Plot odds ratio by groups with ggplot for many variables - r

Question

I am making a plot for 17 symptoms by age group. So far I got what I want when I use my code for just one symptom ( code and plot below), but when running the code through all the variables, I am getting a worng plot and still have no idea where I got it wrong.
This is my data:
x <- data.frame(symptoms=c("symptom1: 0 to 9","symptom1: 10 to 19","symptom1: 20 to 49","symptom1: 50+","symptom2: 0 to 9","symptom2: : 10 to 19",
"symptom2: : 20 to 49","symptom2: 50+","symptom3: 0 to 9","symptom3: 10 to 19","symptom3: 20 to 49","symptom3: 50+",
"symptom4: 0 to 9",
"symptom4: 10 to 19","symptom4: 20 to 49","symptom4:50+","symptom5: 0 to 9","symptom5: 10 to 19","symptom5: 20 to 49",
"symptom5: 50+",
"symptom6: 0 to 9","symptom6: 10 to 19","symptom6: 20 to 49","symptom6: 50+","symptom7: 0 to 9","symptom7: 10 to 19","symptom7: 20 to 49",
"symptom7: 50+", "symptom8: 0 to 9","symptom8: 10 to 19","symptom8: 20 to 49","symptom8: 50+",
"symptom9: 0 to 9","symptom9: 10 to 19","symptom9: 20 to 49","symptom9: 50+","symptom10: 0 to 9","symptom10: 10 to 19",
"symptom10: 20 to 49","symptom10: 50+","symptom11: 0 to 9","symptom11: 10 to 19","symptom11: 20 to 49",
"symptom11: 50+","symptom12: 0 to 9","symptom12: 10 to 19","symptom12: 20 to 49","symptom12: 50+","symptom13: 0 to 9",
"symptom13: 10 to 19","symptom13: 20 to 49","symptom13: 50+","symptom14: 0 to 9","symptom14: 10 to 19",
"symptom14: 20 to 49","symptom14: 50+","symptom15: 0 to 9","symptom15: 10 to 19","symptom15: 20 to 49","symptom15: 50+",
"symptom16:0 to 9","symptom16:10 to 19","symptom16:20 to 49","symptom16:50+","symptom17: 0 to 9","symptom17: 10 to 19",
"symptom17: 20 to 49","symptom17: 50+"),
OR=c(3.1,3,0.6,0.2,2,2.5,5,1.8,7.4,4.2,6.9,2.3,3.7,2.7,3.7,5.1,6.8,3.4,4.4,8.3,14540102.8,1036435.3,8070307.6,565044.8,2.9,1.7,2.6,4.2,3.4,1.3,2.5,2.9,1,1.6,48.4,2.6,1.3,1.9,2.6,4.5,0.8,0.7,3.6,0,7.5,14.8,2.7,3.8,1.5,3.2,3.1,0.8,2.4,12,4.5,1.7,2.8,1.8,3.1,1.9,3.3,25,5,1.4,430072.7,5.8,2.8,1.5),
Lower=c(1.3,1.6,0.2,0,1.6,1.7,1.6,0.7,2.2,1.3,2.6,0.3,1.9,1.8,1.4,2,3.3,2.2,2.2,3.2,0,0,0,0,1.5,1.2,1.3,1.5,1.8,0.9,1.3,1.2,0.3,0.6,1.3,0.4,0.9,1.2,1.3,1.7,0.2,0.3,0.4,NA,3.8,8,1.4,1.5,0.7,1.6,1.3,0.3,1.2,9.1,2.2,0.7,0.7,0.6,1.1,0.3,1.3,9,1.5,0.4,0,2.5,0.9,0.1),
Upper=c(8.7,6.3,2.2,4.2,6.1,3.8,8,4.7,26,7.9,19,14.7,7.6,4,6,15.1,14.1,5.3,8.8,22.8,NA,5.463E+98,NA,NA,5.5,4.6,5.2,15.5,6.6,2,5,7.5,3.2,4.2,165.4,22.4,3.5,2.8,5,12.3,2.6,1.6,76.8,2.0619295829016E+205,15.1,30.1,5.4,10.2,3,6.7,9.4,2.1,4.6,28,9.7,7.3,9.9,4.8,8,4.7,11,46.4,23.1,5.6,NA,16,9.1,38.8),
group=rep(c("0-9 years", "10-19 years", "20-49 years", "50+ years"), 17))
This is the code for just the first symptom:
ggplot(x[1:4,] , aes(x = OR, y = 4:1, group=group)) +
geom_vline(aes(xintercept = 1), size = .25, linetype = "dashed") +
geom_errorbarh(aes(xmax = Upper, xmin = Lower), size = 1, height = .1, color = "blue") +
geom_point(aes(shape=group, color=group), size = 5) +
scale_shape_manual(values=c(15,15,15,15)) +
scale_color_manual(values=c('red','green', 'orange', "grey")) +
theme_bw() +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank()) +
scale_y_continuous(breaks = 4:1, labels = x$symptoms[1:4]) +
scale_x_continuous(breaks = seq(0,20,1) ) +
ylab("") +
xlab("Odds ratio") +
ggtitle("Odd ratios (OR) with 95% COnfidence Interval")
and this is the plot that I got with just the first symptom by age group:
When I repeat this for all symptoms so I can have everything in one plot, the plot is a mess. See below code:
ggplot(x , aes(x = OR, y = 68:1, group=group)) +
geom_vline(aes(xintercept = 1), size = .25, linetype = "dashed") +
geom_errorbarh(aes(xmax = Upper, xmin = Lower), size = 1, height = .1, color = "blue") +
geom_point(aes(shape=group, color=group), size = 5) +
scale_shape_manual(values=c(15,15,15,15)) +
scale_color_manual(values=c('red','green', 'orange', "grey")) +
theme_bw() +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank()) +
scale_y_continuous(breaks = 68:1, labels = x$symptoms) +
scale_x_continuous(breaks = seq(0,20,1) ) +
ylab("") +
xlab("Odds ratio") +
ggtitle("Odd ratios (OR) with 95% Confidence Interval")
This is the ugly plot:
At the end, I should have something like the figure below, with the frequency tables at the left and values for OR with 95%CI. I haven't try that one yet (to add all the numbers etc), but suggestions are more than welcome.
Thanks a lot for helping me to debug my code

