R/drotr

Doubly robust optimal treatment rule estimation

Author: Allison Codi

Description

drotr is a package that uses doubly-robust methods to estimate optimal treatment rules (OTRs).

We define OTRs through the use of conditional average treatment effects (CATEs). The CATE can be interpreted as the expected difference in outcome under treatment vs placebo. A doubly-robust learner is used to estimate the CATE. All individuals with CATE estimates that exceed a specified threshold 't' are treated under the OTR.

Once observations are assigned treatment under the OTR, we can use various estimands to describe the rule: (1) the proportion treated, (2) the overall Average Treatment effect under the Rule ($ATR_d$), (3) the Average Treatment effect in the subgroup Recommended Treatment under the rule ($ATRT_d$), (4) the Average Treatment effect in the subgroup Not Recommended Treatment under the rule ($ATNRT_d$), and (5) the difference between treatment recommendation subgroups ($ATRT_d - ATNRT_d$). These are estimated using Augmented Inverse Probability of Treatment Weight (AIPTW) estimators.

The full procedure is implemented using nested cross-validation and Super Learning.

Installation:

A developmental release may be installed from GitHub via devtools with:

devtools::install_github("allicodi/drotr")

Usage:

Here we demonstrate calls to drotr using a simulated data set of length-for-age z-score 90 days post enrollment for a diarrheal antibiotics trial.

library(drotr)

# 1. Load data
data(abcd_data)

# 2. Set input parameters

# outcome variable
Y <- "lazd90"

# treatment variable
A <- "an_grp_01"

# covariates 
W <- c("rotavirus_new", "rotavirus_bin", "norovirus_new", "norovirus_bin", "adenovirus_new",
      "adenovirus_bin", "sapovirus_new","sapovirus_bin", "astrovirus_new", "astrovirus_bin",
      "st_etec_new", "st_etec_bin", "shigella_new", "shigella_bin", "campylobacter_new",
      "campylobacter_bin", "tepec_new", "tepec_bin", "v_cholerae_new", "v_cholerae_bin",
      "salmonella_new", "salmonella_bin", "cryptosporidium_new", "cryptosporidium_bin",
      "dy1_scrn_vomitall", "dy1_scrn_lstools", "dy1_scrn_sstools", "dy1_scrn_diardays",
      "dy1_scrn_dehydr", "avemuac", "wfazscore", "lfazscore", "wflzscore", "site",
      "dy1_ant_sex", "agemchild", "an_ses_quintile", "an_tothhlt5", "month_en", "rotaseason")

# subset of covariates to use to make rule
Z <- c("avemuac", "wfazscore", "wflzscore", "lfazscore", "dy1_ant_sex", "agemchild", "an_ses_quintile")

# treatment threshold
t <- 0.05

# List of libraries to use for nuisance and CATE models
sl.library.outcome <- c("SL.glm", "SL.ranger", "SL.earth")
sl.library.treatment <- c("SL.mean", "SL.glm")
sl.library.missingness <- c("SL.mean", "SL.glm")
sl.library.CATE <- c("SL.glm", "SL.ranger", "SL.earth")

set.seed(12345)

results_host <- estimate_OTR(df = abcd_data,
                             Y_name = Y, 
                             A_name = A, 
                             W_list = W, 
                             Z_list = Z, 
                             id_name = "pid",
                             sl.library.outcome = sl.library.outcome,
                             sl.library.treatment = sl.library.treatment,
                             sl.library.missingness = sl.library.missingness,
                             sl.library.CATE = sl.library.CATE,
                             threshold = t,
                             k_folds = 10,
                             outcome_type = "gaussian")
                             
print(results_host)
                             

                               Results for  threshold =  0.05  Aggregated Across k =  10  folds 
--------------------------------------------------------------------------------------------------------------------------------------- 
                                                            Estimate            Standard Error      95% CI: Lower       95% CI: Upper       
--------------------------------------------------------------------------------------------------------------------------------------- 
E[Y(1) | d(Z) = 1]                                          -1.6637             0.0265              -1.7156             -1.6117             
E[Y(0) | d(Z) = 1]                                          -1.709              0.026               -1.7599             -1.6581             
E[d(Z) = 1]                                                 0.4516              0.0061              0.4397              0.4635              
E[Y(1) - Y(0) | d(Z) = 1]                                   0.0453              0.0209              0.0043              0.0863              
E[Y(1) - Y(0) | d(Z) = 0]                                   0.0338              0.0193              -0.004              0.0715              
E[Y(d) - Y(0)]                                              0.019               0.0095              5e-04               0.0375              
E[Y(1) - Y(0) | d(Z) = 1] - E[Y(1) - Y(0) | d(Z) = 0]       0.0116              0.0285              -0.0442             0.0673              

Covariates used in decision rule:  avemuac, wfazscore, wflzscore, lfazscore, dy1_ant_sex, agemchild, an_ses_quintile 

Additional information regarding rule interpretation, comparison, and averaging across seeds can be found in vignettes/drotr

Issues

If you encounter any bugs or have feature requests, please file an issue.

Citation

After using the drotr R package, please cite the following:

@Manual{drotr_package,
  title = {drotr: Doubly-robust optimal treatment rule estimation},
  author = {Allison Codi},
  note = {R packag version 1.0.0}
}