我想在一个漏斗图上绘制三组数据点（R 的 metafor）
I would like to plot three sets of data points on one funnel plot (metafor for R)
我可以为漏斗图中的每组数据点分配不同的指标吗？我正在使用 metafor 包。我的数据集是导入到 R 中的 excel sheet。我创建了一个列 ("EffectType") 并为每个研究指定了一个值（1、2 或 3），指的是某种效应类型.
首先，我做了一个荟萃分析，现在我想创建一个包含所有数据点的漏斗图，但能够使用不同的指标来区分效果类型。
# load package
library(meta)
library(metafor)
# code for meta-analysis with sub-groups based on EffectType column
region.subgroup<-update.meta(m.dl,
byvar=EffectType,
comb.random = TRUE,
comb.fixed = FALSE)
# funnel plot
funnel(region.subgroup, xlab="Study Effect", pch=EffectType)
#dput(region.subgroup)
structure(list(studlab = c("Datta et al., 2011 [17] 1", "Nandi et al., 2017 [14] 1",
"Okmen et al., 2017 [33] 1", "Cohen et al., 2012 [28] 1", "Ghahreman et al., 2010 [27] 1",
"Karppinen et al., 2001 [25] 1", "Datta et al., 2011 [17] 2",
"Manchikanti et al., 2012 [29] 2", "Nandi et al., 2017 [14] 2",
"Ghai et al., 2015 [32] 2", "Manchikanti et al., 2014 [30] 2",
"Okmen et al., 2017 [33] 2", "Karppinen et al., 2001 [25] 2",
"Manchikanti et al., 2014 [31] 2", "Tafazal et al., 2009 [26] 2",
"Manchikanti et al., 2012 [29] 3", "Ghai et al., 2015 [32] 3",
"Manchikanti et al., 2014 [30] 3", "Okmen et al., 2017 [33] 3",
"Karppinen et al., 2001 [25] 3", "Manchikanti et al., 2014 [31] 3"
), TE = c(-4.7, -17.9, -9, -12.6, -19.1, -2.3, -12.2, -7, -12.6,
-13.9, -4, -20, 0.5, -1, -3.1, -4, -13.6, -6, -20, 16.2, 2),
seTE = c(1.4, 3.2, 2.6, 7.8, 6.6, 5.6, 1.7, 3.2, 4.1, 4.7,
2.4, 2.7, 5.9, 3, 7.2, 3.2, 5.1, 2.4, 2.3, 5.4, 2.9), lower = c(-7.44394957835607,
-24.1718847505282, -14.0959063598041, -27.8877190794124,
-32.0357622979644, -13.2757983134243, -15.5319387737181,
-13.2718847505282, -20.6358523366142, -23.1118307273383,
-8.70391356289613, -25.2919027582581, -11.0637875087863,
-6.87989195362016, -17.2117406886884, -10.2718847505282,
-23.5958163211543, -10.7039135628961, -24.5079171644421,
5.61619448348371, -3.68389555516616), upper = c(-1.95605042164393,
-11.6281152494718, -3.90409364019586, 2.68771907941242, -6.16423770203565,
8.6757983134243, -8.86806122628191, -0.728115249471828, -4.56414766338578,
-4.68816927266175, 0.703913562896129, -14.7080972417419,
12.0637875087863, 4.87989195362016, 11.0117406886884, 2.27188475052817,
-3.60418367884573, -1.29608643710387, -15.4920828355579,
26.7838055165163, 7.68389555516616), zval = c(-3.35714285714286,
-5.59375, -3.46153846153846, -1.61538461538462, -2.89393939393939,
-0.410714285714286, -7.17647058823529, -2.1875, -3.07317073170732,
-2.95744680851064, -1.66666666666667, -7.40740740740741,
0.0847457627118644, -0.333333333333333, -0.430555555555556,
-1.25, -2.66666666666667, -2.5, -8.69565217391304, 3, 0.689655172413793
), pval = c(0.000787524113356462, 2.22216865721181e-08, 0.000537097369288895,
0.106227429780004, 0.00380441585463475, 0.681282050502955,
7.15340318600297e-13, 0.0287060432176033, 0.00211797275208796,
0.00310198263536442, 0.0955807045456294, 1.28792344844862e-13,
0.932463513418651, 0.738882680363527, 0.666791563255851,
