1 Data format

The input dataset for the clinical data review should be a data.frame with clinical data. Such data is typically imported from SAS data file or xpt data file.
Such dataset can be imported for multiple files at once via the clinUtils::loadDataADaMSDTM function.

The label of the variables stored in the SAS datasets is also used for title/captions.

A few ADaM datasets are included in the clinUtils package for the demonstration, via the dataset dataADaMCDISCP01 and corresponding variable labels.

library(clinUtils)

data(dataADaMCDISCP01)
labelVars <- attr(dataADaMCDISCP01, "labelVars")

dataLB <- dataADaMCDISCP01$ADLBC
dataDM <- dataADaMCDISCP01$ADSL
dataAE <- dataADaMCDISCP01$ADAE

2 Annotate data

The annotateData enables to add metadata for a specific domain/dataset.

dataLBAnnot <- annotateData(
    data = dataLB, 
    annotations = list(data = dataDM, vars = c("ETHNIC", "ARM")), 
    verbose = TRUE
)

## Data annotated with variable(s): ETHNIC ('ETHNIC'), ARM ('ARM') from the 'custom' dataset based on the variable(s):  USUBJID ('USUBJID').

pander(
    head(dataLBAnnot), 
    caption = paste("Laboratory parameters annotated with",
        "demographics information with the `annotatedData` function"
    )
)

3 Filter data

The filterData enables to filter a dataset.

dataLBAnnotTreatment <- filterData(
    data = dataLBAnnot, 
    filters = list(var = "ARM", value = "Placebo", rev = TRUE), 
    verbose = TRUE
)

## 354 records with ARM ('ARM') %in% 'Placebo' are filtered in data.

pander(
    unique(dataLBAnnotTreatment[, c("USUBJID", "ARM")]), 
    caption = paste("Subset of laboratory parameters filtered",
        "with placebo patients"
    )
)

4 Transform data

The transformData enables to convert data to a different format.

For example, the laboratory data is converted from a long format, containing one record per endpoint * visit * subject to a wide format containing one record per visit * subject. The endpoints are included in different columns.

eDishData <- transformData(
    data = subset(dataLB, PARAMCD %in% c("ALT", "BILI")),
    transformations = list(
        type = "pivot_wider",
        varsID = c("USUBJID", "VISIT"), 
        varsValue = c("LBSTRESN", "LBNRIND"),
        varPivot = "PARAMCD"
    ),
    verbose = TRUE,
    labelVars = labelVars
)

## Warning in reshapeWide(data, idvar = idvar, timevar = timevar, varying = varying, : some constant variables
## (AVISIT,AVISITN,PARAM,PARAMN,AVAL,BASE,CHG,A1LO,A1HI,R2A1LO,R2A1HI,BR2A1LO,BR2A1HI,ANL01FL,ALBTRVAL,LBSEQ) are really varying

## Warning in reshapeWide(data, idvar = idvar, timevar = timevar, varying = varying, : multiple rows match for PARAMCD=BILI: first taken

## Warning in reshapeWide(data, idvar = idvar, timevar = timevar, varying = varying, : multiple rows match for PARAMCD=ALT: first taken

## Data is converted to a wide format with variables: 'Numeric Result/Finding in Standard Units', 'Reference Range Indicator' for different: 'Parameter Code' by 'Unique Subject Identifier', 'Visit Name' pivoted to different columns.

pander(head(eDishData))

5 Process data

The processData function executes all the pre-processing steps described in the previous section at once.

dataLBAnnotTreatment2 <- processData(
    data = dataLB,
    processing = list(
        list(annotate = list(data = dataDM, vars = c("ETHNIC", "ARM"))),
        list(filter = list(var = "ARM", value = "Placebo", rev = TRUE))
    ),
    verbose = TRUE
)

identical(dataLBAnnotTreatment, dataLBAnnotTreatment2)

[1] TRUE

6 Appendix