Table - Descriptive analysis
Minh Tri
2022-07-30
knitr::opts_chunk$set(
echo = TRUE,
message = TRUE,
warning = FALSE
)
### Create datable() with Filter:
options(DT.options = list(pageLength = 5,
lengthMenu = c(5, 10, 20, 50, 100), # Adjust option in Show entries
autoWidth = TRUE,
language = list(search = 'Filter:'))) Note: This analysis is used for my own study purpose. In this section, I will summerize some basic commands for making table based on several online sources.
R Packages used in this practice:
# Load the required library
library(tidyverse) # Data Wrangling## -- Attaching packages --------------------------------------- tidyverse 1.3.2 --
## v ggplot2 3.3.6 v purrr 0.3.4
## v tibble 3.1.8 v dplyr 1.0.10
## v tidyr 1.2.0 v stringr 1.4.1
## v readr 2.1.2 v forcats 0.5.2
## -- Conflicts ------------------------------------------ tidyverse_conflicts() --
## x dplyr::filter() masks stats::filter()
## x dplyr::lag() masks stats::lag()library(conflicted) # Dealing with conflict package
library(readxl) # Read csv file
library(DT) # For using datatable()
library(table1) # For create table list (table1)
library(ggpubr) # For combing plot ggarrange()
library(compareGroups) # For create table list (createTable)
library(gmodels) # For Chi-square test
Dealing with Conflicts
There is a lot of packages here, and sometimes individual functions are
in conflict. R’s default conflict resolution system gives precedence to
the most recently loaded package. This can make it hard to detect
conflicts, particularly when introduced by an update to an existing
package.
Using the code below helps the entire book run properly. You may or may
not need to look into the conflicted package for your work.
conflict_prefer("filter", "dplyr")## [conflicted] Will prefer dplyr::filter over any other packageconflict_prefer("select", "dplyr")## [conflicted] Will prefer dplyr::select over any other packageconflict_prefer("Predict", "rms")## [conflicted] Will prefer rms::Predict over any other packageconflict_prefer("impute_median", "simputation")## [conflicted] Will prefer simputation::impute_median over any other packageconflict_prefer("summarize", "dplyr")## [conflicted] Will prefer dplyr::summarize over any other package
Data used in this notes
fakestroke.csv
(Source: https://github.com/THOMASELOVE/432-data/blob/master/data/fakestroke.csv)
Loading the data, adjust the character of each column:
# Loading the data, adjust the character of each column
df <- read_excel("D:/Statistics/R/R data/fakestroke.xlsx",
col_types = c("text", "text", "numeric",
"text", "numeric", "text", "text",
"numeric", "numeric", "text", "numeric",
"text", "numeric", "numeric", "text",
"numeric", "numeric", "numeric"))
1 Example from Journal Article
1.1 Example 1 - New England Journal of Medicine
A typical Table 1 involves a group comparison, for example in this excerpt from Roy et al. (2008). This Table 1 describes a multi-center randomized clinical trial comparing two different approaches to caring for patients with heart failure and atrial fibrillation:
# To add the image to R markdown:
## {width=50%}:
### {height=30%}
The article provides percentages, means and standard deviations across groups, but note that it does not provide p values for the comparison of baseline characteristics. This is a common feature of NEJM reports on randomized clinical trials, where we anticipate that the two groups will be well matched at baseline.
1.2 Example 2: The MR CLEAN trial
Berkhemer et al. (2015) reported on the MR CLEAN trial, involving 500
patients with acute ischemic stroke caused by a proximal intracranial
arterial occlusion. The trial was conducted at 16 medical centers in the
Netherlands, where 233 were randomly assigned to the intervention
(intraarterial treatment plus usual care) and 267 to control (usual care
alone.)
Here’s the Table 1 from Berkhemer et al. (2015).
###  
2 Simulated fakestroke data
The fakestroke.csv file contains the following 18 variables for 500 patients:
# To make a table in R markdown: 1st row: header, 2nd row: Alignment; the remaining row: for content
## |Variable | Description |
## |:------- | :---------- |
## |studyid | Study ID # (z001 through z500) | | Variable | Description |
|---|---|
| studyid | Study ID # (z001 through z500) |
| trt | Treatment group (Intervention or Control) |
| age | Age in years |
| sex | Male or Female |
| nihss | NIH Stroke Scale Score (can range from 0-42; higher scores indicate more severe neurological deficits) |
| location | Stroke Location - Left or Right Hemisphere |
| hx.isch | History of Ischemic Stroke (Yes/No) |
| afib | Atrial Fibrillation (1 = Yes, 0 = No) |
| dm | Diabetes Mellitus (1 = Yes, 0 = No) |
| mrankin | Pre-stroke modified Rankin scale score (0, 1, 2 or > 2) indicating functional disability - complete range is 0 (no symptoms) to 6 (death) |
| sbp | Systolic blood pressure, in mm Hg |
| iv.altep | Treatment with IV alteplase (Yes/No) |
| time.iv | Time from stroke onset to start of IV alteplase (minutes) if iv.altep=Yes |
| aspects | Alberta Stroke Program Early Computed Tomography score, which measures extent of stroke from 0 - 10; higher scores indicate fewer early ischemic changes |
| ia.occlus | Intracranial arterial occlusion, based on vessel imaging - five categories3 |
| extra.ica | Extracranial ICA occlusion (1 = Yes, 0 = No) |
| time.rand | Time from stroke onset to study randomization, in minutes |
| time.punc | Time from stroke onset to groin puncture, in minutes (only if Intervention) |
A quick look at the simulated data in fakestroke
df## # A tibble: 500 x 18
## studyid trt age sex nihss locat~1 hx.isch afib dm mrankin sbp
## <chr> <chr> <dbl> <chr> <dbl> <chr> <chr> <dbl> <dbl> <chr> <dbl>
## 1 z001 Control 53 Male 21 Right No 0 0 2 127
## 2 z002 Interven~ 51 Male 23 Left No 1 0 0 137
## 3 z003 Control 68 Fema~ 11 Right No 0 0 0 138
## 4 z004 Control 28 Male 22 Left No 0 0 0 122
## 5 z005 Control 91 Male 24 Right No 0 0 0 162
## 6 z006 Control 34 Fema~ 18 Left No 0 0 2 166
## 7 z007 Interven~ 75 Male 25 Right No 0 0 0 140
## 8 z008 Control 89 Fema~ 18 Right No 0 0 0 157
## 9 z009 Control 75 Male 25 Left No 1 0 2 129
## 10 z010 Interven~ 26 Fema~ 27 Right No 0 0 0 143
## # ... with 490 more rows, 7 more variables: iv.altep <chr>, time.iv <dbl>,
## # aspects <dbl>, ia.occlus <chr>, extra.ica <dbl>, time.rand <dbl>,
## # time.punc <dbl>, and abbreviated variable name 1: locationstr(df)## tibble [500 x 18] (S3: tbl_df/tbl/data.frame)
## $ studyid : chr [1:500] "z001" "z002" "z003" "z004" ...
## $ trt : chr [1:500] "Control" "Intervention" "Control" "Control" ...
## $ age : num [1:500] 53 51 68 28 91 34 75 89 75 26 ...
## $ sex : chr [1:500] "Male" "Male" "Female" "Male" ...
## $ nihss : num [1:500] 21 23 11 22 24 18 25 18 25 27 ...
## $ location : chr [1:500] "Right" "Left" "Right" "Left" ...
## $ hx.isch : chr [1:500] "No" "No" "No" "No" ...
## $ afib : num [1:500] 0 1 0 0 0 0 0 0 1 0 ...
## $ dm : num [1:500] 0 0 0 0 0 0 0 0 0 0 ...
## $ mrankin : chr [1:500] "2" "0" "0" "0" ...
## $ sbp : num [1:500] 127 137 138 122 162 166 140 157 129 143 ...
## $ iv.altep : chr [1:500] "Yes" "Yes" "No" "Yes" ...
## $ time.iv : num [1:500] 63 68 NA 78 121 78 97 NA 49 99 ...
## $ aspects : num [1:500] 10 10 10 10 8 5 10 9 6 10 ...
## $ ia.occlus: chr [1:500] "M1" "M1" "ICA with M1" "ICA with M1" ...
## $ extra.ica: num [1:500] 0 1 1 1 0 0 0 0 0 0 ...
## $ time.rand: num [1:500] 139 118 178 160 214 154 122 147 271 141 ...
## $ time.punc: num [1:500] NA 281 NA NA NA NA 268 NA NA 259 ...summary(df)## studyid trt age sex
## Length:500 Length:500 Min. :23.00 Length:500
## Class :character Class :character 1st Qu.:55.00 Class :character
## Mode :character Mode :character Median :65.75 Mode :character
## Mean :64.71
## 3rd Qu.:76.00
## Max. :96.00
##
## nihss location hx.isch afib
## Min. :10.00 Length:500 Length:500 Min. :0.00
## 1st Qu.:14.00 Class :character Class :character 1st Qu.:0.00
## Median :18.00 Mode :character Mode :character Median :0.00
## Mean :18.03 Mean :0.27
## 3rd Qu.:22.00 3rd Qu.:1.00
## Max. :28.00 Max. :1.00
##
## dm mrankin sbp iv.altep
## Min. :0.000 Length:500 Min. : 78.0 Length:500
## 1st Qu.:0.000 Class :character 1st Qu.:128.5 Class :character
## Median :0.000 Mode :character Median :145.0 Mode :character
## Mean :0.126 Mean :145.5
## 3rd Qu.:0.000 3rd Qu.:162.5
## Max. :1.000 Max. :231.0
## NA's :1
## time.iv aspects ia.occlus extra.ica
## Min. : 42.00 Min. : 5.000 Length:500 Min. :0.0000
## 1st Qu.: 67.00 1st Qu.: 7.000 Class :character 1st Qu.:0.0000
## Median : 86.00 Median : 9.000 Mode :character Median :0.0000
## Mean : 92.64 Mean : 8.506 Mean :0.2906
## 3rd Qu.:115.00 3rd Qu.:10.000 3rd Qu.:1.0000
## Max. :218.00 Max. :10.000 Max. :1.0000
## NA's :55 NA's :4 NA's :1
## time.rand time.punc
## Min. :100.0 Min. :180
## 1st Qu.:151.2 1st Qu.:212
## Median :201.5 Median :260
## Mean :208.6 Mean :263
## 3rd Qu.:257.8 3rd Qu.:313
## Max. :360.0 Max. :360
## NA's :2 NA's :2673 Building table: compareGroups() function - preferable
Loading the data: fakestroke data
# Loading the data, adjust the character of each column
df <- read_excel("D:/Statistics/R/R data/fakestroke.xlsx",
col_types = c("text", "text", "numeric",
"text", "numeric", "text", "text",
"numeric", "numeric", "text", "numeric",
"text", "numeric", "numeric", "text",
"numeric", "numeric", "numeric"))
3.1 Step 1: Identify categorical or numeric variables
Before constructing a table, I must check numeric/ categorical
variables.
For example: in our data:
- Numeric variable: age; nihss; sbp; time.iv; aspects; time.rand;
time.punc.
- Character variable: sex; location; hx.isch; afib; dm; mrankin; iv.altep; ia.occlus; extra.ica.
- In the excel file, afib, dm, extra.ica are noted as
“1” and “0”.
Therefore, I must convert them back to “Yes-No” for character variables:
# Changing coding variable to categorical:
df$afib <- factor(df$afib , levels=0:1, labels=c("No", "Yes"))
df$dm <- factor(df$dm , levels=0:1, labels=c("No", "Yes"))
df$extra.ica <- factor(df$extra.ica, levels=0:1, labels=c("No", "Yes"))- Reorder the level of some character variables:
- trt: Intervention (baseline) , Control
- mrankin: 0, 1, 2, > 2.
