[Feature Request]: `g_forest()` to support point estimates and confidence intervals stored in a single column

[wwojciech]

Feature description

Feature description

Currently, g_forest() requires the point estimate and confidence interval to be provided in separate columns:

• the column indicated by col_x must contain a scalar value representing the point estimate;
• the column indicated by col_ci must contain a pair of numbers representing the lower and upper confidence interval limits.

However, some tables store the point estimate and confidence interval limits together in a single column rather than in separate columns.

For example:

library(tern)

d <- formatters::ex_adlb
d$ABLFL_logical <- d$ABLFL == "Y"

results <- basic_table() |>
    split_cols_by("ARM", split_fun = keep_split_levels("A: Drug X")) |>
    split_rows_by("AVISIT", split_fun = keep_split_levels(c(
        "BASELINE",
        "WEEK 1 DAY 8",
        "WEEK 2 DAY 15"
    ))) |>
    summarize_change(
        "CHG",
        variables = list(value = "AVAL", baseline_flag = "ABLFL_logical"),
        .stats = "mean_ci_3d"
    ) |>
    build_table(d)

results
                        A: Drug X      
———————————————————————————————————————
BASELINE                               
  Mean (95% CI)   49.48 (48.66 - 50.31)
WEEK 1 DAY 8                           
  Mean (95% CI)    0.47 (-0.68 - 1.63) 
WEEK 2 DAY 15                          
  Mean (95% CI)    0.48 (-0.70 - 1.66) 

Requested enhancement

Extend g_forest() to support tables with a column containing all three values in the following format

est (ci_lwr, ci_upr).

(or some variation of this, that is: est (ci_lwr - ci_upr), est (ci_lwr; ci_upr), ....

g_forest() should be able to extract these three components internally and use them to generate the forest plot.

Code of Conduct

• I agree to follow this project's Code of Conduct.

Contribution Guidelines

• I agree to follow this project's Contribution Guidelines.

Security Policy

• I agree to follow this project's Security Policy.

[wwojciech]

If backward compatibility is not a concern, the most elegant solution would be to replace the existing arguments col_x and col_ci with a single argument, col_x_ci, which would accept a vector of length 1 or 2. If the vector has length:

• 1, both the point estimate and the confidence interval are stored in the same column,
• 2, the point estimate and the confidence interval are stored in separate columns.

[wwojciech]

Probably the best approach is to support it through the existing col_x and col_ci arguments. If both have the same value, it means that the point estimate and confidence interval are stored in one column.