Analyses for comparisons between second-generation supraglottic airways and endotracheal tubes in adults undergoing general anaesthesia for abdominopelvic surgeries
Clístenes Crístian de Carvalho
7/12/2022
#Packages to be used
library(meta)## Loading 'meta' package (version 4.12-0).
## Type 'help(meta)' for a brief overview.library(dmetar)## Extensive documentation for the dmetar package can be found at:
## www.bookdown.org/MathiasHarrer/Doing_Meta_Analysis_in_R/library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(metafor)## Loading required package: Matrix## Loading 'metafor' package (version 2.4-0). For an overview
## and introduction to the package please type: help(metafor).##
## Attaching package: 'metafor'## The following objects are masked from 'package:meta':
##
## baujat, forest, funnel, funnel.default, labbe, radial, trimfill#Pairwise analyses: Second generation supraglottic airways together vs endotracheal tube
Sore throat
sore<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Sore.csv")
length(sore$sore.e1)## [1] 31#Number of comparisons with zero sore throats in both arms
sore_zeros<-dplyr::filter(sore,sore$sore.e1==0 & sore$sore.e2==0)
length(sore_zeros$sore.e1)## [1] 0#Table for Meta-analysis of sore throat
sore_analysis<-dplyr::filter(sore,sore$sore.e1>0 | sore$sore.e2>0)
#Number of comparisons and patients meta-analized for sore throat
length(sore_analysis$sore.e1)## [1] 31sum(sore_analysis$sore.t1,sore_analysis$sore.t2)## [1] 3455#Meta-analysis for sore throat
mbin_sore_random<-meta::metabin(sore.e1,sore.t1,sore.e2,sore.t2,data = sore_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_sore_random## RR 95%-CI %W(random)
## Abdi 2010 0.5909 [0.3246; 1.0756] 4.5
## Ahn 2021 0.1818 [0.0706; 0.4682] 3.6
## Badheka 2015 0.1429 [0.0077; 2.6497] 1.0
## Abdel-Ghaffar 2022 0.3879 [0.1354; 1.1114] 3.3
## Baik 2003 0.8735 [0.4075; 1.8727] 4.1
## Bhushan 2022 0.3793 [0.2666; 0.5398] 5.0
## Carron 2012 0.3246 [0.0353; 2.9807] 1.5
## Du 2019 0.5000 [0.0989; 2.5270] 2.2
## Griffiths 2013 0.6820 [0.3585; 1.2973] 4.4
## Hartmann 2001 1.1895 [0.4667; 3.0316] 3.6
## Hohlrieder 2007 0.5000 [0.0959; 2.6074] 2.2
## Hong 2011 0.8571 [0.3495; 2.1022] 3.7
## Jeong 2004 0.6923 [0.3498; 1.3704] 4.3
## Kang 2019 0.3333 [0.1007; 1.1034] 3.0
## Khan 2020 0.2000 [0.0101; 3.9626] 0.9
## Kim 2021 0.9062 [0.6906; 1.1892] 5.2
## Koo 2003 1.0000 [0.1558; 6.4198] 1.9
## Lai 2017 0.3333 [0.1055; 1.0530] 3.1
## Lim 2007 0.7200 [0.4235; 1.2242] 4.6
## Lorenz 2009 0.7778 [0.3014; 2.0072] 3.6
## Ng 2021 0.6034 [0.2766; 1.3164] 4.0
## Oczenski 1999 3.0000 [1.1186; 8.0454] 3.5
## Panneer 2017 0.1333 [0.0517; 0.3436] 3.6
## Parikh 2017 0.0588 [0.0035; 0.9748] 1.0
## Saraswat 2011 0.4286 [0.1223; 1.5022] 2.9
## Tosh 2019 0.1481 [0.0773; 0.2840] 4.4
## Uerpairojkit 2009 0.6216 [0.4169; 0.9268] 4.9
## Yao 2019 0.6000 [0.2653; 1.3572] 3.9
## Ye 2020 0.0667 [0.0162; 0.2744] 2.6
## Ahmed 2015 0.3333 [0.0140; 7.9424] 0.8
## Saini 2016 0.0909 [0.0234; 0.3529] 2.7
##
## Number of studies combined: k = 31
##
## RR 95%-CI t p-value
## Random effects model 0.4690 [0.3471; 0.6337] -5.14 < 0.0001
## Prediction interval [0.1103; 1.9950]
##
## Quantifying heterogeneity:
## tau^2 = 0.4794 [0.1329; 0.9015]; tau = 0.6924 [0.3646; 0.9495];
## I^2 = 65.7% [49.9%; 76.5%]; H = 1.71 [1.41; 2.06]
##
## Test of heterogeneity:
## Q d.f. p-value
## 87.48 30 < 0.0001
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Estimated probability of sore throat with endotracheal tubes
meta::metaprop(event = sore.e2,n = sore.t2 ,studlab = paste(study),data = sore,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Abdi 2010 0.3188 [0.2117; 0.4421]
## Ahn 2021 0.6875 [0.4999; 0.8388]
## Badheka 2015 0.1000 [0.0211; 0.2653]
## Abdel-Ghaffar 2022 0.3125 [0.1612; 0.5001]
## Baik 2003 0.3030 [0.1559; 0.4871]
## Bhushan 2022 0.8788 [0.7751; 0.9462]
## Carron 2012 0.0811 [0.0170; 0.2191]
## Du 2019 0.1333 [0.0376; 0.3072]
## Griffiths 2013 0.2982 [0.1843; 0.4340]
## Hartmann 2001 0.1373 [0.0570; 0.2626]
## Hohlrieder 2007 0.0800 [0.0222; 0.1923]
## Hong 2011 0.3500 [0.1539; 0.5922]
## Jeong 2004 0.4333 [0.2546; 0.6257]
## Kang 2019 0.3214 [0.1588; 0.5235]
## Khan 2020 0.0800 [0.0098; 0.2603]
## Kim 2021 0.7442 [0.5883; 0.8648]
## Koo 2003 0.1000 [0.0123; 0.3170]
## Lai 2017 0.4500 [0.2306; 0.6847]
## Lim 2007 0.2778 [0.1885; 0.3822]
## Lorenz 2009 0.0900 [0.0420; 0.1640]
## Ng 2021 0.4000 [0.2266; 0.5940]
## Oczenski 1999 0.1600 [0.0454; 0.3608]
## Panneer 2017 0.7500 [0.5880; 0.8731]
## Parikh 2017 0.2667 [0.1228; 0.4589]
## Saraswat 2011 0.2333 [0.0993; 0.4228]
## Tosh 2019 0.9000 [0.7949; 0.9624]
## Uerpairojkit 2009 0.5362 [0.4120; 0.6572]
## Yao 2019 0.0326 [0.0184; 0.0532]
## Ye 2020 0.4545 [0.2811; 0.6365]
## Ahmed 2015 0.0250 [0.0006; 0.1316]
## Saini 2016 0.7333 [0.5411; 0.8772]
##
## Number of studies combined: k = 31
##
## proportion 95%-CI
## Random effects model 0.2976 [0.2022; 0.4147]
##
## Quantifying heterogeneity:
## tau^2 = 1.9349; tau = 1.3910; I^2 = 93.0%; H = 3.78
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 384.90 30 < 0.0001 Wald-type
## 656.48 30 < 0.0001 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies#Forest plot for sore throat
meta::forest(mbin_sore_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for sore throat
dmetar::find.outliers(mbin_sore_random)## Identified outliers (random-effects model)
## ------------------------------------------
## "Kim 2021", "Oczenski 1999", "Panneer 2017", "Tosh 2019", "Ye 2020"
##
## Results with outliers removed
## -----------------------------
## RR 95%-CI %W(random) exclude
## Abdi 2010 0.5909 [0.3246; 1.0756] 6.2
## Ahn 2021 0.1818 [0.0706; 0.4682] 4.3
## Badheka 2015 0.1429 [0.0077; 2.6497] 0.8
## Abdel-Ghaffar 2022 0.3879 [0.1354; 1.1114] 3.9
## Baik 2003 0.8735 [0.4075; 1.8727] 5.3
## Bhushan 2022 0.3793 [0.2666; 0.5398] 7.7
## Carron 2012 0.3246 [0.0353; 2.9807] 1.3
## Du 2019 0.5000 [0.0989; 2.5270] 2.2
## Griffiths 2013 0.6820 [0.3585; 1.2973] 6.0
## Hartmann 2001 1.1895 [0.4667; 3.0316] 4.4
## Hohlrieder 2007 0.5000 [0.0959; 2.6074] 2.1
## Hong 2011 0.8571 [0.3495; 2.1022] 4.6
## Jeong 2004 0.6923 [0.3498; 1.3704] 5.7
## Kang 2019 0.3333 [0.1007; 1.1034] 3.3
## Khan 2020 0.2000 [0.0101; 3.9626] 0.8
## Kim 2021 0.9062 [0.6906; 1.1892] 0.0 *
## Koo 2003 1.0000 [0.1558; 6.4198] 1.8
## Lai 2017 0.3333 [0.1055; 1.0530] 3.5
## Lim 2007 0.7200 [0.4235; 1.2242] 6.6
## Lorenz 2009 0.7778 [0.3014; 2.0072] 4.3
## Ng 2021 0.6034 [0.2766; 1.3164] 5.2
## Oczenski 1999 3.0000 [1.1186; 8.0454] 0.0 *
## Panneer 2017 0.1333 [0.0517; 0.3436] 0.0 *
## Parikh 2017 0.0588 [0.0035; 0.9748] 0.9
## Saraswat 2011 0.4286 [0.1223; 1.5022] 3.1
## Tosh 2019 0.1481 [0.0773; 0.2840] 0.0 *
## Uerpairojkit 2009 0.6216 [0.4169; 0.9268] 7.4
## Yao 2019 0.6000 [0.2653; 1.3572] 5.0
## Ye 2020 0.0667 [0.0162; 0.2744] 0.0 *
## Ahmed 2015 0.3333 [0.0140; 7.9424] 0.7
## Saini 2016 0.0909 [0.0234; 0.3529] 2.8
##
## Number of studies combined: k = 26
##
## RR 95%-CI t p-value
## Random effects model 0.5152 [0.4084; 0.6499] -5.88 < 0.0001
## Prediction interval [0.1875; 1.4152]
##
## Quantifying heterogeneity:
## tau^2 = 0.2270 [0.0000; 0.3594]; tau = 0.4764 [0.0000; 0.5995];
## I^2 = 17.8% [0.0%; 49.2%]; H = 1.10 [1.00; 1.40]
##
## Test of heterogeneity:
## Q d.f. p-value
## 30.41 25 0.2094
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Influence Analysis for sore throat
inf_analysis_sore<-dmetar::InfluenceAnalysis(mbin_sore_random,random = TRUE)## [===========================================================================] DONEplot(inf_analysis_sore,"baujat")
#Meta-regression for sore throat
#Controling for risk of bias
meta::metareg(mbin_sore_random,sore.bias)##
## Mixed-Effects Model (k = 31; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.4838 (SE = 0.1543)
## tau (square root of estimated tau^2 value): 0.6956
## I^2 (residual heterogeneity / unaccounted variability): 76.23%
## H^2 (unaccounted variability / sampling variability): 4.21
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 28) = 85.6221, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 28) = 0.7956, p-val = 0.4613
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt -0.2078 0.4644 -0.4473 0.6581 -1.1591 0.7436
## sore.biaslow risk -0.5572 0.5842 -0.9538 0.3483 -1.7540 0.6395
## sore.biassome concerns -0.6258 0.4962 -1.2612 0.2176 -1.6421 0.3906
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for operator experience
meta::metareg(mbin_sore_random,intervention.experience)##
## Mixed-Effects Model (k = 31; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.4673 (SE = 0.1516)
## tau (square root of estimated tau^2 value): 0.6836
## I^2 (residual heterogeneity / unaccounted variability): 74.89%
## H^2 (unaccounted variability / sampling variability): 3.98
## R^2 (amount of heterogeneity accounted for): 2.52%
##
## Test for Residual Heterogeneity:
## QE(df = 28) = 78.3067, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 28) = 1.5442, p-val = 0.2311
##
## Model Results:
##
## estimate se tval pval
## intrcpt -0.9947 0.2002 -4.9686 <.0001
## intervention.experienceexperienced 0.4795 0.2955 1.6229 0.1158
## intervention.experienceinexperienced 0.7434 0.7878 0.9436 0.3534
## ci.lb ci.ub
## intrcpt -1.4048 -0.5846 ***
## intervention.experienceexperienced -0.1257 1.0847
## intervention.experienceinexperienced -0.8704 2.3571
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mbin_sore_random,population)##
## Mixed-Effects Model (k = 31; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.4285 (SE = 0.1462)
## tau (square root of estimated tau^2 value): 0.6546
## I^2 (residual heterogeneity / unaccounted variability): 73.89%
## H^2 (unaccounted variability / sampling variability): 3.83
## R^2 (amount of heterogeneity accounted for): 10.61%
##
## Test for Residual Heterogeneity:
## QE(df = 27) = 71.2778, p-val < .0001
##
## Test of Moderators (coefficients 2:4):
## F(df1 = 3, df2 = 27) = 2.3193, p-val = 0.0978
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt -0.6782 0.2250 -3.0139 0.0056 -1.1398 -0.2165 **
## populationgeneral 0.0551 0.2927 0.1881 0.8522 -0.5455 0.6556
## populationobese -0.4471 1.1887 -0.3762 0.7097 -2.8861 1.9919
## populationpregnant -1.3780 0.5672 -2.4294 0.0221 -2.5418 -0.2142 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Subgroup analysis of studies with low risk of bias
sore.bias<-dplyr::filter(sore_analysis,sore.bias=="low risk")
length(sore.bias$sore.e1)## [1] 6sum(sore.bias$sore.t1,sore.bias$sore.t2)## [1] 1282mbin_sore.bias_random<-meta::metabin(sore.e1,sore.t1,sore.e2,sore.t2,data = sore.bias,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_sore.bias_random## RR 95%-CI %W(random)
## Griffiths 2013 0.6820 [0.3585; 1.2973] 36.5
## Hohlrieder 2007 0.5000 [0.0959; 2.6074] 7.5
## Kang 2019 0.3333 [0.1007; 1.1034] 13.5
## Khan 2020 0.2000 [0.0101; 3.9626] 2.4
## Lai 2017 0.3333 [0.1055; 1.0530] 14.5
## Yao 2019 0.6000 [0.2653; 1.3572] 25.7
##
## Number of studies combined: k = 6
##
## RR 95%-CI t p-value
## Random effects model 0.5124 [0.3505; 0.7492] -4.52 0.0063
## Prediction interval [0.2457; 1.0685]
##
## Quantifying heterogeneity:
## tau^2 = 0.0482 [0.0000; 0.5235]; tau = 0.2196 [0.0000; 0.7236];
## I^2 = 0.0% [0.0%; 44.7%]; H = 1.00 [1.00; 1.34]
##
## Test of heterogeneity:
## Q d.f. p-value
## 2.29 5 0.8072
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Subgroup analysis of studies with pregnant women
sore.pregnant<-dplyr::filter(sore_analysis,population=="pregnant")
length(sore.pregnant$sore.e1)## [1] 3sum(sore.pregnant$sore.t1,sore.pregnant$sore.t2)## [1] 220mbin_sore.pregnant_random<-meta::metabin(sore.e1,sore.t1,sore.e2,sore.t2,data = sore.pregnant,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_sore.pregnant_random## RR 95%-CI %W(random)
## Panneer 2017 0.1333 [0.0517; 0.3436] 60.1
## Ahmed 2015 0.3333 [0.0140; 7.9424] 6.8
## Saini 2016 0.0909 [0.0234; 0.3529] 33.2
##
## Number of studies combined: k = 3
##
## RR 95%-CI t p-value
## Random effects model 0.1249 [0.0474; 0.3290] -9.24 0.0115
## Prediction interval [0.0015; 10.2238]
##
## Quantifying heterogeneity:
## tau^2 = 0.0695 [0.0000; 15.8349]; tau = 0.2637 [0.0000; 3.9793];
## I^2 = 0.0% [0.0%; 65.2%]; H = 1.00 [1.00; 1.69]
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.60 2 0.7419
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Funnel Plot for sore throat
meta::funnel(mbin_sore_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::metabias(mbin_sore_random, method.bias = "linreg")##
