Defining your SEQopts()

Behind SEQopts()

SEQopts() provides the primary API to changing internal options of SEQuential(). While the documentation should go over their use, we will expand on them here in greater detail:

The options

General Options

Option Name Description Input Type Example
bootstrap Whether bootstrapping should take place Logical TRUE
bootstrap.CI Defines the confidence interval after bootstrapping (should be within [0, 1]) Numeric double 0.95
bootstrap.CI_method Selects which way to calculate bootstraps confidence intervals ("se", "percentile") Character "se"
bootstrap.nboot Number of bootstraps to run in addition to the full model Integer 100L
bootstrap.sample Subsample of data to use when bootstrapping Numeric double [0, 1] 0.8
compevent Column name for competing event variable Character "death"
covariates If provided, forces covariates for outcome models Character "X1+X2*X3+X4"
data.return Whether to return expanded data as output Logical TRUE
end_of_fup Estimate an outcome measured at a single follow-up time as a weighted per-arm average, instead of fitting a survival outcome model. Incompatible with km.curves, hazard, compevent and the dose-response method Logical TRUE
end_of_fup.time The follow-up time k at which the end-of-follow-up outcome is evaluated (required when end_of_fup = TRUE) Numeric 12
end_of_fup.type Type of end-of-follow-up outcome: "binary" reports the weighted proportion per arm, "continuous" the weighted mean Character "binary"
end_of_fup.window Half-width of the window used when a trial-period has no measurement at exactly end_of_fup.time; it then falls back to the measurement nearest to k within [k - window, k + window] (ties broken toward the later one, so at least k of follow-up has elapsed), and is censored if there is none Numeric 3
expand.only Return the expanded dataset and skip analysis Logical TRUE
followup.class Whether to expand followup values to an indicator matrix Logical FALSE
followup.include Whether to include followup and followup_squared in outcome models Logical TRUE
followup.max Maximum value of followup per trial Numeric 60
followup.min Minimum value of followup per trial Numeric 0
followup.spline Whether to format followup values to a spline Logical FALSE
followup.spline.df Degrees of freedom passed to splines::ns() when followup.spline = TRUE Integer 4L
hazard Whether to estimate the hazard ratio Logical FALSE
km.curves Whether to estimate Kaplan-Meier survival curves and data Logical TRUE
multinomial Whether to expect more than 2 treatment types Logical FALSE
ncores Number of cores to use in parallel multisession Integer 8L
nthreads Number of threads to use for data.table manipulations Integer 16L
parallel Whether the SEQuential process should be run in parallel Logical TRUE
risk.times Follow-up times (in follow-up units) at which to report risk difference and risk ratio when km.curves = TRUE; each is snapped to the latest follow-up at or before it and the final time is always included (NA reports only the final time) Numeric vector c(36, 120)
seed Starting seed for random number generators Integer 1636
selection.first_trial Selects only the first eligible trial in the expanded dataset Logical TRUE
selection.prob Subsample of total IDs to select Numeric double [0, 1] 0.8
selection.random Whether to randomly select IDs with replacement to run analysis Logical FALSE
subgroup Column name for subgroup analysis Character "sex"
survival.max Maximum value for Risk/Survival curves Numeric double 60
trial.include Whether to include trial and trial_squared in outcome models Logical TRUE
treat.level Treatment levels to compare List c(0, 1)
visit Column name for visit indicator Character "visit"
visit.denominator Visit denominator covariates to the right hand side of a formula object Character "X1+X2"
visit.numerator Visit numerator covariates to the right hand side of a formula object Character "X1+X2"

Weighting Options

In general these only affect analytic methods of ‘dose-response’ and ‘censoring’. However, providing cense will allow ITT to be weighted in the case of a censoring variable like loss-to-followup.

Option Name Description Input Type Example
cense Column name for censoring variable Character "LTFU"
cense.denominator If provided, forces denominator covariates for censoring models Character "A1+A2*A3+A4"
cense.eligible Column name for indicator column defining which rows to use for censoring model Character "eligible_cense"
cense.numerator If provided, forces numerator covariates for censoring models Character "Z1+Z2*Z3+Z4"
denominator If provided, forces denominator covariates for weight models Character "C1+C2*C3+B4"
numerator If provided, forces numerator covariates for weight models Character "B1+B2*B3+B4"
weight.eligible_cols List of column names for indicator columns defining which weights are eligible for weight models Character list list("eligible1", "eligible2")
weight.lag_condition Whether weights should be conditioned on treatment lag value Logical TRUE
weight.lower Lower truncation for weight values Numeric double 0.0
weight.p99 Whether to truncate weights at the 1st and 99th percentiles Logical TRUE
weight.preexpansion Whether weighting should be done on pre-expanded data Logical TRUE
weight.upper Upper truncation for weight values Numeric double 10.0
weighted Whether the analysis should be weighted Logical TRUE

Plot Options (km.curves = TRUE)

All of these can be changed after SEQuential() has finished analysis - if you forget to add them, no need to stop and restart. Additionally, SEQuential() will output the data used for the plot if you would like to customize it yourself through your preferred plotting software.

Option Name Description Input Type Example
plot.colors Figure colors for output plot List list("red", "blue")
plot.labels Legend labels for output plot List list("Tx 0", "Tx 1")
plot.subtitle Subtitle for plot output Character "My Plot Subtitle"
plot.title Title for plot output Character "My Plot Title"
plot.type Type of output plot ("survival", "risk", "inc") Character "survival"

Special Cases

Option Name Description Input Type Example
excused When censoring, whether there is an excused condition Logical TRUE
excused.cols List of column names for treatment switch excuses List list("excuse1", "excuse2")

Internal Options

These are internal options that change output name of covariates or the decomposition method when estimating coefficients.

Option Name Description Input Type Example
fastglm.method Method for decomposition by fastglm Integer 3L
glm.package Package to use for fitting GLMs Character either "fastglm" (default) or "parglm"
indicator.baseline Identifier for baseline variables Character "_bas"
indicator.squared Identifier for squared variables Character "_sq"
parglm.control A control object from parglm::parglm.control() to pass to parglm::parglm.fit() List parglm::parglm.control(method = "LINPACK")