- ia.occlus: rearrange the ia.occlus variable to the order presented in Berkhemer et al. (2015).
# Relevel factors:
df$trt = factor(df$trt, levels=c("Intervention", "Control"))
df$mrankin = factor(df$mrankin, levels=c("0", "1", "2", "> 2"))
df$ia.occlus = factor(df$ia.occlus, levels=c("Intracranial ICA", "ICA with M1",
"M1", "M2", "A1 or A2"))
# Check order of factor after relevel:
df$trt[1:2]## [1] Control Intervention
## Levels: Intervention Controldf$mrankin[1:2]## [1] 2 0
## Levels: 0 1 2 > 2df$ia.occlus[1:3]## [1] M1 M1 ICA with M1
## Levels: Intracranial ICA ICA with M1 M1 M2 A1 or A2
3.2 Step 2: First draft: compareGroups()
Our goal, then, is to take the data in fakestroke.csv and use it to generate a Table for the study that compares the 233 patients in the Intervention group to the 267 patients in the Control group, on all of the other variables (except study ID #) available.
I’ll use the compareGroups package of functions available in R to help me complete this task.
The command is as follow:
createTable( compareGroups( group ~ var1 + var2 + …., data = …
))
- var1; var2: numeric or categorical variables
** For numeric var: Mean (SD) or Median (Q1, Q3)
** For categorical var: arragne in order: count (%)
- group: name of the main group(s) you want to summerize/ put to header.
createTable( compareGroups( trt ~ age + sex + nihss + location + hx.isch + afib + dm +
mrankin + sbp + iv.altep + time.iv + aspects + ia.occlus +
extra.ica + time.rand + time.punc,
data = df ))##
## --------Summary descriptives table by 'trt'---------
##
## _______________________________________________________
## Intervention Control p.overall
## N=233 N=267
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
## age 63.9 (18.1) 65.4 (16.1) 0.347
## sex: 0.917
## Female 98 (42.1%) 110 (41.2%)
## Male 135 (57.9%) 157 (58.8%)
## nihss 18.0 (5.04) 18.1 (4.32) 0.790
## location: 0.111
## Left 116 (49.8%) 153 (57.3%)
## Right 117 (50.2%) 114 (42.7%)
## hx.isch: 0.335
## No 204 (87.6%) 242 (90.6%)
## Yes 29 (12.4%) 25 (9.36%)
## afib: 0.601
## No 167 (71.7%) 198 (74.2%)
## Yes 66 (28.3%) 69 (25.8%)
## dm: 1.000
## No 204 (87.6%) 233 (87.3%)
## Yes 29 (12.4%) 34 (12.7%)
## mrankin: 0.922
## 0 190 (81.5%) 214 (80.1%)
## 1 21 (9.01%) 29 (10.9%)
## 2 12 (5.15%) 13 (4.87%)
## > 2 10 (4.29%) 11 (4.12%)
## sbp 146 (26.0) 145 (24.4) 0.649
## iv.altep: 0.267
## No 30 (12.9%) 25 (9.36%)
## Yes 203 (87.1%) 242 (90.6%)
## time.iv 98.2 (45.5) 88.0 (26.0) 0.005
## aspects 8.35 (1.64) 8.65 (1.47) 0.034
## ia.occlus: 0.819
## Intracranial ICA 1 (0.43%) 3 (1.13%)
## ICA with M1 59 (25.3%) 75 (28.2%)
## M1 154 (66.1%) 165 (62.0%)
## M2 18 (7.73%) 21 (7.89%)
## A1 or A2 1 (0.43%) 2 (0.75%)
## extra.ica: 0.179
## No 158 (67.8%) 196 (73.7%)
## Yes 75 (32.2%) 70 (26.3%)
## time.rand 203 (57.3) 214 (70.3) 0.048
## time.punc 263 (54.2) . (.) .
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Discuss
The p-value here is just for normal distribution - continuous variables; not non-normal. Therefore, we must check the distribution of continuous variable - Step 3 or go directly to Step 4 !!
For numeric variable: Mean (SD) or Median (Q1-Q3)?
** Normal distribution: Mean (SD)
** Non-normal: Median (Q1-Q3)
3.3 Step 3: Testing normality - Shapiro–Wilk test (can skip this step)
3.3.1 Shapiro-Wilk test
Mechanism: it compares the scores in the sample to a normally distributed set of scores with the same mean and standard deviation. In compareGroups() function, we have the command to help us checking directly, so we can skip this step.
- Null hypothesis (p> 0.05): the data are normally
distributed.
- Alternative hypothesis (p <0.05): the data are not normally distributed.
Limitations: because with large sample sizes it is
very easy to get significant results from small deviations from
normality, and so a significant test doesn’t necessarily tell us whether
the deviation from normality is enough to bias any statistical
procedures that I apply to the data.
=> Plot your data (histogram / Q-Q plots) as well and try to make an
informed decision about the extent of non-normality (and the values of
skew and kurtosis.).
- For 1 variable that have many groups:
shapiro.test(variable)
- For many groups in 1 variable:
by(numeric variable, group/categorical variable, name of test)
3.3.2 Drawing Q-Q plot
We have 2 ways to command:
Package: ggplot2
- ggplot(data = , aes(sample = )) + geom_qq() + geom_qq_line() + labs( x = “Theoretical”, y = “Sample Quantiles -”)
Package: ggpubr
- ggqqplot(variable, ylab = ” “)
3.3.3 Practice
Let check the numeric variable: age; nihss; sbp; time.iv; aspects; time.rand; time.punc.
- age
# Calculate p-value
by(df$age, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.97333, p-value = 0.0002228
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.96996, p-value = 2.132e-05# Draw Q-Q plot for age
a = ggplot(data = df, aes(sample = age)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - Age")
# Draw Q-Q plot for Age - Intervention vS Control
b = ggplot(data = df, aes(sample = age)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - Age")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for Age",
color = "red", face = "bold", size = 12))
A quicker way to draw Q-Q plot:
ggqqplot(df$age, ylab = "age ")
- nihss
# Calculate p-value
by(df$nihss, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.94435, p-value = 9.121e-08
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.93951, p-value = 5.145e-09# Draw Q-Q plot for nihss
a = ggplot(data = df, aes(sample = nihss)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - nihss")
# Draw Q-Q plot for nihss - Intervention vS Control
b = ggplot(data = df, aes(sample = nihss)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - nihss")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for nihss",
color = "red", face = "bold", size = 12))
- sbp
# Calculate p-value
by(df$sbp, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.99584, p-value = 0.7869
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.99676, p-value = 0.8671# Draw Q-Q plot for sbp
a = ggplot(data = df, aes(sample = sbp)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - sbp")
# Draw Q-Q plot for sbp - Intervention vS Control
b = ggplot(data = df, aes(sample = sbp)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - sbp")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for sbp",
color = "red", face = "bold", size = 12))
- time.iv
# Calculate p-value
by(df$time.iv, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.88032, p-value = 1.31e-11
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.93338, p-value = 5.276e-09# Draw Q-Q plot for time.iv
a = ggplot(data = df, aes(sample = time.iv)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - time.iv")
# Draw Q-Q plot for time.iv - Intervention vS Control
b = ggplot(data = df, aes(sample = time.iv)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - time.iv")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for time.iv",
color = "red", face = "bold", size = 12))
- aspects
# Calculate p-value
by(df$aspects, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.85145, p-value = 3.364e-14
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.82188, p-value < 2.2e-16# Draw Q-Q plot for aspects
a = ggplot(data = df, aes(sample = aspects)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - aspects")
# Draw Q-Q plot for aspects - Intervention vS Control
b = ggplot(data = df, aes(sample = aspects)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - aspects")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for aspects",
color = "red", face = "bold", size = 12))
- time.rand
# Calculate p-value
by(df$time.rand, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.95684, p-value = 2.032e-06
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.92567, p-value = 2.703e-10# Draw Q-Q plot for time.rand
a = ggplot(data = df, aes(sample = time.rand)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - time.rand")
# Draw Q-Q plot for time.rand - Intervention vS Control
b = ggplot(data = df, aes(sample = time.rand)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - time.rand")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for time.rand",
color = "red", face = "bold", size = 12))
- time.punc.
# Because time.punc has values in only one group - Intervention
shapiro.test(df$time.punc)##
## Shapiro-Wilk normality test
##
## data: df$time.punc
## W = 0.93577, p-value = 1.435e-08# Draw Q-Q plot for time.punc
plot = ggplot(data = df, aes(sample = time.punc)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - time.punc")
annotate_figure(plot, bottom = text_grob("The Q-Q plot for time.punc",
color = "red", face = "bold", size = 12))
Discuss
Based on the p value + Q-Q plot for Intervention and Control groups:
sbp has p-value > 0.05 => normal distribution => Use Mean [SD].
Other variables has p-value < 0.05 => not normal distribution => Use Median-[Q1-Q3].
3.4 Step 4: Displaying different statistics for different variables:
By default continuous variables are analyzed as normal-distributed. When a table is built, continuous variables sbp will be described with mean and standard deviation. To change default options, e.g., age, nihss, time.iv, aspects, time.rand, time.punc used as non-normal distributed:
compareGroups(group variable ~ numeric/categorical variable, data = , method = c(numeric variable = 2))
Possible values in methods statement are:
- 1: forces analysis as normal-distributed
- 2: forces analysis as continuous non-normal
- 3: forces analysis as categorical
- NA: performs a Shapiro-Wilks test to decide between normal or non-normal (preferable).
If the method argument is stated as NA for a variable, then a Shapiro-Wilk test for normality is used to decide if the variable is normal or non-normal distributed.
In case, I already know about distribution of data (From step 3- normal or non-normal):
createTable( compareGroups( trt ~ age + sex + nihss + location + hx.isch + afib + dm +