0.21129954733371, 0.00766076113517946, 0.0124193306515523,
3.44843305532707e-18, 0.00269979606326019, 0.49041106256261
), w.fixed = c(0.510204081632653, 0.09765625, 0.14792899408284,
0.0164365548980934, 0.0229568411386593, 0.0318877551020408,
0.346020761245675, 0.09765625, 0.0594883997620464, 0.0452693526482571,
0.173611111111111, 0.137174211248285, 0.0287273771904625,
0.111111111111111, 0.0192901234567901, 0.09765625, 0.0384467512495194,
0.173611111111111, 0.189035916824197, 0.0342935528120713,
0.118906064209275), w.random = c(0.0214176925138593, 0.0181916183177061,
0.0194211070756584, 0.00947235474026367, 0.0113262634833647,
0.0131422674465617, 0.0209994165994997, 0.0181916183177061,
0.0162494959147883, 0.0149654961747124, 0.0198057567991695,
0.0192232386898817, 0.0125722321000861, 0.0186114480430381,
0.0103551430341111, 0.0181916183177061, 0.0141361996281868,
0.0198057567991695, 0.019991855094554, 0.0135335632021998,
0.0188180848095829), TE.fixed = -8.41570672745868, seTE.fixed = 0.632788615741458,
lower.fixed = -9.65594962413889, upper.fixed = -7.17546383077847,
zval.fixed = -13.2993965411305, pval.fixed = 2.33343994878356e-40,
TE.random = -8.03110782264546, seTE.random = 1.69413132899102,
lower.random = -11.3515442125488, upper.random = -4.71067143274209,
zval.random = -4.74054619332764, pval.random = 2.13142849171202e-06,
null.effect = 0, seTE.predict = 6.89930804641291, lower.predict = -22.4715255225605,
upper.predict = 6.40930987726956, level.predict = 0.95, k = 21L,
Q = 121.264278025894, df.Q = 20L, pval.Q = 1.66241937202041e-16,
tau2 = 44.730370559429, se.tau2 = 21.3861120417447, lower.tau2 = 24.5614440042434,
upper.tau2 = 127.202454097413, tau = 6.68807674592846, lower.tau = 4.95595036337567,
upper.tau = 11.2784065407048, method.tau.ci = "J", sign.lower.tau = "",
sign.upper.tau = "", H = 2.4623594175698, lower.H = 2.03762882436406,
upper.H = 2.97562236497465, I2 = 0.835070967925695, lower.I2 = 0.759148226308052,
upper.I2 = 0.887060887266879, Rb = 0.742145491876444, lower.Rb = 0.586934294476571,
upper.Rb = 0.897356689276318, approx.TE = c("", "", "", "",
"", "", "", "", "", "", "", "", "", "", "", "", "", "", "",
"", ""), approx.seTE = c("", "", "", "", "", "", "", "",
"", "", "", "", "", "", "", "", "", "", "", "", ""), sm = "MD",
method = "Inverse", level = 0.95, level.comb = 0.95, comb.fixed = FALSE,
comb.random = TRUE, overall = TRUE, overall.hetstat = TRUE,
hakn = FALSE, df.hakn = NULL, method.tau = "DL", method.tau.ci = "J",
tau.preset = NULL, TE.tau = NULL, tau.common = FALSE, prediction = FALSE,
method.bias = "linreg", n.e = NULL, n.c = NULL, title = "",
complab = "", outclab = "", label.e = "Experimental", label.c = "Control",
label.left = "", label.right = "", data = structure(list(
Author = c("Datta et al., 2011 [17] 1", "Nandi et al., 2017 [14] 1",
"Okmen et al., 2017 [33] 1", "Cohen et al., 2012 [28] 1",
"Ghahreman et al., 2010 [27] 1", "Karppinen et al., 2001 [25] 1",
"Datta et al., 2011 [17] 2", "Manchikanti et al., 2012 [29] 2",
