Posts
In clinical studies, especially in RCTs, It is important to present the flow diagram of the patients. There are some packages like {visR} in producing diagram.
In this tutorial, I preferred manual diagram as I do in my practice.
Selected article:
Title: Molnupiravir for Oral Treatment of Covid-19 in Nonhospitalized Patients
Journal: The New England Journal of Medicine
Authors: Bernal AJ, Gomes da Silva MM, Musungaie DB et al.
Year: 2022
PMID: 34914868
DOI: 10.1056/NEJMoa2116044
The original figure
Import libraries
suppressPackageStartupMessages(library(tidyverse))
suppressPackageStartupMessages(library(fabricatr))
suppressPackageStartupMessages(library(glue))
suppressPackageStartupMessages(library(ggtext))
theme_set(theme_linedraw(base_family = "Source Sans Pro"))
Prepare fabricated data
# Molnupiravir group
set.seed(2022)
N1 <- 716
molnupiravir <- tibble(
allocated = "randomization",
group = "molnupiravir",
completed = c(rep("yes", 680), rep("no", N1-680)),
adverse = c(rep("yes", 10), rep("no", N1-10)),
not_adherent = c(rep("yes", 8), rep("no", N1-8)),
withdrew = c(rep("yes", 8), rep("no", N1-8)),
other = c(rep("yes", 4), rep("no", N1-4)),
not_received = c(rep("yes", 6), rep("no", N1-6)),
completed_29 = c(rep("yes", 680), rep("no", N1-680)),
withdrew_trial = c(rep("yes", 22), rep("no", N1-22)),
died = c(rep("yes", 1), rep("no", N1-1)),
other_trial = c(rep("yes", 7), rep("no", N1-7)),
mod_itt = c(rep("yes", 709), rep("no", N1-709)),
hospitalized = c(rep("yes", 1), rep("no", N1-1)),
not_received_trial = c(rep("yes", 6), rep("no", N1-6))
)
# Placebo group
set.seed(2022)
N2 <- 717
placebo <- tibble(
allocated = "randomization",
group = "placebo",
completed = c(rep("yes", 665), rep("no", N2-665)),
adverse = c(rep("yes", 20), rep("no", N2-20)),
not_adherent = c(rep("yes", 7), rep("no", N2-7)),
withdrew = c(rep("yes", 5), rep("no", N2-5)),
other = c(rep("yes", 4), rep("no", N2-4)),
not_received = c(rep("yes", 16), rep("no", N2-16)),
completed_29 = c(rep("yes", 672), rep("no", N2-672)),
withdrew_trial = c(rep("yes", 15), rep("no", N2-15)),
died = c(rep("yes", 9), rep("no", N2-9)),
other_trial = c(rep("yes", 5), rep("no", N2-5)),
mod_itt = c(rep("yes", 699), rep("no", N2-699)),
hospitalized = c(rep("yes", 2), rep("no", N2-2)),
not_received_trial = c(rep("yes", 16), rep("no", N2-16))
)
# Excluded group
N3 <- 204
not_randomized <- tibble(
allocated = c(rep("excluded", N3)),
group = c(rep("not_randomized", N3))
)
combined_dataset <- bind_rows(
not_randomized,
placebo,
molnupiravir
)
A part of fake dataset
combined_dataset %>%
sample_n(10)
## # A tibble: 10 × 15
## allocated group completed adverse not_adherent withdrew other not_received
## <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr>
## 1 randomiza… placebo yes no no no no no
## 2 randomiza… molnup… yes no no no no no
## 3 randomiza… placebo yes no no no no no
## 4 randomiza… placebo yes no no no no no
## 5 randomiza… placebo yes no no no no no
## 6 randomiza… placebo yes no no no no no
## 7 randomiza… placebo yes no no no no no
## 8 excluded not_ra… <NA> <NA> <NA> <NA> <NA> <NA>
## 9 randomiza… placebo yes no no no no no
## 10 randomiza… molnup… yes no no no no no
## # … with 7 more variables: completed_29 <chr>, withdrew_trial <chr>,
## # died <chr>, other_trial <chr>, mod_itt <chr>, hospitalized <chr>,
## # not_received_trial <chr>
Possible strategy:
In a original study dataset, categorization is different than the presented one. This is just for an example.