## Linear regression test of funnel plot asymmetry
##
## data: mbin_sore_random
## t = -2.0025, df = 29, p-value = 0.05465
## alternative hypothesis: asymmetry in funnel plot
## sample estimates:
## bias se.bias intercept
## -1.0796890 0.5391595 -0.2335660dmetar::eggers.test(mbin_sore_random)## Intercept ConfidenceInterval t p
## Egger's test -1.08 -2.06--0.1 -2.003 0.05465meta::trimfill(mbin_sore_random)## RR 95%-CI %W(random)
## Abdi 2010 0.5909 [0.3246; 1.0756] 3.3
## Ahn 2021 0.1818 [0.0706; 0.4682] 2.9
## Badheka 2015 0.1429 [0.0077; 2.6497] 1.1
## Abdel-Ghaffar 2022 0.3879 [0.1354; 1.1114] 2.8
## Baik 2003 0.8735 [0.4075; 1.8727] 3.1
## Bhushan 2022 0.3793 [0.2666; 0.5398] 3.5
## Carron 2012 0.3246 [0.0353; 2.9807] 1.6
## Du 2019 0.5000 [0.0989; 2.5270] 2.1
## Griffiths 2013 0.6820 [0.3585; 1.2973] 3.3
## Hartmann 2001 1.1895 [0.4667; 3.0316] 2.9
## Hohlrieder 2007 0.5000 [0.0959; 2.6074] 2.1
## Hong 2011 0.8571 [0.3495; 2.1022] 3.0
## Jeong 2004 0.6923 [0.3498; 1.3704] 3.2
## Kang 2019 0.3333 [0.1007; 1.1034] 2.6
## Khan 2020 0.2000 [0.0101; 3.9626] 1.1
## Kim 2021 0.9062 [0.6906; 1.1892] 3.5
## Koo 2003 1.0000 [0.1558; 6.4198] 1.9
## Lai 2017 0.3333 [0.1055; 1.0530] 2.7
## Lim 2007 0.7200 [0.4235; 1.2242] 3.4
## Lorenz 2009 0.7778 [0.3014; 2.0072] 2.9
## Ng 2021 0.6034 [0.2766; 1.3164] 3.1
## Oczenski 1999 3.0000 [1.1186; 8.0454] 2.9
## Panneer 2017 0.1333 [0.0517; 0.3436] 2.9
## Parikh 2017 0.0588 [0.0035; 0.9748] 1.2
## Saraswat 2011 0.4286 [0.1223; 1.5022] 2.5
## Tosh 2019 0.1481 [0.0773; 0.2840] 3.3
## Uerpairojkit 2009 0.6216 [0.4169; 0.9268] 3.5
## Yao 2019 0.6000 [0.2653; 1.3572] 3.1
## Ye 2020 0.0667 [0.0162; 0.2744] 2.4
## Ahmed 2015 0.3333 [0.0140; 7.9424] 1.0
## Saini 2016 0.0909 [0.0234; 0.3529] 2.4
## Filled: Carron 2012 1.3772 [0.1500; 12.6483] 1.6
## Filled: Khan 2020 2.2350 [0.1128; 44.2822] 1.1
## Filled: Ahn 2021 2.4585 [0.9546; 6.3315] 2.9
## Filled: Tosh 2019 3.0173 [1.5742; 5.7831] 3.3
## Filled: Badheka 2015 3.1290 [0.1687; 58.0376] 1.1
## Filled: Panneer 2017 3.3525 [1.3008; 8.6405] 2.9
## Filled: Saini 2016 4.9170 [1.2668; 19.0848] 2.4
## Filled: Ye 2020 6.7050 [1.6288; 27.6011] 2.4
## Filled: Parikh 2017 7.5990 [0.4586; 125.9244] 1.2
##
## Number of studies combined: k = 40 (with 9 added studies)
##
## RR 95%-CI t p-value
## Random effects model 0.6602 [0.4612; 0.9451] -2.34 0.0244
## Prediction interval [0.0863; 5.0512]
##
## Quantifying heterogeneity:
## tau^2 = 0.9789 [0.4129; 1.6814]; tau = 0.9894 [0.6425; 1.2967];
## I^2 = 75.1% [66.1%; 81.6%]; H = 2.00 [1.72; 2.33]
##
## Test of heterogeneity:
## Q d.f. p-value
## 156.34 39 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Trim-and-fill method to adjust for funnel plot asymmetryHoarseness
hoar<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Hoarseness.csv")
length(hoar$hoar.e1)## [1] 22#Number of comparisons with zero hoarseness in both arms
hoar_zeros<-dplyr::filter(hoar,hoar$hoar.e1==0 & hoar$hoar.e2==0)
length(hoar_zeros$hoar.e1)## [1] 5#Table for Meta-analysis of hoarseness
hoar_analysis<-dplyr::filter(hoar,hoar$hoar.e1>0 | hoar$hoar.e2>0)
#Number of comparisons and patients meta-analized for hoarseness
length(hoar_analysis$hoar.e1)## [1] 17sum(hoar_analysis$hoar.t1,hoar_analysis$hoar.t2)## [1] 1626#Meta-analysis for hoarseness
mbin_hoar_random<-meta::metabin(hoar.e1,hoar.t1,hoar.e2,hoar.t2,data = hoar_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_hoar_random## RR 95%-CI %W(random)
## Abdi 2010 0.3437 [0.1890; 0.6253] 10.5
## Ahn 2021 0.1111 [0.0062; 1.9813] 1.8
## Baik 2003 0.0971 [0.0131; 0.7165] 3.2
## Carron 2012 0.3247 [0.0137; 7.7209] 1.5
## Du 2019 0.2500 [0.0296; 2.1081] 2.9
## Griffiths 2013 0.9349 [0.6617; 1.3210] 12.4
## Hartmann 2001 0.1486 [0.0079; 2.8041] 1.7
## Kang 2019 0.1111 [0.0151; 0.8198] 3.2
## Kim 2021 0.9333 [0.5158; 1.6890] 10.6
## Kuvaki 2019 0.2417 [0.0762; 0.7663] 6.6
## Lorenz 2009 0.1667 [0.0204; 1.3594] 3.0
## Ng 2021 0.2821 [0.0876; 0.9091] 6.5
## Oczenski 1999 0.2727 [0.0864; 0.8613] 6.6
## Ozbilgin 2021 0.2222 [0.0505; 0.9774] 4.9
## Tosh 2019 0.1111 [0.0270; 0.4580] 5.2
## Uerpairojkit 2009 0.7541 [0.6252; 0.9096] 13.2
## Ye 2020 0.1364 [0.0408; 0.4556] 6.3
##
## Number of studies combined: k = 17
##
## RR 95%-CI t p-value
## Random effects model 0.3442 [0.2283; 0.5190] -5.51 < 0.0001
## Prediction interval [0.0957; 1.2382]
##
## Quantifying heterogeneity:
## tau^2 = 0.3232 [0.0644; 0.8734]; tau = 0.5685 [0.2537; 0.9346];
## I^2 = 64.7% [40.8%; 78.9%]; H = 1.68 [1.30; 2.18]
##
## Test of heterogeneity:
## Q d.f. p-value
## 45.31 16 0.0001
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Estimated probability of hoarseness with endotracheal tubes
meta::metaprop(event = hoar.e2,n = hoar.t2 ,studlab = paste(study),data = hoar,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Abdi 2010 0.4638 [0.3428; 0.5880]
## Ahn 2021 0.1250 [0.0351; 0.2899]
## Baik 2003 0.3030 [0.1559; 0.4871]
## Carron 2012 0.0270 [0.0007; 0.1416]
## Du 2019 0.1333 [0.0376; 0.3072]
## Griffiths 2013 0.5439 [0.4066; 0.6764]
## Hartmann 2001 0.0588 [0.0123; 0.1624]
## Hohlrieder 2007 0.0000 [0.0000; 0.0711]
## Ibrahim 2011 0.0000 [0.0000; 0.1157]
## Kang 2019 0.3214 [0.1588; 0.5235]
## Khan 2020 0.0000 [0.0000; 0.1372]
## Kim 2021 0.3488 [0.2101; 0.5093]
## Kuvaki 2019 0.3103 [0.1954; 0.4454]
## Lorenz 2009 0.0600 [0.0223; 0.1260]
## Ng 2021 0.3667 [0.1993; 0.5614]
## Oczenski 1999 0.4400 [0.2440; 0.6507]
## Ozbilgin 2021 0.1800 [0.0858; 0.3144]
## Saraswat 2011 0.0000 [0.0000; 0.1157]
## Tosh 2019 0.3000 [0.1885; 0.4321]
## Uerpairojkit 2009 0.8841 [0.7843; 0.9486]
## Yao 2019 0.0000 [0.0000; 0.0080]
## Ye 2020 0.3333 [0.1796; 0.5183]
##
## Number of studies combined: k = 22
##
## proportion 95%-CI
## Random effects model 0.1205 [0.0494; 0.2652]
##
## Quantifying heterogeneity:
## tau^2 = 4.5522; tau = 2.1336; I^2 = 96.4%; H = 5.26
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 129.64 21 < 0.0001 Wald-type
## 538.19 21 < 0.0001 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)#Forest plot for hoarseness
meta::forest(mbin_hoar_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for hoarseness
dmetar::find.outliers(mbin_hoar_random)## Identified outliers (random-effects model)
## ------------------------------------------
## "Griffiths 2013", "Uerpairojkit 2009"
##
## Results with outliers removed
## -----------------------------
## RR 95%-CI %W(random) exclude
## Abdi 2010 0.3437 [0.1890; 0.6253] 16.9
## Ahn 2021 0.1111 [0.0062; 1.9813] 1.9
## Baik 2003 0.0971 [0.0131; 0.7165] 3.7
## Carron 2012 0.3247 [0.0137; 7.7209] 1.6
## Du 2019 0.2500 [0.0296; 2.1081] 3.3
## Griffiths 2013 0.9349 [0.6617; 1.3210] 0.0 *
## Hartmann 2001 0.1486 [0.0079; 2.8041] 1.9
## Kang 2019 0.1111 [0.0151; 0.8198] 3.7
## Kim 2021 0.9333 [0.5158; 1.6890] 17.0
## Kuvaki 2019 0.2417 [0.0762; 0.7663] 8.7
## Lorenz 2009 0.1667 [0.0204; 1.3594] 3.4
## Ng 2021 0.2821 [0.0876; 0.9091] 8.5
## Oczenski 1999 0.2727 [0.0864; 0.8613] 8.7
## Ozbilgin 2021 0.2222 [0.0505; 0.9774] 6.1
## Tosh 2019 0.1111 [0.0270; 0.4580] 6.5
## Uerpairojkit 2009 0.7541 [0.6252; 0.9096] 0.0 *
## Ye 2020 0.1364 [0.0408; 0.4556] 8.2
##
## Number of studies combined: k = 15
##
## RR 95%-CI t p-value
## Random effects model 0.2717 [0.1840; 0.4012] -7.17 < 0.0001
## Prediction interval [0.1017; 0.7258]
##
## Quantifying heterogeneity:
## tau^2 = 0.1739 [0.0000; 0.5054]; tau = 0.4170 [0.0000; 0.7109];
## I^2 = 32.5% [0.0%; 63.7%]; H = 1.22 [1.00; 1.66]
##
## Test of heterogeneity:
## Q d.f. p-value
## 20.74 14 0.1084
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Influence Analysis for hoarseness
inf_analysis_hoar<-dmetar::InfluenceAnalysis(mbin_hoar_random,random = TRUE)## [===========================================================================] DONEplot(inf_analysis_hoar,"baujat")
#Meta-regression for hoarseness
#Controling for risk of bias
meta::metareg(mbin_hoar_random,hoar.bias)##
## Mixed-Effects Model (k = 17; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.3260 (SE = 0.1574)
## tau (square root of estimated tau^2 value): 0.5710
## I^2 (residual heterogeneity / unaccounted variability): 61.69%
## H^2 (unaccounted variability / sampling variability): 2.61
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 14) = 39.7321, p-val = 0.0003
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 14) = 0.9773, p-val = 0.4006
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt -2.3324 1.0792 -2.1613 0.0485 -4.6470 -0.0178 *
## hoar.biaslow risk 1.5849 1.1611 1.3651 0.1938 -0.9053 4.0751
## hoar.biassome concerns 1.2317 1.1018 1.1179 0.2824 -1.1314 3.5948
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for operator experience
meta::metareg(mbin_hoar_random,intervention.experience)##
## Mixed-Effects Model (k = 17; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.3567 (SE = 0.1627)
## tau (square root of estimated tau^2 value): 0.5972
## I^2 (residual heterogeneity / unaccounted variability): 69.28%
## H^2 (unaccounted variability / sampling variability): 3.25
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 14) = 39.2913, p-val = 0.0003
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 14) = 0.2277, p-val = 0.7993
##
## Model Results:
##
## estimate se tval pval
## intrcpt -1.1068 0.2682 -4.1266 0.0010
## intervention.experienceexperienced 0.1111 0.4215 0.2636 0.7959
## intervention.experienceinexperienced -0.6849 1.1874 -0.5768 0.5732
## ci.lb ci.ub
## intrcpt -1.6821 -0.5316 **
## intervention.experienceexperienced -0.7929 1.0151
## intervention.experienceinexperienced -3.2316 1.8618
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mbin_hoar_random,population)##
## Mixed-Effects Model (k = 17; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.3282 (SE = 0.1567)
## tau (square root of estimated tau^2 value): 0.5729
## I^2 (residual heterogeneity / unaccounted variability): 68.82%
## H^2 (unaccounted variability / sampling variability): 3.21
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 14) = 38.6633, p-val = 0.0004
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 14) = 0.8263, p-val = 0.4579
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt -0.8371 0.2671 -3.1341 0.0073 -1.4100 -0.2642 **
## populationgeneral -0.5092 0.3963 -1.2850 0.2196 -1.3591 0.3407
## populationobese -0.2878 1.6178 -0.1779 0.8613 -3.7576 3.1820
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Subgroup analysis of studies with low risk of bias
hoar.bias<-dplyr::filter(hoar_analysis,hoar.bias=="low risk")
length(hoar.bias$hoar.e1)## [1] 3sum(hoar.bias$hoar.t1,hoar.bias$hoar.t2)## [1] 272mbin_hoar.bias_random<-meta::metabin(hoar.e1,hoar.t1,hoar.e2,hoar.t2,data = hoar.bias,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_hoar.bias_random## RR 95%-CI %W(random)
## Griffiths 2013 0.9349 [0.6617; 1.3210] 48.6
## Kang 2019 0.1111 [0.0151; 0.8198] 21.9
## Ozbilgin 2021 0.2222 [0.0505; 0.9774] 29.4
##
## Number of studies combined: k = 3
##
## RR 95%-CI t p-value
## Random effects model 0.3838 [0.0248; 5.9370] -1.50 0.2714
## Prediction interval [0.0000; 435532.0395]
##
## Quantifying heterogeneity:
## tau^2 = 0.7988 [0.0001; 46.1956]; tau = 0.8937 [0.0113; 6.7967];
## I^2 = 72.9% [8.9%; 91.9%]; H = 1.92 [1.05; 3.52]
##
## Test of heterogeneity:
## Q d.f. p-value
## 7.38 2 0.0250
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model#Funnel Plot for hoarseness
meta::funnel(mbin_hoar_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::metabias(mbin_hoar_random, method.bias = "linreg")##
## Linear regression test of funnel plot asymmetry
##
## data: mbin_hoar_random
## t = -5.5934, df = 15, p-value = 5.128e-05
## alternative hypothesis: asymmetry in funnel plot
## sample estimates:
## bias se.bias intercept
## -1.82939498 0.32706025 -0.04816175dmetar::eggers.test(mbin_hoar_random)## Intercept ConfidenceInterval t p
## Egger's test -1.829 -2.417--1.241 -5.593 5e-05meta::trimfill(mbin_hoar_random)## RR 95%-CI %W(random)
## Abdi 2010 0.3437 [0.1890; 0.6253] 5.4
## Ahn 2021 0.1111 [0.0062; 1.9813] 2.3
## Baik 2003 0.0971 [0.0131; 0.7165] 3.3
## Carron 2012 0.3247 [0.0137; 7.7209] 2.0
## Du 2019 0.2500 [0.0296; 2.1081] 3.1
## Griffiths 2013 0.9349 [0.6617; 1.3210] 5.6
## Hartmann 2001 0.1486 [0.0079; 2.8041] 2.2
## Kang 2019 0.1111 [0.0151; 0.8198] 3.3
## Kim 2021 0.9333 [0.5158; 1.6890] 5.4
## Kuvaki 2019 0.2417 [0.0762; 0.7663] 4.6
## Lorenz 2009 0.1667 [0.0204; 1.3594] 3.2
## Ng 2021 0.2821 [0.0876; 0.9091] 4.6
## Oczenski 1999 0.2727 [0.0864; 0.8613] 4.6
## Ozbilgin 2021 0.2222 [0.0505; 0.9774] 4.1
## Tosh 2019 0.1111 [0.0270; 0.4580] 4.2
## Uerpairojkit 2009 0.7541 [0.6252; 0.9096] 5.7
## Ye 2020 0.1364 [0.0408; 0.4556] 4.5
## Filled: Kuvaki 2019 2.1887 [0.6903; 6.9397] 4.6
## Filled: Ozbilgin 2021 2.3802 [0.5412; 10.4684] 4.1
## Filled: Lorenz 2009 3.1736 [0.3891; 25.8844] 3.2
## Filled: Hartmann 2001 3.5587 [0.1886; 67.1411] 2.2
## Filled: Ye 2020 3.8788 [1.1610; 12.9592] 4.5
## Filled: Tosh 2019 4.7604 [1.1548; 19.6228] 4.2
## Filled: Ahn 2021 4.7604 [0.2670; 84.8877] 2.3
## Filled: Kang 2019 4.7604 [0.6452; 35.1210] 3.3
## Filled: Baik 2003 5.4496 [0.7382; 40.2292] 3.3
##
## Number of studies combined: k = 26 (with 9 added studies)
##
## RR 95%-CI t p-value
## Random effects model 0.6264 [0.3580; 1.0958] -1.72 0.0973
## Prediction interval [0.0492; 7.9730]
##
## Quantifying heterogeneity:
## tau^2 = 1.4455 [0.4677; 3.0584]; tau = 1.2023 [0.6839; 1.7488];
## I^2 = 68.6% [52.9%; 79.0%]; H = 1.78 [1.46; 2.18]
##
## Test of heterogeneity:
## Q d.f. p-value
## 79.53 25 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Trim-and-fill method to adjust for funnel plot asymmetryDysphagia
dysphagia<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Dysphagia.csv")