"Nandi et al., 2017 [14] 2", "Ghai et al., 2015 [32] 2",
"Manchikanti et al., 2014 [30] 2", "Okmen et al., 2017 [33] 2",
"Karppinen et al., 2001 [25] 2", "Manchikanti et al., 2014 [31] 2",
"Tafazal et al., 2009 [26] 2", "Manchikanti et al., 2012 [29] 3",
"Ghai et al., 2015 [32] 3", "Manchikanti et al., 2014 [30] 3",
"Okmen et al., 2017 [33] 3", "Karppinen et al., 2001 [25] 3",
"Manchikanti et al., 2014 [31] 3"), TE = c(-4.7, -17.9,
-9, -12.6, -19.1, -2.3, -12.2, -7, -12.6, -13.9, -4,
-20, 0.5, -1, -3.1, -4, -13.6, -6, -20, 16.2, 2), seTE = c(1.4,
3.2, 2.6, 7.8, 6.6, 5.6, 1.7, 3.2, 4.1, 4.7, 2.4, 2.7,
5.9, 3, 7.2, 3.2, 5.1, 2.4, 2.3, 5.4, 2.9), RoB = c("High",
"Some", "Some", "Low", "Low", "Low", "High", "High",
"Some", "High", "High", "Some", "Low", "High", "High",
"High", "High", "High", "Some", "Low", "High"), Technique = c("Caudal",
"Caudal", "Interlaminar", "Transforaminal", "Transforaminal",
"Transforaminal", "Caudal", "Caudal", "Caudal", "Interlaminar",
"Interlaminar", "Interlaminar", "Transforaminal", "Transforaminal",
"Transforaminal", "Caudal", "Interlaminar", "Interlaminar",
"Interlaminar", "Transforaminal", "Transforaminal"),
EffectType = c(1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2,
2, 2, 3, 3, 3, 3, 3, 3), .TE = c(-4.7, -17.9, -9, -12.6,
-19.1, -2.3, -12.2, -7, -12.6, -13.9, -4, -20, 0.5, -1,
-3.1, -4, -13.6, -6, -20, 16.2, 2), .seTE = c(1.4, 3.2,
2.6, 7.8, 6.6, 5.6, 1.7, 3.2, 4.1, 4.7, 2.4, 2.7, 5.9,
3, 7.2, 3.2, 5.1, 2.4, 2.3, 5.4, 2.9), .studlab = c("Datta et al., 2011 [17] 1",
"Nandi et al., 2017 [14] 1", "Okmen et al., 2017 [33] 1",
"Cohen et al., 2012 [28] 1", "Ghahreman et al., 2010 [27] 1",
"Karppinen et al., 2001 [25] 1", "Datta et al., 2011 [17] 2",
"Manchikanti et al., 2012 [29] 2", "Nandi et al., 2017 [14] 2",
"Ghai et al., 2015 [32] 2", "Manchikanti et al., 2014 [30] 2",
"Okmen et al., 2017 [33] 2", "Karppinen et al., 2001 [25] 2",
"Manchikanti et al., 2014 [31] 2", "Tafazal et al., 2009 [26] 2",
"Manchikanti et al., 2012 [29] 3", "Ghai et al., 2015 [32] 3",
"Manchikanti et al., 2014 [30] 3", "Okmen et al., 2017 [33] 3",
"Karppinen et al., 2001 [25] 3", "Manchikanti et al., 2014 [31] 3"
), .byvar = c(1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2,
2, 2, 3, 3, 3, 3, 3, 3)), row.names = c(NA, -21L), class = c("tbl_df",
"tbl", "data.frame")), subset = NULL, exclude = NULL, print.byvar = TRUE,
byseparator = " = ", warn = FALSE, call = update.meta(object = m.dl,
comb.fixed = FALSE, comb.random = TRUE, byvar = EffectType),
backtransf = TRUE, pscale = 1, irscale = 1, irunit = "person-years",
control = NULL, version = "4.11-0", byvar = c(1, 1, 1, 1,
1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3), bylab = "EffectType",
bylevs = c(1, 2, 3), TE.fixed.w = c(-7.49184655119754, -9.70033189245231,
-7.58113477771093), seTE.fixed.w = c(1.09958483281334, 0.990950004138082,
1.23849133665371), lower.fixed.w = c(-9.64699322145819, -11.6425582110428,
-10.0085331927171), upper.fixed.w = c(-5.33669988093689,
-7.75810557386185, -5.15373636270479), zval.fixed.w = c(-6.81334111532738,
