Pretty text version of table

If you want a nicer version in your console window then add the text=TRUE option.

summary(tab1, text=TRUE)

## 
## 
## |             | A: IFL (N=428)  | F: FOLFOX (N=691) | G: IROX (N=380) | Total (N=1499)  | p value|
## |:------------|:---------------:|:-----------------:|:---------------:|:---------------:|-------:|
## |sex          |                 |                   |                 |                 |   0.190|
## |-  Male      |   277 (64.7%)   |    411 (59.5%)    |   228 (60.0%)   |   916 (61.1%)   |        |
## |-  Female    |   151 (35.3%)   |    280 (40.5%)    |   152 (40.0%)   |   583 (38.9%)   |        |
## |Age in Years |                 |                   |                 |                 |   0.614|
## |-  Mean (SD) | 59.673 (11.365) |  60.301 (11.632)  | 59.763 (11.499) | 59.985 (11.519) |        |
## |-  Range     | 27.000 - 88.000 |  19.000 - 88.000  | 26.000 - 85.000 | 19.000 - 88.000 |        |

Pretty Rmarkdown version of table

In order for the report to look nice within an R markdown (knitr) report, you just need to specify results="asis" when creating the R chunk. This changes the layout slightly (compresses it) and bolds the variable names.

summary(tab1)

Data frame version of table

If you want a data.frame version, simply use as.data.frame.

as.data.frame(tab1)

##   group.term   group.label strata.term variable     term        label variable.type
## 1        arm Treatment Arm                  sex      sex          sex   categorical
## 2        arm Treatment Arm                  sex countpct         Male   categorical
## 3        arm Treatment Arm                  sex countpct       Female   categorical
## 4        arm Treatment Arm                  age      age Age in Years       numeric
## 5        arm Treatment Arm                  age   meansd    Mean (SD)       numeric
## 6        arm Treatment Arm                  age    range        Range       numeric
##                A: IFL           F: FOLFOX            G: IROX              Total
## 1                                                                              
## 2 277.00000, 64.71963 411.00000, 59.47902            228, 60  916.0000, 61.1074
## 3 151.00000, 35.28037 280.00000, 40.52098            152, 40  583.0000, 38.8926
## 4                                                                              
## 5  59.67290, 11.36454  60.30101, 11.63225 59.76316, 11.49930 59.98532, 11.51877
## 6              27, 88              19, 88             26, 85             19, 88
##                         test   p.value
## 1 Pearson's Chi-squared test 0.1904388
## 2 Pearson's Chi-squared test 0.1904388
## 3 Pearson's Chi-squared test 0.1904388
## 4         Linear Model ANOVA 0.6143859
## 5         Linear Model ANOVA 0.6143859
## 6         Linear Model ANOVA 0.6143859

Summaries using standard R code

## base R frequency example
tmp <- table(Gender=mockstudy$sex, "Study Arm"=mockstudy$arm)
tmp

##         Study Arm
## Gender   A: IFL F: FOLFOX G: IROX
##   Male      277       411     228
##   Female    151       280     152

# Note: The continuity correction is applied by default in R (not used in %table)
chisq.test(tmp)

## 
##  Pearson's Chi-squared test
## 
## data:  tmp
## X-squared = 3.3168, df = 2, p-value = 0.1904

## base R numeric summary example
tapply(mockstudy$age, mockstudy$arm, summary)

## $`A: IFL`
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   27.00   53.00   61.00   59.67   68.00   88.00 
## 
## $`F: FOLFOX`
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##    19.0    52.0    61.0    60.3    69.0    88.0 
## 
## $`G: IROX`
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   26.00   52.00   61.00   59.76   68.00   85.00

summary(aov(age ~ arm, data=mockstudy))

##               Df Sum Sq Mean Sq F value Pr(>F)
## arm            2    129    64.7   0.487  0.614
## Residuals   1496 198628   132.8

Add labels

In the above example, age is shown with a label (Age in Years), but sex is listed “as is” with lower case letters. This is because the data was created in SAS and in the SAS dataset, age had a label but sex did not. The label is stored as an attribute within R.

## Look at one variable's label
attr(mockstudy$age,'label')

## [1] "Age in Years"

## See all the variables with a label
unlist(lapply(mockstudy,'attr','label'))

##                        age                        arm                       race 
##             "Age in Years"            "Treatment Arm"                     "Race" 
##                        bmi 
## "Body Mass Index (kg/m^2)"

# Can also use labels(mockstudy)

If you want to add labels to other variables, there are a couple of options. First, you could add labels to the variables in your dataset.

attr(mockstudy$sex,'label')  <- 'Gender'

tab1 <- tableby(arm ~ sex + age, data=mockstudy)
summary(tab1)

You can also use the built-in data.frame method for labels<-:

labels(mockstudy)  <- c(age = 'Age, yrs', sex = "Gender")

tab1 <- tableby(arm ~ sex + age, data=mockstudy)
summary(tab1)

Another option is to add labels after you have created the table

mylabels <- list(sex = "SEX", age = "Age, yrs")
summary(tab1, labelTranslations = mylabels)