mrankin + sbp + iv.altep + time.iv + aspects + ia.occlus +
extra.ica + time.rand + time.punc,
data = df,
method = c(age=2, nihss=2, sbp = 1, time.iv=2, aspects=2,
time.rand =2, time.punc =2)))##
## --------Summary descriptives table by 'trt'---------
##
## ________________________________________________________________
## Intervention Control p.overall
## N=233 N=267
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
## age 65.8 [54.5;76.0] 65.7 [55.8;76.2] 0.579
## sex: 0.917
## Female 98 (42.1%) 110 (41.2%)
## Male 135 (57.9%) 157 (58.8%)
## nihss 17.0 [14.0;21.0] 18.0 [14.0;22.0] 0.453
## location: 0.111
## Left 116 (49.8%) 153 (57.3%)
## Right 117 (50.2%) 114 (42.7%)
## hx.isch: 0.335
## No 204 (87.6%) 242 (90.6%)
## Yes 29 (12.4%) 25 (9.36%)
## afib: 0.601
## No 167 (71.7%) 198 (74.2%)
## Yes 66 (28.3%) 69 (25.8%)
## dm: 1.000
## No 204 (87.6%) 233 (87.3%)
## Yes 29 (12.4%) 34 (12.7%)
## mrankin: 0.922
## 0 190 (81.5%) 214 (80.1%)
## 1 21 (9.01%) 29 (10.9%)
## 2 12 (5.15%) 13 (4.87%)
## > 2 10 (4.29%) 11 (4.12%)
## sbp 146 (26.0) 145 (24.4) 0.649
## iv.altep: 0.267
## No 30 (12.9%) 25 (9.36%)
## Yes 203 (87.1%) 242 (90.6%)
## time.iv 85.0 [67.0;110] 87.0 [65.0;116] 0.596
## aspects 9.00 [7.00;10.0] 9.00 [8.00;10.0] 0.075
## ia.occlus: 0.819
## Intracranial ICA 1 (0.43%) 3 (1.13%)
## ICA with M1 59 (25.3%) 75 (28.2%)
## M1 154 (66.1%) 165 (62.0%)
## M2 18 (7.73%) 21 (7.89%)
## A1 or A2 1 (0.43%) 2 (0.75%)
## extra.ica: 0.179
## No 158 (67.8%) 196 (73.7%)
## Yes 75 (32.2%) 70 (26.3%)
## time.rand 204 [152;250] 196 [149;266] 0.251
## time.punc 260 [212;313] . [.;.] .
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
If I skip step 3 for normality checking, I can define method = c( numeric variable=NA)
createTable( compareGroups( trt ~ age + sex + nihss + location + hx.isch + afib + dm +
mrankin + sbp + iv.altep + time.iv + aspects + ia.occlus +
extra.ica + time.rand + time.punc,
data = df,
method = c(age=NA, nihss=NA, sbp = NA, time.iv=NA, aspects=NA,
time.rand =NA, time.punc =NA)))##
## --------Summary descriptives table by 'trt'---------
##
## ________________________________________________________________
## Intervention Control p.overall
## N=233 N=267
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
## age 65.8 [54.5;76.0] 65.7 [55.8;76.2] 0.579
## sex: 0.917
## Female 98 (42.1%) 110 (41.2%)
## Male 135 (57.9%) 157 (58.8%)
## nihss 17.0 [14.0;21.0] 18.0 [14.0;22.0] 0.453
## location: 0.111
## Left 116 (49.8%) 153 (57.3%)
## Right 117 (50.2%) 114 (42.7%)
## hx.isch: 0.335
## No 204 (87.6%) 242 (90.6%)
## Yes 29 (12.4%) 25 (9.36%)
## afib: 0.601
## No 167 (71.7%) 198 (74.2%)
## Yes 66 (28.3%) 69 (25.8%)
## dm: 1.000
## No 204 (87.6%) 233 (87.3%)
## Yes 29 (12.4%) 34 (12.7%)
## mrankin: 0.922
## 0 190 (81.5%) 214 (80.1%)
## 1 21 (9.01%) 29 (10.9%)
## 2 12 (5.15%) 13 (4.87%)
## > 2 10 (4.29%) 11 (4.12%)
## sbp 146 (26.0) 145 (24.4) 0.649
## iv.altep: 0.267
## No 30 (12.9%) 25 (9.36%)
## Yes 203 (87.1%) 242 (90.6%)
## time.iv 85.0 [67.0;110] 87.0 [65.0;116] 0.596
## aspects 9.00 [7.00;10.0] 9.00 [8.00;10.0] 0.075
## ia.occlus: 0.819
## Intracranial ICA 1 (0.43%) 3 (1.13%)
## ICA with M1 59 (25.3%) 75 (28.2%)
## M1 154 (66.1%) 165 (62.0%)
## M2 18 (7.73%) 21 (7.89%)
## A1 or A2 1 (0.43%) 2 (0.75%)
## extra.ica: 0.179
## No 158 (67.8%) 196 (73.7%)
## Yes 75 (32.2%) 70 (26.3%)
## time.rand 204 [152;250] 196 [149;266] 0.251
## time.punc 260 [212;313] . [.;.] .
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Discuss:
The 2 table in Step 4 are identical. Therefore, we can skip step 3 and define (method = c(variable=NA) directly.
3.5 Step 5: Odds Ratio
When the response variable is binary, the Odds Ratio (OR) can be printed in the final table. If the response variable is time-to-event, the Hazard Ratio (HR) can be printed instead.
- ref: This statement can be used to change the
reference category of row.
- ref.y: By default (ref.y =1 = 1st reference
category of column) when OR or HR are calculated, the reference
category for the response variable is the first. The reference category
could be changed using the ref.y statement.
https://cran.r-project.org/web/packages/compareGroups/vignettes/compareGroups_vignette.html
Note: This out put show OR of Control, Intervention is the refernce category. This code should be used for visualization only (default ref.y =1)
OR = compareGroups( trt ~ age + sex + nihss + location + hx.isch +
afib + dm + mrankin + sbp + iv.altep +
time.iv + aspects + ia.occlus + extra.ica +
time.rand + time.punc,
data = df,
method = c(age=NA, nihss=NA, sbp = NA, time.iv=NA,
aspects=NA, time.rand =NA, time.punc =NA),
ref = 1,
ref.y=1)
createTable(OR, show.ratio = TRUE)##
## --------Summary descriptives table by 'trt'---------
##
## _________________________________________________________________________________________
## Intervention Control OR p.ratio p.overall
## N=233 N=267
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
## age 65.8 [54.5;76.0] 65.7 [55.8;76.2] 1.01 [0.99;1.02] 0.342 0.579
## sex: 0.917
## Female 98 (42.1%) 110 (41.2%) Ref. Ref.
## Male 135 (57.9%) 157 (58.8%) 1.04 [0.72;1.48] 0.846
## nihss 17.0 [14.0;21.0] 18.0 [14.0;22.0] 1.01 [0.97;1.04] 0.787 0.453
## location: 0.111
## Left 116 (49.8%) 153 (57.3%) Ref. Ref.
## Right 117 (50.2%) 114 (42.7%) 0.74 [0.52;1.05] 0.094
## hx.isch: 0.335
## No 204 (87.6%) 242 (90.6%) Ref. Ref.
## Yes 29 (12.4%) 25 (9.36%) 0.73 [0.41;1.28] 0.273
## afib: 0.601
## No 167 (71.7%) 198 (74.2%) Ref. Ref.
## Yes 66 (28.3%) 69 (25.8%) 0.88 [0.59;1.31] 0.534
## dm: 1.000
## No 204 (87.6%) 233 (87.3%) Ref. Ref.
## Yes 29 (12.4%) 34 (12.7%) 1.03 [0.60;1.75] 0.925
## mrankin: 0.922
## 0 190 (81.5%) 214 (80.1%) Ref. Ref.
## 1 21 (9.01%) 29 (10.9%) 1.22 [0.68;2.25] 0.507
## 2 12 (5.15%) 13 (4.87%) 0.96 [0.42;2.20] 0.924
## > 2 10 (4.29%) 11 (4.12%) 0.98 [0.40;2.42] 0.956
## sbp 146 (26.0) 145 (24.4) 1.00 [0.99;1.01] 0.646 0.649
## iv.altep: 0.267
## No 30 (12.9%) 25 (9.36%) Ref. Ref.
## Yes 203 (87.1%) 242 (90.6%) 1.43 [0.81;2.53] 0.215
## time.iv 85.0 [67.0;110] 87.0 [65.0;116] 0.99 [0.99;1.00] 0.004 0.596
## aspects 9.00 [7.00;10.0] 9.00 [8.00;10.0] 1.13 [1.01;1.27] 0.033 0.075
## ia.occlus: 0.819
## Intracranial ICA 1 (0.43%) 3 (1.13%) Ref. Ref.
## ICA with M1 59 (25.3%) 75 (28.2%) 0.46 [0.02;4.10] 0.513
## M1 154 (66.1%) 165 (62.0%) 0.39 [0.01;3.40] 0.414
## M2 18 (7.73%) 21 (7.89%) 0.43 [0.01;4.06] 0.484
## A1 or A2 1 (0.43%) 2 (0.75%) 0.71 [0.01;38.5] 0.857
## extra.ica: 0.179
## No 158 (67.8%) 196 (73.7%) Ref. Ref.
## Yes 75 (32.2%) 70 (26.3%) 0.75 [0.51;1.11] 0.152
## time.rand 204 [152;250] 196 [149;266] 1.00 [1.00;1.01] 0.051 0.251
## time.punc 260 [212;313] . [.;.] . [.;.] . .
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Note: To show OR of Intervention, define using ref.y =2:
OR <- compareGroups(trt ~ age + sex + nihss + location + hx.isch + afib + dm +
mrankin + sbp + iv.altep + time.iv + aspects + ia.occlus +
extra.ica + time.rand + time.punc,
data = df,
method = c(age=NA, nihss=NA, sbp = NA, time.iv=NA,
aspects=NA, time.rand =NA, time.punc =NA),
ref = c(sex = 2, location = 2),
ref.y = 2
)
