When changing the sign of the estimate, I also wanted to reverse the confidence intervals (CIs). Below is the code:

modify_table_body(
  ~.x %>%
  dplyr::mutate(
    estimate = -estimate,
    temp = conf.low,
    conf.low = -conf.high,
    conf.high = -temp
  ) %>%
  dplyr::select(-temp)
)