Alternatively, you can check the variable labels and manipulate them with a function called labels, which works on the tableby object.

labels(tab1)

##             arm             sex             age 
## "Treatment Arm"        "Gender"      "Age, yrs"

labels(tab1) <- c(arm="Treatment Assignment", age="Baseline Age (yrs)")
labels(tab1)

##                    arm                    sex                    age 
## "Treatment Assignment"               "Gender"   "Baseline Age (yrs)"

summary(tab1)

Change summary statistics globally

Currently the default behavior is to summarize continuous variables with: Number of missing values, Mean (SD), 25th - 75th quantiles, and Minimum-Maximum values with an ANOVA (t-test with equal variances) p-value. For categorical variables the default is to show: Number of missing values and count (column percent) with a chi-square p-value. This behavior can be modified using the tableby.control function. In fact, you can save your standard settings and use that for future tables. Note that test=FALSE and total=FALSE results in the total column and p-value column not being printed.

mycontrols  <- tableby.control(test=FALSE, total=FALSE,
                               numeric.test="kwt", cat.test="chisq",
                               numeric.stats=c("N", "median", "q1q3"),
                               cat.stats=c("countpct"),
                               stats.labels=list(N='Count', median='Median', q1q3='Q1,Q3'))
tab2 <- tableby(arm ~ sex + age, data=mockstudy, control=mycontrols)
summary(tab2)

You can also change these settings directly in the tableby call.

tab3 <- tableby(arm ~ sex + age, data=mockstudy, test=FALSE, total=FALSE, 
                numeric.stats=c("median","q1q3"), numeric.test="kwt")
summary(tab3)

Change summary statistics within the formula

In addition to modifying summary options globally, it is possible to modify the test and summary statistics for specific variables within the formula statement. For example, both the kwt (Kruskal-Wallis rank-based) and anova (asymptotic analysis of variance) tests apply to numeric variables, and we can use one for the variable “age”, another for the variable “bmi”, and no test for the variable “ast”. A list of all the options is shown at the end of the vignette.

The tests function can do a quick check on what tests were performed on each variable in tableby.

tab.test <- tableby(arm ~ kwt(age) + anova(bmi) + notest(ast), data=mockstudy)
tests(tab.test)

##           Group Variable   p.value                       Method
## 1 Treatment Arm      age 0.6390614 Kruskal-Wallis rank sum test
## 2 Treatment Arm      bmi 0.8916552           Linear Model ANOVA
## 3 Treatment Arm      ast        NA                      No test

summary(tab.test)

Summary statistics for any individual variable can also be modified, but it must be done as secondary arguments to the test function. The function names must be strings that are functions already written for tableby, built-in R functions like mean and range, or user-defined functions.

tab.test <- tableby(arm ~ kwt(ast, "Nmiss2","median") + anova(age, "N","mean") +
                    notest(bmi, "Nmiss","median"), data=mockstudy)
summary(tab.test)

Controlling Options for Categorical Tests (Chisq and Fisher’s)

The formal tests for categorical variables against the levels of the by variable, chisq and fe, have options to simulate p-values. We show how to turn on the simulations for these with 500 replicates for the Fisher’s test (fe).

set.seed(100)
tab.catsim <- tableby(arm ~ sex + race, cat.test="fe", simulate.p.value=TRUE, B=500, data=mockstudy)
tests(tab.catsim)

      Group Variable   p.value

1 Treatment Arm sex 0.2195609 2 Treatment Arm race 0.3093812 Method 1 Fisher’s Exact Test for Count Data with simulated p-valuebased on 500 replicates) 2 Fisher’s Exact Test for Count Data with simulated p-valuebased on 500 replicates)

The chi-square test on 2x2 tables applies Yates’ continuity correction by default, so we provide an option to turn off the correction. We show the results with and without the correction that is applied to treatment arm by sex, if we use subset to ignore one of the three treatment arms.

cat.correct <- tableby(arm ~ sex + race, cat.test="chisq", subset = !grepl("^F", arm), data=mockstudy)
tests(cat.correct)

      Group Variable   p.value                     Method

1 Treatment Arm sex 0.1666280 Pearson’s Chi-squared test 2 Treatment Arm race 0.8108543 Pearson’s Chi-squared test

cat.nocorrect <- tableby(arm ~ sex + race, cat.test="chisq", subset = !grepl("^F", arm),
     chisq.correct=FALSE, data=mockstudy)
tests(cat.nocorrect)

      Group Variable   p.value                     Method

1 Treatment Arm sex 0.1666280 Pearson’s Chi-squared test 2 Treatment Arm race 0.8108543 Pearson’s Chi-squared test

Modifying the look & feel in Word documents

You can easily create Word versions of tableby output via an Rmarkdown report and the default options will give you a reasonable table in Word - just select the “Knit Word” option in RStudio.