createTable(OR, show.ratio = TRUE)##
## --------Summary descriptives table by 'trt'---------
##
## _________________________________________________________________________________________
## Intervention Control OR p.ratio p.overall
## N=233 N=267
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
## age 65.8 [54.5;76.0] 65.7 [55.8;76.2] 1.00 [1.01;0.98] 0.342 0.579
## sex: 0.917
## Female 98 (42.1%) 110 (41.2%) 1.04 [0.72;1.48] 0.846
## Male 135 (57.9%) 157 (58.8%) Ref. Ref.
## nihss 17.0 [14.0;21.0] 18.0 [14.0;22.0] 0.99 [1.03;0.96] 0.787 0.453
## location: 0.111
## Left 116 (49.8%) 153 (57.3%) 0.74 [0.52;1.05] 0.094
## Right 117 (50.2%) 114 (42.7%) Ref. Ref.
## hx.isch: 0.335
## No 204 (87.6%) 242 (90.6%) Ref. Ref.
## Yes 29 (12.4%) 25 (9.36%) 1.37 [0.78;2.44] 0.273
## afib: 0.601
## No 167 (71.7%) 198 (74.2%) Ref. Ref.
## Yes 66 (28.3%) 69 (25.8%) 1.13 [0.76;1.69] 0.534
## dm: 1.000
## No 204 (87.6%) 233 (87.3%) Ref. Ref.
## Yes 29 (12.4%) 34 (12.7%) 0.97 [0.57;1.66] 0.925
## mrankin: 0.922
## 0 190 (81.5%) 214 (80.1%) Ref. Ref.
## 1 21 (9.01%) 29 (10.9%) 0.82 [0.44;1.48] 0.507
## 2 12 (5.15%) 13 (4.87%) 1.04 [0.45;2.37] 0.924
## > 2 10 (4.29%) 11 (4.12%) 1.03 [0.41;2.51] 0.956
## sbp 146 (26.0) 145 (24.4) 1.00 [1.01;0.99] 0.646 0.649
## iv.altep: 0.267
## No 30 (12.9%) 25 (9.36%) Ref. Ref.
## Yes 203 (87.1%) 242 (90.6%) 0.70 [0.40;1.23] 0.215
## time.iv 85.0 [67.0;110] 87.0 [65.0;116] 1.01 [1.01;1.00] 0.004 0.596
## aspects 9.00 [7.00;10.0] 9.00 [8.00;10.0] 0.88 [0.99;0.79] 0.033 0.075
## ia.occlus: 0.819
## Intracranial ICA 1 (0.43%) 3 (1.13%) Ref. Ref.
## ICA with M1 59 (25.3%) 75 (28.2%) 2.16 [0.24;63.1] 0.513
## M1 154 (66.1%) 165 (62.0%) 2.57 [0.29;74.2] 0.414
## M2 18 (7.73%) 21 (7.89%) 2.33 [0.25;71.2] 0.484
## A1 or A2 1 (0.43%) 2 (0.75%) 1.41 [0.03;76.8] 0.857
## extra.ica: 0.179
## No 158 (67.8%) 196 (73.7%) Ref. Ref.
## Yes 75 (32.2%) 70 (26.3%) 1.33 [0.90;1.96] 0.152
## time.rand 204 [152;250] 196 [149;266] 1.00 [1.00;0.99] 0.051 0.251
## time.punc 260 [212;313] . [.;.] . [.;.] . .
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Checking the OR with Crosstable() function:
- sex
CrossTable(df$sex, df$trt, fisher = TRUE, chisq = TRUE,expected = TRUE, sresid = TRUE, format = "SPSS")##
## Cell Contents
## |-------------------------|
## | Count |
## | Expected Values |
## | Chi-square contribution |
## | Row Percent |
## | Column Percent |
## | Total Percent |
## | Std Residual |
## |-------------------------|
##
## Total Observations in Table: 500
##
## | df$trt
## df$sex | Intervention | Control | Row Total |
## -------------|--------------|--------------|--------------|
## Female | 98 | 110 | 208 |
## | 96.928 | 111.072 | |
## | 0.012 | 0.010 | |
## | 47.115% | 52.885% | 41.600% |
## | 42.060% | 41.199% | |
## | 19.600% | 22.000% | |
## | 0.109 | -0.102 | |
## -------------|--------------|--------------|--------------|
## Male | 135 | 157 | 292 |
## | 136.072 | 155.928 | |
## | 0.008 | 0.007 | |
## | 46.233% | 53.767% | 58.400% |
## | 57.940% | 58.801% | |
## | 27.000% | 31.400% | |
## | -0.092 | 0.086 | |
## -------------|--------------|--------------|--------------|
## Column Total | 233 | 267 | 500 |
## | 46.600% | 53.400% | |
## -------------|--------------|--------------|--------------|
##
##
## Statistics for All Table Factors
##
##
## Pearson's Chi-squared test
## ------------------------------------------------------------
## Chi^2 = 0.03801773 d.f. = 1 p = 0.8454075
##
## Pearson's Chi-squared test with Yates' continuity correction
## ------------------------------------------------------------
## Chi^2 = 0.01082402 d.f. = 1 p = 0.9171387
##
##
## Fisher's Exact Test for Count Data
## ------------------------------------------------------------
## Sample estimate odds ratio: 1.03604
##
## Alternative hypothesis: true odds ratio is not equal to 1
## p = 0.8561188
## 95% confidence interval: 0.7138417 1.50323
##
## Alternative hypothesis: true odds ratio is less than 1
## p = 0.6126348
## 95% confidence interval: 0 1.419597
##
## Alternative hypothesis: true odds ratio is greater than 1
## p = 0.4584408
## 95% confidence interval: 0.7559892 Inf
##
##
##
## Minimum expected frequency: 96.928- location
CrossTable(df$location, df$trt, fisher = TRUE, chisq = TRUE,expected = TRUE, sresid = TRUE, format = "SPSS")##
## Cell Contents
## |-------------------------|
## | Count |
## | Expected Values |
## | Chi-square contribution |
## | Row Percent |
## | Column Percent |
## | Total Percent |
## | Std Residual |
## |-------------------------|
##
## Total Observations in Table: 500
##
## | df$trt
## df$location | Intervention | Control | Row Total |
## -------------|--------------|--------------|--------------|
## Left | 116 | 153 | 269 |
## | 125.354 | 143.646 | |
## | 0.698 | 0.609 | |
## | 43.123% | 56.877% | 53.800% |
## | 49.785% | 57.303% | |
## | 23.200% | 30.600% | |
## | -0.835 | 0.780 | |
## -------------|--------------|--------------|--------------|
## Right | 117 | 114 | 231 |
## | 107.646 | 123.354 | |
## | 0.813 | 0.709 | |
## | 50.649% | 49.351% | 46.200% |
## | 50.215% | 42.697% | |
## | 23.400% | 22.800% | |
## | 0.902 | -0.842 | |
## -------------|--------------|--------------|--------------|
## Column Total | 233 | 267 | 500 |
## | 46.600% | 53.400% | |
## -------------|--------------|--------------|--------------|
##
##
## Statistics for All Table Factors
##
##
## Pearson's Chi-squared test
## ------------------------------------------------------------
## Chi^2 = 2.829263 d.f. = 1 p = 0.09256083
##
## Pearson's Chi-squared test with Yates' continuity correction
## ------------------------------------------------------------
## Chi^2 = 2.534881 d.f. = 1 p = 0.1113553
##
##
## Fisher's Exact Test for Count Data
## ------------------------------------------------------------
## Sample estimate odds ratio: 0.7391921
##
## Alternative hypothesis: true odds ratio is not equal to 1
## p = 0.105602
## 95% confidence interval: 0.5108763 1.068214
##
## Alternative hypothesis: true odds ratio is less than 1
## p = 0.05565712
## 95% confidence interval: 0 1.00947
##
## Alternative hypothesis: true odds ratio is greater than 1
## p = 0.9618255
## 95% confidence interval: 0.5408447 Inf
##
##
##
## Minimum expected frequency: 107.646Discuss
p.ratio: we can ignore this parameter.
** For numeric variable: It reflects the p-value of parametric test.
** For categorical variable: it is p-value of Chi-squared test.p.overall:
** For numeric variable: it reflects p-value of parametric or non-parametric tests, depend on what we define the method = c(variable=NA)
** For categorical variable: it reflects p-value of Chi-squared with Yates’ continuity correction.OR: results of Fisher exact test.
OR of Fisher exact test similar to OR created by compareGroups()
3.6 Step 6: Exporting the table (for CompareGroups() only)
Tables can be exported to CSV, HTML, LaTeX, PDF, Markdown, Word or Excel:
- export2csv(restab, file=‘table1.csv’), exports to CSV format
- export2html(restab, file=‘table1.html’), exports to HTML
format
- export2pdf(restab, file=‘table1.pdf’), exports to PDF format
- export2word(restab, file=‘table1.docx’), exports to Word
format
- export2xls(restab, file=‘table1.xlsx’), exports to Excel format
restab = createTable(OR, show.ratio = TRUE)
export2csv(restab, file='table1.csv')
3.7 Extra: Dealing with missing value
Many times, it is important to be aware of the missingness contained in each variable, possibly by groups. Althought “available” table shows the number of the non-missing values for each row-variable and in each group, it would be desirable to test whether the frequency of non-available data is different between groups. For this porpose, a new function has been implemented in the compareGroups package, which is called missingTable. This function applies to both compareGroups and createTable class objects. This last option is useful when the table is already created.
The following code missingTable() is preferable as it can show missing value in both numeric and categorical variables.
# Identify both numeric and categorical variables:
OR <- compareGroups(trt ~ age + sex + nihss + location + hx.isch + afib + dm +
mrankin + sbp + iv.altep + time.iv + aspects + ia.occlus +
extra.ica + time.rand + time.punc,
data = df,
method = c(age=NA, nihss=NA, sbp = NA, time.iv=NA,
aspects=NA, time.rand =NA, time.punc =NA),
ref = c(sex = 2, location = 2),
ref.y = 2,
)
missingTable(OR)##
## --------Missingness table by 'trt'---------
##
## ___________________________________________
## Intervention Control p.overall
## N=233 N=267
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
## age 0 (0.00%) 0 (0.00%) .
## sex 0 (0.00%) 0 (0.00%) .
## nihss 0 (0.00%) 0 (0.00%) .
## location 0 (0.00%) 0 (0.00%) .
## hx.isch 0 (0.00%) 0 (0.00%) .
## afib 0 (0.00%) 0 (0.00%) .
## dm 0 (0.00%) 0 (0.00%) .
## mrankin 0 (0.00%) 0 (0.00%) .
## sbp 0 (0.00%) 1 (0.37%) 1.000
## iv.altep 0 (0.00%) 0 (0.00%) .
## time.iv 30 (12.9%) 25 (9.36%) 0.267
## aspects 0 (0.00%) 4 (1.50%) 0.127
## ia.occlus 0 (0.00%) 1 (0.37%) 1.000
## extra.ica 0 (0.00%) 1 (0.37%) 1.000
## time.rand 2 (0.86%) 0 (0.00%) 0.217
## time.punc 0 (0.00%) 267 (100%) <0.001
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯Discuss
p.overall: show if the difference is significant between Intervention and Control.
** Except time.punc, other variables p>0.05: The missing value was not significant different between 2 groups.
** time.punc: p<0.05: because we only measure in Intervention group, no apply in Control group.
4 Building table: table1() function
Loading the data: fakestroke data
# Loading the data, adjust the character of each column
df <- read_excel("D:/Statistics/R/R data/fakestroke.xlsx",
col_types = c("text", "text", "numeric",
"text", "numeric", "text", "text",
"numeric", "numeric", "text", "numeric",
"text", "numeric", "numeric", "text",
"numeric", "numeric", "numeric")) 4.1 Step 1: Identify categorical or numeric variables
Before constructing a table, I must check numeric/ categorical
variables.
For example: in our data:
- Numeric variable: age; nihss; sbp; time.iv; aspects; time.rand;
time.punc.
- Character variable: sex; location; hx.isch; afib; dm; mrankin; iv.altep; ia.occlus; extra.ica.
- In the excel file, afib, dm, extra.ica are noted as
“1” and “0”.
Therefore, I must convert them back to “Yes-No” for character variables:
# Changing coding variable to categorical:
df$afib <- factor(df$afib , levels=0:1, labels=c("No", "Yes"))
df$dm <- factor(df$dm , levels=0:1, labels=c("No", "Yes"))
df$extra.ica <- factor(df$extra.ica, levels=0:1, labels=c("No", "Yes"))- Reorder the level of some character variables:
- trt: Intervention (baseline) , Control
- mrankin: 0, 1, 2, > 2.
- ia.occlus: rearrange the ia.occlus variable to the order presented in Berkhemer et al. (2015).
# Relevel factors:
df$trt = factor(df$trt, levels=c("Intervention", "Control"))
df$mrankin = factor(df$mrankin, levels=c("0", "1", "2", "> 2"))
df$ia.occlus = factor(df$ia.occlus, levels=c("Intracranial ICA", "ICA with M1",
"M1", "M2", "A1 or A2"))
# Check order of factor after relevel:
df$trt[1:2]## [1] Control Intervention
## Levels: Intervention Controldf$mrankin[1:2]## [1] 2 0
## Levels: 0 1 2 > 2df$ia.occlus[1:3]## [1] M1 M1 ICA with M1
## Levels: Intracranial ICA ICA with M1 M1 M2 A1 or A2
4.2 Step 2: First draft: table1()
Our goal, then, is to take the data in fakestroke.csv and use it to generate a Table for the study that compares the 233 patients in the Intervention group to the 267 patients in the Control group, on all of the other variables (except study ID #) available.
I’ll use the table1 package of functions available in R to help me complete this task.
The command is as follow:
table1( ~ var1 + var2 + … | group, data = …)
table1( ~ . | group, data = …)
- var1; var2: numeric or categorical variables
** For numeric var: Mean (SD) – Median (Q1, Q3)
** For categorical var: arragne in order: count (%)
- group: name of the main group(s) you want to summerize/ put to header.
