demo/sec_4_1_ex.r

In olssol/psrwe: PS-Integrated Methods for Incorporating RWE in Clinical Studies

### Example of Section 4.1.
suppressMessages(library(psrwe, quietly = TRUE))
options(digits = 3)
data(ex_dta)

### First parts of Data.
head(ex_dta)

### Obtain PSs.
dta_ps_single <- psrwe_est(ex_dta,
                     v_covs = paste("V", 1:7, sep = ""),
                     v_grp = "Group", cur_grp_level = "current",
                     ps_method = "logistic", nstrata = 5)

### Balance assessment of PS stratification.
plot(dta_ps_single, "balance")
plot(dta_ps_single, "ps")
plot(dta_ps_single, "diff")
plot(dta_ps_single, "diff", metric = "astd", avg_only = TRUE)

### Obtain discounting parameters.
ps_bor_single <- psrwe_borrow(dta_ps_single, total_borrow = 30)
ps_bor_single

### PSPP, single arm study, binary outcome.
options(mc.cores = 1)
.msg <- capture.output({ suppressWarnings({
rst_pp <- psrwe_powerp(ps_bor_single,
                       outcome_type = "binary",
                       v_outcome    = "Y_Bin",
                       seed         = 1234)
}) })
rst_pp

### Plot PSPP results.
plot(rst_pp)
plot(rst_pp, add_stratum = TRUE)

### Outcome analysis.
oa_pp <- psrwe_outana(rst_pp, mu = 0.4)
oa_pp

### PSCL, single arm study, binary outcome.
rst_cl <- psrwe_compl(ps_bor_single,
                      outcome_type = "binary",
                      v_outcome    = "Y_Bin")
rst_cl

### Outcome analysis.
oa_cl <- psrwe_outana(rst_cl, mu = 0.4)
oa_cl

### Use simple Jackknife stderr. This may take a while longer.
rst_cl_jko <- psrwe_compl(ps_bor_single,
                          outcome_type = "binary",
                          v_outcome = "Y_Bin",
                          stderr_method = "sjk")
oa_cl_jko <- psrwe_outana(rst_cl_jko, mu = 0.4)
oa_cl_jko

### With score method (binary outcomes and single arm only.)
oa_cl_score <- psrwe_outana(rst_cl, method_ci = "wilson", mu = 0.40,
                            method_pval = "score")
oa_cl_score

### With score_weighted method (binary outcomes and single arm only.)
oa_cl_score_wt <- psrwe_outana(rst_cl, method_ci = "wilson", mu = 0.40,
                               method_pval = "score_weighted")
oa_cl_score_wt