The functionality listed in this next paragraph is coming soon but needs an upgraded version of RStudio If you want to modify fonts used for the table, then you’ll need to add an extra line to your header at the beginning of your file. You can take the WordStylesReference01.docx file and modify the fonts (storing the format preferences in your project directory). To see how this works, run your report once using WordStylesReference01.docx and then WordStylesReference02.docx.

output: word_document
  reference_docx: /projects/bsi/gentools/R/lib320/arsenal/doc/WordStylesReference01.docx 

For more informating on changing the look/feel of your Word document, see the Rmarkdown documentation website.

1. Summarize without a group/by variable

tab.noby <- tableby(~ bmi + sex + age, data=mockstudy)
summary(tab.noby)

2. Display footnotes indicating which “test” was used

summary(tab.test, pfootnote=TRUE)

1. Kruskal-Wallis rank sum test
2. Linear Model ANOVA

3. Summarize an ordered factor

When comparing groups of ordered data there are a couple of options. The default uses a general independence test available from the coin package. For two-group comparisons, this is essentially the Armitage trend test. The other option is to specify the Kruskal Wallis test. The example below shows both options.

mockstudy$age.ordnew <- ordered(c("a",NA,as.character(mockstudy$age.ord[-(1:2)])))
table(mockstudy$age.ord, mockstudy$sex)

##        
##         Male Female
##   10-19    1      0
##   20-29    8     11
##   30-39   37     30
##   40-49  127     83
##   50-59  257    179
##   60-69  298    170
##   70-79  168    101
##   80-89   20      9

table(mockstudy$age.ordnew, mockstudy$sex)

##        
##         Male Female
##   10-19    1      0
##   20-29    8     11
##   30-39   37     30
##   40-49  127     83
##   50-59  257    179
##   60-69  297    170
##   70-79  168    100
##   80-89   20      9
##   a        1      0

class(mockstudy$age.ord)

## [1] "ordered" "factor"

summary(tableby(sex ~ age.ordnew, data = mockstudy), pfootnote = TRUE)

1. Trend test for ordinal variables

summary(tableby(sex ~ age.ord, data = mockstudy), pfootnote = TRUE)

1. Trend test for ordinal variables

4. Summarize a survival variable

First look at the information that is presented by the survfit() function, then see how the same results can be seen with tableby. The default is to show the median survival (time at which the probability of survival = 50%).

survfit(Surv(fu.time, fu.stat)~sex, data=mockstudy)

## Call: survfit(formula = Surv(fu.time, fu.stat) ~ sex, data = mockstudy)
## 
##              n events median 0.95LCL 0.95UCL
## sex=Male   916    829    550     515     590
## sex=Female 583    527    543     511     575

survdiff(Surv(fu.time, fu.stat)~sex, data=mockstudy)

## Call:
## survdiff(formula = Surv(fu.time, fu.stat) ~ sex, data = mockstudy)
## 
##              N Observed Expected (O-E)^2/E (O-E)^2/V
## sex=Male   916      829      830  0.000370  0.000956
## sex=Female 583      527      526  0.000583  0.000956
## 
##  Chisq= 0  on 1 degrees of freedom, p= 1

summary(tableby(sex ~ Surv(fu.time, fu.stat), data=mockstudy))

It is also possible to obtain summaries of the % survival at certain time points (say the probability of surviving 1-year).

summary(survfit(Surv(fu.time/365.25, fu.stat)~sex, data=mockstudy), times=1:5)

## Call: survfit(formula = Surv(fu.time/365.25, fu.stat) ~ sex, data = mockstudy)
## 
##                 sex=Male 
##  time n.risk n.event survival std.err lower 95% CI upper 95% CI
##     1    626     286   0.6870  0.0153       0.6576       0.7177
##     2    309     311   0.3437  0.0158       0.3142       0.3761
##     3    152     151   0.1748  0.0127       0.1516       0.2015
##     4     57      61   0.0941  0.0104       0.0759       0.1168
##     5     24      16   0.0628  0.0095       0.0467       0.0844
## 
##                 sex=Female 
##  time n.risk n.event survival std.err lower 95% CI upper 95% CI
##     1    380     202   0.6531  0.0197       0.6155        0.693
##     2    190     189   0.3277  0.0195       0.2917        0.368
##     3     95      90   0.1701  0.0157       0.1420        0.204
##     4     51      32   0.1093  0.0133       0.0861        0.139
##     5     18      12   0.0745  0.0126       0.0534        0.104

summary(tableby(sex ~ Surv(fu.time/365.25, fu.stat), data=mockstudy, times=1:5, surv.stats=c("NeventsSurv","NriskSurv")))

5. Summarize date variables

Date variables by default are summarized with the number of missing values, the median, and the range. For example purposes we’ve created a random date. Missing values are introduced for impossible February dates.

set.seed(100)
N <- nrow(mockstudy)
mockstudy$dtentry <- mdy.Date(month=sample(1:12,N,replace=T), day=sample(1:29,N,replace=T), 
                              year=sample(2005:2009,N,replace=T))
summary(tableby(sex ~ dtentry, data=mockstudy))

6. Summarize multiple variables without typing them out

Often one wants to summarize a number of variables. Instead of typing by hand each individual variable, an alternative approach is to create a formula using the paste command with the collapse="+" option.

