Skip to contents

If only care about the higher subgroup (above cutoff), only need ate.rmtl.high and hr.high so set "onlyhigh" to be TRUE Scores are adjusted to the opposite sign if higher.y == FALSE; scores stay the same if higher.y == FALSE; this is because estsurv() function always takes the subgroup of the top highest adjusted scores, and higher adjusted scores should always represent high responders of trt=1

Usage

estsurv.bilevel.subgroups(
  y,
  d,
  x.cate,
  x.ps,
  x.ipcw,
  trt,
  yf,
  tau0 = tau0,
  score,
  higher.y,
  prop,
  onlyhigh,
  surv.min = 0.025,
  ps.method = "glm",
  minPS = 0.01,
  maxPS = 0.99,
  ipcw.method = "breslow"
)

Arguments

y

Observed survival or censoring time; vector of size n.

d

The event indicator, normally 1 = event, 0 = censored; vector of size n.

x.cate

Matrix of p.cate baseline covariates specified in the outcome model; dimension n by p.cate.

x.ps

Matrix of p.ps baseline covariates specified in the propensity score model; dimension n by p.ps.

x.ipcw

Matrix of p.ipw baseline covariate specified in inverse probability of censoring weighting; dimension n by p.ipw.

trt

Treatment received; vector of size n with treatment coded as 0/1.

yf

Follow-up time, interpreted as the potential censoring time; vector of size n if the potential censoring time is known.

tau0

The truncation time for defining restricted mean time lost.

score

Estimated log CATE scores for all n observations from one of the five methods (random forest, boosting, naive Poisson, two regressions, contrast regression); vector of size n.

higher.y

A logical value indicating whether higher (TRUE) or lower (FALSE)

prop

Proportions corresponding to percentiles in the estimated log CATE scores that define subgroups to calculate ATE for; vector of floats in `(0, 1]` (if onlyhigh=TRUE) or in `(0, 1)` (if onlyhigh=FALSE): Each element of prop represents the high/low cutoff in each bi-level subgroup and the length of prop is number of bi-level subgroups

onlyhigh

Indicator of returning only the ATEs in the higher-than-cutoff category of the bi-level subgroups; boolean.

surv.min

Lower truncation limit for probability of being censored (positive and very close to 0).

ps.method

A character value for the method to estimate the propensity score. Allowed values include one of: 'glm' for logistic regression with main effects only (default), or 'lasso' for a logistic regression with main effects and LASSO penalization on two-way interactions (added to the model if interactions are not specified in ps.model). Relevant only when ps.model has more than one variable.

minPS

A numerical value (in `[0, 1]`) below which estimated propensity scores should be truncated. Default is 0.01.

maxPS

A numerical value (in `(0, 1]`) above which estimated propensity scores should be truncated. Must be strictly greater than minPS. Default is 0.99.

ipcw.method

The censoring model. Allowed values are: 'breslow' (Cox regression with Breslow estimator of the baseline survivor function), 'aft (exponential)', 'aft (weibull)', 'aft (lognormal)' or 'aft (loglogistic)'. Default is 'breslow'.

Value

ate.rmtl.high: estimated ATEs (ratio of RMTL) in the multiple bi-level subgroups that are in the higher-than-cutoff category; vector of size equal to the length of prop; always returned. ate.rmtl.low: estimated ATEs (ratio of RMTL) in the multiple bi-level subgroups that are in the lower-than-cutoff category; vector of size equal to the length of prop; returned only when onlyhigh = TRUE. hr.high: unadjusted hazard ratio in the multiple bi-level subgroups that are in the higher-than-cutoff category; vector of size equal to the length of prop; always returned. hr.low: unadjusted hazard ratio in the multiple bi-level subgroups that are in the lower-than-cutoff category; vector of size equal to the length of prop; returned only when onlyhigh = TRUE