length(dysphagia$dysphagia.e1)## [1] 15#Number of comparisons with zero dysphagia in both arms
dysphagia_zeros<-dplyr::filter(dysphagia,dysphagia$dysphagia.e1==0 & dysphagia$dysphagia.e2==0)
length(dysphagia_zeros$dysphagia.e1)## [1] 3#Table for Meta-analysis of dysphagia
dysphagia_analysis<-dplyr::filter(dysphagia,dysphagia$dysphagia.e1>0 | dysphagia$dysphagia.e2>0)
#Number of comparisons and patients meta-analized for dysphagia
length(dysphagia_analysis$dysphagia.e1)## [1] 12sum(dysphagia_analysis$dysphagia.t1,dysphagia_analysis$dysphagia.t2)## [1] 1224#Meta-analysis for dysphagia
mbin_dysphagia_random<-meta::metabin(dysphagia.e1,dysphagia.t1,dysphagia.e2,dysphagia.t2,data = dysphagia_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_dysphagia_random## RR 95%-CI %W(random)
## Abdi 2010 0.6111 [0.3122; 1.1964] 11.9
## Abdel-Ghaffar 2022 0.4848 [0.1324; 1.7751] 9.2
## Baik 2003 1.6985 [0.5480; 5.2647] 9.9
## Griffiths 2013 0.4831 [0.1539; 1.5160] 9.9
## Hartmann 2001 2.0816 [0.3992; 10.8550] 7.7
## Hohlrieder 2007 0.3333 [0.0139; 7.9896] 3.6
## Khan 2020 3.0000 [0.1281; 70.2318] 3.6
## Kuvaki 2019 0.2727 [0.0809; 0.9191] 9.5
## Lorenz 2009 2.0000 [0.1843; 21.7051] 5.3
## Oczenski 1999 5.6667 [1.8956; 16.9396] 10.1
## Ozbilgin 2021 0.0625 [0.0086; 0.4535] 6.5
## Uerpairojkit 2009 0.7895 [0.5601; 1.1127] 12.9
##
## Number of studies combined: k = 12
##
## RR 95%-CI t p-value
## Random effects model 0.8105 [0.3876; 1.6948] -0.63 0.5435
## Prediction interval [0.0801; 8.2028]
##
## Quantifying heterogeneity:
## tau^2 = 0.9668 [0.1209; 3.5234]; tau = 0.9833 [0.3477; 1.8771];
## I^2 = 60.9% [26.5%; 79.2%]; H = 1.60 [1.17; 2.19]
##
## Test of heterogeneity:
## Q d.f. p-value
## 28.12 11 0.0031
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Estimated probability of dysphagia with endotracheal tubes
meta::metaprop(event = dysphagia.e2,n = dysphagia.t2 ,studlab = paste(study),data = dysphagia,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Abdi 2010 0.2609 [0.1625; 0.3806]
## Abdel-Ghaffar 2022 0.1875 [0.0721; 0.3644]
## Baik 2003 0.1212 [0.0340; 0.2820]
## Carron 2012 0.0000 [0.0000; 0.0949]
## Griffiths 2013 0.1404 [0.0626; 0.2579]
## Hartmann 2001 0.0392 [0.0048; 0.1346]
## Hohlrieder 2007 0.0200 [0.0005; 0.1065]
## Khan 2020 0.0000 [0.0000; 0.1372]
## Kuvaki 2019 0.2200 [0.1153; 0.3596]
## Lorenz 2009 0.0100 [0.0003; 0.0545]
## Oczenski 1999 0.1200 [0.0255; 0.3122]
## Ozbilgin 2021 0.3200 [0.1952; 0.4670]
## Panneer 2017 0.0000 [0.0000; 0.0881]
## Saraswat 2011 0.0000 [0.0000; 0.1157]
## Uerpairojkit 2009 0.5507 [0.4262; 0.6708]
##
## Number of studies combined: k = 15
##
## proportion 95%-CI
## Random effects model 0.0637 [0.0242; 0.1575]
##
## Quantifying heterogeneity:
## tau^2 = 2.8834; tau = 1.6981; I^2 = 92.9%; H = 3.75
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 67.64 14 < 0.0001 Wald-type
## 160.80 14 < 0.0001 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)#Forest plot for dysphagia
meta::forest(mbin_dysphagia_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for dysphagia
dmetar::find.outliers(mbin_dysphagia_random)## Identified outliers (random-effects model)
## ------------------------------------------
## "Oczenski 1999"
##
## Results with outliers removed
## -----------------------------
## RR 95%-CI %W(random) exclude
## Abdi 2010 0.6111 [0.3122; 1.1964] 14.3
## Abdel-Ghaffar 2022 0.4848 [0.1324; 1.7751] 10.1
## Baik 2003 1.6985 [0.5480; 5.2647] 11.2
## Griffiths 2013 0.4831 [0.1539; 1.5160] 11.1
## Hartmann 2001 2.0816 [0.3992; 10.8550] 8.1
## Hohlrieder 2007 0.3333 [0.0139; 7.9896] 3.4
## Khan 2020 3.0000 [0.1281; 70.2318] 3.4
## Kuvaki 2019 0.2727 [0.0809; 0.9191] 10.6
## Lorenz 2009 2.0000 [0.1843; 21.7051] 5.2
## Oczenski 1999 5.6667 [1.8956; 16.9396] 0.0 *
## Ozbilgin 2021 0.0625 [0.0086; 0.4535] 6.6
## Uerpairojkit 2009 0.7895 [0.5601; 1.1127] 16.0
##
## Number of studies combined: k = 11
##
## RR 95%-CI t p-value
## Random effects model 0.6518 [0.3416; 1.2436] -1.48 0.1707
## Prediction interval [0.0923; 4.6024]
##
## Quantifying heterogeneity:
## tau^2 = 0.6625 [0.0000; 2.7825]; tau = 0.8140 [0.0000; 1.6681];
## I^2 = 33.9% [0.0%; 67.5%]; H = 1.23 [1.00; 1.76]
##
## Test of heterogeneity:
## Q d.f. p-value
## 15.14 10 0.1271
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Influence Analysis for dysphagia
inf_analysis_dysphagia<-dmetar::InfluenceAnalysis(mbin_dysphagia_random,random = TRUE)## [===========================================================================] DONEplot(inf_analysis_dysphagia,"baujat")
#Meta-regression for dysphagia
#Controling for risk of bias
meta::metareg(mbin_dysphagia_random,dysphagia.bias)##
## Mixed-Effects Model (k = 12; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.9433 (SE = 0.5167)
## tau (square root of estimated tau^2 value): 0.9712
## I^2 (residual heterogeneity / unaccounted variability): 76.60%
## H^2 (unaccounted variability / sampling variability): 4.27
## R^2 (amount of heterogeneity accounted for): 2.43%
##
## Test for Residual Heterogeneity:
## QE(df = 9) = 23.2853, p-val = 0.0056
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 9) = 1.1485, p-val = 0.3595
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt 0.5298 1.0486 0.5052 0.6256 -1.8423 2.9019
## dysphagia.biaslow risk -1.5962 1.2510 -1.2759 0.2339 -4.4261 1.2338
## dysphagia.biassome concerns -0.5485 1.1240 -0.4880 0.6372 -3.0911 1.9941
##
## intrcpt
## dysphagia.biaslow risk
## dysphagia.biassome concerns
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for operator experience
meta::metareg(mbin_dysphagia_random,intervention.experience)##
## Mixed-Effects Model (k = 12; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 1.0090 (SE = 0.5334)
## tau (square root of estimated tau^2 value): 1.0045
## I^2 (residual heterogeneity / unaccounted variability): 74.98%
## H^2 (unaccounted variability / sampling variability): 4.00
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 9) = 26.8463, p-val = 0.0015
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 9) = 0.7122, p-val = 0.5163
##
## Model Results:
##
## estimate se tval pval
## intrcpt -0.6160 0.4969 -1.2398 0.2464
## intervention.experienceexperienced 0.7227 0.7094 1.0187 0.3349
## intervention.experienceinexperienced 1.3092 1.5742 0.8316 0.4271
## ci.lb ci.ub
## intrcpt -1.7400 0.5080
## intervention.experienceexperienced -0.8821 2.3274
## intervention.experienceinexperienced -2.2520 4.8703
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mbin_dysphagia_random,population)## Warning in rma.uni(yi = TE[!exclude], sei = seTE[!exclude], data = dataset, :
## Redundant predictors dropped from the model.##
## Mixed-Effects Model (k = 12; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.3513 (SE = 0.3192)
## tau (square root of estimated tau^2 value): 0.5927
## I^2 (residual heterogeneity / unaccounted variability): 54.02%
## H^2 (unaccounted variability / sampling variability): 2.17
## R^2 (amount of heterogeneity accounted for): 63.67%
##
## Test for Residual Heterogeneity:
## QE(df = 10) = 11.7522, p-val = 0.3020
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 10) = 18.9067, p-val = 0.0014
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt -0.7881 0.2360 -3.3394 0.0075 -1.3139 -0.2623 **
## populationgeneral 1.7953 0.4129 4.3482 0.0014 0.8754 2.7153 **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Subgroup analysis of studies with low risk of bias
dysphagia.bias<-dplyr::filter(dysphagia_analysis,dysphagia.bias=="low risk")
length(dysphagia.bias$dysphagia.e1)## [1] 4sum(dysphagia.bias$dysphagia.t1,dysphagia.bias$dysphagia.t2)## [1] 366mbin_dysphagia.bias_random<-meta::metabin(dysphagia.e1,dysphagia.t1,dysphagia.e2,dysphagia.t2,data = dysphagia.bias,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_dysphagia.bias_random## RR 95%-CI %W(random)
## Griffiths 2013 0.4831 [0.1539; 1.5160] 39.1
## Hohlrieder 2007 0.3333 [0.0139; 7.9896] 16.5
## Khan 2020 3.0000 [0.1281; 70.2318] 16.6
## Ozbilgin 2021 0.0625 [0.0086; 0.4535] 27.8
##
## Number of studies combined: k = 4
##
## RR 95%-CI t p-value
## Random effects model 0.3491 [0.0334; 3.6519] -1.43 0.2490
## Prediction interval [0.0010; 124.5085]
##
## Quantifying heterogeneity:
## tau^2 = 1.3214 [0.0000; 33.2430]; tau = 1.1495 [0.0000; 5.7657];
## I^2 = 39.9% [0.0%; 79.6%]; H = 1.29 [1.00; 2.21]
##
## Test of heterogeneity:
## Q d.f. p-value
## 4.99 3 0.1725
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Subgroup analysis of studies with women undergoing gynecological surgeries
dysphagia.famale<-dplyr::filter(dysphagia_analysis,population=="famale")
length(dysphagia.famale$dysphagia.e1)## [1] 7sum(dysphagia.famale$dysphagia.t1,dysphagia.famale$dysphagia.t2)## [1] 757mbin_dysphagia.famale_random<-meta::metabin(dysphagia.e1,dysphagia.t1,dysphagia.e2,dysphagia.t2,data = dysphagia.famale,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_dysphagia.famale_random## RR 95%-CI %W(random)
## Abdi 2010 0.6111 [0.3122; 1.1964] 21.7
## Abdel-Ghaffar 2022 0.4848 [0.1324; 1.7751] 12.7
## Griffiths 2013 0.4831 [0.1539; 1.5160] 14.5
## Hohlrieder 2007 0.3333 [0.0139; 7.9896] 3.3
## Kuvaki 2019 0.2727 [0.0809; 0.9191] 13.6
## Ozbilgin 2021 0.0625 [0.0086; 0.4535] 7.2
## Uerpairojkit 2009 0.7895 [0.5601; 1.1127] 26.9
##
## Number of studies combined: k = 7
##
## RR 95%-CI t p-value
## Random effects model 0.4578 [0.2388; 0.8778] -2.94 0.0261
## Prediction interval [0.0897; 2.3374]
##
## Quantifying heterogeneity:
## tau^2 = 0.3314 [0.0000; 2.6733]; tau = 0.5757 [0.0000; 1.6350];
## I^2 = 34.9% [0.0%; 72.5%]; H = 1.24 [1.00; 1.91]
##
## Test of heterogeneity:
## Q d.f. p-value
## 9.21 6 0.1621
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Funnel Plot for dysphagia
meta::funnel(mbin_dysphagia_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::metabias(mbin_dysphagia_random, method.bias = "linreg")##
## Linear regression test of funnel plot asymmetry
##
## data: mbin_dysphagia_random
## t = 0.064735, df = 10, p-value = 0.9497
## alternative hypothesis: asymmetry in funnel plot
## sample estimates:
## bias se.bias intercept
## 0.05195878 0.80264209 -0.24128476dmetar::eggers.test(mbin_dysphagia_random)## Intercept ConfidenceInterval t p
## Egger's test 0.052 -1.516-1.62 0.065 0.94966meta::trimfill(mbin_dysphagia_random)## RR 95%-CI %W(random)
## Abdi 2010 0.6111 [0.3122; 1.1964] 11.9
## Abdel-Ghaffar 2022 0.4848 [0.1324; 1.7751] 9.2
## Baik 2003 1.6985 [0.5480; 5.2647] 9.9
## Griffiths 2013 0.4831 [0.1539; 1.5160] 9.9
## Hartmann 2001 2.0816 [0.3992; 10.8550] 7.7
## Hohlrieder 2007 0.3333 [0.0139; 7.9896] 3.6
## Khan 2020 3.0000 [0.1281; 70.2318] 3.6
## Kuvaki 2019 0.2727 [0.0809; 0.9191] 9.5
## Lorenz 2009 2.0000 [0.1843; 21.7051] 5.3
## Oczenski 1999 5.6667 [1.8956; 16.9396] 10.1
## Ozbilgin 2021 0.0625 [0.0086; 0.4535] 6.5
## Uerpairojkit 2009 0.7895 [0.5601; 1.1127] 12.9
##
## Number of studies combined: k = 12 (with 0 added studies)
##
## RR 95%-CI t p-value
## Random effects model 0.8105 [0.3876; 1.6948] -0.63 0.5435
## Prediction interval [0.0801; 8.2028]
##
## Quantifying heterogeneity:
## tau^2 = 0.9668 [0.1209; 3.5234]; tau = 0.9833 [0.3477; 1.8771];
## I^2 = 60.9% [26.5%; 79.2%]; H = 1.60 [1.17; 2.19]
##
## Test of heterogeneity:
## Q d.f. p-value
## 28.12 11 0.0031
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Trim-and-fill method to adjust for funnel plot asymmetryTissue damage
tissue<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Tissue.csv")
length(tissue$tissue.e1)## [1] 20#Number of comparisons with zero tissue damage in both arms
tissue_zeros<-dplyr::filter(tissue,tissue$tissue.e1==0 & tissue$tissue.e2==0)
length(tissue_zeros$tissue.e1)## [1] 1#Table for Meta-analysis of tissue damage
tissue_analysis<-dplyr::filter(tissue,tissue$tissue.e1>0 | tissue$tissue.e2>0)
#Number of comparisons and patients meta-analized for tissue damage
length(tissue_analysis$tissue.e1)## [1] 19sum(tissue_analysis$tissue.t1,tissue_analysis$tissue.t2)## [1] 2365#Meta-analysis for tissue damage
mbin_tissue_random<-meta::metabin(tissue.e1,tissue.t1,tissue.e2,tissue.t2,data = tissue_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_tissue_random## RR 95%-CI %W(random)
## Badheka 2015 0.6667 [0.1198; 3.7087] 5.6
## Abdel-Ghaffar 2022 1.9394 [0.1848; 20.3496] 3.8
## Baik 2003 3.3971 [0.7605; 15.1744] 6.5
## Carron 2012 0.0464 [0.0028; 0.7636] 2.9
## Dunnebier 2017 0.2632 [0.0112; 6.1566] 2.4
## Griffiths 2013 0.7514 [0.3000; 1.8821] 9.3
## Hartmann 2001 1.0408 [0.3211; 3.3735] 8.0
## Hong 2011 3.0000 [0.1297; 69.4167] 2.4
## Lim 2007 1.2000 [0.3799; 3.7909] 8.1
## Ng 2021 0.2586 [0.0307; 2.1785] 4.3
## Oczenski 1999 7.0000 [0.3806; 128.7425] 2.7
## Panneer 2017 0.7500 [0.1792; 3.1383] 6.7
## Parikh 2017 0.3333 [0.0141; 7.8648] 2.4
## Sabuncu 2018 2.2109 [0.1092; 44.7707] 2.6
## Saraswat 2011 0.6000 [0.1573; 2.2889] 7.2
## Uerpairojkit 2009 2.5000 [0.5020; 12.4510] 6.0
## Yao 2019 0.7778 [0.4829; 1.2527] 11.6
## Ahmed 2015 0.2000 [0.0099; 4.0371] 2.6
## Saini 2016 1.0000 [0.1505; 6.6426] 5.0
##
## Number of studies combined: k = 19
##
## RR 95%-CI t p-value
## Random effects model 0.9008 [0.5716; 1.4196] -0.48 0.6352
## Prediction interval [0.1641; 4.9442]
##
## Quantifying heterogeneity:
## tau^2 = 0.6044 [0.0000; 1.3082]; tau = 0.7775 [0.0000; 1.1438];
## I^2 = 0.0% [0.0%; 44.8%]; H = 1.00 [1.00; 1.35]
##
## Test of heterogeneity:
## Q d.f. p-value
## 16.65 18 0.5471
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Estimated probability of tissue damage with endotracheal tubes