## create a vector specifying the variable names
myvars <- names(mockstudy)

## select the 8th through the last variables
## paste them together, separated by the + sign
RHS <- paste(myvars[8:10], collapse="+")
RHS

[1] “ps+hgb+bmi”

## create a formula using the as.formula function
as.formula(paste('arm ~ ', RHS))

arm ~ ps + hgb + bmi

## use the formula in the tableby function
summary(tableby(as.formula(paste('arm ~', RHS)), data=mockstudy))

These steps can also be done using the formulize function.

## The formulize function does the paste and as.formula steps
tmp <- formulize('arm',myvars[8:10])
tmp

arm ~ ps + hgb + bmi

## More complex formulas could also be written using formulize
tmp2 <- formulize('arm',c('ps','hgb^2','bmi'))

## use the formula in the tableby function
summary(tableby(tmp, data=mockstudy))

To change summary statistics or statistical tests en masse, consider using paste0() together with formulize():

varlist1 <- c('age','sex','hgb')
varlist2 <- paste0("anova(", c('bmi','alk.phos','ast'), ", 'meansd')")

summary(tableby(formulize("arm", c(varlist1, varlist2)),
                data = mockstudy, numeric.test = "kwt"), pfootnote = TRUE)

1. Kruskal-Wallis rank sum test
2. Pearson’s Chi-squared test
3. Linear Model ANOVA

7. Subset the dataset used in the analysis

Here are two ways to get the same result (limit the analysis to subjects age>5 and in the F: FOLFOX treatment group).

• The first approach uses the subset function applied to the dataset mockstudy. This example also selects a subset of variables. The tableby function is then applied to this subsetted data.

newdata <- subset(mockstudy, subset=age>50 & arm=='F: FOLFOX', select = c(sex,ps:bmi))
dim(mockstudy)

## [1] 1499   16

table(mockstudy$arm)

## 
##    A: IFL F: FOLFOX   G: IROX 
##       428       691       380

dim(newdata)

## [1] 557   4

names(newdata)

## [1] "sex" "ps"  "hgb" "bmi"

summary(tableby(sex ~ ., data=newdata))

• The second approach does the same analysis but uses the subset argument within tableby to subset the data.

summary(tableby(sex ~ ps + hgb + bmi, subset=age>50 & arm=="F: FOLFOX", data=mockstudy))

8. Create combinations of variables on the fly

## create a variable combining the levels of mdquality.s and sex
with(mockstudy, table(interaction(mdquality.s,sex)))

## 
##   0.Male   1.Male 0.Female 1.Female 
##       77      686       47      437

summary(tableby(arm ~ interaction(mdquality.s,sex), data=mockstudy))

## create a new grouping variable with combined levels of arm and sex
summary(tableby(interaction(mdquality.s, sex) ~  age + bmi, data=mockstudy, subset=arm=="F: FOLFOX"))

9. Transform variables on the fly

Certain transformations need to be surrounded by I() so that R knows to treat it as a variable transformation and not some special model feature. If the transformation includes any of the symbols / - + ^ * then surround the new variable by I().

trans <- tableby(arm ~ I(age/10) + log(bmi) + factor(mdquality.s, levels=0:1, labels=c('N','Y')),
                 data=mockstudy)
summary(trans)

The labels for these variables aren’t exactly what we’d like, so we can change modify those after the fact. Instead of typing out the very long variable names, you can modify specific labels by position.

labels(trans)

##                                                           arm 
##                                               "Treatment Arm" 
##                                                     I(age/10) 
##                                                    "Age, yrs" 
##                                                      log(bmi) 
##                                    "Body Mass Index (kg/m^2)" 
##       factor(mdquality.s, levels = 0:1, labels = c("N", "Y")) 
## "factor(mdquality.s, levels = 0:1, labels = c(\"N\", \"Y\"))"

labels(trans)[2:4] <- c('Age per 10 yrs', 'log(BMI)', 'MD Quality')
labels(trans)

##                                                     arm 
##                                         "Treatment Arm" 
##                                               I(age/10) 
##                                        "Age per 10 yrs" 
##                                                log(bmi) 
##                                              "log(BMI)" 
## factor(mdquality.s, levels = 0:1, labels = c("N", "Y")) 
##                                            "MD Quality"

summary(trans)

Note that if we had not changed mdquality.s to a factor, it would have been summarized as though it were a continuous variable.

class(mockstudy$mdquality.s)

[1] “integer”

summary(tableby(arm~mdquality.s, data=mockstudy))

Another option would be to specify the test and summary statistics. In fact, if I had a set of variables coded 0/1 and that was all I was summarizing, then I could change the global option for continuous variables to use the chi-square test and show countpct.

summary(tableby(arm ~ chisq(mdquality.s, "Nmiss","countpct"), data=mockstudy))

10. Subsetting (change the ordering of the variables, delete a variable, sort by p-value, filter by p-value)

mytab <- tableby(arm ~ sex + alk.phos + age, data=mockstudy)
mytab2 <- mytab[c('age','sex','alk.phos')]
summary(mytab2)

summary(mytab[c('age','sex')], digits = 2)

summary(mytab[c(3,1)], digits = 3)

summary(sort(mytab, decreasing = TRUE))

summary(mytab[mytab < 0.5])

head(mytab, 1) # can also use tail()

tableby Object

Function Call: tableby(formula = arm ~ sex + alk.phos + age, data = mockstudy)

Variable(s): arm ~ sex

11. Merge two tableby objects together

It is possible to combine two tableby objects so that they print out together. Overlapping by-variables will have their x-variables concatenated, and (if all=TRUE) non-overlapping by-variables will have their tables printed separately.

