fix: handle all-NA PPSTRESU groups in descriptive statistics

[Gero1999]

Issue

Closes #1216

Description

When PPSTRESU is NA for all rows in a parameter group (common with partial AUC intervals like RCAMINT where no result was computed), two things break:

1. Descriptive statistics crash — table(PPSTRESU) drops NAs, returning an empty table. names(sort(empty_table, ...)[1])[1] returns NULL, which dplyr::mutate can't combine with character results from other groups.

2. Column names get NA suffix — ifelse(PPSTRESU != "", ...) returns NA when PPSTRESU is NA, producing names like RCAMINT_0-20NA.

Fixes:

• Extract .mode_unit() helper in calculate_summary_stats.R that returns NA_character_ for empty/all-NA input
• Guard PPTESTCD formatting against NA ModeUnit
• Add !is.na(PPSTRESU) guard to all ifelse(PPSTRESU != "", ...) patterns in pivot_wider_pknca_results.R (4 locations) and label_operators.R (2 locations)

Definition of Done

• calculate_summary_stats() no longer crashes with all-NA unit groups
• Manual interval parameters appear as RCAMINT_0-20 (not RCAMINT_0-20NA)
• Parameters with no units appear without a unit suffix in the output
• Existing behavior unchanged for groups with valid units

How to test

1. Upload a dataset with both SERUM and URINE specimens
2. In NCA settings, add partial AUC intervals (AUCINT and RCAMINT)
3. Run NCA computation
4. In NCA Results > Select Parameters — verify RCAMINT appears as RCAMINT_0-20 (no NA suffix)
5. Navigate to Descriptive Statistics — should render without error
6. Verify that RCAMINT columns appear without unit brackets, while other parameters retain their units

Contributor checklist

• Code passes lintr checks
• Code passes all unit tests
• New logic covered by unit tests
• New logic is documented
• App or package changes are reflected in NEWS
• Package version is incremented
• R script works with the new implementation (if applicable)
• Settings upload works with the new implementation (if applicable)
• If any .scss change was done, run data-raw/compile_css.R

Notes to reviewer

First reported in #1169 (review). The root cause is that export_cdisc.R correctly sets PPSTRESU = NA when PPSTRES is NA (CDISC SD0027), but downstream code assumed PPSTRESU is always a non-NA string. The fix adds NA guards at all 7 affected locations across 3 files."

[Shaakon35]

RCAMINT does not appear.

Also can you fix the typo in tab_nca.R:

msg <- paste0(

"Urine Parameters (FREXINT, RCAMINT) calculated for non-urine samples",
"will return NA values and units"

-> use paste to add a " "

[Gero1999]

@Shaakon35 did you include Urine in the NCA setup?

[js3110]

LGTM

[Shaakon35]

I tried again, added Plasma and Urine.

Descriptive stats:

NCA results:
RCAMINT is there two times, but I asked it only once:

Can you fix that please? :)
Except if that is correct? but I doubt

[Gero1999]

hey @Shaakon35, it is "correct". The problem is that right now when you request an interval parameter you do it for all PCSPEC (SERUM, URINE) even if you request urine interval parameters (i.e, RCAMINT). Therefore, the results show two columns, one for the URINE, which has the expected units, and one for the SERUM, which is not possible to calculate and has NA units.

I changed it so now the NA units show in a clean way:

However, notice that this is merely aesthetic. Let me know if you think something else should be done