-9.78892159235577, -6.12126589290035), pval.fixed.w = c(9.53576665434999e-12,
1.25628597075612e-22, 9.283483870295e-10), w.fixed.w = c(0.827070476854287,
1.01834869777374, 0.651949646206173), TE.random.w = c(-10.2458375238752,
-8.71229409104888, -4.62563643882933), seTE.random.w = c(2.84471402157904,
2.3359545201711, 4.58539814394897), lower.random.w = c(-15.8213745524862,
-13.2906808201078, -13.6128516557461), upper.random.w = c(-4.67030049526417,
-4.13390736198999, 4.36157877808746), zval.random.w = c(-3.60171090877808,
-3.72965056289313, -1.00877531102364), pval.random.w = c(0.000316129799539468,
0.000191745502800541, 0.313082404109481), df.hakn.w = NA,
w.random.w = c(0.0929713035774138, 0.150973845672993, 0.104477077851399
), n.harmonic.mean.w = c(NaN, NaN, NaN), t.harmonic.mean.w = c(NaN,
NaN, NaN), n.e.w = c(NA_real_, NA_real_, NA_real_), n.c.w = c(NA_real_,
NA_real_, NA_real_), k.w = c(6, 9, 6), k.all.w = c(6, 9,
6), Q.w = c(19.2741174885702, 36.6059037028299, 62.5437059660418
), pval.Q.w = c(0.00170872673573467, 1.35972605922616e-05,
3.6187724514725e-12), tau2.w = c(30.2319880674129, 34.6927986679139,
112.715864492933), lower.tau2.w = c(0, 7.93291056541169,
40.9875930481943), upper.tau2.w = c(273.001472845768, 164.636965092514,
841.055330841785), tau.w = c(5.49836230776154, 5.8900593093715,
10.6167727908689), lower.tau.w = c(0, 2.81654230669658, 6.4021553439599
), upper.tau.w = c(16.5227562121387, 12.8310936826334, 29.0009539643403
), sign.lower.tau.w = c("", "", ""), sign.upper.tau.w = c(">",
"", ""), H.w = c(1.96337044332292, 2.13909746455222, 3.53676988129117
), lower.H.w = c(1.30025792415368, 1.55608033482686, 2.61352364778506
), upper.H.w = c(2.96466064625071, 2.94055381360684, 4.78615956041163
), I2.w = c(0.740584750354196, 0.781456016905231, 0.920055904542741
), lower.I2.w = c(0.408518750491136, 0.587013389638618, 0.853597958795784
), upper.I2.w = c(0.886224167199427, 0.884351038632625, 0.956345838173359
), Rb.w = c(0.622641738855097, 0.706428253205448, 0.893469422711138
), lower.Rb.w = c(0.239211625891973, 0.444004978891535, 0.756837535911127
), upper.Rb.w = c(1, 0.968851527519361, 1), Q.w.fixed = 118.423727157442,
Q.w.random = NA, df.Q.w = 18, pval.Q.w.fixed = 8.32021898549227e-17,
pval.Q.w.random = NA, Q.b.fixed = 2.84055086845167, Q.b.random = 1.08484237833637,
df.Q.b = 2, pval.Q.b.fixed = 0.241647449751775, pval.Q.b.random = 0.581339015320737,
upper.tau2.resid = NA, lower.tau2.resid = NA, tau2.resid = NA,
upper.tau.resid = NA, lower.tau.resid = NA, tau.resid = NA,
H.resid = 2.5649748445533, lower.H.resid = 2.11081744736669,
upper.H.resid = 3.11684743813296, I2.resid = 0.848003432825,
lower.I2.resid = 0.775560832323316, upper.I2.resid = 0.897063615624055,
call.object = metagen(TE = TE, seTE = seTE, studlab = paste(Author),
data = Funnel_plot_data_Pain, sm = "MD", comb.fixed = FALSE,
comb.random = TRUE, hakn = FALSE, prediction = FALSE)), class = c("metagen",
"meta"))
我知道我可以用 pch 函数以某种方式做到这一点，但我无法让它工作。有什么建议吗？
谢谢！ E.
