In adayim/cttab: Tables of Descriptive Statistics in word

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Introduction

In a study, most of the tables are standard descriptive table. Some times tables will be grouped by population stratified by treatment group. Further in cancer researches, the variables repeated by cycle. This package makes it fairly straightforward to produce such a table using R. The table output is a matrix and with other function can output as a flextable object. The advantage of the flextble object is, it supports HTML output and the word output. This make it easy to write a report with Rmakrdown and output a word report with officedown. It also made it easy to deal with macro data, with some attributes works just like SAS.

# library(cttab)
devtools::load_all()

Read macro data

Assume you have data (test_data_csv.csv) under test_raw folder, along with CLU (test_data_csv_CLU.csv) and DLU (test_data_csv_DLU.csv) file. You can load the data easily as the following, and the date will be converted automatically with %d%m%Y format. The data meta data, CLU and DLU, will be loaded as well.

dt <- read_macro(path = "data_raw", 
                 file_name = "test_data_csv.csv")
str(dt[,c(5:13)])

As you can find out, all the variable has label and value label attributes, and the date imported with date format. This is imported from DLU and CLU file, like label and format in SAS. Variable label can be assigned with var_lab function. A value label can be assigned with val_lab function with a named atomic vector, name as label of the value.

data(mtcars)
mtcars <- mtcars[, c("mpg", "am")]
str(mtcars)

# Add variable label
var_lab(mtcars$mpg) <- "Miles/(US) gallon"
var_lab(mtcars$am) <- "Transmission"
# Add value label
val_lab(mtcars$am) <- c("Automatic" = 0, "Manual"=1)
str(mtcars)

If we want to create a frequency table for Transmission, we can use table function directly or convert this variable to character

table(mtcars$am)
table(as.character(mtcars$am))

But by doing so, we will lost variable label and may take advantage of the value label. So we will use to_character in this example. Other functions like to_factor and to_numeric is provide to keep variable labels. The to_character function will assign the unique non-missing and non-blank value as value label. The to_factor function will apply the value label to the variable and convert it to factor.

str(as.character(mtcars$am)) # variable and value label dropped
str(to_character(mtcars$am)) # all kept
str(to_factor(mtcars$am)) # value label applied and converted to factor
table(mtcars$am)
table(to_factor(mtcars$am))

Note at the last one, this is the advantage of value label.

Descriptive table

For the example we use the macro data. Some variables might be a subset of other variables, and we may not get exact frequency table if summarize as whole. In the following situation, only for the subjects with tumor grade recorded (variable TUMGRADE) has values for tumor grade PTM (only showed P NOTTSCOREP in this example). And the subset can be dong with select parameters, more variables can be given with named atomic vector. But before that, as NOTTSCOREP is numeric in DLU and we want to change it to character. If we use as.character function, we will lost label attributes as stated earlier. We use to_character function here for the conversion.

# Only total
dt$NOTTSCOREP <- to_character(dt$NOTTSCOREP) # Convert to character
res1 <- tab1(vars = c("SEX", "ARM", "HEIGHT", "WEIGHT",
                      "SIDE", "TUBULAR", "TUMGRADE", 
                      "NOTTSCOREP"),
             data = dt[!is.na(dt$REGDAT), ],
             select = c("NOTTSCOREP" = "TUMGRADE == 1"))

write_table(res1)

And we want to summarize table by treatment arm in the following example.

res2 <- tab1(vars = c("SEX", "HEIGHT", "WEIGHT",
                      "SIDE", "TUBULAR", "TUMGRADE",
                      "NOTTSCOREP"),
             group = "ARM",
             data = dt[!is.na(dt$REGDAT), ],
             select = c("NOTTSCOREP" = "TUMGRADE == 1"))

write_table(res2)

For some repeated measures, we want to describe data by time, cycle in cancer example. The repeated variable will automatically have shaded background.

# With cycle
res3 <- tab1(vars = c("BLDTIMPT", "HAEMANAL1", "NEUTROPHILS1", "LYMPHOCYTES1", 
                      "HB", "WBC", "PLTS", "BIOCHEMANAL1", "ECOGPERSTATUS"),
               group = "ARM",
               row_split = "BLDRESCYCLENUM",
               data = dt[!is.na(dt$BLDRESCYCLENUM), ])

write_table(res3,
            caption = c("Table 1. Description by Arm and Cycle",
                        "ASAT population"))

Data listing

Can perform data listing just with write_table function.

write_table(head(dt),
            caption = c("Table 2. Data Listing"))

adayim/cttab documentation built on Dec. 18, 2021, 10:27 p.m.