widen: Function to transform data from person-time format to person...
In confoundr: Diagnostics for Confounding of Time-Varying and Other Joint Exposures

Description Usage Arguments Details Value Examples

Function to transform data from person-time format to person format suitable for lengthen()

1
2
3

widen(input, id, time, exposure, covariate, history = NULL,
  weight.exposure = NULL, weight.censor = NULL, strata = NULL,
  censor = NULL)

`input`	dataframe in long format e.g., a person-time format
`id`	unique identifier at the unit (person) level
`time`	unique index for each observation within each unit
`exposure`	the exposure of interest at time t
`covariate`	a vector of covariates at time t
`history`	variable describing exposure history through time t
`weight.exposure`	inverse probability weight for exposure, at or through time t
`weight.censor`	cumulative inverse probability weight for censoring through time t
`strata`	propensity score strata at time t
`censor`	censoring indicators at time t

Numeric formats are preserved, factors are coerced into character.

A "wide" dataframe where each row uniquely indexes a single subject’s data, so that columns index measurement of each variable at each time. The indices should be indicated with an underscore suffix followed by the time, e.g. variable_1,variable_2.

# Simulate long data set for two subjects
id <- as.numeric(c(1, 1, 1, 2, 2, 2))
time <- as.numeric(c(0, 1, 2, 0, 1, 2))
a <- as.numeric(c(0, 1, 1, 1, 1, 0))
l <- as.numeric(rbinom(6, 1, 0.5))
m <- as.numeric(rbinom(6, 1, 0.5))
n <- as.numeric(rbinom(6, 1, 0.5))

mydata.long <- data.frame(id, time, a, l, m, n)

# Run the widen() function
mydata.wide <- widen(input=mydata.long,
                     id="id", time="time",
                     exposure="a",
                     covariate=c("l","m","n")
                     )