Using built-in styles for table1():
- zebra: alternating shaded and unshaded rows (zebra stripes)
- grid: show all grid lines
- shade: shade the header row(s) in gray
- times: use a serif font
- center: center all columns, including the first which contains the row labels
Adjust the components of table:
- render.continuous = c(.=“Mean (SD)”, .=“Median [Q1, Q3]”)
- render.continuous = c(.=“Mean (SD)”, .=“Median [Mix, Max]”)
More adjustment can be found here: https://cran.r-project.org/web/packages/table1/vignettes/table1-examples.html
table1( ~ age + sex + nihss + location + hx.isch + afib + dm + mrankin +
sbp + iv.altep + time.iv + aspects + ia.occlus + extra.ica + time.rand +
time.punc | trt, data = df, topclass="Rtable1-zebra",
render.continuous = c(.="Mean (SD)", .="Median [Q1, Q3]"),
overall="Total")| Intervention (N=233) |
Control (N=267) |
Total (N=500) |
|
|---|---|---|---|
| age | |||
| Mean (SD) | 63.9 (18.1) | 65.4 (16.1) | 64.7 (17.1) |
| Median [Q1, Q3] | 65.8 [54.5, 76.0] | 65.7 [55.8, 76.2] | 65.8 [55.0, 76.0] |
| sex | |||
| Female | 98 (42.1%) | 110 (41.2%) | 208 (41.6%) |
| Male | 135 (57.9%) | 157 (58.8%) | 292 (58.4%) |
| nihss | |||
| Mean (SD) | 18.0 (5.04) | 18.1 (4.32) | 18.0 (4.67) |
| Median [Q1, Q3] | 17.0 [14.0, 21.0] | 18.0 [14.0, 22.0] | 18.0 [14.0, 22.0] |
| location | |||
| Left | 116 (49.8%) | 153 (57.3%) | 269 (53.8%) |
| Right | 117 (50.2%) | 114 (42.7%) | 231 (46.2%) |
| hx.isch | |||
| No | 204 (87.6%) | 242 (90.6%) | 446 (89.2%) |
| Yes | 29 (12.4%) | 25 (9.4%) | 54 (10.8%) |
| afib | |||
| No | 167 (71.7%) | 198 (74.2%) | 365 (73.0%) |
| Yes | 66 (28.3%) | 69 (25.8%) | 135 (27.0%) |
| dm | |||
| No | 204 (87.6%) | 233 (87.3%) | 437 (87.4%) |
| Yes | 29 (12.4%) | 34 (12.7%) | 63 (12.6%) |
| mrankin | |||
| 0 | 190 (81.5%) | 214 (80.1%) | 404 (80.8%) |
| 1 | 21 (9.0%) | 29 (10.9%) | 50 (10.0%) |
| 2 | 12 (5.2%) | 13 (4.9%) | 25 (5.0%) |
| > 2 | 10 (4.3%) | 11 (4.1%) | 21 (4.2%) |
| sbp | |||
| Mean (SD) | 146 (26.0) | 145 (24.4) | 145 (25.1) |
| Median [Q1, Q3] | 146 [129, 164] | 145 [128, 161] | 145 [129, 163] |
| Missing | 0 (0%) | 1 (0.4%) | 1 (0.2%) |
| iv.altep | |||
| No | 30 (12.9%) | 25 (9.4%) | 55 (11.0%) |
| Yes | 203 (87.1%) | 242 (90.6%) | 445 (89.0%) |
| time.iv | |||
| Mean (SD) | 98.2 (45.5) | 88.0 (26.0) | 92.6 (36.5) |
| Median [Q1, Q3] | 85.0 [67.0, 110] | 87.0 [65.0, 116] | 86.0 [67.0, 115] |
| Missing | 30 (12.9%) | 25 (9.4%) | 55 (11.0%) |
| aspects | |||
| Mean (SD) | 8.35 (1.64) | 8.65 (1.47) | 8.51 (1.56) |
| Median [Q1, Q3] | 9.00 [7.00, 10.0] | 9.00 [8.00, 10.0] | 9.00 [7.00, 10.0] |
| Missing | 0 (0%) | 4 (1.5%) | 4 (0.8%) |
| ia.occlus | |||
| Intracranial ICA | 1 (0.4%) | 3 (1.1%) | 4 (0.8%) |
| ICA with M1 | 59 (25.3%) | 75 (28.1%) | 134 (26.8%) |
| M1 | 154 (66.1%) | 165 (61.8%) | 319 (63.8%) |
| M2 | 18 (7.7%) | 21 (7.9%) | 39 (7.8%) |
| A1 or A2 | 1 (0.4%) | 2 (0.7%) | 3 (0.6%) |
| Missing | 0 (0%) | 1 (0.4%) | 1 (0.2%) |
| extra.ica | |||
| No | 158 (67.8%) | 196 (73.4%) | 354 (70.8%) |
| Yes | 75 (32.2%) | 70 (26.2%) | 145 (29.0%) |
| Missing | 0 (0%) | 1 (0.4%) | 1 (0.2%) |
| time.rand | |||
| Mean (SD) | 203 (57.3) | 214 (70.3) | 209 (64.8) |
| Median [Q1, Q3] | 204 [152, 250] | 196 [149, 266] | 202 [151, 258] |
| Missing | 2 (0.9%) | 0 (0%) | 2 (0.4%) |
| time.punc | |||
| Mean (SD) | 263 (54.2) | NA | 263 (54.2) |
| Median [Q1, Q3] | 260 [212, 313] | NA | 260 [212, 313] |
| Missing | 0 (0%) | 267 (100%) | 267 (53.4%) |
Discuss:
This table provides full detail Mean-Median for numeric variable; Count-% for character variable, missing value.
For numeric variable: Mean (SD) or Median (Q1-Q3)?
** Normal distribution: Mean (SD)
** Non-normal: Median (Q1-Q3)
4.3 Step 3: Testing normality - Shapiro–Wilk test
Mechanism: it compares the scores in the sample to a normally distributed set of scores with the same mean and standard deviation.
- If the test is non-significant (p > .05) = normal
distribution.
- If the test is significant (p < .05) = non-normal distribution.
Limitations: because with large sample sizes it is
very easy to get significant results from small deviations from
normality, and so a significant test doesn’t necessarily tell us whether
the deviation from normality is enough to bias any statistical
procedures that I apply to the data.
=> Plot your data (histogram / Q-Q plots) as well and try to make an
informed decision about the extent of non-normality (and the values of
skew and kurtosis.).
- For 1 variable that have many groups:
shapiro.test(variable)
- For many groups in 1 variable:
by(numeric variable, group/categorical variable, name of test)
Let check the numeric variable: age; nihss; sbp; time.iv; aspects; time.rand; time.punc.
- age
# Calculate p-value
by(df$age, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.97333, p-value = 0.0002228
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.96996, p-value = 2.132e-05# Draw Q-Q plot for age
a = ggplot(data = df, aes(sample = age)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - Age")
# Draw Q-Q plot for Age - Intervention vS Control
b = ggplot(data = df, aes(sample = age)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - Age")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for Age",
color = "red", face = "bold", size = 12))
- nihss
# Calculate p-value
by(df$nihss, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.94435, p-value = 9.121e-08
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.93951, p-value = 5.145e-09# Draw Q-Q plot for nihss
a = ggplot(data = df, aes(sample = nihss)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - nihss")
# Draw Q-Q plot for nihss - Intervention vS Control
b = ggplot(data = df, aes(sample = nihss)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - nihss")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for nihss",
color = "red", face = "bold", size = 12))
- sbp
# Calculate p-value
by(df$sbp, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.99584, p-value = 0.7869
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.99676, p-value = 0.8671# Draw Q-Q plot for sbp
a = ggplot(data = df, aes(sample = sbp)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - sbp")
# Draw Q-Q plot for sbp - Intervention vS Control
b = ggplot(data = df, aes(sample = sbp)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - sbp")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for sbp",
color = "red", face = "bold", size = 12))
- time.iv
# Calculate p-value
by(df$time.iv, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.88032, p-value = 1.31e-11
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.93338, p-value = 5.276e-09# Draw Q-Q plot for time.iv
a = ggplot(data = df, aes(sample = time.iv)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - time.iv")
# Draw Q-Q plot for time.iv - Intervention vS Control
b = ggplot(data = df, aes(sample = time.iv)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - time.iv")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for time.iv",
color = "red", face = "bold", size = 12))
- aspects
# Calculate p-value
by(df$aspects, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.85145, p-value = 3.364e-14
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.82188, p-value < 2.2e-16# Draw Q-Q plot for aspects
a = ggplot(data = df, aes(sample = aspects)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - aspects")
# Draw Q-Q plot for aspects - Intervention vS Control
b = ggplot(data = df, aes(sample = aspects)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - aspects")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for aspects",
color = "red", face = "bold", size = 12))
- time.rand
# Calculate p-value
by(df$time.rand, df$trt, shapiro.test)## df$trt: Intervention
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.95684, p-value = 2.032e-06
##
## ------------------------------------------------------------
## df$trt: Control
##
## Shapiro-Wilk normality test
##
## data: dd[x, ]
## W = 0.92567, p-value = 2.703e-10# Draw Q-Q plot for time.rand
a = ggplot(data = df, aes(sample = time.rand)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - time.rand")
# Draw Q-Q plot for time.rand - Intervention vS Control
b = ggplot(data = df, aes(sample = time.rand)) +
geom_qq() +
geom_qq_line() +
facet_wrap(~trt) +
labs( x = "Theoretical", y = "Sample Quantiles - time.rand")
# Combine 2 plots:
plot = ggarrange(a,b,
ncol=2, nrow=1,
common.legend = FALSE,
legend="right",
labels = c("A","B"))
annotate_figure(plot, bottom = text_grob("The Q-Q plot for time.rand",
color = "red", face = "bold", size = 12))
- time.punc.
# Because time.punc has values in only one group - Intervention
shapiro.test(df$time.punc)##
## Shapiro-Wilk normality test
##
## data: df$time.punc
## W = 0.93577, p-value = 1.435e-08# Draw Q-Q plot for time.punc
plot = ggplot(data = df, aes(sample = time.punc)) +
geom_qq() +
geom_qq_line() +
labs( x = "Theoretical", y = "Sample Quantiles - time.punc")
annotate_figure(plot, bottom = text_grob("The Q-Q plot for time.punc",
color = "red", face = "bold", size = 12))
Discuss
Based on the p value + Q-Q plot for Intervention and Control groups:
sbp has p-value > 0.05 => normal distribution => Use Mean [SD].
Other variables has p-value < 0.05 => not normal distribution => Use Median-[Q1-Q3].
4.4 Step 4: Displaying different statistics for different variables: table1()
# Function for numeric variable
rndr <- function(x, name, ...) {
if (!is.numeric(x)) return(render.categorical.default(x))
what <- switch(name,
age = "Median [Q1, Q3]",
nihss = "Median [Q1, Q3]",
sbp = "Mean (SD)",
time.iv = "Median [Q1, Q3]",
aspects = "Median [Q1, Q3]",
time.rand = "Median [Q1, Q3]",
time.punc = "Median [Q1, Q3]")
parse.abbrev.render.code(c("", what))(x)
}
table1( ~ age + sex + nihss + location + hx.isch + afib + dm + mrankin +
sbp + iv.altep + time.iv + aspects + ia.occlus + extra.ica + time.rand +
time.punc | trt, data = df,
render=rndr,
overall="Total")| Intervention (N=233) |
Control (N=267) |
Total (N=500) |
|
|---|---|---|---|
| age | |||
| Median [Q1, Q3] | 65.8 [54.5, 76.0] | 65.7 [55.8, 76.2] | 65.8 [55.0, 76.0] |
| sex | |||
| Female | 98 (42.1%) | 110 (41.2%) | 208 (41.6%) |
| Male | 135 (57.9%) | 157 (58.8%) | 292 (58.4%) |
| nihss | |||
| Median [Q1, Q3] | 17.0 [14.0, 21.0] | 18.0 [14.0, 22.0] | 18.0 [14.0, 22.0] |
| location | |||
| Left | 116 (49.8%) | 153 (57.3%) | 269 (53.8%) |
| Right | 117 (50.2%) | 114 (42.7%) | 231 (46.2%) |
| hx.isch | |||
| No | 204 (87.6%) | 242 (90.6%) | 446 (89.2%) |
| Yes | 29 (12.4%) | 25 (9.4%) | 54 (10.8%) |
| afib | |||
| No | 167 (71.7%) | 198 (74.2%) | 365 (73.0%) |
| Yes | 66 (28.3%) | 69 (25.8%) | 135 (27.0%) |
| dm | |||
| No | 204 (87.6%) | 233 (87.3%) | 437 (87.4%) |
| Yes | 29 (12.4%) | 34 (12.7%) | 63 (12.6%) |
| mrankin | |||
| 0 | 190 (81.5%) | 214 (80.1%) | 404 (80.8%) |
| 1 | 21 (9.0%) | 29 (10.9%) | 50 (10.0%) |
| 2 | 12 (5.2%) | 13 (4.9%) | 25 (5.0%) |
| > 2 | 10 (4.3%) | 11 (4.1%) | 21 (4.2%) |
| sbp | |||
| Mean (SD) | 146 (26.0) | 145 (24.4) | 145 (25.1) |
| iv.altep | |||
| No | 30 (12.9%) | 25 (9.4%) | 55 (11.0%) |
| Yes | 203 (87.1%) | 242 (90.6%) | 445 (89.0%) |
| time.iv | |||
| Median [Q1, Q3] | 85.0 [67.0, 110] | 87.0 [65.0, 116] | 86.0 [67.0, 115] |
| aspects | |||
| Median [Q1, Q3] | 9.00 [7.00, 10.0] | 9.00 [8.00, 10.0] | 9.00 [7.00, 10.0] |
| ia.occlus | |||
| Intracranial ICA | 1 (0.4%) | 3 (1.1%) | 4 (0.8%) |
| ICA with M1 | 59 (25.3%) | 75 (28.1%) | 134 (26.8%) |
| M1 | 154 (66.1%) | 165 (61.8%) | 319 (63.8%) |
| M2 | 18 (7.7%) | 21 (7.9%) | 39 (7.8%) |
| A1 or A2 | 1 (0.4%) | 2 (0.7%) | 3 (0.6%) |
| extra.ica | |||
| No | 158 (67.8%) | 196 (73.4%) | 354 (70.8%) |
| Yes | 75 (32.2%) | 70 (26.2%) | 145 (29.0%) |
| time.rand | |||
| Median [Q1, Q3] | 204 [152, 250] | 196 [149, 266] | 202 [151, 258] |
| time.punc | |||
| Median [Q1, Q3] | 260 [212, 313] | NA | 260 [212, 313] |
Discuss:
With new table, I don’t display missing value in extra.ica, ia.occlus in Intervention group. (Total observations= 266, not 267).
