package to create descriptive tables
- Overview
- News
- Package installation
- Costumizing the table
- Visual exploration
- Exporting the table
- Stratified tables
- Odds Ratios and Hazard Ratios
- Web-based User Interface
- Citation
compareGroups is an R package available on CRAN which performs descriptive tables displaying means, standard deviation, quantiles or frequencies of several variables. Also, p-value to test equality between groups is computed using the appropiate test.
With a very simple code, nice, compact and ready-to-publish descriptives table are displayed on R console. They can also be exported to different formats, such as Word, Excel, PDF or inserted in a R-Sweave or R-markdown document.
You will find an extensive manual describing all compareGropus capabilities with real examples in the vignette.
Also, compareGroups package has been published in Journal of Statistical Software [Subirana et al, 2014 http://www.jstatsoft.org/v57/i12/.].
Version 4.1
-
Compute confidence intervals of means, medians, proporcions or incidences
-
Proportions can be computed by rows as well as by columns, or by combinations of rows and columns (i.e. to sum up 100%).
-
New argument "position" to place tables justified to the left, centered or to the right using Rmarkdown.
-
When exporting tables to Excel,
export2xlsno longer usesxlsxpackage and useswrite_xlsxfunction from writexl pacakge, instead. -
The web-based user interface ,
cGroupsWUI(), has been improved and updated.
Version 4.0
-
New argument
var.equalto consider unequal variances when performing ANOVA tests. -
Date variables are supported.
-
New
strataTablefunction: to create stratified tables without having to use cbind. -
package vignette improved.
-
New
descrTablefunction that builds a descriptive table in one step. -
New options added in
export2md, to export tables in nicer format. -
New options added in
compareGroupsto control permutated chisquared test.
Install the compareGroups package from CRAN and then load it by typing:
install.packages("compareGroups")
or from github to get the latest version
library(devtools)
devtools::install_github(repo = "isubirana/compareGroups")
In the following table, some variables from the PREDIMED study (http://www.predimed.es/) are analysed. We illustrate the syntax of compareGroups functions to display tables containing descriptives or possible tests to compare groups.
Following, to describe all the variables of the data set just type:
library(compareGroups) # load compareGroups package
data(predimed) # load example data
descrTable(predimed)
--------Summary descriptives table ---------
__________________________________________________
[ALL] N
N=6324
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Intervention group: 6324
Control 2042 (32.3%)
MedDiet + Nuts 2100 (33.2%)
MedDiet + VOO 2182 (34.5%)
Sex: 6324
Male 2679 (42.4%)
Female 3645 (57.6%)
Age 67.0 (6.17) 6324
Smoking: 6324
Never 3892 (61.5%)
Current 858 (13.6%)
Former 1574 (24.9%)
Body mass index 30.0 (3.82) 6324
Waist circumference 100 (10.6) 6324
Waist-to-height ratio 0.63 (0.07) 6324
Hypertension: 6324
No 1089 (17.2%)
Yes 5235 (82.8%)
Type-2 diabetes: 6324
No 3322 (52.5%)
Yes 3002 (47.5%)
Dyslipidemia: 6324
No 1746 (27.6%)
Yes 4578 (72.4%)
Family history of premature CHD: 6324
No 4895 (77.4%)
Yes 1429 (22.6%)
Hormone-replacement therapy: 5661
No 5564 (98.3%)
Yes 97 (1.71%)
MeDiet Adherence score 8.68 (1.94) 6324
follow-up to main event (years) 4.36 (1.69) 6324
AMI, stroke, or CV Death: 6324
No 6072 (96.0%)
Yes 252 (3.98%)
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Example
In the following table, variables are described by intervention group.
Some variables such as Waist-to-height ratio (wth) and MeDiet Adherence score (p14) have been treated as non-normal distributed, and medians and quantiles within square brackets instead of means and standard deviations within round brackets are displayed.
Also, individuals older than 55 years old are selected. Appropiate tests to compare means, medians or proportions are performed.