你很接近，pch 有效。在 metabin() 中，您定义了变量 "byvar"，因此您还需要在绘图的 pch= 参数中对该变量进行 select pch。我不确定 class "byvar" 是哪个，但 as.numeric(as.factor(m1$byvar)) 应该是可靠的。 示例：
library("meta")
data(Olkin95)
# add toy effect type
set.seed(42)
Olkin95 <- transform(Olkin95, EffectType=sample(letters[1:6], nrow(Olkin95), replace=T))
m1 <- metabin(event.e, n.e, event.c, n.c,
data=Olkin95, subset=c(41, 47, 51, 59),
studlab=paste(author, year),
byvar=EffectType, ## defining byvar
sm="RR", method="I")
op <- par(mfrow=c(1, 2))
meta::funnel(m1, pch=as.numeric(as.factor(m1$byvar)), main="w/ pch") ## using byvar for pch
meta::funnel(m1, main="w/o pch")
par(op)
注意， 我在使用 funnel() 时遇到了一些问题，因为奇怪的是这两个包共享同名函数：有一个 metafor::funnel 以及 meta::funnel 函数。我这里用的是meta::funnel
我可以为漏斗图中的每组数据点分配不同的指标吗？我正在使用 metafor 包。我的数据集是导入到 R 中的 excel sheet。我创建了一个列 ("EffectType") 并为每个研究指定了一个值（1、2 或 3），指的是某种效应类型. 首先，我做了一个荟萃分析，现在我想创建一个包含所有数据点的漏斗图，但能够使用不同的指标来区分效果类型。
# load package
library(meta)
library(metafor)
# code for meta-analysis with sub-groups based on EffectType column
region.subgroup<-update.meta(m.dl,
byvar=EffectType,
comb.random = TRUE,
comb.fixed = FALSE)
# funnel plot
funnel(region.subgroup, xlab="Study Effect", pch=EffectType)
#dput(region.subgroup)
structure(list(studlab = c("Datta et al., 2011 [17] 1", "Nandi et al., 2017 [14] 1",
"Okmen et al., 2017 [33] 1", "Cohen et al., 2012 [28] 1", "Ghahreman et al., 2010 [27] 1",
"Karppinen et al., 2001 [25] 1", "Datta et al., 2011 [17] 2",
"Manchikanti et al., 2012 [29] 2", "Nandi et al., 2017 [14] 2",
"Ghai et al., 2015 [32] 2", "Manchikanti et al., 2014 [30] 2",
"Okmen et al., 2017 [33] 2", "Karppinen et al., 2001 [25] 2",
"Manchikanti et al., 2014 [31] 2", "Tafazal et al., 2009 [26] 2",
"Manchikanti et al., 2012 [29] 3", "Ghai et al., 2015 [32] 3",
"Manchikanti et al., 2014 [30] 3", "Okmen et al., 2017 [33] 3",
"Karppinen et al., 2001 [25] 3", "Manchikanti et al., 2014 [31] 3"
), TE = c(-4.7, -17.9, -9, -12.6, -19.1, -2.3, -12.2, -7, -12.6,
-13.9, -4, -20, 0.5, -1, -3.1, -4, -13.6, -6, -20, 16.2, 2),
seTE = c(1.4, 3.2, 2.6, 7.8, 6.6, 5.6, 1.7, 3.2, 4.1, 4.7,
2.4, 2.7, 5.9, 3, 7.2, 3.2, 5.1, 2.4, 2.3, 5.4, 2.9), lower = c(-7.44394957835607,
-24.1718847505282, -14.0959063598041, -27.8877190794124,
-32.0357622979644, -13.2757983134243, -15.5319387737181,
-13.2718847505282, -20.6358523366142, -23.1118307273383,
-8.70391356289613, -25.2919027582581, -11.0637875087863,
-6.87989195362016, -17.2117406886884, -10.2718847505282,
-23.5958163211543, -10.7039135628961, -24.5079171644421,
5.61619448348371, -3.68389555516616), upper = c(-1.95605042164393,
-11.6281152494718, -3.90409364019586, 2.68771907941242, -6.16423770203565,
8.6757983134243, -8.86806122628191, -0.728115249471828, -4.56414766338578,
-4.68816927266175, 0.703913562896129, -14.7080972417419,
12.0637875087863, 4.87989195362016, 11.0117406886884, 2.27188475052817,
-3.60418367884573, -1.29608643710387, -15.4920828355579,
26.7838055165163, 7.68389555516616), zval = c(-3.35714285714286,
-5.59375, -3.46153846153846, -1.61538461538462, -2.89393939393939,
-0.410714285714286, -7.17647058823529, -2.1875, -3.07317073170732,
-2.95744680851064, -1.66666666666667, -7.40740740740741,
0.0847457627118644, -0.333333333333333, -0.430555555555556,
-1.25, -2.66666666666667, -2.5, -8.69565217391304, 3, 0.689655172413793
), pval = c(0.000787524113356462, 2.22216865721181e-08, 0.000537097369288895,