Compare to the image of Berkhemer et al 2015 (as below), our table from fakestroke data display all the essential elements.
4.5 Step 5: Exporting the table (for table1)
a= table1( ~ age + sex + nihss + location + hx.isch + afib + dm + mrankin +
sbp + iv.altep + time.iv + aspects + ia.occlus + extra.ica + time.rand +
time.punc | trt, data = df,
render=rndr,
overall="Total")
write.table (a, "test_file.csv", col.names = T, row.names=F, append= T, sep=',')
4.6 Extra: Adding extra column: p-value (not recommend)
A user asked if it was possible to add a column to the table showing
the p-value associated with a univariate test for differences in each
variable across strata. This can be accomplished using the
extra.col feature.
However, I personally prefer compareGroups function as
they can calculate non-normal distributed data.
Next, I create a function to compute the p-value for continuous or categorical variables.
- t-test: continuous variables
- Chi-squared: categorical variables
#Writing function for p-value column (numeric - t-test; categorical - Chi-squared test)
pvalue <- function(x, ...) {
x <- x[-length(x)] # Remove "overall" group
# Construct vectors of data y, and groups (strata) g
y <- unlist(x)
g <- factor(rep(1:length(x), times=sapply(x, length)))
if (is.numeric(y)) {
# For numeric variables, perform a standard 2-sample t-test
p <- t.test(y ~ g)$p.value
} else {
# For categorical variables, perform a chi-squared test of independence
p <- chisq.test(table(y, g))$p.value
}
# Format the p-value, using an HTML entity for the less-than sign.
# The initial empty string places the output on the line below the variable label.
c("", sub("<", "<", format.pval(p, digits=3, eps=0.001)))
}
# Discard "time.punc" because it has value in "Intervention" only - cannot calculate:
table1( ~ age + sex + nihss + location + hx.isch + afib +
dm + mrankin + sbp + iv.altep + time.iv + aspects +
ia.occlus + extra.ica + time.rand | trt,
data = df,
render=rndr,
overall="Total",
extra.col=list(`Valor-p`=pvalue))| Intervention (N=233) |
Control (N=267) |
Total (N=500) |
Valor-p | |
|---|---|---|---|---|
| age | ||||
| 0.347 | ||||
| Median [Q1, Q3] | 65.8 [54.5, 76.0] | 65.7 [55.8, 76.2] | 65.8 [55.0, 76.0] | |
| sex | ||||
| Female | 98 (42.1%) | 110 (41.2%) | 208 (41.6%) | 0.917 |
| Male | 135 (57.9%) | 157 (58.8%) | 292 (58.4%) | |
| nihss | ||||
| 0.79 | ||||
| Median [Q1, Q3] | 17.0 [14.0, 21.0] | 18.0 [14.0, 22.0] | 18.0 [14.0, 22.0] | |
| location | ||||
| Left | 116 (49.8%) | 153 (57.3%) | 269 (53.8%) | 0.111 |
| Right | 117 (50.2%) | 114 (42.7%) | 231 (46.2%) | |
| hx.isch | ||||
| No | 204 (87.6%) | 242 (90.6%) | 446 (89.2%) | 0.335 |
| Yes | 29 (12.4%) | 25 (9.4%) | 54 (10.8%) | |
| afib | ||||
| No | 167 (71.7%) | 198 (74.2%) | 365 (73.0%) | 0.601 |
| Yes | 66 (28.3%) | 69 (25.8%) | 135 (27.0%) | |
| dm | ||||
| No | 204 (87.6%) | 233 (87.3%) | 437 (87.4%) | 1 |
| Yes | 29 (12.4%) | 34 (12.7%) | 63 (12.6%) | |
| mrankin | ||||
| 0 | 190 (81.5%) | 214 (80.1%) | 404 (80.8%) | 0.922 |
| 1 | 21 (9.0%) | 29 (10.9%) | 50 (10.0%) | |
| 2 | 12 (5.2%) | 13 (4.9%) | 25 (5.0%) | |
| > 2 | 10 (4.3%) | 11 (4.1%) | 21 (4.2%) | |
| sbp | ||||
| 0.649 | ||||
| Mean (SD) | 146 (26.0) | 145 (24.4) | 145 (25.1) | |
| iv.altep | ||||
| No | 30 (12.9%) | 25 (9.4%) | 55 (11.0%) | 0.267 |
| Yes | 203 (87.1%) | 242 (90.6%) | 445 (89.0%) | |
| time.iv | ||||
| 0.00471 | ||||
| Median [Q1, Q3] | 85.0 [67.0, 110] | 87.0 [65.0, 116] | 86.0 [67.0, 115] | |
| aspects | ||||
| 0.0338 | ||||
| Median [Q1, Q3] | 9.00 [7.00, 10.0] | 9.00 [8.00, 10.0] | 9.00 [7.00, 10.0] | |
| ia.occlus | ||||
| Intracranial ICA | 1 (0.4%) | 3 (1.1%) | 4 (0.8%) | 0.795 |
| ICA with M1 | 59 (25.3%) | 75 (28.1%) | 134 (26.8%) | |
| M1 | 154 (66.1%) | 165 (61.8%) | 319 (63.8%) | |
| M2 | 18 (7.7%) | 21 (7.9%) | 39 (7.8%) | |
| A1 or A2 | 1 (0.4%) | 2 (0.7%) | 3 (0.6%) | |
| extra.ica | ||||
| No | 158 (67.8%) | 196 (73.4%) | 354 (70.8%) | 0.179 |
| Yes | 75 (32.2%) | 70 (26.2%) | 145 (29.0%) | |
| time.rand | ||||
| 0.0475 | ||||
| Median [Q1, Q3] | 204 [152, 250] | 196 [149, 266] | 202 [151, 258] |
OR I can use ANOVA instead of t-test (if outcome variable has more than 2 groups):
- ANOVA: continuous variables
- Chi-squared: categorical variables
https://github.com/benjaminrich/table1/issues/52
#Writing function for p-value column (numeric - ANOVA; categorical - Chi-squared test)
pvalue <- function(x, ...) {
x <- x[-length(x)] # Remove "overall" group
# Construct vectors of data y, and groups (strata) g
y <- unlist(x)
g <- factor(rep(1:length(x), times=sapply(x, length)))
if (is.numeric(y)) {
# For numeric variables, perform an ANOVA
p <- summary(aov(y ~ g))[[1]][["Pr(>F)"]][1]
} else {
# For categorical variables, perform a chi-squared test of independence
p <- chisq.test(table(y, g))$p.value
}
# Format the p-value, using an HTML entity for the less-than sign.
# The initial empty string places the output on the line below the variable label.
c("", sub("<", "<", format.pval(p, digits=3, eps=0.001)))
}
# Discard "time.punc" because it has value in "Intervention" only - cannot calculate:
table1( ~ age + sex + nihss + location + hx.isch + afib +
dm + mrankin + sbp + iv.altep + time.iv + aspects +
ia.occlus + extra.ica + time.rand | trt,
data = df,
render=rndr,
overall="Total",
extra.col=list(`Valor-p`=pvalue))| Intervention (N=233) |
Control (N=267) |
Total (N=500) |
Valor-p | |
|---|---|---|---|---|
| age | ||||
| 0.343 | ||||
| Median [Q1, Q3] | 65.8 [54.5, 76.0] | 65.7 [55.8, 76.2] | 65.8 [55.0, 76.0] | |
| sex | ||||
| Female | 98 (42.1%) | 110 (41.2%) | 208 (41.6%) | 0.917 |
| Male | 135 (57.9%) | 157 (58.8%) | 292 (58.4%) | |
| nihss | ||||
| 0.787 | ||||
| Median [Q1, Q3] | 17.0 [14.0, 21.0] | 18.0 [14.0, 22.0] | 18.0 [14.0, 22.0] | |
| location | ||||
| Left | 116 (49.8%) | 153 (57.3%) | 269 (53.8%) | 0.111 |
| Right | 117 (50.2%) | 114 (42.7%) | 231 (46.2%) | |
| hx.isch | ||||
| No | 204 (87.6%) | 242 (90.6%) | 446 (89.2%) | 0.335 |
| Yes | 29 (12.4%) | 25 (9.4%) | 54 (10.8%) | |
| afib | ||||
| No | 167 (71.7%) | 198 (74.2%) | 365 (73.0%) | 0.601 |
| Yes | 66 (28.3%) | 69 (25.8%) | 135 (27.0%) | |
| dm | ||||
| No | 204 (87.6%) | 233 (87.3%) | 437 (87.4%) | 1 |
| Yes | 29 (12.4%) | 34 (12.7%) | 63 (12.6%) | |
| mrankin | ||||
| 0 | 190 (81.5%) | 214 (80.1%) | 404 (80.8%) | 0.922 |
| 1 | 21 (9.0%) | 29 (10.9%) | 50 (10.0%) | |
| 2 | 12 (5.2%) | 13 (4.9%) | 25 (5.0%) | |
| > 2 | 10 (4.3%) | 11 (4.1%) | 21 (4.2%) | |
| sbp | ||||
| 0.647 | ||||
| Mean (SD) | 146 (26.0) | 145 (24.4) | 145 (25.1) | |
| iv.altep | ||||
| No | 30 (12.9%) | 25 (9.4%) | 55 (11.0%) | 0.267 |
| Yes | 203 (87.1%) | 242 (90.6%) | 445 (89.0%) | |
| time.iv | ||||
| 0.00307 | ||||
| Median [Q1, Q3] | 85.0 [67.0, 110] | 87.0 [65.0, 116] | 86.0 [67.0, 115] | |
| aspects | ||||
| 0.0327 | ||||
| Median [Q1, Q3] | 9.00 [7.00, 10.0] | 9.00 [8.00, 10.0] | 9.00 [7.00, 10.0] | |
| ia.occlus | ||||
| Intracranial ICA | 1 (0.4%) | 3 (1.1%) | 4 (0.8%) | 0.795 |
| ICA with M1 | 59 (25.3%) | 75 (28.1%) | 134 (26.8%) | |
| M1 | 154 (66.1%) | 165 (61.8%) | 319 (63.8%) | |
| M2 | 18 (7.7%) | 21 (7.9%) | 39 (7.8%) | |
| A1 or A2 | 1 (0.4%) | 2 (0.7%) | 3 (0.6%) | |
| extra.ica | ||||
| No | 158 (67.8%) | 196 (73.4%) | 354 (70.8%) | 0.179 |
| Yes | 75 (32.2%) | 70 (26.2%) | 145 (29.0%) | |
| time.rand | ||||
| 0.0508 | ||||
| Median [Q1, Q3] | 204 [152, 250] | 196 [149, 266] | 202 [151, 258] |
Discuss
I must exclude the time.punc because it has value only in “Intervention” => cannot calculate p-value.
