byGroupHR: Calculate Cox regression by groups
In n8thangreen/HESmanip: Hospital length of stay data preprocessing and analysis

Description Usage Arguments Details Value Author(s) See Also Examples

Batch calculate all the Cox regressions for each group in turn. Can select which type of hazard ratio to calculate.

1	byGroupHR(survData.list, naive = T, causespec = T, subdist = T, tofile = FALSE)

`survData.list`	List of survData arrays
`naive`	Time-independent regression.
`causespec`	Cause-specific Cox regression.
`subdist`	Fine & Grey subdistribution Cox regression.
`tofile`	Should the output be sent to file.

The time-independent regression does not account for time of infection (or generally the timing of an intermediate event); the cause-specific Cox regression considers the competing risks to be censored times and the Fine & Grey subdistribution approach retains the competing risk in the risk set.

For each element in the list returns

`timeindnaive`	Time-independent model
`timedeptcausespec`	Time-dependent cause-specific model
`timedeptsubdistn`	Time-dependent Fine & Grey subdistribution model

Within timeindnaive,

`fit.disch`	Cox regression results with discharge alive event of interest
`fit.dead`	Cox regression results with death in-hospital event of interest
`fit.both`	Cox regression results with combined event of interest

Within timedeptcausespec and timedeptsubdistn

`fit.alive.timeonly`	Discharge alive as event of interest; only time of infection as covariate
`fit.alive.full`	Discharge alive as event of interest; time of infection and other risk factors as covariates
`fit.dead.timeonly`	Death in hospital as event of interest; only time of infection as covariate
`fit.dead.full`	Death in hospital as event of interest; time of infection and other risk factors as covariates
`fit.both.timeonly`	Combined event of interest; only time of infection as covariate
`fit.both.full`	Combined event of interest; time of infection and other risk factors as covariates

Within each of these, coxph return values.

N Green

hr.naive, hr.tv.causespecific, hr.tv.subdistribution, coxph, byGroupHR.boxplot, HRboxplot.batch

require(survival)
data(survDataByGender)
foo <- byGroupHR(survDataByGender)
foo$Male$timedeptcausespec$fit.alive.full


## The function is currently defined as
function (survData.list, naive = T, causespec = T, subdist = T, 
    tofile = FALSE) 
{
    startTime <- Sys.time()
    namesGroup <- names(survData.list)
    results <- list()
    progressbar <- txtProgressBar(min = 0, max = length(namesGroup), 
        style = 3)
    counter = 0
    for (i in namesGroup) {
        counter = counter + 1
        setTxtProgressBar(progressbar, counter)
        timeindnaive <- timedeptcausespec <- timedeptsubdistn <- NA
        if (naive == T) {
            timeindnaive = hr.naive(survData.list[[i]])
        }
        if (causespec == T) {
            timedeptcausespec = hr.tv.causespecific(survData.list[[i]])
        }
        if (subdist == T) {
            timedeptsubdistn = hr.tv.subdistribution(survData.list[[i]])
        }
        results[[i]] <- list(timeindnaive = timeindnaive, timedeptcausespec = timedeptcausespec, 
            timedeptsubdistn = timedeptsubdistn)
    }
    close(progressbar)
    cat(Sys.time() - startTime)
    if (tofile == TRUE) {
        sink(".\output\byOrgamismHR.txt")
        results
        sink()
    }
    results
  }