 <- c(324, 1279, 501)
# Marginal Odds Ratio
library(MargRates)
set.seed(2013)
out <- CompMargRates(
y0 = y0,
n0 = n0,
y1 = y1,
n1 = n1,
alpha = 0.05,
boot = TRUE,
perm = TRUE,
reps = 2e3
)
show(out)## Marginal Rates:
## Arm N Rates
## 1 0 4321 0.2567286
## 2 1 2104 0.2292085
##
##
## Risk Difference:
## Method Stat Est SE Lower Upper P
## 1 Asymptotic RiskDiff -0.0275201 0.01121997 -0.04951084 -0.005529367 0.01417575
## 4 Bootstrap RiskDiff -0.0275201 0.01123516 -0.04957270 -0.005407053 0.01799100
##
##
## Risk Ratio:
## Method Stat Est SE Lower Upper P
## 2 Asymptotic RiskRatio 0.8928047 0.04218031 0.8138450 0.9794251 0.01639499
## 5 Bootstrap RiskRatio 0.8928047 0.04233066 0.8123457 0.9782870 0.01799100
##
##
## Odds Ratio:
## Method Stat Est SE Lower Upper P
## 3 Asymptotic OddsRatio 0.8609283 0.05334131 0.7624797 0.9720881 0.01565469
## 6 Bootstrap OddsRatio 0.8609283 0.05358627 0.7608769 0.9712936 0.01799100
##
##
## Permutation test:
## Stat Est P
## 1 RiskDiff -0.0275201 0.01299350
## 2 RiskRatio 0.8928047 0.01149425
## 3 OddsRatio 0.8609283 0.01199400