For character variable: Chi-squared or Fisher?
** Expected frequencies > 5: Chi-squared
** Expected frequencies < 5: Fisher
For numeric variable: t-test/ANOVA or Wilcoxon rank-sum test?
** t-test and ANOVA test show nearly similar results. t-test should be used for sbp because outcome variable (trt) has 2 groups.
** Wilcoxon rank-sum test should applied to all other numeric variables - because they are non-normal distributed. (calculate outside or using compareGroups)
** Let compare t.test vs Wilcoxon rank-sum test: sbp, age, nihss :sbp - normal distribution
# For sbp variable
t.test(sbp ~ trt, data =df, paired = FALSE, var.equal= FALSE)##
## Welch Two Sample t-test
##
## data: sbp by trt
## t = 0.45607, df = 478.62, p-value = 0.6485
## alternative hypothesis: true difference in means between group Intervention and group Control is not equal to 0
## 95 percent confidence interval:
## -3.420249 5.487854
## sample estimates:
## mean in group Intervention mean in group Control
## 146.0300 144.9962wilcox.test(sbp ~ trt, data = df, pair = FALSE)##
## Wilcoxon rank sum test with continuity correction
##
## data: sbp by trt
## W = 32039, p-value = 0.5137
## alternative hypothesis: true location shift is not equal to 0- age - non-normal distribiution
# For AGE variable
t.test(age ~ trt, data =df, paired = FALSE, var.equal= FALSE)##
## Welch Two Sample t-test
##
## data: age by trt
## t = -0.94205, df = 468.4, p-value = 0.3467
## alternative hypothesis: true difference in means between group Intervention and group Control is not equal to 0
## 95 percent confidence interval:
## -4.480512 1.576662
## sample estimates:
## mean in group Intervention mean in group Control
## 63.93047 65.38240wilcox.test(age ~ trt, data = df, pair = FALSE)##
## Wilcoxon rank sum test with continuity correction
##
## data: age by trt
## W = 30211, p-value = 0.5788
## alternative hypothesis: true location shift is not equal to 0- nhiss - non-normal distribiution
# For nihss variable
t.test(nihss ~ trt, data =df, paired = FALSE, var.equal= FALSE)##
## Welch Two Sample t-test
##
## data: nihss by trt
## t = -0.26692, df = 459.95, p-value = 0.7896
## alternative hypothesis: true difference in means between group Intervention and group Control is not equal to 0
## 95 percent confidence interval:
## -0.9448105 0.7188376
## sample estimates:
## mean in group Intervention mean in group Control
## 17.96567 18.07865wilcox.test(nihss ~ trt, data = df, pair = FALSE)##
## Wilcoxon rank sum test with continuity correction
##
## data: nihss by trt
## W = 29900, p-value = 0.4533
## alternative hypothesis: true location shift is not equal to 0
Discuss:
There are differences between parametric (t-test) and non-parametric test (Wilcoxon rank sum test).
=> We must check the Normality assumption, Homogeneity of variance to know which test we should apply.
If we want to know p-value, should use compareGroups() function. It is more optimal than table1().
4.7 More comparisons: table with trt*sex variable
More format style can be found here: https://cran.r-project.org/web/packages/table1/vignettes/table1-examples.html
In this section, I temp to display a table based on trt x sex. The step by step analysis can be followed as what I already showed previously:
- Step 1: Identify categorical variables / numeric variables.
- Step 2: First draft: table1()
- Step 3: Testing normality
- Step 4: Displaying different statistics for different variables: table1().
table1( ~ age + nihss + location + hx.isch + afib + dm + mrankin +
sbp + iv.altep + time.iv + aspects + ia.occlus + extra.ica + time.rand +
time.punc | trt*sex, data = df,
render.continuous = c(.="Mean (SD)", .="Median [Q1, Q3]"),
overall="Total")Intervention |
Control |
Total |
||||
|---|---|---|---|---|---|---|
| Female (N=98) |
Male (N=135) |
Female (N=110) |
Male (N=157) |
Female (N=208) |
Male (N=292) |
|
| age | ||||||
| Mean (SD) | 65.0 (18.9) | 63.2 (17.5) | 66.2 (17.2) | 64.8 (15.3) | 65.6 (18.0) | 64.0 (16.4) |
| Median [Q1, Q3] | 66.0 [54.3, 80.5] | 65.0 [54.8, 75.0] | 66.5 [56.0, 80.8] | 65.0 [55.5, 75.0] | 66.0 [55.0, 81.0] | 65.0 [55.0, 75.0] |
| nihss | ||||||
| Mean (SD) | 18.3 (5.44) | 17.7 (4.74) | 17.6 (4.16) | 18.4 (4.41) | 17.9 (4.81) | 18.1 (4.57) |
| Median [Q1, Q3] | 17.0 [14.0, 23.5] | 17.0 [14.0, 21.0] | 18.0 [14.0, 21.8] | 18.0 [14.0, 22.0] | 17.0 [14.0, 22.0] | 18.0 [14.0, 22.0] |
| location | ||||||
| Left | 46 (46.9%) | 70 (51.9%) | 64 (58.2%) | 89 (56.7%) | 110 (52.9%) | 159 (54.5%) |
| Right | 52 (53.1%) | 65 (48.1%) | 46 (41.8%) | 68 (43.3%) | 98 (47.1%) | 133 (45.5%) |
| hx.isch | ||||||
| No | 80 (81.6%) | 124 (91.9%) | 106 (96.4%) | 136 (86.6%) | 186 (89.4%) | 260 (89.0%) |
| Yes | 18 (18.4%) | 11 (8.1%) | 4 (3.6%) | 21 (13.4%) | 22 (10.6%) | 32 (11.0%) |
| afib | ||||||
| No | 69 (70.4%) | 98 (72.6%) | 84 (76.4%) | 114 (72.6%) | 153 (73.6%) | 212 (72.6%) |
| Yes | 29 (29.6%) | 37 (27.4%) | 26 (23.6%) | 43 (27.4%) | 55 (26.4%) | 80 (27.4%) |
| dm | ||||||
| No | 84 (85.7%) | 120 (88.9%) | 99 (90.0%) | 134 (85.4%) | 183 (88.0%) | 254 (87.0%) |
| Yes | 14 (14.3%) | 15 (11.1%) | 11 (10.0%) | 23 (14.6%) | 25 (12.0%) | 38 (13.0%) |
| mrankin | ||||||
| 0 | 80 (81.6%) | 110 (81.5%) | 91 (82.7%) | 123 (78.3%) | 171 (82.2%) | 233 (79.8%) |
| 1 | 12 (12.2%) | 9 (6.7%) | 9 (8.2%) | 20 (12.7%) | 21 (10.1%) | 29 (9.9%) |
| 2 | 4 (4.1%) | 8 (5.9%) | 8 (7.3%) | 5 (3.2%) | 12 (5.8%) | 13 (4.5%) |
| > 2 | 2 (2.0%) | 8 (5.9%) | 2 (1.8%) | 9 (5.7%) | 4 (1.9%) | 17 (5.8%) |
| sbp | ||||||
| Mean (SD) | 148 (26.0) | 144 (26.0) | 141 (25.9) | 148 (23.0) | 145 (26.1) | 146 (24.4) |
| Median [Q1, Q3] | 150 [133, 163] | 143 [127, 164] | 141 [123, 157] | 146 [131, 162] | 145 [126, 162] | 146 [129, 163] |
| Missing | 0 (0%) | 0 (0%) | 1 (0.9%) | 0 (0%) | 1 (0.5%) | 0 (0%) |
| iv.altep | ||||||
| No | 13 (13.3%) | 17 (12.6%) | 11 (10.0%) | 14 (8.9%) | 24 (11.5%) | 31 (10.6%) |
| Yes | 85 (86.7%) | 118 (87.4%) | 99 (90.0%) | 143 (91.1%) | 184 (88.5%) | 261 (89.4%) |
| time.iv | ||||||
| Mean (SD) | 102 (44.2) | 95.5 (46.4) | 90.2 (26.3) | 86.4 (25.8) | 95.6 (36.1) | 90.5 (36.8) |
| Median [Q1, Q3] | 90.0 [68.0, 128] | 83.0 [64.8, 108] | 89.0 [67.5, 117] | 85.0 [63.0, 113] | 90.0 [68.0, 118] | 83.0 [64.0, 110] |
| Missing | 13 (13.3%) | 17 (12.6%) | 11 (10.0%) | 14 (8.9%) | 24 (11.5%) | 31 (10.6%) |
| aspects | ||||||
| Mean (SD) | 8.33 (1.67) | 8.36 (1.62) | 8.67 (1.38) | 8.63 (1.53) | 8.51 (1.53) | 8.50 (1.58) |
| Median [Q1, Q3] | 9.00 [7.00, 10.0] | 9.00 [7.00, 10.0] | 9.00 [8.00, 10.0] | 9.00 [8.00, 10.0] | 9.00 [8.00, 10.0] | 9.00 [7.00, 10.0] |
| Missing | 0 (0%) | 0 (0%) | 0 (0%) | 4 (2.5%) | 0 (0%) | 4 (1.4%) |
| ia.occlus | ||||||
| Intracranial ICA | 0 (0%) | 1 (0.7%) | 1 (0.9%) | 2 (1.3%) | 1 (0.5%) | 3 (1.0%) |
| ICA with M1 | 21 (21.4%) | 38 (28.1%) | 30 (27.3%) | 45 (28.7%) | 51 (24.5%) | 83 (28.4%) |
| M1 | 68 (69.4%) | 86 (63.7%) | 68 (61.8%) | 97 (61.8%) | 136 (65.4%) | 183 (62.7%) |
| M2 | 8 (8.2%) | 10 (7.4%) | 9 (8.2%) | 12 (7.6%) | 17 (8.2%) | 22 (7.5%) |
| A1 or A2 | 1 (1.0%) | 0 (0%) | 1 (0.9%) | 1 (0.6%) | 2 (1.0%) | 1 (0.3%) |
| Missing | 0 (0%) | 0 (0%) | 1 (0.9%) | 0 (0%) | 1 (0.5%) | 0 (0%) |
| extra.ica | ||||||
| No | 70 (71.4%) | 88 (65.2%) | 77 (70.0%) | 119 (75.8%) | 147 (70.7%) | 207 (70.9%) |
| Yes | 28 (28.6%) | 47 (34.8%) | 32 (29.1%) | 38 (24.2%) | 60 (28.8%) | 85 (29.1%) |
| Missing | 0 (0%) | 0 (0%) | 1 (0.9%) | 0 (0%) | 1 (0.5%) | 0 (0%) |
| time.rand | ||||||
| Mean (SD) | 201 (55.2) | 203 (59.0) | 206 (62.6) | 219 (75.0) | 204 (59.2) | 212 (68.4) |