For those binary variables of type "yes/no", you may desire to show only the proportion of "yes" category without showing "yes" but only the variable name or label. For example, for diabetes you may want to see simply "diabetes" instead of "diabetes: yes". This is possible by hide.no argument.
Note the simplicity of the syntax. Also, note the use of formula to select the variables, and the use of subset to filter some individuals as usual in many other R funcions.
tab <- descrTable(group ~ . , predimed, hide.no="no", method=c(wth=2, p14=2), subset=age>55)
tab____________________________________________________________________________________________
Control MedDiet + Nuts MedDiet + VOO p.overall
N=2001 N=2059 N=2141
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Sex: <0.001
Male 780 (39.0%) 930 (45.2%) 865 (40.4%)
Female 1221 (61.0%) 1129 (54.8%) 1276 (59.6%)
Age 67.6 (6.06) 66.9 (5.83) 67.3 (6.01) 0.001
Smoking: 0.490
Never 1268 (63.4%) 1250 (60.7%) 1340 (62.6%)
Current 255 (12.7%) 282 (13.7%) 272 (12.7%)
Former 478 (23.9%) 527 (25.6%) 529 (24.7%)
Body mass index 30.3 (3.95) 29.7 (3.77) 29.9 (3.69) <0.001
Waist circumference 101 (10.7) 100 (10.6) 100 (10.4) 0.024
Waist-to-height ratio 0.63 [0.59;0.68] 0.62 [0.58;0.67] 0.63 [0.58;0.67] <0.001
Hypertension 1683 (84.1%) 1709 (83.0%) 1753 (81.9%) 0.162
Type-2 diabetes 950 (47.5%) 933 (45.3%) 1065 (49.7%) 0.016
Dyslipidemia 1446 (72.3%) 1504 (73.0%) 1532 (71.6%) 0.559
Family history of premature CHD 454 (22.7%) 444 (21.6%) 501 (23.4%) 0.358
Hormone-replacement therapy 31 (1.71%) 30 (1.64%) 36 (1.87%) 0.854
MeDiet Adherence score 9.00 [7.00;10.0] 9.00 [8.00;10.0] 9.00 [8.00;10.0] <0.001
follow-up to main event (years) 4.10 (1.74) 4.32 (1.70) 4.64 (1.59) <0.001
AMI, stroke, or CV Death 95 (4.75%) 70 (3.40%) 84 (3.92%) 0.088
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Also, number of decimals and much more options can be changed to costumize the table as desired (see the package manual)
With compareGroups it is also possible to visualize the distribution of analysed variables. This can be done by the plot function applied on the table:
plot(tab["sex"]) # barplot
plot(tab["age"]) # histogram and normality plot |
 |
Once the table is created, it can be printed on the R console in a nice and compact format, or it can be exported to different formats, such as PDF, Excel, Word or LaTex code.
export2pdf(tab, file = "example.pdf")
export2xls(tab, file = "example.xlsx")
export2word(tab, file = "example.docx")
export2latex(tab, file = "example.tex")This is how the table looks like in PDF:
Also, by using export2md function a descriptive table can be inserted in a Rmarkdown chunk to be compiled in HTML, PDF or Word report. Here there is an example of a Rmarkdown compiled to HTML.