Answer 1

This is a forest plot. Your main problem is that a couple of your values are several orders of magnitude greater than the rest. Typically with a forest plot, you want a log scale for the odds ratio to make it symmetrical around one. However, even that won't be enough here to resolve the details on your plot, so I have simply filtered out the outliers (which appear nonsensical)
Since you effectively have nested factor levels, I have "silently" faceted the plot.
library(dplyr)
x %>%
mutate(Upper = replace(Upper, abs(Upper) > 100, NA),
Lower = replace(Lower, abs(Lower) > 100, NA),
OR = replace(OR, abs(OR) > 100, NA),
symptoms = factor(gsub(":.*$", "", symptoms),
levels = paste0("symptom", 1:17))) %>%
ggplot(aes(x = OR, y = group)) +
geom_rect(aes(xmin = 0.001, xmax = 1000,
ymin = -Inf, ymax = Inf, fill = symptoms)) +
geom_errorbarh(aes(xmin = Lower, xmax = Upper)) +
geom_point(aes(colour = group, shape = group), size = 5 ) +
geom_vline(aes(xintercept = 1), linetype = 2) +
scale_shape_manual(values = rep(15, 5)) +
scale_fill_manual(values = rep(c("#ffffff00", "#f0f0f090"), 9)[-1],
guide = "none") +
scale_x_log10() +
coord_cartesian(xlim = c(0.01, 100)) +
facet_grid(symptoms~., switch = "y") +
theme_bw() +
theme(panel.spacing.y = unit(0, "points"),
panel.border = element_blank(),
axis.text.y = element_blank(),
axis.ticks.length.y = unit(0, "points"),
strip.text.y.left = element_text(angle = 0),
strip.background.y = element_blank(),
strip.placement = "outside",
axis.line = element_line()
)
You may also wish to check out the ggforest package.