library(vaccine)
#> vaccine (version 0.1.0).
#> Type ?vaccine to get started.
The load_data
function takes in raw data and creates a
data object that can be accepted by various estimation functions. We use
publicly-avaliable data from the HVTN 505 HIV vaccine efficacy trial as
our example.
data(hvtn505)
<- load_data(
dat time = "HIVwk28preunblfu",
event = "HIVwk28preunbl",
vacc = "trt",
marker = "IgG_env",
covariates = c("age","BMI","bhvrisk"),
weights = "wt",
ph2 = "casecontrol",
data = hvtn505
)
The summary_stats
function gives us some useful
summaries of the dataset.
summary_stats(dat)
#> Number of subjects (vaccine group, phase-1): 1161
#> Number of subjects (placebo group, phase-1): 1141
#> Number of subjects (vaccine group, phase-2): 150
#> Number of subjects (placebo group, phase-2): 39
#> Number of events (vaccine group, phase-1): 27
#> Number of events (placebo group, phase-1): 21
#> Number of events (vaccine group, phase-2): 25
#> Number of events (placebo group, phase-2): 19
#> Proportion of subjects with an event (vaccine group, phase- 1): 0.02326
#> Proportion of subjects with an event (placebo group, phase- 1): 0.0184
#> Proportion of subjects with an event (vaccine group, phase- 2): 0.16667
#> Proportion of subjects with an event (placebo group, phase- 2): 0.48718
The est_overall
function allows us to estimate overall
risk in the placebo and vaccine groups, as well as estimate vaccine
efficacy, using either a nonparametric Kaplan-Meier estimator or a
marginalized Cox model.
est_overall(dat=dat, t_0=578, method="KM")
#> stat group est se ci_lower ci_upper
#> 1 risk vaccine 0.04067009 0.008230842 0.02506853 0.05602199
#> 2 risk placebo 0.02879861 0.006563785 0.01622360 0.04121288
#> 3 ve both -0.41222411 0.430451788 -1.56659984 0.22294979
est_overall(dat=dat, t_0=578, method="Cox")
#> stat group est se ci_lower ci_upper
#> 1 risk vaccine 0.04177642 0.008111679 0.02847302 0.06090588
#> 2 risk placebo 0.02938706 0.006486545 0.01901930 0.04514638
#> 3 ve both -0.42159246 0.417915188 -1.52937491 0.20101796
The est_ce
function allows us to compute controlled
effects curves; see Gilbert,
Fong, Kenny, and Carone 2022 for more detail.
<- est_ce(dat=dat, type="Cox", t_0=578)
ests_cox <- est_ce(dat=dat, type="NP", t_0=578)
ests_np #> Warning: package 'SuperLearner' was built under R version
#> 4.2.3
#> Loading required package: nnls
#> Loading required package: gam
#> Warning: package 'gam' was built under R version 4.2.3
#> Loading required package: splines
#> Loading required package: foreach
#> Loaded gam 1.22-2
#> Super Learner
#> Version: 2.0-28
#> Package created on 2021-05-04
The plot_ce
function produces basic plots of CR or CVE
curves. These are returned as ggplot2 objects, so they can be further
customized as needed.
plot_ce(ests_cox, ests_np)
plot of chunk unnamed-chunk-6