Propensity score survival analysis — ps

Run a complete pipeline required to estimate average treatment effects (ATEs) with survival data using propensity score methods. The pipeline (1) fits the propensity score model with fit_ps() or fit_ps_mi(), (2) predict the propensity score with predict_ps(), (3) weights patients using the propensity scores with psweight(), and estimates ATEs with surv_ate(). ps_surv() implements the pipeline for a single group while map_ps_surv() iterates over multiple groups.

ps_surv(xdata, formula = NULL, ydata, response, id = "patient_id", ...)

map_ps_surv(
  xdata_list,
  formula = NULL,
  ydata_list,
  response,
  id = "patient_id",
  grp_id,
  integer_grp_id = FALSE,
  ...
)

Arguments

xdata, xdata_list	A dataset containing the variables for propensity score modeling. This can either be an object produced by `make_xdata()` or a standard data frame. If a standard data frame, then `formuls` cannot be `NULL` and `make_xdata()` is run using `formula`.
formula	The propensity score formula as described in `fit_ps()`. Only required if `data` was not produced by `make_xdata()`.
ydata, ydata_list	A data frame or list of data frames containing the response variables as described in `surv_ate()`.
id	The name of the column in `xdata` and `ydata` identifying each patient.
...	Additional arguments to pass to `make_xdata()`.
A	string containing the response for a survival model as described in `surv_ate()`.

Value

ps_surv() returns an object of class ps_surv and map_ps_surv() returnes an object of class grouped_ps_surv. Both are lists containing the elements:

psfit: The propensity score model fit using fit_ps() or fit_ps_mi().
psweight: The propensity score weights returned by psweight().
ate: The average treatment effects returned by surv_ate().
call: The function call returned by match.call().