add type = "geomean"

R/estimate_outcome_continuous.R

@@ -96,6 +96,13 @@ estimate_outcome_continuous <- function(
             digits = digits),
           ")"))
     },
+    "geomean" = {
+      data %>%
+        dplyr::summarize(
+          res = format_round(
+            exp(mean(log(.data$.outcome))),
+            digits = digits))
+    },
     "median" = {
       data %>%
         dplyr::summarize(res = format_round(

R/fill_cells.R

@@ -171,6 +171,7 @@ fill_cells <- function(
     "mean (sd)" =,
     "sd" =,
     "mean (ci)" =,
+    "geomean" =,
     "median" =,
     "median (iqr)" =,
     "range" = "outcome_continuous",

R/rifttable.R

@@ -195,9 +195,10 @@
 #'        survival for censoring.
 #'      * \code{"medfu (iqr)"} Median and interquartile range for follow-up.
 #'      * \code{"maxfu"} Maximum follow-up time.
-#'      * \code{"mean"} Mean.
+#'      * \code{"mean"} Mean (arithmetic mean).
 #'      * \code{"mean (ci)"} Mean and 95% CI.
 #'      * \code{"mean (sd)"} Mean and standard deviation.
+#'      * \code{"geomean"} Geometric mean.
 #'      * \code{"median"} Median.
 #'      * \code{"median (iqr)"} Median and interquartile range.
 #'      * \code{"range"} Range: Minimum to maximum value.

vignettes/estimators.Rmd

@@ -30,9 +30,10 @@ knitr::opts_chunk$set(
 `type` | Description | Options (`arguments = `)
 -------+-------------+--------------
 `"range"` | Range: Minimum to maximum value.
-`"mean"` | Mean.
+`"mean"` | Mean (arithmetic mean).
 `"mean (ci)"` | Mean and 95% CI.
 `"mean (sd)"` | Mean and standard deviation.
+`"geomean"` | Geometric mean.
 `"median"` | Median.
 `"median (iqr)"` | Median and interquartile range.
 

vignettes/faq.Rmd

@@ -325,19 +325,22 @@ Three types of ratios for continuous outcomes are implemented in `rifttable()`:
 
 `type =` | Ratio of ... | Model | Use
 -------|----------|-----------------------|-----------------------------
-`"fold"` | Arithmetric means | Generalized linear model with Gaussian distribution and log link: `glm(y ~ x, family = gaussian(link = "log"))` | Tolerates `0` in the `outcome` variable. Best fit if outcome is normally distributed without transformation.
-`"irrrob"` | Arithmetric means | Generalized linear model with Poisson distribution and log link: `glm(y ~ x, family = poisson())`, with robust (sandwich) standard errors | Tolerates `0` in the `outcome` variable. (Note, `stats::glm()` may issue warnings that the outcome is "non-integer." This warning can be ignored here.)
-`"foldlog"` | Geometric means | Linear model with log-transformed outcome: `lm(log(y) ~ x)` | Does not tolerate `0` in the `outcome` variable, as `log(0)` is undefined (R returns `-Inf`). Best fit if outcome is normally distributed after log transformation.
+`"fold"` | Arithmetric means | Generalized linear model with Gaussian distribution and log link: `glm(y ~ x, family = gaussian(link = "log"))` | Tolerates `0` in the `outcome` variable. May be informative if outcome is normally distributed without transformation.
+`"irrrob"` | Arithmetric means | Generalized linear model with Poisson distribution and log link: `glm(y ~ x, family = poisson())`, with robust (sandwich) standard errors | Tolerates `0` in the `outcome` variable. May be informative if outcome is normally distributed without transformation.
+`"foldlog"` | Geometric means | Linear model with log-transformed outcome: `lm(log(y) ~ x)` | Does not tolerate `0` in the `outcome` variable, as `log(0)` is undefined (R returns `-Inf`). May be informative if outcome is normally distributed after log transformation.
 
-In all models, beta coefficients can be interpreted as ratios after exponentiation. `rifttable`()` automatically does all nessary transformations.
+In all models, after exponentiation, beta coefficients can be interpreted as ratios. rifttable automatically does all necessary transformations.
 
-In the `cancer` data (see previous question), ratios of usual (arithmetric) means of age could be considered informative, given that `hist(cancer$age)` etc. does not show a major skew in this outcome.
+In the `cancer` data (see previous question), ratios of usual (arithmetic) means of age could be considered informative, given that `hist(cancer$age)` etc. does not show a major skew in this outcome.
 
 ```{r table2_ratio}
 tribble(
   ~label,                                       ~type,
   "Observations",                               "total",
-  "Mean age",                                   "mean",
+  "Mean age",                                   "", 
+  "  Arithmetic mean",                          "mean",
+  "  Geometric mean",                           "geomean",
+  "Mean ratio",                                 "",
   "  Ratio of arithmetric means",               "fold",
   "  Ratio of arithmetric means, empirical SE", "irrrob",
   "  Ratio of geometric means",                 "foldlog") %>%
@@ -350,4 +353,4 @@ tribble(
   rt_gt()  # obtain formatted output
 ```
 
-Patients with ECOG performance status of 2, compared to those with performance status 0, are on average 8% older, based on comparing arithmetric means (9%  based on geometric means).
+Patients with ECOG performance status of 2, compared to those with performance status 0, are on average 8% older (or 9%, if based on geometric means).