To make a reproducible diagram, all numbers should be defined with codes.
nrow(filter()) is my preference. Various ways can be used.
Then they can be added to an “empty canvas” with annotate(), geom_text(), geom_richtext() or etc.
# molnupiravir group
n_completed<- combined_dataset %>%
filter(group == "molnupiravir",
completed == "yes") %>% nrow()
n_adverse <- combined_dataset %>%
filter(group == "molnupiravir",
adverse == "yes") %>% nrow()
n_not_adherent <- combined_dataset %>%
filter(group == "molnupiravir",
not_adherent == "yes") %>% nrow()
n_withdrew <- combined_dataset %>%
filter(group == "molnupiravir",
withdrew == "yes") %>% nrow()
n_other <- combined_dataset %>%
filter(group == "molnupiravir",
other == "yes") %>% nrow()
n_discontinued <- n_adverse + n_not_adherent + n_withdrew + n_other
n_not_received <- combined_dataset %>%
filter(group == "molnupiravir",
not_received == "yes") %>% nrow()
n_molnupiravir <- n_completed + n_discontinued + n_not_received
n_completed_29 <- combined_dataset %>%
filter(group == "molnupiravir",
completed_29 == "yes") %>% nrow()
n_withdrew_trial <- combined_dataset %>%
filter(group == "molnupiravir",
withdrew_trial == "yes") %>% nrow()
n_died <- combined_dataset %>%
filter(group == "molnupiravir",
died == "yes") %>% nrow()
n_other_trial <- combined_dataset %>%
filter(group == "molnupiravir",
other_trial == "yes") %>% nrow()
n_discontinued_trial <- n_withdrew_trial + n_died + n_other_trial
n_mod_itt <- combined_dataset %>%
filter(group == "molnupiravir",
mod_itt == "yes") %>% nrow()
n_hospitalized <- combined_dataset %>%
filter(group == "molnupiravir",
hospitalized == "yes") %>% nrow()
n_not_received_trial <- combined_dataset %>%
filter(group == "molnupiravir",
not_received_trial == "yes") %>% nrow()
n_excluded <- n_hospitalized + n_not_received_trial
# placebo group
n_completed_placebo <- combined_dataset %>%
filter(group == "placebo",
completed == "yes") %>% nrow()
n_adverse_placebo <- combined_dataset %>%
filter(group == "placebo",
adverse == "yes") %>% nrow()
n_not_adherent_placebo <- combined_dataset %>%
filter(group == "placebo",
not_adherent == "yes") %>% nrow()
n_withdrew_placebo <- combined_dataset %>%
filter(group == "placebo",
withdrew == "yes") %>% nrow()
n_other_placebo <- combined_dataset %>%
filter(group == "placebo",
other == "yes") %>% nrow()
n_discontinued_placebo <- n_adverse_placebo + n_not_adherent_placebo + n_withdrew_placebo + n_other_placebo
n_not_received_placebo <- combined_dataset %>%
filter(group == "placebo",
not_received == "yes") %>% nrow()
n_placebo <- n_completed_placebo + n_discontinued_placebo + n_not_received_placebo
n_completed_29_placebo <- combined_dataset %>%
filter(group == "placebo",
completed_29 == "yes") %>% nrow()
n_withdrew_trial_placebo <- combined_dataset %>%
filter(group == "placebo",
withdrew_trial == "yes") %>% nrow()
n_died_placebo <- combined_dataset %>%
filter(group == "placebo",
died == "yes") %>% nrow()
n_other_trial_placebo <- combined_dataset %>%
filter(group == "placebo",
other_trial == "yes") %>% nrow()
n_discontinued_trial_placebo <- n_withdrew_trial_placebo + n_died_placebo + n_other_trial_placebo
n_mod_itt_placebo <- combined_dataset %>%
filter(group == "placebo",
mod_itt == "yes") %>% nrow()
n_hospitalized_placebo <- combined_dataset %>%
filter(group == "placebo",
hospitalized == "yes") %>% nrow()
n_not_received_trial_placebo <- combined_dataset %>%