## demographics
tab1 <- tableby(arm ~ sex + age, data=mockstudy,
                control=tableby.control(numeric.stats=c("Nmiss","meansd"), total=FALSE))
## lab data
tab2 <- tableby(arm ~ hgb + alk.phos, data=mockstudy,
                control=tableby.control(numeric.stats=c("Nmiss","median","q1q3"),
                                        numeric.test="kwt", total=FALSE))
tab12 <- merge(tab1, tab2)
class(tab12)

[1] “tableby” “arsenal_table”

summary(tab12)

For tables with two different outcomes, consider the all=TRUE argument:

summary(merge(
  tableby(sex ~ age, data = mockstudy),
  tableby(arm ~ bmi, data = mockstudy),
  all = TRUE
))

12. Add a title to the table

When creating a pdf the tables are automatically numbered and the title appears below the table. In Word and HTML, the titles appear un-numbered and above the table.

t1 <- tableby(arm ~ sex + age, data=mockstudy)
summary(t1, title='Demographics')

13. Modify how missing values are displayed

Depending on the report you are writing you have the following options:

• Show how many subjects have each variable

• Show how many subjects are missing each variable

• Show how many subjects are missing each variable only if there are any missing values

• Don’t indicate missing values at all

## look at how many missing values there are for each variable
apply(is.na(mockstudy),2,sum)

##        case         age         arm         sex        race     fu.time     fu.stat          ps 
##           0           0           0           0           7           0           0         266 
##         hgb         bmi    alk.phos         ast mdquality.s     age.ord  age.ordnew     dtentry 
##         266          33         266         266         252           0           1           5

## Show how many subjects have each variable (non-missing)
summary(tableby(sex ~ ast + age, data=mockstudy,
                control=tableby.control(numeric.stats=c("N","median"), total=FALSE)))

## Always list the number of missing values
summary(tableby(sex ~ ast + age, data=mockstudy,
                control=tableby.control(numeric.stats=c("Nmiss2","median"), total=FALSE)))

## Only show the missing values if there are some (default)
summary(tableby(sex ~ ast + age, data=mockstudy, 
                control=tableby.control(numeric.stats=c("Nmiss","mean"),total=FALSE)))

## Don't show N at all
summary(tableby(sex ~ ast + age, data=mockstudy, 
                control=tableby.control(numeric.stats=c("mean"),total=FALSE)))

One might also consider the use of includeNA() to include NAs in the counts and percents for categorical variables.

mockstudy$ps.cat <- factor(mockstudy$ps)
attr(mockstudy$ps.cat, "label") <- "ps"
summary(tableby(sex ~ includeNA(ps.cat), data = mockstudy, cat.stats = "countpct"))

14. Modify the number of digits used

Within tableby.control function there are 4 options for controlling the number of significant digits shown.

• digits: controls the number of digits after the decimal place for continuous values

• digits.count: controls the number of digits after the decimal point for counts

• digits.pct: controls the number of digits after the decimal point for percents

• digits.p: controls the number of digits after the decimal point for p-values

summary(tableby(arm ~ sex + age + fu.time, data=mockstudy), digits=4, digits.p=2, digits.pct=1)

With the exception of digits.p, all of these can be specified on a per-variable basis using the in-formula functions that specify which tests are run:

summary(tableby(arm ~ chisq(sex, digits.pct=1) + anova(age, digits=4) +
                  anova(fu.time, digits = 1), data=mockstudy))

15. Create a user-defined summary statistic

For purposes of this example, the code below creates a trimmed mean function (trims 10%) and use that to summarize the data. Note the use of the ... which tells R to pass extra arguments on - this is required for user-defined functions. In this case, na.rm=T is passed to myfunc. The weights argument is also required, even though it isn’t passed on to the internal function in this particular example.

trim10 <- function(x, weights=rep(1,length(x)), ...){
  mean(x, trim=.1, ...)
}

summary(tableby(sex ~ hgb, data=mockstudy, 
                control=tableby.control(numeric.stats=c("Nmiss","trim10"), numeric.test="kwt",
                    stats.labels=list(Nmiss='Missing values', trim10="Trimmed Mean, 10%"))))

