森林图，排序和汇总多个变量 [英] Forest Plot, ordering and summarizing multiple variables

问题描述

我有以下数据:

DF<-structure(list(ref = structure(c(15L, 15L, 16L, 19L, 2L, 12L, 
11L, 23L, 6L, 21L, 5L, 13L, 8L, 22L, 26L, 27L, 20L, 17L, 9L, 
7L, 24L, 25L, 18L, 1L, 3L, 14L, 16L, 12L, 23L, 6L, 21L, 8L, 22L, 
26L, 27L, 20L, 17L, 9L, 7L, 24L, 25L, 18L, 4L, 1L, 14L, 16L, 
19L, 2L, 11L, 23L, 21L, 8L, 26L, 27L, 17L, 9L, 7L, 24L, 1L, 10L, 
14L), .Label = c("Bob 2012", "Bob 2016", "Arnez 2004", 
"Smithy 2013", "Smithy 2014", "Smithy 2016", "Carole 2011", "Craig 2014", 
"Fansa 2008", "Johnson 2010", "Joel 2017", "Joelo 2016", 
"Bob2 2017", "Bob2 2020", "Hunter 2015", "Hush 2016", 
"Lock 2012", "Mcdoo 2012", "Nick 2018", "Park 2015", "Joe 2012", 
"Sour 2017", "Shoe 2008", "Vega 2009", "West 2004", 
"West2016", "Smith 2016"), class = "factor"), yi = c(1, 
0.909090909090909, 1, 1, 0.98780487804878, 0.933333333333333, 
0.882352941176471, 0.980519480519481, 0.977272727272727, 1, 1, 
0.98019801980198, 0.959183673469388, 1, 1, 0.982758620689655, 
0.96969696969697, 0.6875, 1, 1, 1, 1, 1, 1, 0.75, 0.969811320754717, 
0, 0.0333333333333333, 0.064935064935065, 0.0227272727272727, 
0, 0.0204081632653061, 0.142857142857143, 0.0384615384615384, 
0.120689655172414, 0.0303030303030303, 0.0625, 0, 0.0625000000000001, 
0.148148148148148, 0.333333333333333, 0.0322580645161291, 0.0625, 
0, 0.0150943396226415, 0, 0.027027027027027, 0.0182926829268293, 
0.0588235294117647, 0.0324675324675325, 0.0416666666666667, 0.0408163265306122, 
0.192307692307692, 0.103448275862069, 0.0625, 0.03125, 0, 0, 
0.037037037037037, 0.0526315789473685, 0.0264150943396226), ci.lb = c(0.968401784273333, 
0.745137584391619, 0.957452087056599, 0.954039295289784, 0.963597464688465, 
0.809439442909756, 0.67719312002544, 0.951199930155904, 0.905001120558666, 
0.929555376052338, 0.880663089089027, 0.941246506281999, 0.880901216198665, 
0.880663089089027, 0.934891169467222, 0.927453022366531, 0.874486623056924, 
0.435962472420225, 0.946947080517241, 0.946947080517241, 0.937267052265125, 
0.861434988827223, 0.945257646596841, 0.937267052265125, 0.384687131024181, 
0.945252480837292, 0, 0, 0.0306637200529119, 0, 0, 0, 0.00329369106613314, 
0, 0.0474215778277017, 0, 0, 0, 0.00112833931883988, 0.0347070885129207, 
0.0895601878163022, 0, 0, 0, 0.00321663954072449, 0, 0, 0.00226571557474109, 
0, 0.00919930409127839, 0, 0.000687698884629828, 0.0597984369364536, 
0.0359775093204114, 0, 0, 0, 0, 0, 0, 0.00995385402759386), ci.ub = c(1, 
0.998207039140277, 1, 1, 0.999812850010077, 0.998780552481617, 
0.997483360196549, 0.997584224395838, 1, 1, 1, 0.999688395336243, 
0.99931230111537, 1, 1, 1, 1, 0.895437964404381, 1, 1, 1, 1, 
1, 1, 0.992197756884658, 0.987557818958737, 0.0425479129434015, 
0.137561603224075, 0.11002605111172, 0.0949988794413338, 0.070444623947662, 
0.0855442043005818, 0.384071178226987, 0.157747957353967, 0.21886405934029, 
0.125513376943076, 0.249041832299857, 0.0530529194827593, 0.179245839830917, 
0.311107006224451, 0.628630049521593, 0.133296666133586, 0.249041832299857, 
0.062732947734875, 0.0340807242282984, 0.0425479129434015, 0.112387275591248, 
0.0458586358986253, 0.235428911558493, 0.0674429497369029, 0.170235847270992, 
0.119098783801335, 0.36946549674197, 0.197004260641943, 0.249041832299857, 
0.129288064181111, 0.0530529194827593, 0.062732947734875, 0.152166113984736, 
0.212219059832308, 0.0497506289906541), TypeTwo = structure(c(2L, 
1L, 1L, 2L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 1L, 2L, 2L, 2L, 2L, 2L, 
2L, 2L, 2L, 2L, 2L, 2L, 1L, 2L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 1L, 1L, 1L, 2L, 1L, 1L, 
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 1L, 2L, 1L), .Label = c("BAR", 
"FOO"), class = "factor"), Variable = c("Death", "Death", 
"Death", "Death", "Death", "Death", "Death", "Death", 
"Death", "Death", "Death", "Death", "Death", "Death", 
"Death", "Death", "Death", "Death", "Death", "Death", 
"Death", "Death", "Death", "Death", "Death", "Death", 
"Vein Problems", "Vein Problems", "Vein Problems", 
"Vein Problems", "Vein Problems", "Vein Problems", 
"Vein Problems", "Vein Problems", "Vein Problems", 
"Vein Problems", "Vein Problems", "Vein Problems", 
"Vein Problems", "Vein Problems", "Vein Problems", 
"Vein Problems", "Vein Problems", "Vein Problems", 
"Vein Problems", "Gas", "Gas", "Gas", 
"Gas", "Gas", "Gas", "Gas", "Gas", "Gas", 
"Gas", "Gas", "Gas", "Gas", "Gas", "Gas", 
"Gas")), row.names = c(NA, -61L), yi.names = "yi", ci.lb.names = "ci.lb", ci.ub.names = "ci.ub", digits = c(est = 4, 
se = 4, test = 4, pval = 4, ci = 4, var = 4, sevar = 4, fit = 4, 
het = 4), class = c("escalc", "data.frame"))