export2md(tab, strip = TRUE, first = TRUE,
header.background = "blue", header.color = "white",
caption = "Description of variables by intervention groups")
After creating a table you may want to repeat the descriptives within stratas. For example, you may want to compare the groups for men and for women. This is very easy using the strataTable function:
# remove sex (first variable)
tab <- tab[-1]
# stratify by sex
tabestr <- strataTable(tab, strata="sex")
tabestr--------Summary descriptives table ---------
__________________________________________________________________________________________________________________________________________________________
Male Female
____________________________________________________________ ____________________________________________________________
Control MedDiet + Nuts MedDiet + VOO p.overall Control MedDiet + Nuts MedDiet + VOO p.overall
N=780 N=930 N=865 N=1221 N=1129 N=1276
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Age 66.8 (6.32) 66.3 (6.10) 66.6 (6.31) 0.181 68.1 (5.84) 67.5 (5.54) 67.7 (5.75) 0.023
Smoking: 0.674 0.923
Never 196 (25.1%) 259 (27.8%) 229 (26.5%) 1072 (87.8%) 991 (87.8%) 1111 (87.1%)
Current 192 (24.6%) 229 (24.6%) 203 (23.5%) 63 (5.16%) 53 (4.69%) 69 (5.41%)
Former 392 (50.3%) 442 (47.5%) 433 (50.1%) 86 (7.04%) 85 (7.53%) 96 (7.52%)
Body mass index 29.5 (3.46) 29.1 (3.27) 29.2 (3.24) 0.014 30.8 (4.18) 30.2 (4.07) 30.4 (3.90) 0.002
Waist circumference 104 (9.73) 103 (9.38) 103 (9.63) 0.251 99.1 (10.9) 97.8 (11.0) 98.0 (10.5) 0.009
Waist-to-height ratio 0.62 [0.58;0.65] 0.61 [0.58;0.65] 0.62 [0.58;0.65] 0.313 0.64 [0.60;0.69] 0.63 [0.58;0.68] 0.63 [0.59;0.68] 0.002
Hypertension 626 (80.3%) 726 (78.1%) 656 (75.8%) 0.097 1057 (86.6%) 983 (87.1%) 1097 (86.0%) 0.733
Type-2 diabetes 415 (53.2%) 480 (51.6%) 472 (54.6%) 0.455 535 (43.8%) 453 (40.1%) 593 (46.5%) 0.007
Dyslipidemia 506 (64.9%) 621 (66.8%) 566 (65.4%) 0.691 940 (77.0%) 883 (78.2%) 966 (75.7%) 0.346
Family history of premature CHD 129 (16.5%) 157 (16.9%) 152 (17.6%) 0.849 325 (26.6%) 287 (25.4%) 349 (27.4%) 0.560
Hormone-replacement therapy 0 (0.00%) 0 (0.00%) 0 (0.00%) . 31 (2.66%) 30 (2.82%) 36 (2.97%) 0.903
MeDiet Adherence score 9.00 [7.00;10.0] 9.00 [8.00;10.0] 9.00 [8.00;10.0] <0.001 8.00 [7.00;10.0] 9.00 [8.00;10.0] 9.00 [7.00;10.0] <0.001
follow-up to main event (years) 4.07 (1.79) 4.40 (1.73) 4.54 (1.64) <0.001 4.12 (1.71) 4.26 (1.67) 4.72 (1.56) <0.001
AMI, stroke, or CV Death 56 (7.18%) 41 (4.41%) 51 (5.90%) 0.048 39 (3.19%) 29 (2.57%) 33 (2.59%) 0.567
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
or when is complied in HTML
export2md(tabestr, strip = TRUE, first = TRUE,
header.background = "blue", header.color = "white", size=11)
Using compareGroups package you can compute Odds Ratios for transversal or case-control studies, or Hazard Ratios for cohort studies
- Example of case-control study: Odds Ratios
data(SNPs)
descrTable(casco ~ .-id, SNPs, show.ratio=TRUE, show.p.overall=FALSE)[1:4]--------Summary descriptives table by 'casco'---------
_______________________________________________________________
0 1 OR p.ratio
N=47 N=110
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
sex:
Male 21 (44.7%) 54 (49.1%) Ref. Ref.
Female 26 (55.3%) 56 (50.9%) 0.84 [0.42;1.67] 0.619
blood.pre 13.1 (0.88) 12.9 (1.03) 0.78 [0.55;1.11] 0.174
protein 39938 (19770) 44371 (24897) 1.00 [1.00;1.00] 0.280
snp10001:
CC 2 (4.26%) 10 (9.09%) Ref. Ref.