For statistics to be formatted appropriately, you may want to use as.tbstat() or as.countpct(). For example, suppose you want to create a trimmed mean function that trims by both 5 and 10 percent. The first example shows them separated by a comma; the second puts the 10% trimmed mean in brackets

trim510comma <- function(x, weights=rep(1,length(x)), ...){
  tmp <- c(mean(x, trim = 0.05, ...), mean(x, trim = 0.1, ...))
  as.tbstat(tmp, sep = ", ")
}
trim510bracket <- function(x, weights=rep(1,length(x)), ...){
  tmp <- c(mean(x, trim = 0.05, ...), mean(x, trim = 0.1, ...))
  as.tbstat(tmp, sep = " ", parens = c("[", "]"))
}

summary(tableby(sex ~ hgb, data=mockstudy, numeric.stats=c("Nmiss", "trim510comma"), test = FALSE))

summary(tableby(sex ~ hgb, data=mockstudy, numeric.stats=c("Nmiss", "trim510bracket"), test = FALSE))

Or perhaps it’s useful to put the amount of trimming in parentheses. Since it is a percent, we can flag it as such:

trim10pct <- function(x, weights=rep(1,length(x)), ...){
  tmp <- mean(x, trim = 0.05, ...)
  as.countpct(c(tmp, 10), sep = " ", parens = c("(", ")"), which.count = 0, which.pct = 2, pct = "%")
}
summary(tableby(sex ~ hgb, data=mockstudy, numeric.stats=c("Nmiss", "trim10pct"),
                digits = 2, digits.pct = 0, test = FALSE))

16. Use case-weights for creating summary statistics

When comparing groups, they are often unbalanced when it comes to nuisances such as age and sex. The tableby function allows you to create weighted summary statistics. If this option us used then p-values are not calculated (test=FALSE).

##create fake group that is not balanced by age/sex 
set.seed(200)
mockstudy$fake_arm <- ifelse(mockstudy$age>60 & mockstudy$sex=='Female',sample(c('A','B'),replace=T, prob=c(.2,.8)),
                            sample(c('A','B'),replace=T, prob=c(.8,.4)))

mockstudy$agegp <- cut(mockstudy$age, breaks=c(18,50,60,70,90), right=FALSE)

## create weights based on agegp and sex distribution
tab1 <- with(mockstudy,table(agegp, sex))
tab2 <- with(mockstudy, table(agegp, sex, fake_arm))
tab2

## , , fake_arm = A
## 
##          sex
## agegp     Male Female
##   [18,50)   73     62
##   [50,60)  128     94
##   [60,70)  139      7
##   [70,90)  102      0
## 
## , , fake_arm = B
## 
##          sex
## agegp     Male Female
##   [18,50)   79     48
##   [50,60)  130     84
##   [60,70)  156    166
##   [70,90)  109    122

gpwts <- rep(tab1, length(unique(mockstudy$fake_arm)))/tab2
gpwts[gpwts>50] <- 30

## apply weights to subjects
index <- with(mockstudy, cbind(as.numeric(agegp), as.numeric(sex), as.numeric(as.factor(fake_arm)))) 
mockstudy$wts <- gpwts[index]

## show weights by treatment arm group
tapply(mockstudy$wts,mockstudy$fake_arm, summary)

## $A
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   1.774   1.894   2.069   2.276   2.082  24.714 
## 
## $B
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   1.000   1.042   1.924   1.677   1.985   2.292

orig <- tableby(fake_arm ~ age + sex + Surv(fu.time/365, fu.stat), data=mockstudy, test=FALSE)
summary(orig, title='No Case Weights used')

tab1 <- tableby(fake_arm ~ age + sex + Surv(fu.time/365, fu.stat), data=mockstudy, weights=wts)
summary(tab1, title='Case Weights used')

17. Create your own p-value and add it to the table

When using weighted summary statistics, it is often desirable to then show a p-value from a model that corresponds to the weighted analysis. It is possible to add your own p-value and modify the column title for that new p-value. Another use for this would be to add standardized differences or confidence intervals instead of a p-value.

To add the p-value, you simply need to create a data frame and use the function modpval.tableby(). The first few columns in the data.frame are required: (1) the by-variable, (2) the strata value (if the table has a strata term), (3) the x-variable, and (4) the new p-value (or test statistic). Another optional column can be used to indicate what method was used to calculate the p-value. If you specify use.pname=TRUE then the column name indicating the p-value will be also be used in the tableby summary.

mypval <- data.frame(
  byvar = "fake_arm",
  variable = c('age','sex','Surv(fu.time/365, fu.stat)'), 
  adj.pvalue = c(.953,.811,.01), 
  method = c('Age/Sex adjusted model results')
)
tab2 <- modpval.tableby(tab1, mypval, use.pname=TRUE)
summary(tab2, title='Case Weights used, p-values added', pfootnote=TRUE)

1. Age/Sex adjusted model results

18. For two-level categorical variables or one-line numeric variables, simplify the output.

If the cat.simplify option is set to TRUE, then only the second level of two-level categorical varialbes is shown. In the example below, sex has two levels, and “Female” is the second level, hence only the counts and percents for Female are shown. Similarly, “mdquality.s” was turned to a factor, and “1” is the second level, but since there are missings, the table ignores cat.simplify and displays all levels (since the output can no longer be displayed on one line).