meta::metaprop(event = tissue.e2,n = tissue.t2 ,studlab = paste(study),data = tissue,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Badheka 2015 0.1000 [0.0211; 0.2653]
## Abdel-Ghaffar 2022 0.0312 [0.0008; 0.1622]
## Baik 2003 0.0606 [0.0074; 0.2023]
## Carron 2012 0.2703 [0.1379; 0.4412]
## Dunnebier 2017 0.0455 [0.0012; 0.2284]
## Griffiths 2013 0.1579 [0.0748; 0.2787]
## Hartmann 2001 0.0980 [0.0326; 0.2141]
## Hong 2011 0.0000 [0.0000; 0.1684]
## Ibrahim 2011 0.0000 [0.0000; 0.1157]
## Lim 2007 0.0556 [0.0183; 0.1249]
## Ng 2021 0.1333 [0.0376; 0.3072]
## Oczenski 1999 0.0000 [0.0000; 0.1372]
## Panneer 2017 0.1000 [0.0279; 0.2366]
## Parikh 2017 0.0333 [0.0008; 0.1722]
## Sabuncu 2018 0.0000 [0.0000; 0.1089]
## Saraswat 2011 0.1667 [0.0564; 0.3472]
## Uerpairojkit 2009 0.0290 [0.0035; 0.1008]
## Yao 2019 0.0783 [0.0554; 0.1067]
## Ahmed 2015 0.0500 [0.0061; 0.1692]
## Saini 2016 0.0667 [0.0082; 0.2207]
##
## Number of studies combined: k = 20
##
## proportion 95%-CI
## Random effects model 0.0659 [0.0437; 0.0983]
##
## Quantifying heterogeneity:
## tau^2 = 0.3999; tau = 0.6324; I^2 = 58.8%; H = 1.56
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 27.33 19 0.0971 Wald-type
## 44.94 19 0.0007 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)#Forest plot for tissue damage
meta::forest(mbin_tissue_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for tissue damage
dmetar::find.outliers(mbin_tissue_random)## No outliers detected (random-effects model).#Influence Analysis for tissue damage
inf_analysis_tissue<-dmetar::InfluenceAnalysis(mbin_tissue_random,random = TRUE)## [===========================================================================] DONEplot(inf_analysis_tissue,"baujat")
#Meta-regression for tissue damage
#Controling for risk of bias
meta::metareg(mbin_tissue_random,tissue.bias)##
## Mixed-Effects Model (k = 19; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.5612 (SE = 0.3116)
## tau (square root of estimated tau^2 value): 0.7491
## I^2 (residual heterogeneity / unaccounted variability): 50.54%
## H^2 (unaccounted variability / sampling variability): 2.02
## R^2 (amount of heterogeneity accounted for): 7.16%
##
## Test for Residual Heterogeneity:
## QE(df = 17) = 13.3292, p-val = 0.7139
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 17) = 2.9124, p-val = 0.1061
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt 1.2229 0.8036 1.5217 0.1465 -0.4726 2.9184
## tissue.biassome concerns -1.4180 0.8309 -1.7066 0.1061 -3.1711 0.3350
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for operator experience
meta::metareg(mbin_tissue_random,intervention.experience)##
## Mixed-Effects Model (k = 19; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.6387 (SE = 0.3247)
## tau (square root of estimated tau^2 value): 0.7992
## I^2 (residual heterogeneity / unaccounted variability): 50.35%
## H^2 (unaccounted variability / sampling variability): 2.01
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 17) = 16.5125, p-val = 0.4878
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 17) = 0.0461, p-val = 0.8326
##
## Model Results:
##
## estimate se tval pval ci.lb
## intrcpt -0.0631 0.2978 -0.2119 0.8347 -0.6914
## intervention.experienceexperienced -0.0969 0.4515 -0.2146 0.8326 -1.0495
## ci.ub
## intrcpt 0.5652
## intervention.experienceexperienced 0.8557
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mbin_tissue_random,population)##
## Mixed-Effects Model (k = 19; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.4358 (SE = 0.3095)
## tau (square root of estimated tau^2 value): 0.6601
## I^2 (residual heterogeneity / unaccounted variability): 44.07%
## H^2 (unaccounted variability / sampling variability): 1.79
## R^2 (amount of heterogeneity accounted for): 27.90%
##
## Test for Residual Heterogeneity:
## QE(df = 14) = 11.4217, p-val = 0.6526
##
## Test of Moderators (coefficients 2:5):
## F(df1 = 4, df2 = 14) = 2.2535, p-val = 0.1153
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt 0.1268 0.2814 0.4506 0.6592 -0.4767 0.7303
## populationgeneral -0.0978 0.4103 -0.2383 0.8151 -0.9777 0.7821
## populationmale -1.4618 1.2878 -1.1351 0.2754 -4.2239 1.3003
## populationobese -3.1976 1.1722 -2.7279 0.0163 -5.7117 -0.6836 *
## populationpregnant -0.5391 0.5705 -0.9449 0.3608 -1.7627 0.6846
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Subgroup analysis of studies with low risk of bias
tissue.bias<-dplyr::filter(tissue_analysis,tissue.bias=="low risk")
length(tissue.bias$tissue.e1)## [1] 0sum(tissue.bias$tissue.t1,tissue.bias$tissue.t2)## [1] 0#Subgroup analysis of studies with obese patients
tissue.obese<-dplyr::filter(tissue_analysis,population=="obese")
length(tissue.obese$tissue.e1)## [1] 1sum(tissue.obese$tissue.t1,tissue.obese$tissue.t2)## [1] 75mbin_tissue.obese_random<-meta::metabin(tissue.e1,tissue.t1,tissue.e2,tissue.t2,data = tissue.obese,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_tissue.obese_random## RR 95%-CI z p-value
## 0.0464 [0.0028; 0.7643] -2.15 0.0317
##
## Details:
## - Mantel-Haenszel method
## - Continuity correction of 0.5 in studies with zero cell frequencies#Funnel Plot for tissue damage
meta::funnel(mbin_tissue_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::metabias(mbin_tissue_random, method.bias = "linreg")##
## Linear regression test of funnel plot asymmetry
##
## data: mbin_tissue_random
## t = 0.10392, df = 17, p-value = 0.9184
## alternative hypothesis: asymmetry in funnel plot
## sample estimates:
## bias se.bias intercept
## 0.04321397 0.41584155 -0.16462174dmetar::eggers.test(mbin_tissue_random)## Intercept ConfidenceInterval t p
## Egger's test 0.043 -0.741-0.827 0.104 0.91845meta::trimfill(mbin_tissue_random)## RR 95%-CI %W(random)
## Badheka 2015 0.6667 [0.1198; 3.7087] 5.6
## Abdel-Ghaffar 2022 1.9394 [0.1848; 20.3496] 3.8
## Baik 2003 3.3971 [0.7605; 15.1744] 6.5
## Carron 2012 0.0464 [0.0028; 0.7636] 2.9
## Dunnebier 2017 0.2632 [0.0112; 6.1566] 2.4
## Griffiths 2013 0.7514 [0.3000; 1.8821] 9.3
## Hartmann 2001 1.0408 [0.3211; 3.3735] 8.0
## Hong 2011 3.0000 [0.1297; 69.4167] 2.4
## Lim 2007 1.2000 [0.3799; 3.7909] 8.1
## Ng 2021 0.2586 [0.0307; 2.1785] 4.3
## Oczenski 1999 7.0000 [0.3806; 128.7425] 2.7
## Panneer 2017 0.7500 [0.1792; 3.1383] 6.7
## Parikh 2017 0.3333 [0.0141; 7.8648] 2.4
## Sabuncu 2018 2.2109 [0.1092; 44.7707] 2.6
## Saraswat 2011 0.6000 [0.1573; 2.2889] 7.2
## Uerpairojkit 2009 2.5000 [0.5020; 12.4510] 6.0
## Yao 2019 0.7778 [0.4829; 1.2527] 11.6
## Ahmed 2015 0.2000 [0.0099; 4.0371] 2.6
## Saini 2016 1.0000 [0.1505; 6.6426] 5.0
##
## Number of studies combined: k = 19 (with 0 added studies)
##
## RR 95%-CI t p-value
## Random effects model 0.9008 [0.5716; 1.4196] -0.48 0.6352
## Prediction interval [0.1641; 4.9442]
##
## Quantifying heterogeneity:
## tau^2 = 0.6044 [0.0000; 1.3082]; tau = 0.7775 [0.0000; 1.1438];
## I^2 = 0.0% [0.0%; 44.8%]; H = 1.00 [1.00; 1.35]
##
## Test of heterogeneity:
## Q d.f. p-value
## 16.65 18 0.5471
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Trim-and-fill method to adjust for funnel plot asymmetryCough
cough<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Cough.csv")
length(cough$cough.e1)## [1] 10#Number of comparisons with zero cough in both arms
cough_zeros<-dplyr::filter(cough,cough$cough.e1==0 & cough$cough.e2==0)
length(cough_zeros$cough.e1)## [1] 2#Table for Meta-analysis of cough
cough_analysis<-dplyr::filter(cough,cough$cough.e1>0 | cough$cough.e2>0)
#Number of comparisons and patients meta-analized for cough
length(cough_analysis$cough.e1)## [1] 8sum(cough_analysis$cough.t1,cough_analysis$cough.t2)## [1] 600#Meta-analysis for cough
mbin_cough_random<-meta::metabin(cough.e1,cough.t1,cough.e2,cough.t2,data = cough_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_cough_random## RR 95%-CI %W(random)
## Ahn 2021 0.0769 [0.0107; 0.5538] 9.1
## Baik 2003 0.5392 [0.2939; 0.9892] 16.7
## Carron 2012 0.1007 [0.0336; 0.3023] 13.9
## Kang 2019 0.1429 [0.0487; 0.4187] 14.1
## Maltby 2002 0.0458 [0.0117; 0.1789] 12.4
## Ng 2021 0.3448 [0.0757; 1.5715] 11.5
## Saraswat 2011 2.0000 [0.1914; 20.8980] 7.6
## Tosh 2019 0.0741 [0.0286; 0.1916] 14.8
##
## Number of studies combined: k = 8
##
## RR 95%-CI t p-value
## Random effects model 0.1709 [0.0654; 0.4461] -4.35 0.0033
## Prediction interval [0.0120; 2.4269]
##
## Quantifying heterogeneity:
## tau^2 = 1.0113 [0.1385; 5.7226]; tau = 1.0056 [0.3722; 2.3922];
## I^2 = 72.9% [44.7%; 86.7%]; H = 1.92 [1.34; 2.75]
##
## Test of heterogeneity:
## Q d.f. p-value
## 25.85 7 0.0005
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model#Estimated probability of cough with endotracheal tubes
meta::metaprop(event = cough.e2,n = cough.t2 ,studlab = paste(study),data = cough,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Ahn 2021 0.4062 [0.2370; 0.5936]
## Badheka 2015 0.0000 [0.0000; 0.1157]
## Baik 2003 0.5455 [0.3635; 0.7189]
## Carron 2012 0.7838 [0.6179; 0.9017]
## Kang 2019 0.8400 [0.6392; 0.9546]
## Lai 2017 0.0000 [0.0000; 0.1684]
## Maltby 2002 0.8727 [0.7552; 0.9473]
## Ng 2021 0.2000 [0.0771; 0.3857]
## Saraswat 2011 0.0333 [0.0008; 0.1722]
## Tosh 2019 0.9000 [0.7949; 0.9624]
##
## Number of studies combined: k = 10
##
## proportion 95%-CI
## Random effects model 0.3347 [0.0831; 0.7364]
##
## Quantifying heterogeneity:
## tau^2 = 6.7237; tau = 2.5930; I^2 = 96.6%; H = 5.40
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 74.22 9 < 0.0001 Wald-type
## 216.69 9 < 0.0001 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)#Forest plot for cough
meta::forest(mbin_cough_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for cough
dmetar::find.outliers(mbin_cough_random)## No outliers detected (random-effects model).#Influence Analysis for cough
inf_analysis_cough<-dmetar::InfluenceAnalysis(mbin_cough_random,random = TRUE)## [===========================================================================] DONEplot(inf_analysis_cough,"baujat")
#Meta-regression for cough
#Controling for risk of bias
meta::metareg(mbin_cough_random,cough.bias)##
## Mixed-Effects Model (k = 8; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.9387 (SE = 0.6191)
## tau (square root of estimated tau^2 value): 0.9689
## I^2 (residual heterogeneity / unaccounted variability): 67.42%
## H^2 (unaccounted variability / sampling variability): 3.07
## R^2 (amount of heterogeneity accounted for): 7.18%
##
## Test for Residual Heterogeneity:
## QE(df = 6) = 10.6517, p-val = 0.0998
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 6) = 1.8500, p-val = 0.2227
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt -0.6176 0.9282 -0.6655 0.5305 -2.8887 1.6535
## cough.biassome concerns -1.3851 1.0183 -1.3601 0.2227 -3.8768 1.1067
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for operator experience
meta::metareg(mbin_cough_random,intervention.experience)##
## Mixed-Effects Model (k = 8; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.7456 (SE = 0.5313)
## tau (square root of estimated tau^2 value): 0.8635
## I^2 (residual heterogeneity / unaccounted variability): 68.15%
## H^2 (unaccounted variability / sampling variability): 3.14
## R^2 (amount of heterogeneity accounted for): 26.27%
##
## Test for Residual Heterogeneity:
## QE(df = 6) = 17.0305, p-val = 0.0092
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 6) = 3.3237, p-val = 0.1181
##
## Model Results:
##
## estimate se tval pval ci.lb
## intrcpt -1.3102 0.4290 -3.0543 0.0224 -2.3599
## intervention.experienceexperienced -1.3266 0.7277 -1.8231 0.1181 -3.1072
## ci.ub
## intrcpt -0.2606 *
## intervention.experienceexperienced 0.4539
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mbin_cough_random,population)## Warning in rma.uni(yi = TE[!exclude], sei = seTE[!exclude], data = dataset, :
## Redundant predictors dropped from the model.##
## Mixed-Effects Model (k = 8; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 1.1344 (SE = 0.6967)
## tau (square root of estimated tau^2 value): 1.0651
## I^2 (residual heterogeneity / unaccounted variability): 76.46%
## H^2 (unaccounted variability / sampling variability): 4.25
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 6) = 23.8949, p-val = 0.0005
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 6) = 0.2428, p-val = 0.6397
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt -2.2954 1.1609 -1.9772 0.0954 -5.1361 0.5453 .