filter(group == "placebo",
not_received_trial == "yes") %>% nrow()
n_excluded_placebo <- n_hospitalized_placebo + n_not_received_trial_placebo
####
n_randomized <- n_molnupiravir + n_placebo
n_not_randomized <- combined_dataset %>%
filter (allocated == "excluded") %>% nrow()
n_eligible <- n_randomized + n_not_randomized
R codes for the figure
empty_canvas <- tibble(x= 1:100, y= 1:100) %>%
ggplot(aes(x, y))
flow_diagram <- empty_canvas +
# eligibility
geom_rect(xmin = 35, xmax=65, ymin=90, ymax=98, color="black", fill="white", size=0.5) +
annotate("text", x= 50, y=94, size=4.2, color = "black", family = "Source Sans Pro",
label= paste0(n_eligible, " Participants were assessed \nfor eligibility")) +
# not-randomized
geom_rect(xmin = 67, xmax=87.5, ymin=80.5, ymax=87.5, color="black", fill="white", size=0.5) +
annotate("text", x= 77, y=84, size=4.2, color = "black", family = "Source Sans Pro",
label= paste0(n_not_randomized, " Were screened but did\n not undergo randomization")) +
# randomized
geom_rect(xmin = 35, xmax=65, ymin=70, ymax=78, color="black", fill="white", size=0.5) +
annotate("text", x= 50, y=74,size=4.2, color = "black", family = "Source Sans Pro",
label= paste0(n_randomized, " Underwent randomization")) +
# left-side
geom_rect(xmin = 10, xmax=42, ymin=29, ymax=59, color="black", fill="white", size=0.5) +
annotate("text", x= 12, y=44, size=4.2, color = "black", family = "Source Sans Pro", hjust = 0,
label= paste0(n_molnupiravir, " Were assigned to receive molnupiravir\n",
" ", n_completed, " Completed 5-day intervention period\n",
" ", n_discontinued, " Discontinued molnupiravir\n",
" ", n_adverse, " Had adverse event\n",
" ", n_not_adherent, " Were not adherent to assigned \n regimen\n",
" ", n_withdrew, " Withdrew\n",
" ", n_other, " Had other reason\n",
" ", n_not_received, " Did not receive molnupiravir")) +
geom_rect(xmin = 2, xmax=25, ymin=5, ymax=21.5, color="black", fill="white", size=0.5) +
annotate("text", x= 3, y=13, size=4.2, color = "black", family = "Source Sans Pro", hjust = 0,
label= paste0(n_completed_29, " Completed 29-day trial period\n",
" ", n_discontinued_trial, " Discontinued trial\n",
" ", n_withdrew_trial, " Withdrew\n",
" ", n_died, " Died\n",
" ", n_other_trial, " Had other reason")) +
geom_rect(xmin = 26, xmax=49, ymin=2, ymax=21.5, color="black", fill="white", size=0.5) +
annotate("text", x= 27, y=11.5, size=4.2, color = "black", family = "Source Sans Pro", hjust = 0,
label= paste0(n_mod_itt, " Were included in the modified\n intention-to-treat population\n",
" ", n_excluded, " Were excluded\n",
" ", n_hospitalized, " Was hospitalized before\n dose 1\n",
" ", n_not_received_trial, " Did not receive molnupiravir")) +
# right-side
geom_rect(xmin = 60, xmax=92, ymin=29, ymax=59, color="black", fill="white", size=0.5) +
annotate("text", x= 62, y=44, size=4.2, color = "black", family = "Source Sans Pro", hjust = 0,
label= paste0(n_placebo, " Were assigned to receive placebo\n",
" ", n_completed_placebo, " Completed 5-day intervention period\n",
" ", n_discontinued_placebo, " Discontinued placebo\n",
" ", n_adverse_placebo, " Had adverse event\n",
" ", n_not_adherent_placebo, " Were not adherent to assigned \n regimen\n",
" ", n_withdrew_placebo, " Withdrew\n",
" ", n_other_placebo, " Had other reason\n",
" ", n_not_received_placebo, " Did not receive placebo")) +
geom_rect(xmin = 52, xmax=75, ymin=2, ymax=21.5, color="black", fill="white", size=0.5) +