0.106227429780004, 0.00380441585463475, 0.681282050502955,
7.15340318600297e-13, 0.0287060432176033, 0.00211797275208796,
0.00310198263536442, 0.0955807045456294, 1.28792344844862e-13,
0.932463513418651, 0.738882680363527, 0.666791563255851,
0.21129954733371, 0.00766076113517946, 0.0124193306515523,
3.44843305532707e-18, 0.00269979606326019, 0.49041106256261
), w.fixed = c(0.510204081632653, 0.09765625, 0.14792899408284,
0.0164365548980934, 0.0229568411386593, 0.0318877551020408,
0.346020761245675, 0.09765625, 0.0594883997620464, 0.0452693526482571,
0.173611111111111, 0.137174211248285, 0.0287273771904625,
0.111111111111111, 0.0192901234567901, 0.09765625, 0.0384467512495194,
0.173611111111111, 0.189035916824197, 0.0342935528120713,
0.118906064209275), w.random = c(0.0214176925138593, 0.0181916183177061,
0.0194211070756584, 0.00947235474026367, 0.0113262634833647,
0.0131422674465617, 0.0209994165994997, 0.0181916183177061,
0.0162494959147883, 0.0149654961747124, 0.0198057567991695,
0.0192232386898817, 0.0125722321000861, 0.0186114480430381,
0.0103551430341111, 0.0181916183177061, 0.0141361996281868,
0.0198057567991695, 0.019991855094554, 0.0135335632021998,
0.0188180848095829), TE.fixed = -8.41570672745868, seTE.fixed = 0.632788615741458,
lower.fixed = -9.65594962413889, upper.fixed = -7.17546383077847,
zval.fixed = -13.2993965411305, pval.fixed = 2.33343994878356e-40,
TE.random = -8.03110782264546, seTE.random = 1.69413132899102,
lower.random = -11.3515442125488, upper.random = -4.71067143274209,
zval.random = -4.74054619332764, pval.random = 2.13142849171202e-06,
null.effect = 0, seTE.predict = 6.89930804641291, lower.predict = -22.4715255225605,
upper.predict = 6.40930987726956, level.predict = 0.95, k = 21L,
Q = 121.264278025894, df.Q = 20L, pval.Q = 1.66241937202041e-16,
tau2 = 44.730370559429, se.tau2 = 21.3861120417447, lower.tau2 = 24.5614440042434,
upper.tau2 = 127.202454097413, tau = 6.68807674592846, lower.tau = 4.95595036337567,
upper.tau = 11.2784065407048, method.tau.ci = "J", sign.lower.tau = "",
sign.upper.tau = "", H = 2.4623594175698, lower.H = 2.03762882436406,
upper.H = 2.97562236497465, I2 = 0.835070967925695, lower.I2 = 0.759148226308052,
upper.I2 = 0.887060887266879, Rb = 0.742145491876444, lower.Rb = 0.586934294476571,
upper.Rb = 0.897356689276318, approx.TE = c("", "", "", "",
"", "", "", "", "", "", "", "", "", "", "", "", "", "", "",
"", ""), approx.seTE = c("", "", "", "", "", "", "", "",
"", "", "", "", "", "", "", "", "", "", "", "", ""), sm = "MD",
method = "Inverse", level = 0.95, level.comb = 0.95, comb.fixed = FALSE,
comb.random = TRUE, overall = TRUE, overall.hetstat = TRUE,
hakn = FALSE, df.hakn = NULL, method.tau = "DL", method.tau.ci = "J",
tau.preset = NULL, TE.tau = NULL, tau.common = FALSE, prediction = FALSE,
method.bias = "linreg", n.e = NULL, n.c = NULL, title = "",
complab = "", outclab = "", label.e = "Experimental", label.c = "Control",
label.left = "", label.right = "", data = structure(list(
Author = c("Datta et al., 2011 [17] 1", "Nandi et al., 2017 [14] 1",
"Okmen et al., 2017 [33] 1", "Cohen et al., 2012 [28] 1",
"Ghahreman et al., 2010 [27] 1", "Karppinen et al., 2001 [25] 1",
"Datta et al., 2011 [17] 2", "Manchikanti et al., 2012 [29] 2",
"Nandi et al., 2017 [14] 2", "Ghai et al., 2015 [32] 2",