## populationgeneral 0.6164 1.2508 0.4928 0.6397 -2.4443 3.6771
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Subgroup analysis of studies with low risk of bias
cough.bias<-dplyr::filter(cough_analysis,cough.bias=="low risk")
length(cough.bias$cough.e1)## [1] 0sum(cough.bias$cough.t1,cough.bias$cough.t2)## [1] 0#Funnel Plot for cough
meta::funnel(mbin_cough_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::metabias(mbin_cough_random, method.bias = "linreg")## Warning in print.metabias(x): Number of studies (k=8) too small to test for small study effects (k.min=10). Change argument 'k.min' if appropriate.dmetar::eggers.test(mbin_cough_random)## Warning in dmetar::eggers.test(mbin_cough_random): Your meta-analysis contains k
## = 8 studies. Egger's test may lack the statistical power to detect bias when the
## number of studies is small (i.e., k<10).## Intercept ConfidenceInterval t p
## Egger's test -1.511 -5.039-2.017 -0.841 0.4326meta::trimfill(mbin_cough_random)## RR 95%-CI %W(random)
## Ahn 2021 0.0769 [0.0107; 0.5538] 7.9
## Baik 2003 0.5392 [0.2939; 0.9892] 12.2
## Carron 2012 0.1007 [0.0336; 0.3023] 10.8
## Kang 2019 0.1429 [0.0487; 0.4187] 10.8
## Maltby 2002 0.0458 [0.0117; 0.1789] 9.9
## Ng 2021 0.3448 [0.0757; 1.5715] 9.4
## Saraswat 2011 2.0000 [0.1914; 20.8980] 6.8
## Tosh 2019 0.0741 [0.0286; 0.1916] 11.2
## Filled: Tosh 2019 1.2634 [0.4884; 3.2685] 11.2
## Filled: Maltby 2002 2.0419 [0.5232; 7.9688] 9.9
##
## Number of studies combined: k = 10 (with 2 added studies)
##
## RR 95%-CI t p-value
## Random effects model 0.2759 [0.1009; 0.7542] -2.90 0.0177
## Prediction interval [0.0126; 6.0364]
##
## Quantifying heterogeneity:
## tau^2 = 1.5927 [0.4773; 6.3270]; tau = 1.2620 [0.6908; 2.5154];
## I^2 = 80.3% [64.5%; 89.0%]; H = 2.25 [1.68; 3.02]
##
## Test of heterogeneity:
## Q d.f. p-value
## 45.60 9 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Trim-and-fill method to adjust for funnel plot asymmetryPostoperative nausea and vomiting
ponv<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA PONV.csv")
length(ponv$ponv.e1)## [1] 15#Number of comparisons with zero PONV in both arms
ponv_zeros<-dplyr::filter(ponv,ponv$ponv.e1==0 & ponv$ponv.e2==0)
length(ponv_zeros$ponv.e1)## [1] 3#Table for Meta-analysis of PONV
ponv_analysis<-dplyr::filter(ponv,ponv$ponv.e1>0 | ponv$ponv.e2>0)
#Number of comparisons and patients meta-analized for PONV
length(ponv_analysis$ponv.e1)## [1] 12sum(ponv_analysis$ponv.t1,ponv_analysis$ponv.t2)## [1] 965#Meta-analysis for PONV
mbin_ponv_random<-meta::metabin(ponv.e1,ponv.t1,ponv.e2,ponv.t2,data = ponv_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_ponv_random## RR 95%-CI %W(random)
## Abdel-Ghaffar 2022 0.1940 [0.0097; 3.8885] 1.9
## Carron 2012 0.1217 [0.0160; 0.9258] 3.5
## Du 2019 0.3333 [0.0730; 1.5213] 5.4
## Griffiths 2013 1.0019 [0.6828; 1.4700] 13.8
## Hohlrieder 2007 0.2000 [0.0461; 0.8670] 5.6
## Jeong 2004 1.0667 [0.6535; 1.7411] 12.9
## Kang 2019 0.3333 [0.1223; 0.9089] 8.5
## Kim 2021 1.0370 [0.7549; 1.4245] 14.3
## Ng 2021 1.0345 [0.0678; 15.7732] 2.2
## Parikh 2017 0.4000 [0.1409; 1.1354] 8.2
## Uerpairojkit 2009 1.1250 [0.6268; 2.0193] 12.1
## Ye 2020 0.5833 [0.3093; 1.1001] 11.6
##
## Number of studies combined: k = 12
##
## RR 95%-CI t p-value
## Random effects model 0.6334 [0.4128; 0.9719] -2.35 0.0387
## Prediction interval [0.1674; 2.3966]
##
## Quantifying heterogeneity:
## tau^2 = 0.3188 [0.0000; 1.1762]; tau = 0.5646 [0.0000; 1.0845];
## I^2 = 45.7% [0.0%; 72.2%]; H = 1.36 [1.00; 1.90]
##
## Test of heterogeneity:
## Q d.f. p-value
## 20.25 11 0.0420
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Estimated probability of PONV with endotracheal tubes
meta::metaprop(event = ponv.e2,n = ponv.t2 ,studlab = paste(study),data = ponv,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Abdel-Ghaffar 2022 0.0625 [0.0077; 0.2081]
## Carron 2012 0.2162 [0.0983; 0.3821]
## Du 2019 0.2000 [0.0771; 0.3857]
## Griffiths 2013 0.4737 [0.3398; 0.6103]
## Hohlrieder 2007 0.2000 [0.1003; 0.3372]
## Jeong 2004 0.5000 [0.3130; 0.6870]
## Kang 2019 0.4286 [0.2446; 0.6282]
## Kim 2021 0.6279 [0.4673; 0.7702]
## Ng 2021 0.0333 [0.0008; 0.1722]
## Panneer 2017 0.0000 [0.0000; 0.0881]
## Parikh 2017 0.3333 [0.1729; 0.5281]
## Saraswat 2011 0.0000 [0.0000; 0.1157]
## Uerpairojkit 2009 0.2319 [0.1387; 0.3491]
## Ye 2020 0.4000 [0.2266; 0.5940]
## Saini 2016 0.0000 [0.0000; 0.1157]
##
## Number of studies combined: k = 15
##
## proportion 95%-CI
## Random effects model 0.1648 [0.0759; 0.3218]
##
## Quantifying heterogeneity:
## tau^2 = 2.4219; tau = 1.5562; I^2 = 93.2%; H = 3.84
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 50.78 14 < 0.0001 Wald-type
## 135.07 14 < 0.0001 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)#Forest plot for PONV
meta::forest(mbin_ponv_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for PONV
dmetar::find.outliers(mbin_ponv_random)## No outliers detected (random-effects model).#Influence Analysis for PONV
inf_analysis_ponv<-dmetar::InfluenceAnalysis(mbin_ponv_random,random = TRUE)## [===========================================================================] DONEplot(inf_analysis_ponv,"baujat")
#Meta-regression for PONV
#Controling for risk of bias
meta::metareg(mbin_ponv_random,ponv.bias)##
## Mixed-Effects Model (k = 12; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.3441 (SE = 0.2090)
## tau (square root of estimated tau^2 value): 0.5866
## I^2 (residual heterogeneity / unaccounted variability): 65.68%
## H^2 (unaccounted variability / sampling variability): 2.91
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 9) = 19.3627, p-val = 0.0223
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 9) = 0.5719, p-val = 0.5837
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt 0.0645 0.5706 0.1131 0.9124 -1.2262 1.3552
## ponv.biaslow risk -0.7325 0.6889 -1.0633 0.3153 -2.2909 0.8259
## ponv.biassome concerns -0.5500 0.6291 -0.8744 0.4047 -1.9731 0.8730
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for operator experience
meta::metareg(mbin_ponv_random,intervention.experience)##
## Mixed-Effects Model (k = 12; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.3107 (SE = 0.1865)
## tau (square root of estimated tau^2 value): 0.5574
## I^2 (residual heterogeneity / unaccounted variability): 66.41%
## H^2 (unaccounted variability / sampling variability): 2.98
## R^2 (amount of heterogeneity accounted for): 2.56%
##
## Test for Residual Heterogeneity:
## QE(df = 10) = 20.2478, p-val = 0.0270
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 10) = 0.9843, p-val = 0.3445
##
## Model Results:
##
## estimate se tval pval ci.lb
## intrcpt -0.3224 0.2348 -1.3732 0.1997 -0.8456
## intervention.experienceexperienced -0.4144 0.4177 -0.9921 0.3445 -1.3451
## ci.ub
## intrcpt 0.2007
## intervention.experienceexperienced 0.5163
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mbin_ponv_random,population)## Warning in rma.uni(yi = TE[!exclude], sei = seTE[!exclude], data = dataset, :
## Redundant predictors dropped from the model.##
## Mixed-Effects Model (k = 12; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.2884 (SE = 0.1877)
## tau (square root of estimated tau^2 value): 0.5370
## I^2 (residual heterogeneity / unaccounted variability): 68.24%
## H^2 (unaccounted variability / sampling variability): 3.15
## R^2 (amount of heterogeneity accounted for): 9.54%
##
## Test for Residual Heterogeneity:
## QE(df = 9) = 16.5738, p-val = 0.0558
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 9) = 1.4058, p-val = 0.2943
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt -0.3642 0.2766 -1.3167 0.2205 -0.9900 0.2615
## populationgeneral -0.0405 0.3791 -0.1068 0.9173 -0.8981 0.8171
## populationobese -1.7419 1.0477 -1.6626 0.1308 -4.1119 0.6282
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Subgroup analysis of studies with low risk of bias
ponv.bias<-dplyr::filter(ponv_analysis,ponv.bias=="low risk")
length(ponv.bias$ponv.e1)## [1] 3sum(ponv.bias$ponv.t1,ponv.bias$ponv.t2)## [1] 272mbin_ponv.bias_random<-meta::metabin(ponv.e1,ponv.t1,ponv.e2,ponv.t2,data = ponv.bias,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_ponv.bias_random## RR 95%-CI %W(random)
## Griffiths 2013 1.0019 [0.6828; 1.4700] 45.9
## Hohlrieder 2007 0.2000 [0.0461; 0.8670] 22.4
## Kang 2019 0.3333 [0.1223; 0.9089] 31.7
##
## Number of studies combined: k = 3
##
## RR 95%-CI t p-value
## Random effects model 0.4923 [0.0623; 3.8925] -1.47 0.2783
## Prediction interval [0.0000; 19321.6854]
##
## Quantifying heterogeneity:
## tau^2 = 0.4621 [0.0062; 26.5457]; tau = 0.6798 [0.0786; 5.1523];
## I^2 = 74.0% [13.1%; 92.2%]; H = 1.96 [1.07; 3.58]
##
## Test of heterogeneity:
## Q d.f. p-value
## 7.68 2 0.0215
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model#Funnel Plot for PONV
meta::funnel(mbin_ponv_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::metabias(mbin_ponv_random, method.bias = "linreg")##
## Linear regression test of funnel plot asymmetry
##
## data: mbin_ponv_random
## t = -4.1226, df = 10, p-value = 0.002068
## alternative hypothesis: asymmetry in funnel plot
## sample estimates:
## bias se.bias intercept
## -1.8306767 0.4440585 0.3495620dmetar::eggers.test(mbin_ponv_random)## Intercept ConfidenceInterval t p
## Egger's test -1.831 -2.615--1.047 -4.123 0.00207meta::trimfill(mbin_ponv_random)## RR 95%-CI %W(random)
## Abdel-Ghaffar 2022 0.1940 [0.0097; 3.8885] 2.6
## Carron 2012 0.1217 [0.0160; 0.9258] 4.3
## Du 2019 0.3333 [0.0730; 1.5213] 5.6
## Griffiths 2013 1.0019 [0.6828; 1.4700] 9.1
## Hohlrieder 2007 0.2000 [0.0461; 0.8670] 5.8
## Jeong 2004 1.0667 [0.6535; 1.7411] 8.8
## Kang 2019 0.3333 [0.1223; 0.9089] 7.3
## Kim 2021 1.0370 [0.7549; 1.4245] 9.2
## Ng 2021 1.0345 [0.0678; 15.7732] 3.0
## Parikh 2017 0.4000 [0.1409; 1.1354] 7.2
## Uerpairojkit 2009 1.1250 [0.6268; 2.0193] 8.6
## Ye 2020 0.5833 [0.3093; 1.1001] 8.5
## Filled: Kang 2019 2.6609 [0.9759; 7.2550] 7.3
## Filled: Hohlrieder 2007 4.4348 [1.0230; 19.2243] 5.8
## Filled: Abdel-Ghaffar 2022 4.5712 [0.2281; 91.6107] 2.6
## Filled: Carron 2012 7.2874 [0.9581; 55.4297] 4.3
##
## Number of studies combined: k = 16 (with 4 added studies)
##
## RR 95%-CI t p-value
## Random effects model 0.8377 [0.4766; 1.4723] -0.67 0.5135
## Prediction interval [0.1029; 6.8225]
##
## Quantifying heterogeneity:
## tau^2 = 0.8862 [0.0938; 2.5613]; tau = 0.9414 [0.3062; 1.6004];
## I^2 = 56.4% [23.7%; 75.1%]; H = 1.51 [1.14; 2.00]
##
## Test of heterogeneity:
## Q d.f. p-value
## 34.40 15 0.0030
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Trim-and-fill method to adjust for funnel plot asymmetryAbdominopelvic pain
abd.pain<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Pain.csv")
#Number of comparisons and patients meta-analized for abdominopelvic pain
length(abd.pain$abd.pain.m1)## [1] 6sum(abd.pain$abd.pain.f1,abd.pain$abd.pain.f2)## [1] 525#Meta-analysis for abdominopelvic pain
mcont_abd.pain<-meta::metacont(abd.pain.f1,abd.pain.m1,abd.pain.sd1,abd.pain.f2,abd.pain.m2,abd.pain.sd2,data=abd.pain,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,prediction = TRUE,sm="MD")
mcont_abd.pain## MD 95%-CI %W(random)
## Abdi 2010 0.0330 [-0.2672; 0.3332] 25.2
## Carron 2012 -0.4000 [-1.2278; 0.4278] 16.1
## Griffiths 2013 0.5000 [-0.2647; 1.2647] 17.2
## Hohlrieder 2007 -0.7000 [-1.2324; -0.1676] 21.3
## Kang 2019 -1.7000 [-2.8013; -0.5987] 12.2
## Koo 2003 -0.2000 [-1.7606; 1.3606] 7.8
##
## Number of studies combined: k = 6
##
## MD 95%-CI z p-value
## Random effects model -0.3435 [-0.8602; 0.1733] -1.30 0.1927
## Prediction interval [-1.9178; 1.2308]
##
## Quantifying heterogeneity:
## tau^2 = 0.2520 [0.0090; 2.8040]; tau = 0.5020 [0.0951; 1.6745];
## I^2 = 69.2% [27.5%; 86.9%]; H = 1.80 [1.17; 2.76]
##
## Test of heterogeneity:
## Q d.f. p-value
## 16.22 5 0.0062
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau#Forest plot for abdominopelvic pain
meta::forest(mcont_abd.pain,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for abdominopelvic pain
dmetar::find.outliers(mcont_abd.pain)## No outliers detected (random-effects model).#Meta-regression for mean abdominopelvic pain
#Controling for risk of bias
meta::metareg(mcont_abd.pain,abd.pain.bias)##
## Mixed-Effects Model (k = 6; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 0.4170 (SE = 0.4619)
## tau (square root of estimated tau^2 value): 0.6457
## I^2 (residual heterogeneity / unaccounted variability): 76.19%
## H^2 (unaccounted variability / sampling variability): 4.20
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 3) = 12.6017, p-val = 0.0056
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 0.3946, p-val = 0.8209
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -0.2000 1.0252 -0.1951 0.8453 -2.2093 1.8093
## abd.pain.biaslow risk -0.3559 1.1157 -0.3190 0.7497 -2.5426 1.8308
## abd.pain.biassome concerns 0.0489 1.1420 0.0428 0.9659 -2.1894 2.2871
##
## intrcpt
## abd.pain.biaslow risk
## abd.pain.biassome concerns
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for operator experience
meta::metareg(mcont_abd.pain,intervention.experience)##
## Mixed-Effects Model (k = 6; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 0.3458 (SE = 0.3874)
## tau (square root of estimated tau^2 value): 0.5881
## I^2 (residual heterogeneity / unaccounted variability): 75.31%
## H^2 (unaccounted variability / sampling variability): 4.05
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 4) = 16.2032, p-val = 0.0028