annotate("text", x= 53, y=11.5, size=4.2, color = "black", family = "Source Sans Pro", hjust = 0,
label= paste0(n_mod_itt_placebo, " Were included in the modified\n intention-to-treat population\n",
" ", n_excluded_placebo, " Were excluded\n",
" ", n_hospitalized_placebo, " Were hospitalized before\n dose 1\n",
" ", n_not_received_trial_placebo, " Did not receive placebo")) +
geom_rect(xmin = 76, xmax=99, ymin=5, ymax=21.5, color="black", fill="white", size=0.5) +
annotate("text", x= 77, y=13, size=4.2, color = "black", family = "Source Sans Pro", hjust = 0,
label= paste0(n_completed_29_placebo, " Completed 29-day trial period\n",
" ", n_discontinued_trial_placebo, " Discontinued trial\n",
" ", n_withdrew_trial_placebo, " Withdrew\n",
" ", n_died_placebo, " Died\n",
" ", n_other_trial_placebo, " Had other reason")) +
# arrows
geom_segment(x=50, xend=50, y=90, yend=78.5, size=0.4, linejoin = "mitre", lineend = "butt", arrow = arrow(length = unit(3, "mm"), type= "closed", angle=15)) +
geom_segment(x=50, xend=66.8, y=84, yend=84, size=0.4, linejoin = "mitre", lineend = "butt", arrow = arrow(length = unit(3, "mm"), type= "closed", angle=15)) +
geom_segment(x=50, xend=50, y=70, yend=64.5, size=0.4, linejoin = "mitre", lineend = "butt") +
geom_segment(x=25.5, xend=75.5, y=64.5, yend=64.5, size=0.4, linejoin = "mitre", lineend = "butt") +
geom_segment(x=25.5, xend=25.5, y=64.5, yend=59.5, size=0.4, linejoin = "mitre", lineend = "butt", arrow = arrow(length = unit(3, "mm"), type= "closed", angle=15)) +
geom_segment(x=75.5, xend=75.5, y=64.5, yend=59.5, size=0.4, linejoin = "mitre", lineend = "butt", arrow = arrow(length = unit(3, "mm"), type= "closed", angle=15)) +
geom_segment(x=25.5, xend=25.5, y=29, yend=26, size=0.4, linejoin = "mitre", lineend = "butt") +
geom_segment(x=75.5, xend=75.5, y=29, yend=26, size=0.4, linejoin = "mitre", lineend = "butt") +
geom_segment(x=13.5, xend=37.5, y=26, yend=26, size=0.4, linejoin = "mitre", lineend = "butt") +
geom_segment(x=63.5, xend=87.5, y=26, yend=26, size=0.4, linejoin = "mitre", lineend = "butt") +
geom_segment(x=13.5, xend=13.5, y=26, yend=22, size=0.4, linejoin = "mitre", lineend = "butt", arrow = arrow(length = unit(3, "mm"), type= "closed", angle=15)) +
geom_segment(x=63.5, xend=63.5, y=26, yend=22, size=0.4, linejoin = "mitre", lineend = "butt", arrow = arrow(length = unit(3, "mm"), type= "closed", angle=15)) +
geom_segment(x=37.5, xend=37.5, y=26, yend=22, size=0.4, linejoin = "mitre", lineend = "butt", arrow = arrow(length = unit(3, "mm"), type= "closed", angle=15)) +
geom_segment(x=87.5, xend=87.5, y=26, yend=22, size=0.4, linejoin = "mitre", lineend = "butt", arrow = arrow(length = unit(3, "mm"), type= "closed", angle=15)) +
theme_void() +
theme(panel.border = element_rect(color = "black", fill = NA, size = 1))
### SAVE FIGURE
ggsave(
flow_diagram,
file = "w5_replica.jpeg",
width = 12,
height = 9,
dpi = 144
)
Final replica
Some personal comments:
- I want to hear about the packages to make a flow diagram. I ll share another with visR in the future.
Citation
For attribution, please cite this work as
Ali Guner (Feb 14, 2022) Week-5. Retrieved from https://datavizmed.com/blog/2022-02-14-week-5/
BibTeX citation
@misc{ 2022-week-5,
author = { Ali Guner },
title = { Week-5 },
url = { https://datavizmed.com/blog/2022-02-14-week-5/ },
year = { 2022 }
updated = { Feb 14, 2022 }
}