Absolute risk estimation of time-to-event data with competing risks

Absolute risk estimation and comparison of two or more groups with a time-to-event outcome. Wrapper for the prodlim and ate-functions in riskRegression.

Usage

estimatR(
  data,
  timevar,
  event,
  group,
  survscale = "AM",
  method = "cox",
  vars,
  event.form = NULL,
  time = 120,
  breaks = 12,
  cause = 1,
  pl = T,
  digits = 4,
  event.digits = 2,
  alpha = 0.05,
  survtime = T,
  proportions = F,
  conditional = F,
  diagnostics = F,
  plot = T
)

Arguments

data: dataset
timevar: Time variable
event: Status indicator
group: Grouping variable
survscale: Whether overall survial should be estimated as survival or all-cause mortality (1-survival)
vars: Only applicable when "select" is chosen as type. The variables are added to the model as an additive model
time: Time horizon of interest. Defaults to 60 (e.g. 5-years)
breaks: Interim time points of interest. Defaults to 12 months (1-year gaps)
cause: cause of interest, default = 1
pl: Whether product.limit in ATE shoulde be T or F (default = T)
digits: number of digits risk estimates in the returned results
event.digits: Rounding of event times. Default is 2 to preserve exact times
alpha: alpha level for the estimation of confidence intervals and p-values. Default = 0.05
survtime: Whether median time to event should be calculated (default = T)
proportions: Whether risk of event in different windows should be estimated (default = F)
conditional: Whether conditional risk at the time horizon should be calculated (default = F)
diagnostics: whether Scaled Schoenfield residuals should be visualized (default = F)
plot: whether estimates for plotR should be performed (default = T)
type: Model specification, Can be univariate ("uni"), Age- and sex standardized ("age-sex"), Multivariate with variable selection ("select"). In this case vars should be a vector of the covariates. "custom" allows for free modelling where the form-argument contains the formula.
form: Only applicable when "custom" is chosen as type. Free specification of the model as the right-hand side of the formula.

Value

List of class "estimatR" containing the following: table: Event table
models: Model objects (cause1 and cause2)
risks: Absolute risk estimates at the specified time points
plot_data: Data for plotting CIF curves
time_to_event: Median survival time
hr: Hazard ratio between the groups
difference: Absolute risk difference at the specified time horizons
ratio: Absolute risk ratios at the the specified time horizon
counts: Event and group counts in the contrasted groups
info: information on arguments for extraction Optional: diag: Diagnostics for assessing proportionality
proportions:
"Before" estimates the risk of event within a certain timepoint (e.g. x% of the events occurred within).
"After" estimates the risk of residual events between timepoint x and ten years.
"Window" estimates the percentage of events within six month windows (e.g. x% of the events occurred between t1 and t2)
conditional: Conditional risk estimates

Examples

estimatR(analysis_df, ttt, event, X1, type = "select", vars = c(X6,X7))
#> Error in estimatR(analysis_df, ttt, event, X1, type = "select", vars = c(X6,     X7)): unused argument (type = "select")