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.0084, p-val = 0.9270
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt -0.3898 0.4693 -0.8307 0.4062 -1.3097
## intervention.experienceexperienced 0.0552 0.6026 0.0917 0.9270 -1.1259
## ci.ub
## intrcpt 0.5300
## intervention.experienceexperienced 1.2363
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mcont_abd.pain,population)##
## Mixed-Effects Model (k = 6; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 0.2271 (SE = 0.2976)
## tau (square root of estimated tau^2 value): 0.4766
## I^2 (residual heterogeneity / unaccounted variability): 70.96%
## H^2 (unaccounted variability / sampling variability): 3.44
## R^2 (amount of heterogeneity accounted for): 9.88%
##
## Test for Residual Heterogeneity:
## QE(df = 3) = 10.3316, p-val = 0.0159
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 2.4679, p-val = 0.2911
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -0.0881 0.3170 -0.2778 0.7811 -0.7094 0.5333
## populationgeneral -1.0320 0.6584 -1.5675 0.1170 -2.3223 0.2584
## populationobese -0.3119 0.7113 -0.4385 0.6610 -1.7061 1.0823
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Estimating the mean abdominopelvic pain for endotracheal tubes
meta::metamean(n = abd.pain$abd.pain.f2,mean= abd.pain$abd.pain.m2, sd=abd.pain$abd.pain.sd2,studlab = study,data = abd.pain,method.tau = "SJ",sm = "MRAW",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## mean 95%-CI %W(random)
## Abdi 2010 2.8570 [2.6574; 3.0566] 18.7
## Carron 2012 2.7000 [2.0233; 3.3767] 16.7
## Griffiths 2013 3.0000 [2.4808; 3.5192] 17.5
## Hohlrieder 2007 4.0000 [3.5842; 4.4158] 18.0
## Kang 2019 5.3000 [4.4851; 6.1149] 15.9
## Koo 2003 4.3000 [3.0729; 5.5271] 13.2
##
## Number of studies combined: k = 6
##
## mean 95%-CI
## Random effects model 3.6393 [2.5782; 4.7004]
##
## Quantifying heterogeneity:
## tau^2 = 0.8966 [0.2709; 6.1243]; tau = 0.9469 [0.5205; 2.4747];
## I^2 = 91.1% [83.5%; 95.3%]; H = 3.36 [2.46; 4.59]
##
## Test of heterogeneity:
## Q d.f. p-value
## 56.47 5 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Untransformed (raw) means#Funnel Plot for abdominopelvic pain
meta::funnel(mcont_abd.pain,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
trimfill_abd.pain<-meta::trimfill(mcont_abd.pain)
trimfill_abd.pain## MD 95%-CI %W(random)
## Abdi 2010 0.0330 [-0.2672; 0.3332] 20.3
## Carron 2012 -0.4000 [-1.2278; 0.4278] 14.7
## Griffiths 2013 0.5000 [-0.2647; 1.2647] 15.4
## Hohlrieder 2007 -0.7000 [-1.2324; -0.1676] 18.0
## Kang 2019 -1.7000 [-2.8013; -0.5987] 11.8
## Koo 2003 -0.2000 [-1.7606; 1.3606] 8.1
## Filled: Kang 2019 1.4898 [ 0.3885; 2.5911] 11.8
##
## Number of studies combined: k = 7 (with 1 added studies)
##
## MD 95%-CI z p-value
## Random effects model -0.1422 [-0.7041; 0.4197] -0.50 0.6199
## Prediction interval [-1.8937; 1.6093]
##
## Quantifying heterogeneity:
## tau^2 = 0.3821 [0.1238; 3.5885]; tau = 0.6181 [0.3518; 1.8943];
## I^2 = 75.6% [48.5%; 88.5%]; H = 2.03 [1.39; 2.94]
##
## Test of heterogeneity:
## Q d.f. p-value
## 24.63 6 0.0004
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau
## - Trim-and-fill method to adjust for funnel plot asymmetryRegurgitation
regurgitation<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Regurgitation.csv")
length(regurgitation$regurgitation.e1)## [1] 20sum(regurgitation$regurgitation.t1,regurgitation$regurgitation.t2)## [1] 2212#Number of comparisons with zero regurgitation in both arms
regurgitation_zeros<-dplyr::filter(regurgitation,regurgitation$regurgitation.e1==0 & regurgitation$regurgitation.e2==0)
length(regurgitation_zeros$regurgitation.e1)## [1] 18#Table for Meta-analysis of regurgitation
regurgitation_analysis<-dplyr::filter(regurgitation,regurgitation$regurgitation.e1>0 | regurgitation$regurgitation.e2>0)
#Number of comparisons and patients meta-analized for regurgitation
length(regurgitation_analysis$regurgitation.e1)## [1] 2sum(regurgitation_analysis$regurgitation.t1,regurgitation_analysis$regurgitation.t2)## [1] 120#Meta-analysis for regurgitation
mbin_regurgitation_random<-meta::metabin(regurgitation.e1,regurgitation.t1,regurgitation.e2,regurgitation.t2,data = regurgitation_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_regurgitation_random## RR 95%-CI %W(random)
## Koo 2003 0.5000 [0.0492; 5.0831] 2.5
## Tosh 2021 0.9722 [0.8316; 1.1366] 97.5
##
## Number of studies combined: k = 2
##
## RR 95%-CI t p-value
## Random effects model 0.9563 [0.2569; 3.5596] -0.43 0.7405
##
## Quantifying heterogeneity:
## tau^2 = 0.0300; tau = 0.1733; I^2 = 0.0%; H = 1.00
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.31 1 0.5750
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Hartung-Knapp adjustment for random effects model#Estimated probability of regurgitation with endotracheal tubes
meta::metaprop(event = regurgitation.e2,n = regurgitation.t2 ,studlab = paste(study),data = regurgitation,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Badheka 2015 0.0000 [0.0000; 0.1157]
## Abdel-Ghaffar 2022 0.0000 [0.0000; 0.1089]
## Baik 2003 0.0000 [0.0000; 0.1058]
## Bhushan 2022 0.0000 [0.0000; 0.0544]
## Carron 2012 0.0000 [0.0000; 0.0949]
## Gulec 2012 0.0000 [0.0000; 0.1122]
## Hartmann 2001 0.0000 [0.0000; 0.0698]
## Hong 2011 0.0000 [0.0000; 0.1684]
## Ibrahim 2011 0.0000 [0.0000; 0.1157]
## Khan 2020 0.0000 [0.0000; 0.1372]
## Koo 2003 0.1000 [0.0123; 0.3170]
## Kuvaki 2019 0.0000 [0.0000; 0.0711]
## Lai 2017 0.0000 [0.0000; 0.1684]
## Maharhan 2013 0.0000 [0.0000; 0.1157]
## Panneer 2017 0.0000 [0.0000; 0.0881]
## Parikh 2017 0.0000 [0.0000; 0.1157]
## Saraswat 2011 0.0000 [0.0000; 0.1157]
## Tosh 2021 0.9000 [0.7634; 0.9721]
## Yao 2019 0.0000 [0.0000; 0.0080]
## Saini 2016 0.0000 [0.0000; 0.1157]
##
## Number of studies combined: k = 20
##
## proportion 95%-CI
## Random effects model 0.0000 [0.0000; 0.7152]
##
## Quantifying heterogeneity:
## tau^2 = 132.0910; tau = 11.4931; I^2 = 98.9%; H = 9.50
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 23.17 19 0.2298 Wald-type
## 291.79 19 < 0.0001 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)meta::metaprop(event = regurgitation.e1,n = regurgitation.t1 ,studlab = paste(study),data = regurgitation,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Badheka 2015 0.0000 [0.0000; 0.1157]
## Abdel-Ghaffar 2022 0.0000 [0.0000; 0.1058]
## Baik 2003 0.0000 [0.0000; 0.1028]
## Bhushan 2022 0.0000 [0.0000; 0.0544]
## Carron 2012 0.0000 [0.0000; 0.0925]
## Gulec 2012 0.0000 [0.0000; 0.1089]
## Hartmann 2001 0.0000 [0.0000; 0.0725]
## Hong 2011 0.0000 [0.0000; 0.1684]
## Ibrahim 2011 0.0000 [0.0000; 0.1157]
## Khan 2020 0.0000 [0.0000; 0.1372]
## Koo 2003 0.0500 [0.0013; 0.2487]
## Kuvaki 2019 0.0000 [0.0000; 0.0711]
## Lai 2017 0.0000 [0.0000; 0.1684]
## Maharhan 2013 0.0000 [0.0000; 0.1157]
## Panneer 2017 0.0000 [0.0000; 0.0881]
## Parikh 2017 0.0000 [0.0000; 0.1157]
## Saraswat 2011 0.0000 [0.0000; 0.1157]
## Tosh 2021 0.8750 [0.7320; 0.9581]
## Yao 2019 0.0000 [0.0000; 0.0080]
## Saini 2016 0.0000 [0.0000; 0.1157]
##
## Number of studies combined: k = 20
##
## proportion 95%-CI
## Random effects model 0.0000 [0.0000; 0.7146]
##
## Quantifying heterogeneity:
## tau^2 = 105.2047; tau = 10.2569; I^2 = 98.6%; H = 8.37
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 18.67 19 0.4784 Wald-type
## 279.40 19 < 0.0001 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)#Forest plot for regurgitation
meta::forest(mbin_regurgitation_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for regurgitation
dmetar::find.outliers(mbin_regurgitation_random)## No outliers detected (random-effects model).#Funnel Plot for regurgitation
meta::funnel(mbin_regurgitation_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
Pulmonary aspiration
aspiration<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Aspiration.csv")
length(aspiration$aspiration.e1)## [1] 20sum(aspiration$aspiration.t1,aspiration$aspiration.t2)## [1] 2189#Number of comparisons with zero aspiration in both arms
aspiration_zeros<-dplyr::filter(aspiration,aspiration$aspiration.e1==0 & aspiration$aspiration.e2==0)
length(aspiration_zeros$aspiration.e1)## [1] 20#Table for Meta-analysis of aspiration
aspiration_analysis<-dplyr::filter(aspiration,aspiration$aspiration.e1>0 | aspiration$aspiration.e2>0)
#Number of comparisons and patients meta-analized for aspiration
length(aspiration_analysis$aspiration.e1)## [1] 0sum(aspiration_analysis$aspiration.t1,aspiration_analysis$aspiration.t2)## [1] 0#Meta-analysis for aspiration
mbin_aspiration_random<-meta::metabin(aspiration.e1,aspiration.t1,aspiration.e2,aspiration.t2,data = aspiration,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_aspiration_random## RR 95%-CI %W(random)
## Badheka 2015 NA 0.0
## Baik 2003 NA 0.0
## Gulec 2012 NA 0.0
## Hong 2011 NA 0.0
## Ibrahim 2011 NA 0.0
## Khan 2020 NA 0.0
## Kim 2021 NA 0.0
## Koo 2003 NA 0.0
## Kuvaki 2019 NA 0.0
## Lai 2017 NA 0.0
## Maharhan 2013 NA 0.0
## Ng 2021 NA 0.0
## Panneer 2017 NA 0.0
## Parikh 2017 NA 0.0
## Sabuncu 2018 NA 0.0
## Saraswat 2011 NA 0.0
## Tosh 2021 NA 0.0
## Yao 2019 NA 0.0
## Ye 2020 NA 0.0
## Saini 2016 NA 0.0
##
## Number of studies combined: k = 0
##
## RR 95%-CI t p-value
## Random effects model NA -- --
##
## Quantifying heterogeneity:
## tau^2 = NA; tau = NA; I^2 = NA; H = NA
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Hartung-Knapp adjustment for random effects model#Forest plot for aspiration
meta::forest(mbin_aspiration_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
Major complications
major<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Major.csv")
length(major$major.e1)## [1] 16#Number of comparisons with zero major complications in both arms
major_zeros<-dplyr::filter(major,major$major.e1==0 & major$major.e2==0)
length(major_zeros$major.e1)## [1] 9#Table for Meta-analysis of major complications
major_analysis<-dplyr::filter(major,major$major.e1>0 | major$major.e2>0)
#Number of comparisons and patients meta-analized for major complications
length(major_analysis$major.e1)## [1] 7sum(major_analysis$major.t1,major_analysis$major.t2)## [1] 515#Meta-analysis for major complications
mbin_major_random<-meta::metabin(major.e1,major.t1,major.e2,major.t2,data = major_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_major_random## RR 95%-CI %W(random)
## Badheka 2015 0.3247 [0.0137; 7.7209] 8.5
## Abdel-Ghaffar 2022 0.3333 [0.0141; 7.8648] 8.6
## Baik 2003 0.5000 [0.1404; 1.7808] 43.5
## Bhushan 2022 1.0000 [0.0646; 15.4775] 11.2
## Dunnebier 2017 0.2068 [0.0104; 4.1280] 9.5
## Griffiths 2013 0.1474 [0.0062; 3.5226] 8.5
## Hohlrieder 2007 0.1111 [0.0062; 1.9974] 10.2
##
## Number of studies combined: k = 7
##
## RR 95%-CI t p-value
## Random effects model 0.3580 [0.1899; 0.6750] -3.96 0.0074
## Prediction interval [0.1198; 1.0697]
##
## Quantifying heterogeneity:
## tau^2 = 0.1141 [0.0000; 0.8180]; tau = 0.3378 [0.0000; 0.9045];
## I^2 = 0.0% [0.0%; 6.2%]; H = 1.00 [1.00; 1.03]
##
## Test of heterogeneity:
## Q d.f. p-value
## 1.87 6 0.9315
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Estimated probability of major complications with endotracheal tubes
meta::metaprop(event = major.e2,n = major.t2 ,studlab = paste(study),data = major,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Abdi 2010 0.0000 [0.0000; 0.1089]
## Ahn 2021 0.0000 [0.0000; 0.1157]
## Badheka 2015 0.0270 [0.0007; 0.1416]
## Abdel-Ghaffar 2022 0.0333 [0.0008; 0.1722]
## Baik 2003 0.2400 [0.0936; 0.4513]
## Bhushan 2022 0.0233 [0.0006; 0.1229]
## Biswas 2015 0.0000 [0.0000; 0.1684]
## Carron 2012 0.0000 [0.0000; 0.0402]
## Du 2019 0.0000 [0.0000; 0.1157]
## Dunnebier 2017 0.0667 [0.0082; 0.2207]
## Gombar 2012 0.0000 [0.0000; 0.0881]
## Griffiths 2013 0.0312 [0.0008; 0.1622]
## Gulec 2012 0.0000 [0.0000; 0.1157]
## Hartmann 2001 0.0000 [0.0000; 0.0080]
## Hohlrieder 2007 0.1000 [0.0279; 0.2366]
## Hong 2011 0.0000 [0.0000; 0.1157]
##
## Number of studies combined: k = 16
##
## proportion 95%-CI
## Random effects model 0.0080 [0.0018; 0.0346]
##
## Quantifying heterogeneity:
## tau^2 = 3.6704; tau = 1.9158; I^2 = 79.7%; H = 2.22
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 12.22 15 0.6622 Wald-type
## 59.42 15 < 0.0001 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)#Forest plot for major complications
meta::forest(mbin_major_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for major complications
dmetar::find.outliers(mbin_major_random)## No outliers detected (random-effects model).#Influence Analysis for major complications
inf_analysis_major<-dmetar::InfluenceAnalysis(mbin_major_random,random = TRUE)## [===========================================================================] DONEplot(inf_analysis_major,"baujat")
#Meta-regression for major complications
#Controling for risk of bias
meta::metareg(mbin_major_random,major.bias)## Warning in rma.uni(yi = TE[!exclude], sei = seTE[!exclude], data = dataset, :