table2 <- tableby(arm~sex + factor(mdquality.s), data=mockstudy, cat.simplify=TRUE)
summary(table2, labelTranslations=c(sex="Female", "factor(mdquality.s)"="MD Quality"))

Similarly, if numeric.simplify is set to TRUE, then any numerics which only have one row of summary statistics are simplified into a single row. Note again that ast has missing values and so is not simplified to a single row.

summary(tableby(arm ~ age + ast, data = mockstudy,
                numeric.simplify=TRUE, numeric.stats=c("Nmiss", "meansd")))

The in-formula functions to change which tests are run can also be used to specify these options for each variable at a time.

summary(tableby(arm ~ anova(age, "meansd", numeric.simplify=TRUE) +
                  chisq(sex, cat.simplify=TRUE), data = mockstudy))

19. Use tableby within an Sweave document

For those users who wish to create tables within an Sweave document, the following code seems to work.

\documentclass{article}

\usepackage{longtable}
\usepackage{pdfpages}

\begin{document}

\section{Read in Data}
<<echo=TRUE>>=
require(arsenal)
require(knitr)
require(rmarkdown)
data(mockstudy)

tab1 <- tableby(arm~sex+age, data=mockstudy)
@

\section{Convert Summary.Tableby to LaTeX}
<<echo=TRUE, results='hide', message=FALSE>>=
capture.output(summary(tab1), file="Test.md")

## Convert R Markdown Table to LaTeX
render("Test.md", pdf_document(keep_tex=TRUE))
@ 

\includepdf{Test.pdf}

\end{document}

20. Export tableby object to a .CSV file

When looking at multiple variables it is sometimes useful to export the results to a csv file. The as.data.frame function creates a data frame object that can be exported or further manipulated within R.

tab1 <- summary(tableby(arm~sex+age, data=mockstudy), text = NULL)
as.data.frame(tab1)

##              A: IFL (N=428) F: FOLFOX (N=691) G: IROX (N=380)  Total (N=1499) p value
## 1    Gender                                                                     0.190
## 2      Male     277 (64.7%)       411 (59.5%)     228 (60.0%)     916 (61.1%)        
## 3    Female     151 (35.3%)       280 (40.5%)     152 (40.0%)     583 (38.9%)        
## 4  Age, yrs                                                                     0.614
## 5 Mean (SD) 59.673 (11.365)   60.301 (11.632) 59.763 (11.499) 59.985 (11.519)        
## 6     Range 27.000 - 88.000   19.000 - 88.000 26.000 - 85.000 19.000 - 88.000

# write.csv(tab1, '/my/path/here/my_table.csv')

21. Write tableby object to a separate Word or HTML file

## write to an HTML document
tab1 <- tableby(arm ~ sex + age, data=mockstudy)
write2html(tab1, "~/trash.html")

## write to a Word document
write2word(tab1, "~/trash.doc", title="My table in Word")

22. Use tableby in R Shiny

The easiest way to output a tableby() object in an R Shiny app is to use the tableOutput() UI in combination with the renderTable() server function and as.data.frame(summary(tableby())):

# A standalone shiny app
library(shiny)
library(arsenal)
data(mockstudy)

shinyApp(
  ui = fluidPage(tableOutput("table")),
  server = function(input, output) {
    output$table <- renderTable({
      as.data.frame(summary(tableby(sex ~ age, data = mockstudy), text = "html"))
    }, sanitize.text.function = function(x) x)
  }
)

This can be especially powerful if you feed the selections from a selectInput(multiple = TRUE) into formulize() to make the table dynamic!

23. Use tableby in bookdown

Since the backbone of tableby() is knitr::kable(), tables still render well in bookdown. However, print.summary.tableby() doesn’t use the caption= argument of kable(), so some tables may not have a properly numbered caption. To fix this, use the method described on the bookdown site to give the table a tag/ID.

summary(tableby(sex ~ age, data = mockstudy), title="(\\#tab:mytableby) Caption here")

24. Adjust tableby for multiple p-values

The padjust() function is a new S3 generic piggybacking off of p.adjust(). It works on both tableby and summary.tableby objects:

tab <- summary(tableby(sex ~ age + fu.time + bmi + mdquality.s, data = mockstudy))
tab

padjust(tab, method = "bonferroni")

25. Tabulate multiple endpoints

You can now use list() on the left-hand side of tableby() to give multiple endpoints.

summary(tableby(list(sex, mdquality.s, ps) ~ age + bmi, data = mockstudy))

To avoid confusion about which table is which endpoint, you can set term.name=TRUE in summary(). This takes the labels for each by-variable and puts them in the top-left of the table.

summary(tableby(list(sex, mdquality.s, ps) ~ age + bmi, data = mockstudy), term.name = TRUE)

26. Tabulate data by a non-test group (strata)

You can also specify a second grouping variable that doesn’t get tested (but instead separates results): a strata variable.

summary(tableby(list(sex, ps) ~ age + bmi, strata = arm, data = mockstudy))

Summary statistics

The default summary statistics, by varible type, are:

• numeric.stats: Continuous variables will show by default Nmiss, meansd, range
• cat.stats: Categorical and factor variables will show by default Nmiss, countpct
• ordered.stats: Ordered factors will show by default Nmiss, countpct
• surv.stats: Survival variables will show by default Nmiss, Nevents, medsurv
• date.stats: Date variables will show by default Nmiss, median, range

There are a number of extra functions defined specifically for the tableby function.

