Unanchored Survival Analysis
2024-11-22
Source:vignettes/unanchored_survival.Rmd
unanchored_survival.Rmd
Illustration using example data
This example reads in centered_ipd_sat
data that was
created in calculating_weights
vignette and uses
adtte_sat
and pseudo_ipd_sat
outcome datasets
to run survival analysis using the maic_unanchored
function
by specifying endpoint_type = "tte"
.
Note that parameters ipd
and pseudo_ipd
in
the maic_unanchored
function for survival data analysis
needs to have the following columns: USUBJID, ARM, EVENT, and TIME.
USUBJID in ipd
needs to match USUBJID in
weights_object
. USUBJID in pseudo_ipd
is not
strictly required, as if left unspecified, subject numbers are assigned
using the row indexes.
Furthermore, TIME in both these datasets have to be in days. For instance, when we digitize a KM plot where time is recorded in months, we need to multiply the months by the appropriate factor (i.e. 30.4375) to get the time in days.
If you would like to run an analysis using scaled weights, set
normalize_weight
to TRUE. One clear benefit of using scaled
weights is that the risk table at time = 0 in the Kaplan-Meier curve
will match the IPD sample size.
data(centered_ipd_sat)
data(adtte_sat)
data(pseudo_ipd_sat)
centered_colnames <- c("AGE", "AGE_SQUARED", "SEX_MALE", "ECOG0", "SMOKE", "N_PR_THER_MEDIAN")
centered_colnames <- paste0(centered_colnames, "_CENTERED")
weighted_data <- estimate_weights(
data = centered_ipd_sat,
centered_colnames = centered_colnames
)
result <- maic_unanchored(
weights_object = weighted_data,
ipd = adtte_sat,
pseudo_ipd = pseudo_ipd_sat,
trt_ipd = "A",
trt_agd = "B",
normalize_weight = FALSE,
endpoint_name = "Overall Survival",
endpoint_type = "tte",
eff_measure = "HR",
time_scale = "month",
km_conf_type = "log-log"
)
There are two summaries available in the result: descriptive and
inferential. In the descriptive section, we have summaries from fitting
survfit
function. Note that restricted mean (rmean),
median, and 95% CI are summarized in the time_scale
specified.
result$descriptive$summary
## trt_ind treatment type records n.max n.start events
## 1 B B Before matching 300 300.0000 300.0000 178.00000
## 2 A A Before matching 500 500.0000 500.0000 190.00000
## 3 B B After matching 300 300.0000 300.0000 178.00000
## 4 A A After matching 500 199.4265 199.4265 65.68878
## rmean se(rmean) median 0.95LCL 0.95UCL
## 1 4.303551 0.3367260 2.746131 2.261125 3.320857
## 2 8.709690 0.3551477 7.587627 6.278691 10.288538
## 3 4.303551 0.3367260 2.746131 2.261125 3.320857
## 4 10.166029 0.5499915 11.900015 7.815275 14.873786
# Not shown due to long output
# result$descriptive$survfit_before
# result$descriptive$survfit_after
In the inferential section, we have the fitted models stored
(i.e. survfit
and coxph
) and the results from
the coxph
models (i.e. hazard ratios and CI). Note that the
p-values summarized are from coxph
model Wald test and not
from a log-rank test. Here is the overall summary.
result$inferential$summary
## case HR LCL UCL pval
## 1 AB 0.3748981 0.3039010 0.4624815 5.245204e-20
## 2 adjusted_AB 0.2834780 0.2074664 0.3873387 2.473442e-15
Here are models and results before adjustment.
result$inferential$fit$km_before
## Call: survfit(formula = Surv(TIME, EVENT) ~ ARM, data = dat, conf.type = km_conf_type)
##
## n events median 0.95LCL 0.95UCL
## ARM=B 300 178 83.6 68.8 101
## ARM=A 500 190 230.9 191.1 313
result$inferential$fit$model_before
## Call:
## coxph(formula = Surv(TIME, EVENT) ~ ARM, data = dat)
##
## coef exp(coef) se(coef) z p
## ARMA -0.9811 0.3749 0.1071 -9.159 <2e-16
##
## Likelihood ratio test=80.62 on 1 df, p=< 2.2e-16
## n= 800, number of events= 368
result$inferential$fit$res_AB_unadj
## $est
## [1] 0.3748981
##
## $se
## [1] 0.0405065
##
## $ci_l
## [1] 0.303901
##
## $ci_u
## [1] 0.4624815
##
## $pval
## [1] 5.245204e-20
Here are models and results after adjustment.
result$inferential$fit$km_after
## Call: survfit(formula = Surv(TIME, EVENT) ~ ARM, data = dat, weights = dat$weights,
## conf.type = km_conf_type)
##
## records n events median 0.95LCL 0.95UCL
## ARM=B 300 300 178.0 83.6 68.8 101
## ARM=A 500 199 65.7 362.2 237.9 453
result$inferential$fit$model_after
## Call:
## coxph(formula = Surv(TIME, EVENT) ~ ARM, data = dat, weights = weights,
## robust = TRUE)
##
## coef exp(coef) se(coef) robust se z p
## ARMA -1.2606 0.2835 0.1504 0.1593 -7.915 2.47e-15
##
## Likelihood ratio test=80.4 on 1 df, p=< 2.2e-16
## n= 800, number of events= 368
result$inferential$fit$res_AB
## $est
## [1] 0.283478
##
## $se
## [1] 0.04601759
##
## $ci_l
## [1] 0.2074664
##
## $ci_u
## [1] 0.3873387
##
## $pval
## [1] 2.473442e-15
Using bootstrap to calculate standard errors
If bootstrap standard errors are preferred, we need to specify the
number of bootstrap iteration (n_boot_iteration
) in
estimate_weights
function and proceed fitting
maic_unanchored
function. Then, the outputs include
bootstrapped CI. Different types of bootstrap CI can be found by
specifying the parameter boot_ci_type
. See
boot.ci
in boot
package for more details.
weighted_data2 <- estimate_weights(
data = centered_ipd_sat,
centered_colnames = centered_colnames,
n_boot_iteration = 100,
set_seed_boot = 1234
)
result_boot <- maic_unanchored(
weights_object = weighted_data2,
ipd = adtte_sat,
pseudo_ipd = pseudo_ipd_sat,
trt_ipd = "A",
trt_agd = "B",
normalize_weight = FALSE,
endpoint_name = "Overall Survival",
endpoint_type = "tte",
eff_measure = "HR",
boot_ci_type = "perc",
time_scale = "month",
km_conf_type = "log-log"
)
result_boot$inferential$fit$boot_res_AB
## $est
## [1] 0.283478
##
## $se
## [1] NA
##
## $ci_l
## [1] 0.2144978
##
## $ci_u
## [1] 0.3740624
##
## $pval
## [1] NA