## Redundant predictors dropped from the model.##
## Random-Effects Model (k = 7; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of total heterogeneity): 0.1141 (SE = 0.2335)
## tau (square root of estimated tau^2 value): 0.3378
## I^2 (total heterogeneity / total variability): 6.34%
## H^2 (total variability / sampling variability): 1.07
##
## Test for Heterogeneity:
## Q(df = 6) = 1.8667, p-val = 0.9315
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## -1.0272 0.2592 -3.9631 0.0074 -1.6615 -0.3930 **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for operator experience
meta::metareg(mbin_major_random,intervention.experience)##
## Mixed-Effects Model (k = 7; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.1123 (SE = 0.2563)
## tau (square root of estimated tau^2 value): 0.3351
## I^2 (residual heterogeneity / unaccounted variability): 5.91%
## H^2 (unaccounted variability / sampling variability): 1.06
## R^2 (amount of heterogeneity accounted for): 1.59%
##
## Test for Residual Heterogeneity:
## QE(df = 5) = 1.6773, p-val = 0.8917
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 5) = 0.7106, p-val = 0.4377
##
## Model Results:
##
## estimate se tval pval ci.lb
## intrcpt -1.1674 0.3136 -3.7224 0.0137 -1.9736
## intervention.experienceexperienced 0.4970 0.5896 0.8430 0.4377 -1.0187
## ci.ub
## intrcpt -0.3612 *
## intervention.experienceexperienced 2.0127
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mbin_major_random,population)## Warning in rma.uni(yi = TE[!exclude], sei = seTE[!exclude], data = dataset, :
## Redundant predictors dropped from the model.##
## Mixed-Effects Model (k = 7; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.0561 (SE = 0.2782)
## tau (square root of estimated tau^2 value): 0.2368
## I^2 (residual heterogeneity / unaccounted variability): 3.04%
## H^2 (unaccounted variability / sampling variability): 1.03
## R^2 (amount of heterogeneity accounted for): 50.87%
##
## Test for Residual Heterogeneity:
## QE(df = 4) = 0.5715, p-val = 0.9662
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 4) = 4.4585, p-val = 0.0959
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt -1.7642 0.3409 -5.1757 0.0066 -2.7105 -0.8178 **
## populationgeneral 1.1327 0.4038 2.8053 0.0485 0.0116 2.2537 *
## populationmale 0.1881 0.6700 0.2807 0.7929 -1.6721 2.0482
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Subgroup analysis of studies with low risk of bias
major.bias<-dplyr::filter(major_analysis,major.bias=="low risk")
length(major.bias$major.e1)## [1] 0sum(major.bias$major.t1,major.bias$major.t2)## [1] 0#Subgroup analysis of studies with women undergoing gynecological surgeries
major.women<-dplyr::filter(major_analysis,population=="famale")
length(major.women$major.e1)## [1] 3sum(major.women$major.t1,major.women$major.t2)## [1] 245mbin_major.women_random<-meta::metabin(major.e1,major.t1,major.e2,major.t2,data = major.women,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_major.women_random## RR 95%-CI %W(random)
## Abdel-Ghaffar 2022 0.3333 [0.0141; 7.8648] 31.4
## Griffiths 2013 0.1474 [0.0062; 3.5226] 31.1
## Hohlrieder 2007 0.1111 [0.0062; 1.9974] 37.5
##
## Number of studies combined: k = 3
##
## RR 95%-CI t p-value
## Random effects model 0.1713 [0.0416; 0.7052] -5.36 0.0330
## Prediction interval [0.0016; 18.1255]
##
## Quantifying heterogeneity:
## tau^2 = 0.0264 [0.0000; 10.3937]; tau = 0.1626 [0.0000; 3.2239];
## I^2 = 0.0% [0.0%; 21.6%]; H = 1.00 [1.00; 1.13]
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.27 2 0.8758
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Funnel Plot for major complications
meta::funnel(mbin_major_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::metabias(mbin_major_random, method.bias = "linreg")## Warning in print.metabias(x): Number of studies (k=7) too small to test for small study effects (k.min=10). Change argument 'k.min' if appropriate.dmetar::eggers.test(mbin_major_random)## Warning in dmetar::eggers.test(mbin_major_random): Your meta-analysis contains k
## = 7 studies. Egger's test may lack the statistical power to detect bias when the
## number of studies is small (i.e., k<10).## Intercept ConfidenceInterval t p
## Egger's test -0.729 -1.709-0.251 -1.392 0.22277meta::trimfill(mbin_major_random)## RR 95%-CI %W(random)
## Badheka 2015 0.3247 [0.0137; 7.7209] 7.3
## Abdel-Ghaffar 2022 0.3333 [0.0141; 7.8648] 7.3
## Baik 2003 0.5000 [0.1404; 1.7808] 28.4
## Bhushan 2022 1.0000 [0.0646; 15.4775] 9.4
## Dunnebier 2017 0.2068 [0.0104; 4.1280] 8.0
## Griffiths 2013 0.1474 [0.0062; 3.5226] 7.3
## Hohlrieder 2007 0.1111 [0.0062; 1.9974] 8.6
## Filled: Dunnebier 2017 1.2315 [0.0617; 24.5851] 8.0
## Filled: Griffiths 2013 1.7277 [0.0723; 41.2925] 7.3
## Filled: Hohlrieder 2007 2.2919 [0.1275; 41.2016] 8.6
##
## Number of studies combined: k = 10 (with 3 added studies)
##
## RR 95%-CI t p-value
## Random effects model 0.5042 [0.2558; 0.9938] -2.28 0.0483
## Prediction interval [0.1119; 2.2712]
##
## Quantifying heterogeneity:
## tau^2 = 0.3360 [0.0000; 1.3890]; tau = 0.5797 [0.0000; 1.1786];
## I^2 = 0.0% [0.0%; 21.7%]; H = 1.00 [1.00; 1.13]
##
## Test of heterogeneity:
## Q d.f. p-value
## 4.33 9 0.8887
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Trim-and-fill method to adjust for funnel plot asymmetryFailed first attempt
first<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA First.csv")
length(first$first.e1)## [1] 25#Number of comparisons with zero failed first attempts in both arms
first_zeros<-dplyr::filter(first,first$first.e1==0 & first$first.e2==0)
length(first_zeros$first.e1)## [1] 1#Table for Meta-analysis of failed first attempt
first_analysis<-dplyr::filter(first,first$first.e1>0 | first$first.e2>0)
#Number of comparisons and patients meta-analized for failed first attempt
length(first_analysis$first.e1)## [1] 24sum(first_analysis$first.t1,first_analysis$first.t2)## [1] 3069#Meta-analysis for failed first attempt
mbin_first_random<-meta::metabin(first.e1,first.t1,first.e2,first.t2,data = first_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_first_random## RR 95%-CI %W(random)
## Badheka 2015 11.0000 [0.6355; 190.3981] 2.3
## Abdel-Ghaffar 2022 8.7313 [0.4893; 155.7940] 2.3
## Bhushan 2022 0.7273 [0.3126; 1.6920] 6.5
## Carron 2012 0.9737 [0.3071; 3.0873] 5.7
## Dunnebier 2017 1.3750 [0.4603; 4.1070] 5.9
## Griffiths 2013 1.2881 [0.3016; 5.5025] 4.9
## Gulec 2012 0.6458 [0.1157; 3.6055] 4.2
## Hartmann 2001 0.1486 [0.0079; 2.8041] 2.2
## Ibrahim 2011 4.0000 [0.4744; 33.7292] 3.4
## Kang 2019 1.5000 [0.4742; 4.7449] 5.7
## Kuvaki 2019 0.8333 [0.2718; 2.5546] 5.8
## Lim 2007 0.3333 [0.0138; 8.0745] 2.0
## Lorenz 2009 0.7143 [0.2345; 2.1753] 5.8
## Ng 2021 0.6897 [0.1241; 3.8316] 4.3
## Panneer 2017 0.0588 [0.0035; 0.9856] 2.4
## Sabuncu 2018 0.1474 [0.0062; 3.5226] 2.0
## Saraswat 2011 0.8000 [0.2377; 2.6922] 5.5
## Tosh 2019 9.0000 [2.1833; 37.0991] 5.0
## Tosh 2021 3.0000 [0.8760; 10.2735] 5.5
## Uerpairojkit 2009 0.5000 [0.0464; 5.3871] 3.0
## Yao 2019 1.0000 [0.2516; 3.9744] 5.1
## Zhang 2015 0.6250 [0.1591; 2.4558] 5.1
## Ahmed 2015 0.1429 [0.0076; 2.6781] 2.3
## Saini 2016 0.5000 [0.0479; 5.2245] 3.0
##
## Number of studies combined: k = 24
##
## RR 95%-CI t p-value
## Random effects model 0.9979 [0.6268; 1.5889] -0.01 0.9927
## Prediction interval [0.1333; 7.4716]
##
## Quantifying heterogeneity:
## tau^2 = 0.8918 [0.0000; 1.8151]; tau = 0.9443 [0.0000; 1.3473];
## I^2 = 26.7% [0.0%; 55.6%]; H = 1.17 [1.00; 1.50]
##
## Test of heterogeneity:
## Q d.f. p-value
## 31.38 23 0.1137
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Estimated probability of failed first attempt with endotracheal tubes
meta::metaprop(event = first.e2,n = first.t2 ,studlab = paste(study),data = first,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Badheka 2015 0.0000 [0.0000; 0.1157]
## Abdel-Ghaffar 2022 0.0000 [0.0000; 0.1089]
## Bhushan 2022 0.1667 [0.0862; 0.2787]
## Carron 2012 0.1351 [0.0454; 0.2877]
## Dunnebier 2017 0.1818 [0.0519; 0.4028]
## Griffiths 2013 0.0526 [0.0110; 0.1462]
## Gulec 2012 0.0968 [0.0204; 0.2575]
## Hartmann 2001 0.0588 [0.0123; 0.1624]
## Ibrahim 2011 0.0333 [0.0008; 0.1722]
## Kang 2019 0.1429 [0.0403; 0.3267]
## Kuvaki 2019 0.1200 [0.0453; 0.2431]
## Lim 2007 0.0111 [0.0003; 0.0604]
## Lorenz 2009 0.0700 [0.0286; 0.1389]
## Ng 2021 0.1000 [0.0211; 0.2653]
## Panneer 2017 0.2000 [0.0905; 0.3565]
## Parikh 2017 0.0000 [0.0000; 0.1157]
## Sabuncu 2018 0.0312 [0.0008; 0.1622]
## Saraswat 2011 0.1667 [0.0564; 0.3472]
## Tosh 2019 0.0333 [0.0041; 0.1153]
## Tosh 2021 0.0750 [0.0157; 0.2039]
## Uerpairojkit 2009 0.0290 [0.0035; 0.1008]
## Yao 2019 0.0087 [0.0024; 0.0221]
## Zhang 2015 0.1000 [0.0211; 0.2653]
## Ahmed 2015 0.0750 [0.0157; 0.2039]
## Saini 2016 0.0667 [0.0082; 0.2207]
##
## Number of studies combined: k = 25
##
## proportion 95%-CI
## Random effects model 0.0589 [0.0397; 0.0866]
##
## Quantifying heterogeneity:
## tau^2 = 0.6549; tau = 0.8093; I^2 = 66.1%; H = 1.72
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 54.35 24 0.0004 Wald-type
## 89.49 24 < 0.0001 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)#Forest plot for failed first attempt
meta::forest(mbin_first_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for failed first attempt
dmetar::find.outliers(mbin_first_random)## Identified outliers (random-effects model)
## ------------------------------------------
## "Tosh 2019"
##
## Results with outliers removed
## -----------------------------
## RR 95%-CI %W(random) exclude
## Badheka 2015 11.0000 [0.6355; 190.3981] 2.3
## Abdel-Ghaffar 2022 8.7313 [0.4893; 155.7940] 2.3
## Bhushan 2022 0.7273 [0.3126; 1.6920] 7.3
## Carron 2012 0.9737 [0.3071; 3.0873] 6.2
## Dunnebier 2017 1.3750 [0.4603; 4.1070] 6.4
## Griffiths 2013 1.2881 [0.3016; 5.5025] 5.2
## Gulec 2012 0.6458 [0.1157; 3.6055] 4.4
## Hartmann 2001 0.1486 [0.0079; 2.8041] 2.2
## Ibrahim 2011 4.0000 [0.4744; 33.7292] 3.4
## Kang 2019 1.5000 [0.4742; 4.7449] 6.2
## Kuvaki 2019 0.8333 [0.2718; 2.5546] 6.3
## Lim 2007 0.3333 [0.0138; 8.0745] 1.9
## Lorenz 2009 0.7143 [0.2345; 2.1753] 6.3
## Ng 2021 0.6897 [0.1241; 3.8316] 4.4
## Panneer 2017 0.0588 [0.0035; 0.9856] 2.3
## Sabuncu 2018 0.1474 [0.0062; 3.5226] 2.0
## Saraswat 2011 0.8000 [0.2377; 2.6922] 6.0
## Tosh 2019 9.0000 [2.1833; 37.0991] 0.0 *
## Tosh 2021 3.0000 [0.8760; 10.2735] 5.9
## Uerpairojkit 2009 0.5000 [0.0464; 5.3871] 3.0
## Yao 2019 1.0000 [0.2516; 3.9744] 5.4
## Zhang 2015 0.6250 [0.1591; 2.4558] 5.4
## Ahmed 2015 0.1429 [0.0076; 2.6781] 2.2
## Saini 2016 0.5000 [0.0479; 5.2245] 3.0
##
## Number of studies combined: k = 23
##
## RR 95%-CI t p-value
## Random effects model 0.8970 [0.5887; 1.3667] -0.54 0.5979
## Prediction interval [0.1479; 5.4413]
##
## Quantifying heterogeneity:
## tau^2 = 0.7102 [0.0000; 1.3245]; tau = 0.8427 [0.0000; 1.1509];
## I^2 = 0.3% [0.0%; 45.5%]; H = 1.00 [1.00; 1.35]
##
## Test of heterogeneity:
## Q d.f. p-value
## 22.06 22 0.4562
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Influence Analysis for failed first attempt
inf_analysis_first<-dmetar::InfluenceAnalysis(mbin_first_random,random = TRUE)## [===========================================================================] DONEplot(inf_analysis_first,"baujat")
#Meta-regression for failed first attempt
#Controling for risk of bias
meta::metareg(mbin_first_random,first.bias)##
## Mixed-Effects Model (k = 24; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.9252 (SE = 0.3613)
## tau (square root of estimated tau^2 value): 0.9619
## I^2 (residual heterogeneity / unaccounted variability): 60.65%
## H^2 (unaccounted variability / sampling variability): 2.54
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 22) = 30.8133, p-val = 0.1000
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 22) = 0.2222, p-val = 0.6420
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt -0.4700 1.0157 -0.4628 0.6481 -2.5763 1.6363
## first.biassome concerns 0.4914 1.0426 0.4713 0.6420 -1.6708 2.6537
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for operator experience
meta::metareg(mbin_first_random,intervention.experience)##
## Mixed-Effects Model (k = 24; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.9728 (SE = 0.3800)
## tau (square root of estimated tau^2 value): 0.9863
## I^2 (residual heterogeneity / unaccounted variability): 60.39%
## H^2 (unaccounted variability / sampling variability): 2.52
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 21) = 29.8705, p-val = 0.0946
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 21) = 0.0717, p-val = 0.9311
##
## Model Results:
##
## estimate se tval pval
## intrcpt 0.0449 0.3133 0.1435 0.8873
## intervention.experienceexperienced -0.0774 0.4978 -0.1555 0.8779
## intervention.experienceinexperienced -0.3814 1.0351 -0.3685 0.7162
## ci.lb ci.ub
## intrcpt -0.6066 0.6965
## intervention.experienceexperienced -1.1126 0.9578
## intervention.experienceinexperienced -2.5341 1.7713
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mbin_first_random,population)##
## Mixed-Effects Model (k = 24; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.8327 (SE = 0.3716)
## tau (square root of estimated tau^2 value): 0.9125
## I^2 (residual heterogeneity / unaccounted variability): 56.60%
## H^2 (unaccounted variability / sampling variability): 2.30
## R^2 (amount of heterogeneity accounted for): 6.62%
##
## Test for Residual Heterogeneity:
## QE(df = 19) = 26.0791, p-val = 0.1280
##
## Test of Moderators (coefficients 2:5):
## F(df1 = 4, df2 = 19) = 1.3162, p-val = 0.2998
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt 0.0391 0.4535 0.0863 0.9321 -0.9101 0.9884