• N: a count of the number of observations for a particular group
• Nmiss: only show the count of the number of missing values if there are some missing values
• Nmiss2: always show a count of the number of missing values for a variable within each group
• meansd: print the mean and standard deviation in the format mean(sd)
• meanCI: print the mean and a (t) confidence interval
• count: print the number of values in a category
• countN: print the number of values in a category plus the total N for the group in the format N/Total
• countpct: print the number of values in a category plus the column-percentage in the format N (%)
• countrowpct: print the number of values in a category plus the row-percentage in the format N (%)
• countcellpct: print the number of values in a category plus the cell-percentage in the format N (%)
• binomCI: print the proportion in a category plus a binomial confidence interval.
• rowbinomCI: print the row proportion in a category plus a binomial confidence interval.
• medianq1q3: print the median, 25th, and 75th quantiles median (Q1, Q3)
• q1q3: print the 25th and 75th quantiles Q1, Q3
• iqr: print the inter-quartile range.
• medianrange: print the median, minimum and maximum values median (minimum, maximum)
• medianmad: print the median and median absolute deviation (mad)
• Nevents: print number of events for a survival object within each grouping level
• medSurv: print the median survival
• NeventsSurv: print number of events and survival at given times
• NriskSurv: print the number still at risk and survival at given times
• Nrisk: print the number still at risk at given times
• medTime: print the median follow-up time
• sum
• max
• min
• mean
• sd
• var
• median
• range
• gmean, gsd, gmeansd, gmeanCI: geometric means, sds, and confidence intervals.

Testing options

The tests used to calculate p-values differ by the variable type, but can be specified explicitly in the formula statement or in the control function.

The following tests are accepted:

• anova: analysis of variance test; the default test for continuous variables. When the grouping variable has two levels, it is equivalent to the two-sample t-test with equal variance.

• kwt: Kruskal-Wallis test, optional test for continuous variables. When the grouping variable has two levels, it is equivalent to the Wilcoxon Rank Sum test.

• chisq: chi-square goodness of fit test for equal counts of a categorical variable across categories; the default for categorical or factor variables

• fe: Fisher’s exact test for categorical variables; optional

• logrank: log-rank test, the default test for time-to-event variables

• trend: The independence_test function from the coin is used to test for trends. Whenthe grouping variable has two levels, it is equivalent to the Armitage trend test. This is the default for ordered factors

• notest: Don’t perform a test.

tableby.control settings

A quick way to see what arguments are possible to utilize in a function is to use the args() command. Settings involving the number of digits can be set in tableby.control or in summary.tableby.

args(tableby.control)

## function (test = TRUE, total = TRUE, test.pname = NULL, numeric.simplify = FALSE, 
##     cat.simplify = FALSE, ordered.simplify = FALSE, date.simplify = FALSE, 
##     numeric.test = "anova", cat.test = "chisq", ordered.test = "trend", 
##     surv.test = "logrank", date.test = "kwt", test.always = FALSE, 
##     numeric.stats = c("Nmiss", "meansd", "range"), cat.stats = c("Nmiss", 
##         "countpct"), ordered.stats = c("Nmiss", "countpct"), 
##     surv.stats = c("Nmiss", "Nevents", "medSurv"), date.stats = c("Nmiss", 
##         "median", "range"), stats.labels = list(Nmiss = "N-Miss", 
##         Nmiss2 = "N-Miss", meansd = "Mean (SD)", medianrange = "Median (Range)", 
##         median = "Median", medianq1q3 = "Median (Q1, Q3)", q1q3 = "Q1, Q3", 
##         iqr = "IQR", mean = "Mean", sd = "SD", var = "Var", max = "Max", 
##         min = "Min", meanCI = "Mean (CI)", sum = "Sum", gmean = "Geom Mean", 
##         gsd = "Geom SD", gmeansd = "Geom Mean (Geom SD)", gmeanCI = "Geom Mean (CI)", 
##         range = "Range", Npct = "N (Pct)", Nevents = "Events", 
##         medSurv = "Median Survival", medTime = "Median Follow-Up", 
##         medianmad = "Median (MAD)", overall = "Overall", total = "Total", 
##         difference = "Difference"), digits = 3L, digits.count = 0L, 
##     digits.pct = 1L, digits.p = 3L, format.p = TRUE, digits.n = 0L, 
##     conf.level = 0.95, chisq.correct = FALSE, simulate.p.value = FALSE, 
##     B = 2000, times = 1:5, ...) 
## NULL

summary.tableby settings

The summary.tableby function has options that modify how the table appears (such as adding a title or modifying labels).

args(arsenal:::summary.tableby)

## function (object, ..., labelTranslations = NULL, text = FALSE, 
##     title = NULL, pfootnote = FALSE, term.name = "") 
## NULL