"Manchikanti et al., 2014 [30] 2", "Okmen et al., 2017 [33] 2",
"Karppinen et al., 2001 [25] 2", "Manchikanti et al., 2014 [31] 2",
"Tafazal et al., 2009 [26] 2", "Manchikanti et al., 2012 [29] 3",
"Ghai et al., 2015 [32] 3", "Manchikanti et al., 2014 [30] 3",
"Okmen et al., 2017 [33] 3", "Karppinen et al., 2001 [25] 3",
"Manchikanti et al., 2014 [31] 3"), TE = c(-4.7, -17.9,
-9, -12.6, -19.1, -2.3, -12.2, -7, -12.6, -13.9, -4,
-20, 0.5, -1, -3.1, -4, -13.6, -6, -20, 16.2, 2), seTE = c(1.4,
3.2, 2.6, 7.8, 6.6, 5.6, 1.7, 3.2, 4.1, 4.7, 2.4, 2.7,
5.9, 3, 7.2, 3.2, 5.1, 2.4, 2.3, 5.4, 2.9), RoB = c("High",
"Some", "Some", "Low", "Low", "Low", "High", "High",
"Some", "High", "High", "Some", "Low", "High", "High",
"High", "High", "High", "Some", "Low", "High"), Technique = c("Caudal",
"Caudal", "Interlaminar", "Transforaminal", "Transforaminal",
"Transforaminal", "Caudal", "Caudal", "Caudal", "Interlaminar",
"Interlaminar", "Interlaminar", "Transforaminal", "Transforaminal",
"Transforaminal", "Caudal", "Interlaminar", "Interlaminar",
"Interlaminar", "Transforaminal", "Transforaminal"),
EffectType = c(1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2,
2, 2, 3, 3, 3, 3, 3, 3), .TE = c(-4.7, -17.9, -9, -12.6,
-19.1, -2.3, -12.2, -7, -12.6, -13.9, -4, -20, 0.5, -1,
-3.1, -4, -13.6, -6, -20, 16.2, 2), .seTE = c(1.4, 3.2,
2.6, 7.8, 6.6, 5.6, 1.7, 3.2, 4.1, 4.7, 2.4, 2.7, 5.9,
3, 7.2, 3.2, 5.1, 2.4, 2.3, 5.4, 2.9), .studlab = c("Datta et al., 2011 [17] 1",
"Nandi et al., 2017 [14] 1", "Okmen et al., 2017 [33] 1",
"Cohen et al., 2012 [28] 1", "Ghahreman et al., 2010 [27] 1",
"Karppinen et al., 2001 [25] 1", "Datta et al., 2011 [17] 2",
"Manchikanti et al., 2012 [29] 2", "Nandi et al., 2017 [14] 2",
"Ghai et al., 2015 [32] 2", "Manchikanti et al., 2014 [30] 2",
"Okmen et al., 2017 [33] 2", "Karppinen et al., 2001 [25] 2",
"Manchikanti et al., 2014 [31] 2", "Tafazal et al., 2009 [26] 2",
"Manchikanti et al., 2012 [29] 3", "Ghai et al., 2015 [32] 3",
"Manchikanti et al., 2014 [30] 3", "Okmen et al., 2017 [33] 3",
"Karppinen et al., 2001 [25] 3", "Manchikanti et al., 2014 [31] 3"
), .byvar = c(1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2,
2, 2, 3, 3, 3, 3, 3, 3)), row.names = c(NA, -21L), class = c("tbl_df",
"tbl", "data.frame")), subset = NULL, exclude = NULL, print.byvar = TRUE,
byseparator = " = ", warn = FALSE, call = update.meta(object = m.dl,
comb.fixed = FALSE, comb.random = TRUE, byvar = EffectType),
backtransf = TRUE, pscale = 1, irscale = 1, irunit = "person-years",
control = NULL, version = "4.11-0", byvar = c(1, 1, 1, 1,
1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3), bylab = "EffectType",
bylevs = c(1, 2, 3), TE.fixed.w = c(-7.49184655119754, -9.70033189245231,
-7.58113477771093), seTE.fixed.w = c(1.09958483281334, 0.990950004138082,
1.23849133665371), lower.fixed.w = c(-9.64699322145819, -11.6425582110428,
-10.0085331927171), upper.fixed.w = c(-5.33669988093689,
-7.75810557386185, -5.15373636270479), zval.fixed.w = c(-6.81334111532738,
-9.78892159235577, -6.12126589290035), pval.fixed.w = c(9.53576665434999e-12,
1.25628597075612e-22, 9.283483870295e-10), w.fixed.w = c(0.827070476854287,
1.01834869777374, 0.651949646206173), TE.random.w = c(-10.2458375238752,
-8.71229409104888, -4.62563643882933), seTE.random.w = c(2.84471402157904,
2.3359545201711, 4.58539814394897), lower.random.w = c(-15.8213745524862,
