[Bug]: Count number of patient with percentage with row-wise denominator column by column

[huanlugwu]

What happened?

                      Planned arm      with any critical or major deviations  
                     TRT A    TRT B          TRT A                TRT B       
——————————————————————————————————————————————————————————————————————————————
ALL                   252      251          5 (2.0)              2 (0.8)      
Argentina              19       17          5 (26.3)             2 (11.8)      
  000001                5        2          5 (100)              2 (100)      

hello guys, I am working on this demanding table to count number of patients:

• in columns: with and without flag
• in rows: by country and site with different fomats, AND with different denominators in row within same ARM.

can anyone shine a light on this? thx!

[Melkiades]

Hi @huanlugwu, could you provide a reproducible example? Avoid screenshots as they are not easily parsable (use {reprex} ;)

[huanlugwu]

Thank you @Melkiades , so I made up these codes:

library(dplyr)
library(rtables)
library(forcats)

adsl_1 <- ex_adsl %>%
  select(USUBJID, ARM, ARMCD, BMEASIFL, COUNTRY, SITEID) %>%
  dplyr::mutate(
    COUNTRY = as.factor(COUNTRY),
    ARM = as.factor(ARM),
    LAYER1 = "Group 1",
    LAYER1 = factor(LAYER1)
  ) %>%
  dplyr::mutate(ARM = forcats::fct_reorder(ARM, rank(ARMCD))) %>%
  arrange(SITEID) %>%
  dplyr::mutate(SITEID = forcats::fct_inorder(SITEID))    

site_in_use <- adsl_1 %>%
  select(COUNTRY, SITEID) %>%
  distinct() %>%
  arrange(COUNTRY, SITEID) %>%
  mutate(COUNTRY=as.character(COUNTRY),
         SITEID=as.character(SITEID))

adsl_2 <- adsl_1 %>%
  filter(BMEASIFL == "Y") %>%
  dplyr::mutate(
    # LAYER1 to display the first row in header
    LAYER1 = "Group 2",
    LAYER1 = factor(LAYER1)
  )

cfun_cntrysite <- function(df, labelstr, .N_col, ...) {
  in_rows(
    rcell(nrow(df), format = "xx"),
    .labels = labelstr
  )
}

tbl_recipe <- basic_table() %>%
  rtables::split_cols_by(var = "LAYER1") %>%
  rtables::split_cols_by(var = "ARM") %>%
  rtables::summarize_row_groups(cfun = cfun_cntrysite) %>%
  rtables::split_rows_by("COUNTRY", split_fun = trim_levels_to_map(map = site_in_use)) %>%
  rtables::summarize_row_groups() %>%
  rtables::split_rows_by("SITEID", split_fun = trim_levels_to_map(map = site_in_use)) %>%
  rtables::summarize_row_groups()

tbl1 <- tbl_recipe %>%
  rtables::build_table(adsl_1)

tbl_recipe2 <- basic_table() %>%
  rtables::split_cols_by(var = "LAYER1") %>%
  rtables::split_cols_by(var = "ARM")%>%
  rtables::summarize_row_groups(cfun = cfun_cntrysite) %>%
  rtables::split_rows_by("COUNTRY", split_fun = trim_levels_to_map(map = site_in_use)) %>%
  rtables::summarize_row_groups() %>%
  rtables::split_rows_by("SITEID", split_fun = trim_levels_to_map(map = site_in_use)) %>%
  rtables::summarize_row_groups() 

tbl2 <- tbl_recipe2 %>%
  rtables::build_table(adsl_2)

t <- cbind_rtables(tbl1, tbl2)
t

Then you should get:

> t
                             Group 1                                    Group 2                 
             A: Drug X    B: Placebo   C: Combination   A: Drug X    B: Placebo   C: Combination
————————————————————————————————————————————————————————————————————————————————————————————————
                134          134            132             68           73             62      
  CHN        74 (55.2%)   81 (60.4%)     64 (48.5%)     35 (51.5%)   44 (60.3%)     25 (40.3%)  
    CHN-1    21 (15.7%)   20 (14.9%)     16 (12.1%)     10 (14.7%)   14 (19.2%)     9 (14.5%)   

What I need is that, in a row, with same ARM, the fraction of group2 is counted by using the number in group1 as denominator. I updated the example in my original post above.

[shajoezhu]

Hi @huanlugwu , many thanks for sharing this interesting use case.

I agree, your approach of creating two tables is correct, and it is to cbind them.

The ask was to compute the fraction following 35 by 35/74 rather than 35/68.

Can you please post this question at StackOverflow with the tag nest-tern, and we will follow up there. closing this issue for now

[huanlugwu]

consider it done: https://stackoverflow.com/staging-ground/79206010

[shajoezhu]

hi @huanlugwu , the link is broken. Thanks