| Median [Q1, Q3] | 200 [158, 244] | 205 [148, 255] | 192 [152, 252] | 208 [149, 274] | 196 [155, 250] | 206 [149, 268] |
| Missing | 2 (2.0%) | 0 (0%) | 0 (0%) | 0 (0%) | 2 (1.0%) | 0 (0%) |
| time.punc | ||||||
| Mean (SD) | 261 (55.4) | 264 (53.6) | NA | NA | 261 (55.4) | 264 (53.6) |
| Median [Q1, Q3] | 253 [208, 319] | 265 [217, 312] | NA | NA | 253 [208, 319] | 265 [217, 312] |
| Missing | 0 (0%) | 0 (0%) | 110 (100%) | 157 (100%) | 110 (52.9%) | 157 (53.8%) |
5 Summary
- For table1():
table1( ~ age + sex + nihss + location + hx.isch + afib + dm + mrankin +
sbp + iv.altep + time.iv + aspects + ia.occlus + extra.ica + time.rand +
time.punc | trt, data = df,
render=rndr,
overall="Total")| Intervention (N=233) |
Control (N=267) |
Total (N=500) |
|
|---|---|---|---|
| age | |||
| Median [Q1, Q3] | 65.8 [54.5, 76.0] | 65.7 [55.8, 76.2] | 65.8 [55.0, 76.0] |
| sex | |||
| Female | 98 (42.1%) | 110 (41.2%) | 208 (41.6%) |
| Male | 135 (57.9%) | 157 (58.8%) | 292 (58.4%) |
| nihss | |||
| Median [Q1, Q3] | 17.0 [14.0, 21.0] | 18.0 [14.0, 22.0] | 18.0 [14.0, 22.0] |
| location | |||
| Left | 116 (49.8%) | 153 (57.3%) | 269 (53.8%) |
| Right | 117 (50.2%) | 114 (42.7%) | 231 (46.2%) |
| hx.isch | |||
| No | 204 (87.6%) | 242 (90.6%) | 446 (89.2%) |
| Yes | 29 (12.4%) | 25 (9.4%) | 54 (10.8%) |
| afib | |||
| No | 167 (71.7%) | 198 (74.2%) | 365 (73.0%) |
| Yes | 66 (28.3%) | 69 (25.8%) | 135 (27.0%) |
| dm | |||
| No | 204 (87.6%) | 233 (87.3%) | 437 (87.4%) |
| Yes | 29 (12.4%) | 34 (12.7%) | 63 (12.6%) |
| mrankin | |||
| 0 | 190 (81.5%) | 214 (80.1%) | 404 (80.8%) |
| 1 | 21 (9.0%) | 29 (10.9%) | 50 (10.0%) |
| 2 | 12 (5.2%) | 13 (4.9%) | 25 (5.0%) |
| > 2 | 10 (4.3%) | 11 (4.1%) | 21 (4.2%) |
| sbp | |||
| Mean (SD) | 146 (26.0) | 145 (24.4) | 145 (25.1) |
| iv.altep | |||
| No | 30 (12.9%) | 25 (9.4%) | 55 (11.0%) |
| Yes | 203 (87.1%) | 242 (90.6%) | 445 (89.0%) |
| time.iv | |||
| Median [Q1, Q3] | 85.0 [67.0, 110] | 87.0 [65.0, 116] | 86.0 [67.0, 115] |
| aspects | |||
| Median [Q1, Q3] | 9.00 [7.00, 10.0] | 9.00 [8.00, 10.0] | 9.00 [7.00, 10.0] |
| ia.occlus | |||
| Intracranial ICA | 1 (0.4%) | 3 (1.1%) | 4 (0.8%) |
| ICA with M1 | 59 (25.3%) | 75 (28.1%) | 134 (26.8%) |
| M1 | 154 (66.1%) | 165 (61.8%) | 319 (63.8%) |
| M2 | 18 (7.7%) | 21 (7.9%) | 39 (7.8%) |
| A1 or A2 | 1 (0.4%) | 2 (0.7%) | 3 (0.6%) |
| extra.ica | |||
| No | 158 (67.8%) | 196 (73.4%) | 354 (70.8%) |
| Yes | 75 (32.2%) | 70 (26.2%) | 145 (29.0%) |
| time.rand | |||
| Median [Q1, Q3] | 204 [152, 250] | 196 [149, 266] | 202 [151, 258] |
| time.punc | |||
| Median [Q1, Q3] | 260 [212, 313] | NA | 260 [212, 313] |
- For compareGroup():
createTable(compareGroups( trt ~ age + sex + nihss + location + hx.isch + afib + dm +
mrankin + sbp + iv.altep + time.iv + aspects + ia.occlus +
extra.ica + time.rand + time.punc, data = df,
method = c(age=NA, nihss=NA, sbp = NA, time.iv=NA, aspects=NA,
time.rand =NA, time.punc =NA)))##
## --------Summary descriptives table by 'trt'---------
##
## ________________________________________________________________
## Intervention Control p.overall
## N=233 N=267
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
## age 65.8 [54.5;76.0] 65.7 [55.8;76.2] 0.579
## sex: 0.917
## Female 98 (42.1%) 110 (41.2%)
## Male 135 (57.9%) 157 (58.8%)
## nihss 17.0 [14.0;21.0] 18.0 [14.0;22.0] 0.453
## location: 0.111
## Left 116 (49.8%) 153 (57.3%)
## Right 117 (50.2%) 114 (42.7%)
## hx.isch: 0.335
## No 204 (87.6%) 242 (90.6%)
## Yes 29 (12.4%) 25 (9.36%)
## afib: 0.601
## No 167 (71.7%) 198 (74.2%)
## Yes 66 (28.3%) 69 (25.8%)
## dm: 1.000
## No 204 (87.6%) 233 (87.3%)
## Yes 29 (12.4%) 34 (12.7%)
## mrankin: 0.922
## 0 190 (81.5%) 214 (80.1%)
## 1 21 (9.01%) 29 (10.9%)
## 2 12 (5.15%) 13 (4.87%)
## > 2 10 (4.29%) 11 (4.12%)
## sbp 146 (26.0) 145 (24.4) 0.649
## iv.altep: 0.267
## No 30 (12.9%) 25 (9.36%)
## Yes 203 (87.1%) 242 (90.6%)
## time.iv 85.0 [67.0;110] 87.0 [65.0;116] 0.596
## aspects 9.00 [7.00;10.0] 9.00 [8.00;10.0] 0.075
## ia.occlus: 0.819
## Intracranial ICA 1 (0.43%) 3 (1.13%)
## ICA with M1 59 (25.3%) 75 (28.2%)
## M1 154 (66.1%) 165 (62.0%)
## M2 18 (7.73%) 21 (7.89%)
## A1 or A2 1 (0.43%) 2 (0.75%)
## extra.ica: 0.179
## No 158 (67.8%) 196 (73.7%)
## Yes 75 (32.2%) 70 (26.3%)
## time.rand 204 [152;250] 196 [149;266] 0.251
## time.punc 260 [212;313] . [.;.] .
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯OR <- compareGroups(trt ~ age + sex + nihss + location + hx.isch + afib + dm +
mrankin + sbp + iv.altep + time.iv + aspects + ia.occlus +
extra.ica + time.rand + time.punc,
data = df,
method = c(age=NA, nihss=NA, sbp = NA, time.iv=NA,
aspects=NA, time.rand =NA, time.punc =NA),
ref = c(sex = 2, location = 2),
ref.y = 2
)
createTable(OR, show.ratio = TRUE)##
## --------Summary descriptives table by 'trt'---------
##
## _________________________________________________________________________________________
## Intervention Control OR p.ratio p.overall
## N=233 N=267
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
## age 65.8 [54.5;76.0] 65.7 [55.8;76.2] 1.00 [1.01;0.98] 0.342 0.579
## sex: 0.917
## Female 98 (42.1%) 110 (41.2%) 1.04 [0.72;1.48] 0.846
## Male 135 (57.9%) 157 (58.8%) Ref. Ref.
## nihss 17.0 [14.0;21.0] 18.0 [14.0;22.0] 0.99 [1.03;0.96] 0.787 0.453
## location: 0.111
## Left 116 (49.8%) 153 (57.3%) 0.74 [0.52;1.05] 0.094
## Right 117 (50.2%) 114 (42.7%) Ref. Ref.
## hx.isch: 0.335
## No 204 (87.6%) 242 (90.6%) Ref. Ref.
## Yes 29 (12.4%) 25 (9.36%) 1.37 [0.78;2.44] 0.273
## afib: 0.601
## No 167 (71.7%) 198 (74.2%) Ref. Ref.
## Yes 66 (28.3%) 69 (25.8%) 1.13 [0.76;1.69] 0.534
## dm: 1.000
## No 204 (87.6%) 233 (87.3%) Ref. Ref.
## Yes 29 (12.4%) 34 (12.7%) 0.97 [0.57;1.66] 0.925
## mrankin: 0.922
## 0 190 (81.5%) 214 (80.1%) Ref. Ref.
## 1 21 (9.01%) 29 (10.9%) 0.82 [0.44;1.48] 0.507
## 2 12 (5.15%) 13 (4.87%) 1.04 [0.45;2.37] 0.924
## > 2 10 (4.29%) 11 (4.12%) 1.03 [0.41;2.51] 0.956
## sbp 146 (26.0) 145 (24.4) 1.00 [1.01;0.99] 0.646 0.649
## iv.altep: 0.267
## No 30 (12.9%) 25 (9.36%) Ref. Ref.
## Yes 203 (87.1%) 242 (90.6%) 0.70 [0.40;1.23] 0.215
## time.iv 85.0 [67.0;110] 87.0 [65.0;116] 1.01 [1.01;1.00] 0.004 0.596
## aspects 9.00 [7.00;10.0] 9.00 [8.00;10.0] 0.88 [0.99;0.79] 0.033 0.075
## ia.occlus: 0.819
## Intracranial ICA 1 (0.43%) 3 (1.13%) Ref. Ref.
## ICA with M1 59 (25.3%) 75 (28.2%) 2.16 [0.24;63.1] 0.513
## M1 154 (66.1%) 165 (62.0%) 2.57 [0.29;74.2] 0.414
## M2 18 (7.73%) 21 (7.89%) 2.33 [0.25;71.2] 0.484
## A1 or A2 1 (0.43%) 2 (0.75%) 1.41 [0.03;76.8] 0.857
## extra.ica: 0.179
## No 158 (67.8%) 196 (73.7%) Ref. Ref.
## Yes 75 (32.2%) 70 (26.3%) 1.33 [0.90;1.96] 0.152
## time.rand 204 [152;250] 196 [149;266] 1.00 [1.00;0.99] 0.051 0.251
## time.punc 260 [212;313] . [.;.] . [.;.] . .
## ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯- Should use both commands:
** table1(): Show the “Total” column.
** compareGroups() - preferable: Show the “p-overall” column.
6 References
- thomaselove.github.io/432-notes
- https://cran.r-project.org/web/packages/table1/vignettes/table1-examples.html
- https://github.com/benjaminrich/table1/issues/52
- Berkhemer, Olvert A., Puck S. S. Fransen, Debbie Buemer, et al. 2015. “A Randomized Trial of Intraarterial Treatment for Acute Ischemic Stroke.” New England Journal of Medicine 372: 11–20. http://www.nejm.org/doi/full/10.1056/NEJMoa1411587.
- Roy, Denis, Mario Talajic, Stanley Nattel, et al. 2008. “Rhythm Control Versus Rate Control for Atrial Fibrillation and Heart Failure.” New England Journal of Medicine 358: 2667–77. http://www.nejm.org/doi/full/10.1056/NEJMoa0708789.