-13.2906808201078, -13.6128516557461), upper.random.w = c(-4.67030049526417,
-4.13390736198999, 4.36157877808746), zval.random.w = c(-3.60171090877808,
-3.72965056289313, -1.00877531102364), pval.random.w = c(0.000316129799539468,
0.000191745502800541, 0.313082404109481), df.hakn.w = NA,
w.random.w = c(0.0929713035774138, 0.150973845672993, 0.104477077851399
), n.harmonic.mean.w = c(NaN, NaN, NaN), t.harmonic.mean.w = c(NaN,
NaN, NaN), n.e.w = c(NA_real_, NA_real_, NA_real_), n.c.w = c(NA_real_,
NA_real_, NA_real_), k.w = c(6, 9, 6), k.all.w = c(6, 9,
6), Q.w = c(19.2741174885702, 36.6059037028299, 62.5437059660418
), pval.Q.w = c(0.00170872673573467, 1.35972605922616e-05,
3.6187724514725e-12), tau2.w = c(30.2319880674129, 34.6927986679139,
112.715864492933), lower.tau2.w = c(0, 7.93291056541169,
40.9875930481943), upper.tau2.w = c(273.001472845768, 164.636965092514,
841.055330841785), tau.w = c(5.49836230776154, 5.8900593093715,
10.6167727908689), lower.tau.w = c(0, 2.81654230669658, 6.4021553439599
), upper.tau.w = c(16.5227562121387, 12.8310936826334, 29.0009539643403
), sign.lower.tau.w = c("", "", ""), sign.upper.tau.w = c(">",
"", ""), H.w = c(1.96337044332292, 2.13909746455222, 3.53676988129117
), lower.H.w = c(1.30025792415368, 1.55608033482686, 2.61352364778506
), upper.H.w = c(2.96466064625071, 2.94055381360684, 4.78615956041163
), I2.w = c(0.740584750354196, 0.781456016905231, 0.920055904542741
), lower.I2.w = c(0.408518750491136, 0.587013389638618, 0.853597958795784
), upper.I2.w = c(0.886224167199427, 0.884351038632625, 0.956345838173359
), Rb.w = c(0.622641738855097, 0.706428253205448, 0.893469422711138
), lower.Rb.w = c(0.239211625891973, 0.444004978891535, 0.756837535911127
), upper.Rb.w = c(1, 0.968851527519361, 1), Q.w.fixed = 118.423727157442,
Q.w.random = NA, df.Q.w = 18, pval.Q.w.fixed = 8.32021898549227e-17,
pval.Q.w.random = NA, Q.b.fixed = 2.84055086845167, Q.b.random = 1.08484237833637,
df.Q.b = 2, pval.Q.b.fixed = 0.241647449751775, pval.Q.b.random = 0.581339015320737,
upper.tau2.resid = NA, lower.tau2.resid = NA, tau2.resid = NA,
upper.tau.resid = NA, lower.tau.resid = NA, tau.resid = NA,
H.resid = 2.5649748445533, lower.H.resid = 2.11081744736669,
upper.H.resid = 3.11684743813296, I2.resid = 0.848003432825,
lower.I2.resid = 0.775560832323316, upper.I2.resid = 0.897063615624055,
call.object = metagen(TE = TE, seTE = seTE, studlab = paste(Author),
data = Funnel_plot_data_Pain, sm = "MD", comb.fixed = FALSE,
comb.random = TRUE, hakn = FALSE, prediction = FALSE)), class = c("metagen",
"meta"))
我知道我可以用 pch 函数以某种方式做到这一点，但我无法让它工作。有什么建议吗？
谢谢！ E.
你很接近，pch 有效。在 metabin() 中，您定义了变量 "byvar"，因此您还需要在绘图的 pch= 参数中对该变量进行 select pch。我不确定 class "byvar" 是哪个，但 as.numeric(as.factor(m1$byvar)) 应该是可靠的。 示例：
library("meta")
data(Olkin95)
# add toy effect type
set.seed(42)
Olkin95 <- transform(Olkin95, EffectType=sample(letters[1:6], nrow(Olkin95), replace=T))
m1 <- metabin(event.e, n.e, event.c, n.c,
data=Olkin95, subset=c(41, 47, 51, 59),
studlab=paste(author, year),
byvar=EffectType, ## defining byvar
sm="RR", method="I")
op <- par(mfrow=c(1, 2))
meta::funnel(m1, pch=as.numeric(as.factor(m1$byvar)), main="w/ pch") ## using byvar for pch
meta::funnel(m1, main="w/o pch")
par(op)
注意， 我在使用 funnel() 时遇到了一些问题，因为奇怪的是这两个包共享同名函数：有一个 metafor::funnel 以及 meta::funnel 函数。我这里用的是meta::funnel