我已经使用以下代码创建了一个森林图:

And I've used this code to create a forest plot:

DF%>%ggplot(aes(x=yi,y=ref,xmin=ci.lb,xmax=ci.ub,color=TypeTwo, group=TypeTwo))+geom_point()+geom_errorbarh(height=.1, size=.5)+geom_vline(xintercept = 0,color="black", linetype="dashed", alpha=.5)+facet_grid(Variable~.,scales="free",space="free")+labs(title="Forest Plot Combined",x="Effect Size",y="Study")

看起来像这样:

我想对图表进行重新排序，以使"Foo"和酒吧"彼此靠近(如下图所示)，并且我有单独的摘要"如果数据在不同的数据集中，那么我想为每个组在每个构面下添加一条线(如果可以的话).如何为此添加新行，但将其保留在构面中?

I'd like to reorder the graphs so that the "Foo" and the "Bar" are near each other, (pictured below), and I have separate "summary" data in a different data set, I'd like to add a line under each facet for, for each group, if that makes sense. How do I add a new line for that but keep it in the facet?

我完全不知道该怎么做，请帮忙！

I'm totally lost on how to do that, please help!

推荐答案

有很多方法可以解决此问题，但这是其中一种.请注意，由于您在多个方面都有相同的研究，并且 TypeTwo 不一致，因此我们必须做一些技巧才能在每个方面进行排序.

There's quite a lot of ways to go about this, but here is one. Note that since you have the same study in multiple facets, and the TypeTwo is not consistent, we have to do some trickery to be able to sort within each facet.

我也按效果大小排序，因为这很令人愉快并且很常见.

I also sorted by effect size, since that's pleasing and quite commonly done.

您在Hunter 2015的数据中有一个错误，该错误针对死亡具有两种效果大小，因此这就是为什么其中有一个红色条和一个绿色条的原因.

There is a mistake in your data for Hunter 2015 which has two effect sizes for Death, so that's why there's a red bar in with the green bars.

使用一些随机数据以获得平均效果:

Using some random data for a mean effect:

library(tidyverse)

avg <- data.frame(
  Variable = c('Death', 'Gas', 'Vein Problems'),
  yi = c(0.9, 0.1, 0.1),
  ci.lb = c(0.5, 0, 0),
  ci.ub = c(1, 0.5, 0.5),
  TypeTwo = 'mean effect',
  ref = ''
)
DF2 <- bind_rows(DF, avg) %>% 
  arrange(desc(TypeTwo), yi) %>% 
  mutate(ref2 = fct_inorder(paste(ref, Variable)))

ggplot(DF2, aes(x=yi,y=ref2,xmin=ci.lb,xmax=ci.ub,color=TypeTwo, group=TypeTwo))+
  geom_point()+
  geom_errorbarh(height=.1, size=.5)+
  geom_vline(xintercept = 0,color="black", linetype="dashed", alpha=.5)+
  facet_grid(Variable~.,scales="free",space="free")+
  scale_y_discrete(breaks = DF2$ref2, labels = DF2$ref) +
  labs(title="Forest Plot Combined",x="Effect Size",y="Study")

这篇关于森林图，排序和汇总多个变量的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！