CT 21 (44.7%) 32 (29.1%) 0.33 [0.04;1.43] 0.147
TT 24 (51.1%) 68 (61.8%) 0.60 [0.08;2.55] 0.521
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
- Example of cohort study: Hazard Ratios
# create a Surv response:
predimed$tevent <- Surv(predimed$toevent, predimed$event=="Yes")
# perform descriptive table placeing the Surv object as the response (left side of ~):
tab <- descrTable(tevent ~ .-toevent-event, predimed, method=c(wth=2,p14=2),
hide.no="no", show.ratio=TRUE, show.p.overall=FALSE)
# print table on R console
print(tab, header.label=c("p.ratio"="p-value")) --------Summary descriptives table by 'tevent'---------
__________________________________________________________________________________________
No event Event HR p-value
N=6072 N=252
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Intervention group:
Control 1945 (32.0%) 97 (38.5%) Ref. Ref.
MedDiet + Nuts 2030 (33.4%) 70 (27.8%) 0.66 [0.48;0.89] 0.008
MedDiet + VOO 2097 (34.5%) 85 (33.7%) 0.70 [0.53;0.94] 0.018
Sex:
Male 2528 (41.6%) 151 (59.9%) Ref. Ref.
Female 3544 (58.4%) 101 (40.1%) 0.49 [0.38;0.63] <0.001
Age 66.9 (6.14) 69.4 (6.65) 1.06 [1.04;1.09] <0.001
Smoking:
Never 3778 (62.2%) 114 (45.2%) Ref. Ref.
Current 809 (13.3%) 49 (19.4%) 1.96 [1.40;2.74] <0.001
Former 1485 (24.5%) 89 (35.3%) 2.02 [1.53;2.67] <0.001
Body mass index 30.0 (3.81) 29.8 (3.92) 0.99 [0.96;1.02] 0.455
Waist circumference 100 (10.6) 102 (10.6) 1.02 [1.01;1.03] 0.003
Waist-to-height ratio 0.63 [0.58;0.67] 0.63 [0.59;0.68] 5.27 [0.83;33.6] 0.079
Hypertension 5025 (82.8%) 210 (83.3%) 1.10 [0.79;1.53] 0.578
Type-2 diabetes 2841 (46.8%) 161 (63.9%) 1.88 [1.46;2.44] <0.001
Dyslipidemia 4427 (72.9%) 151 (59.9%) 0.62 [0.49;0.80] <0.001
Family history of premature CHD 1378 (22.7%) 51 (20.2%) 0.93 [0.68;1.26] 0.640
Hormone-replacement therapy 96 (1.77%) 1 (0.45%) 0.29 [0.04;2.10] 0.223
MeDiet Adherence score 9.00 [7.00;10.0] 8.00 [7.00;10.0] 0.88 [0.83;0.94] <0.001
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
For those not familiar to R syntax, a Web User Interface (WUI) has been implemented using Shiny tools, which can be used remotely just accessing the compareGroups project website
Try the WUI compareGroups here
citation("compareGroups")To cite compareGroups in publications use:
Isaac Subirana, Hector Sanz, Joan Vila (2014). Building Bivariate Tables: The compareGroups Package for R. Journal of Statistical
Software, 57(12), 1-16. URL http://www.jstatsoft.org/v57/i12/.
A BibTeX entry for LaTeX users is
@Article{,
title = {Building Bivariate Tables: The {compareGroups} Package for {R}},
author = {Isaac Subirana and H\'ector Sanz and Joan Vila},
journal = {Journal of Statistical Software},
year = {2014},
volume = {57},
number = {12},
pages = {1--16},
url = {http://www.jstatsoft.org/v57/i12/},
}
Subirana, Isaac, Héctor Sanz, and Joan Vila. 2014. “Building Bivariate Tables: The compareGroups Package for R.” Journal of Statistical Software 57 (12): 1–16. http://www.jstatsoft.org/v57/i12/.