## populationgeneral 0.1414 0.5363 0.2637 0.7948 -0.9810 1.2638
## populationmale 0.2793 1.0024 0.2786 0.7835 -1.8187 2.3773
## populationobese -0.0658 1.0145 -0.0649 0.9490 -2.1892 2.0576
## populationpregnant -1.7629 0.9131 -1.9308 0.0686 -3.6740 0.1482 .
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Subgroup analysis of studies with low risk of bias
first.bias<-dplyr::filter(first_analysis,first.bias=="low risk")
length(first.bias$first.e1)## [1] 0sum(first.bias$first.t1,first.bias$first.t2)## [1] 0#Subgroup analysis of studies with pregnant women
first.pregnant<-dplyr::filter(first_analysis,population=="pregnant")
length(first.pregnant$first.e1)## [1] 3sum(first.pregnant$first.t1,first.pregnant$first.t2)## [1] 220mbin_first.pregnant_random<-meta::metabin(first.e1,first.t1,first.e2,first.t2,data = first.pregnant,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_first.pregnant_random## RR 95%-CI %W(random)
## Panneer 2017 0.0588 [0.0035; 0.9856] 30.4
## Ahmed 2015 0.1429 [0.0076; 2.6781] 28.4
## Saini 2016 0.5000 [0.0479; 5.2245] 41.2
##
## Number of studies combined: k = 3
##
## RR 95%-CI t p-value
## Random effects model 0.1827 [0.0115; 2.8962] -2.65 0.1180
## Prediction interval [0.0000; 12850.9173]
##
## Quantifying heterogeneity:
## tau^2 = 0.3591 [0.0000; 43.9459]; tau = 0.5993 [0.0000; 6.6292];
## I^2 = 0.0% [0.0%; 84.6%]; H = 1.00 [1.00; 2.55]
##
## Test of heterogeneity:
## Q d.f. p-value
## 1.35 2 0.5081
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Funnel Plot for failed first attempt
meta::funnel(mbin_first_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::metabias(mbin_first_random, method.bias = "linreg")##
## Linear regression test of funnel plot asymmetry
##
## data: mbin_first_random
## t = -0.78869, df = 22, p-value = 0.4387
## alternative hypothesis: asymmetry in funnel plot
## sample estimates:
## bias se.bias intercept
## -0.5158383 0.6540440 0.4082336dmetar::eggers.test(mbin_first_random)## Intercept ConfidenceInterval t p
## Egger's test -0.516 -1.888-0.856 -0.789 0.43871meta::trimfill(mbin_first_random)## RR 95%-CI %W(random)
## Badheka 2015 11.0000 [0.6355; 190.3981] 2.4
## Abdel-Ghaffar 2022 8.7313 [0.4893; 155.7940] 2.3
## Bhushan 2022 0.7273 [0.3126; 1.6920] 5.5
## Carron 2012 0.9737 [0.3071; 3.0873] 4.9
## Dunnebier 2017 1.3750 [0.4603; 4.1070] 5.1
## Griffiths 2013 1.2881 [0.3016; 5.5025] 4.4
## Gulec 2012 0.6458 [0.1157; 3.6055] 3.9
## Hartmann 2001 0.1486 [0.0079; 2.8041] 2.3
## Ibrahim 2011 4.0000 [0.4744; 33.7292] 3.3
## Kang 2019 1.5000 [0.4742; 4.7449] 5.0
## Kuvaki 2019 0.8333 [0.2718; 2.5546] 5.0
## Lim 2007 0.3333 [0.0138; 8.0745] 2.1
## Lorenz 2009 0.7143 [0.2345; 2.1753] 5.0
## Ng 2021 0.6897 [0.1241; 3.8316] 3.9
## Panneer 2017 0.0588 [0.0035; 0.9856] 2.4
## Sabuncu 2018 0.1474 [0.0062; 3.5226] 2.1
## Saraswat 2011 0.8000 [0.2377; 2.6922] 4.8
## Tosh 2019 9.0000 [2.1833; 37.0991] 4.5
## Tosh 2021 3.0000 [0.8760; 10.2735] 4.8
## Uerpairojkit 2009 0.5000 [0.0464; 5.3871] 2.9
## Yao 2019 1.0000 [0.2516; 3.9744] 4.5
## Zhang 2015 0.6250 [0.1591; 2.4558] 4.6
## Ahmed 2015 0.1429 [0.0076; 2.6781] 2.3
## Saini 2016 0.5000 [0.0479; 5.2245] 3.0
## Filled: Hartmann 2001 8.9233 [0.4730; 168.3515] 2.3
## Filled: Sabuncu 2018 8.9981 [0.3765; 215.0527] 2.1
## Filled: Ahmed 2015 9.2838 [0.4952; 174.0377] 2.3
## Filled: Panneer 2017 22.5463 [1.3456; 377.7707] 2.4
##
## Number of studies combined: k = 28 (with 4 added studies)
##
## RR 95%-CI t p-value
## Random effects model 1.2270 [0.7395; 2.0361] 0.83 0.4144
## Prediction interval [0.1129; 13.3356]
##
## Quantifying heterogeneity:
## tau^2 = 1.2863 [0.0000; 2.4682]; tau = 1.1341 [0.0000; 1.5711];
## I^2 = 34.9% [0.0%; 58.9%]; H = 1.24 [1.00; 1.56]
##
## Test of heterogeneity:
## Q d.f. p-value
## 41.48 27 0.0370
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Trim-and-fill method to adjust for funnel plot asymmetryFailed insertion
fail<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Fail.csv")
length(fail$fail.e1)## [1] 26#Number of comparisons with zero failed insertions in both arms
fail_zeros<-dplyr::filter(fail,fail$fail.e1==0 & fail$fail.e2==0)
length(fail_zeros$fail.e1)## [1] 22#Table for Meta-analysis of failed insertion
fail_analysis<-dplyr::filter(fail,fail$fail.e1>0 | fail$fail.e2>0)
#Number of comparisons and patients meta-analized for failed insertion
length(fail_analysis$fail.e1)## [1] 4sum(fail_analysis$fail.t1,fail_analysis$fail.t2)## [1] 414#Meta-analysis for failed insertion
mbin_fail_random<-meta::metabin(fail.e1,fail.t1,fail.e2,fail.t2,data = fail_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_fail_random## RR 95%-CI %W(random)
## Dunnebier 2017 19.7368 [1.2341; 315.6452] 27.7
## Nagahisa 2017 3.0000 [0.1242; 72.4465] 24.0
## Uerpairojkit 2009 0.3333 [0.0138; 8.0420] 24.1
## Ahmed 2015 0.3333 [0.0140; 7.9424] 24.2
##
## Number of studies combined: k = 4
##
## RR 95%-CI t p-value
## Random effects model 1.7534 [0.0718; 42.8014] 0.56 0.6149
## Prediction interval [0.0008; 3770.7562]
##
## Quantifying heterogeneity:
## tau^2 = 2.1727 [0.0000; 51.6013]; tau = 1.4740 [0.0000; 7.1834];
## I^2 = 41.4% [0.0%; 80.3%]; H = 1.31 [1.00; 2.25]
##
## Test of heterogeneity:
## Q d.f. p-value
## 5.12 3 0.1630
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Estimated probability of failed insertion with endotracheal tubes
meta::metaprop(event = fail.e2,n = fail.t2 ,studlab = paste(study),data = fail,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Abdi 2010 0.0000 [0.0000; 0.0521]
## Ahn 2021 0.0000 [0.0000; 0.1089]
## Badheka 2015 0.0000 [0.0000; 0.1157]
## Abdel-Ghaffar 2022 0.0000 [0.0000; 0.1058]
## Baik 2003 0.0000 [0.0000; 0.1058]
## Bhushan 2022 0.0000 [0.0000; 0.0544]
## Dunnebier 2017 0.0000 [0.0000; 0.1544]
## Griffiths 2013 0.0000 [0.0000; 0.0627]
## Gulec 2012 0.0000 [0.0000; 0.1122]
## Ibrahim 2011 0.0000 [0.0000; 0.1157]
## Kang 2019 0.0000 [0.0000; 0.1234]
## Kuvaki 2019 0.0000 [0.0000; 0.0711]
## Lim 2007 0.0000 [0.0000; 0.0402]
## Lorenz 2009 0.0000 [0.0000; 0.0362]
## Nagahisa 2017 0.0000 [0.0000; 0.0493]
## Ng 2021 0.0000 [0.0000; 0.1157]
## Panneer 2017 0.0000 [0.0000; 0.0881]
## Parikh 2017 0.0000 [0.0000; 0.1157]
## Sabuncu 2018 0.0000 [0.0000; 0.1089]
## Saraswat 2011 0.0000 [0.0000; 0.1157]
## Tosh 2019 0.0000 [0.0000; 0.0596]
## Tosh 2021 0.0000 [0.0000; 0.0881]
## Uerpairojkit 2009 0.0145 [0.0004; 0.0781]
## Yao 2019 0.0000 [0.0000; 0.0080]
## Ahmed 2015 0.0250 [0.0006; 0.1316]
## Saini 2016 0.0000 [0.0000; 0.1157]
##
## Number of studies combined: k = 26
##
## proportion 95%-CI
## Random effects model 0.0009 [0.0001; 0.0146]
##
## Quantifying heterogeneity:
## tau^2 = 0.7430; tau = 0.8619; I^2 = 28.3%; H = 1.18
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 0.15 25 1.0000 Wald-type
## 10.94 25 0.9932 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)#Forest plot for failed insertion
meta::forest(mbin_fail_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for failed insertion
dmetar::find.outliers(mbin_fail_random)## No outliers detected (random-effects model).#Meta-regression for failed insertion
#Controling for operator experience
meta::metareg(mbin_fail_random,intervention.experience)## Warning in rma.uni(yi = TE[!exclude], sei = seTE[!exclude], data = dataset, :
## Redundant predictors dropped from the model.##
## Random-Effects Model (k = 4; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of total heterogeneity): 2.1727 (SE = 2.0492)
## tau (square root of estimated tau^2 value): 1.4740
## I^2 (total heterogeneity / total variability): 46.98%
## H^2 (total variability / sampling variability): 1.89
##
## Test for Heterogeneity:
## Q(df = 3) = 5.1232, p-val = 0.1630
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## 0.5616 1.0039 0.5594 0.6149 -2.6334 3.7566
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Influence Analysis for failed insertion
inf_analysis_fail<-dmetar::InfluenceAnalysis(mbin_fail_random,random = TRUE)## [===========================================================================] DONEplot(inf_analysis_fail,"baujat")
#Funnel Plot for failed insertion
meta::funnel(mbin_fail_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::trimfill(mbin_fail_random)## RR 95%-CI %W(random)
## Dunnebier 2017 19.7368 [1.2341; 315.6452] 27.7
## Nagahisa 2017 3.0000 [0.1242; 72.4465] 24.0
## Uerpairojkit 2009 0.3333 [0.0138; 8.0420] 24.1
## Ahmed 2015 0.3333 [0.0140; 7.9424] 24.2
##
## Number of studies combined: k = 4 (with 0 added studies)
##
## RR 95%-CI t p-value
## Random effects model 1.7534 [0.0718; 42.8014] 0.56 0.6149
## Prediction interval [0.0008; 3770.7562]
##
## Quantifying heterogeneity:
## tau^2 = 2.1727 [0.0000; 51.6013]; tau = 1.4740 [0.0000; 7.1834];
## I^2 = 41.4% [0.0%; 80.3%]; H = 1.31 [1.00; 2.25]
##
## Test of heterogeneity:
## Q d.f. p-value
## 5.12 3 0.1630
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Trim-and-fill method to adjust for funnel plot asymmetryLeak pressure
seal.pressure<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Seal Pressure.csv", header = TRUE)## Warning in read.table(file = file, header = header, sep = sep, quote = quote, :
## incomplete final line found by readTableHeader on '~/Desktop/Systematic Reviews/
## SR SGA vs TT/Tables for Analyses/SGA Seal Pressure.csv'#Number of comparisons and patients meta-analized for leak pressure
length(seal.pressure$seal.pressure.m1)## [1] 2sum(seal.pressure$seal.pressure.f1,seal.pressure$seal.pressure.f2)## [1] 1120#Meta-analysis for leak pressure
mcont_seal.pressure<-meta::metacont(seal.pressure.f1,seal.pressure.m1,seal.pressure.sd1,seal.pressure.f2,seal.pressure.m2,seal.pressure.sd2,data=seal.pressure,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,prediction = TRUE,sm="MD")
mcont_seal.pressure## MD 95%-CI %W(random)
## Lorenz 2009 1.0000 [-0.2908; 2.2908] 44.2
## Yao 2019 -0.8000 [-1.2783; -0.3217] 55.8
##
## Number of studies combined: k = 2
##
## MD 95%-CI z p-value
## Random effects model -0.0039 [-1.7561; 1.7482] -0.00 0.9965
##
## Quantifying heterogeneity:
## tau^2 = 1.3734; tau = 1.1719; I^2 = 84.8% [37.9%; 96.3%]; H = 2.56 [1.27; 5.18]
##
## Test of heterogeneity:
## Q d.f. p-value
## 6.57 1 0.0104
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2#Forest plot for leak pressure
meta::forest(mcont_seal.pressure,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for leak pressure
dmetar::find.outliers(mcont_seal.pressure)## No outliers detected (random-effects model).#Estimating the mean leak pressure for endotracheal tubes
meta::metamean(n = seal.pressure$seal.pressure.f2,mean= seal.pressure$seal.pressure.m2, sd=seal.pressure$seal.pressure.sd2,studlab = study,data = seal.pressure,method.tau = "SJ",sm = "MRAW",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## mean 95%-CI %W(random)
## Lorenz 2009 34.5000 [33.5396; 35.4604] 49.8
## Yao 2019 27.9000 [27.5710; 28.2290] 50.2
##
## Number of studies combined: k = 2
##
## mean 95%-CI
## Random effects model 31.1838 [-10.7461; 73.1138]
##
## Quantifying heterogeneity:
## tau^2 = 21.5150; tau = 4.6384; I^2 = 99.4% [98.9%; 99.7%]; H = 12.74 [9.51; 17.08]
##
## Test of heterogeneity:
## Q d.f. p-value
## 162.37 1 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Untransformed (raw) means#Funnel Plot for leak pressure
meta::funnel(mcont_seal.pressure,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
trimfill_seal.pressure<-meta::trimfill(mcont_seal.pressure)## Warning in trimfill.meta(mcont_seal.pressure): Minimal number of three studies
## for trim-and-fill methodtrimfill_seal.pressure## NULLLeak fraction
leak.fraction<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Leak Fraction.csv", header = TRUE)## Warning in read.table(file = file, header = header, sep = sep, quote = quote, :
## incomplete final line found by readTableHeader on '~/Desktop/Systematic Reviews/
## SR SGA vs TT/Tables for Analyses/SGA Leak Fraction.csv'#Number of comparisons and patients meta-analized for leak fraction
length(leak.fraction$leak.fraction.m1)## [1] 4sum(leak.fraction$leak.fraction.f1,leak.fraction$leak.fraction.f2)## [1] 240#Meta-analysis for leak fraction
mcont_leak.fraction<-meta::metacont(leak.fraction.f1,leak.fraction.m1,leak.fraction.sd1,leak.fraction.f2,leak.fraction.m2,leak.fraction.sd2,data=leak.fraction,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,prediction = TRUE,sm="MD")
mcont_leak.fraction## MD 95%-CI %W(random)
## Abdel-Ghaffar 2022 4.2900 [-0.2797; 8.8597] 15.7
## Carron 2012 6.4000 [ 3.2042; 9.5958] 21.8
## Lai 2017 0.7000 [-0.7933; 2.1933] 30.5
## Maharhan 2013 -0.1600 [-1.2745; 0.9545] 32.1
##
## Number of studies combined: k = 4
##
## MD 95%-CI z p-value
## Random effects model 2.2290 [-0.2349; 4.6930] 1.77 0.0762
## Prediction interval [-8.4739; 12.9320]
##
## Quantifying heterogeneity:
## tau^2 = 4.6074 [1.1677; >100.0000]; tau = 2.1465 [1.0806; >10.0000];
## I^2 = 82.2% [54.2%; 93.1%]; H = 2.37 [1.48; 3.81]
##
## Test of heterogeneity:
## Q d.f. p-value
## 16.88 3 0.0007
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau#Forest plot for leak fraction
meta::forest(mcont_leak.fraction,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for leak fraction
dmetar::find.outliers(mcont_leak.fraction)## No outliers detected (random-effects model).#Meta-regression for leak fraction
#Controling for operator experience
meta::metareg(mcont_leak.fraction, intervention.experience)##
## Mixed-Effects Model (k = 4; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 0.7477 (SE = 1.6601)
## tau (square root of estimated tau^2 value): 0.8647
## I^2 (residual heterogeneity / unaccounted variability): 48.17%
## H^2 (unaccounted variability / sampling variability): 1.93
## R^2 (amount of heterogeneity accounted for): 83.77%
##
## Test for Residual Heterogeneity:
## QE(df = 2) = 3.8587, p-val = 0.1452
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 8.5820, p-val = 0.0034
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt 0.5797 0.7356 0.7881 0.4307 -0.8620
## intervention.experienceexperienced 5.8203 1.9868 2.9295 0.0034 1.9263
## ci.ub
## intrcpt 2.0213
## intervention.experienceexperienced 9.7144 **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mcont_leak.fraction,population)##
## Mixed-Effects Model (k = 4; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 3.4358 (SE = 9.1133)
## tau (square root of estimated tau^2 value): 1.8536
## I^2 (residual heterogeneity / unaccounted variability): 53.32%
## H^2 (unaccounted variability / sampling variability): 2.14
## R^2 (amount of heterogeneity accounted for): 25.43%
##
## Test for Residual Heterogeneity:
## QE(df = 1) = 2.1421, p-val = 0.1433
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 4.4213, p-val = 0.1096
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt 1.8187 1.6627 1.0938 0.2740 -1.4402 5.0776
## populationgeneral -1.9787 2.5542 -0.7747 0.4385 -6.9848 3.0274
## populationobese 4.5813 2.9764 1.5392 0.1238 -1.2524 10.4149
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Estimating the mean leak fraction for endotracheal tubes
meta::metamean(n = leak.fraction$leak.fraction.f2,mean= leak.fraction$leak.fraction.m2, sd=leak.fraction$leak.fraction.sd2,studlab = study,data = leak.fraction,method.tau = "SJ",sm = "MRAW",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## mean 95%-CI %W(random)
## Abdel-Ghaffar 2022 7.9000 [4.4491; 11.3509] 19.9
## Carron 2012 0.5000 [0.1778; 0.8222] 27.0
## Lai 2017 5.9000 [5.1681; 6.6319] 26.6
## Maharhan 2013 4.6800 [3.8570; 5.5030] 26.5
##
## Number of studies combined: k = 4
##
## mean 95%-CI
## Random effects model 4.5193 [-0.3925; 9.4311]
##
## Quantifying heterogeneity:
## tau^2 = 8.6730 [2.4027; >100.0000]; tau = 2.9450 [1.5501; >10.0000];
## I^2 = 98.8% [98.1%; 99.2%]; H = 8.98 [7.23; 11.17]
##
## Test of heterogeneity:
## Q d.f. p-value
## 242.12 3 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Untransformed (raw) means#Funnel Plot for leak fraction
meta::funnel(mcont_leak.fraction,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
trimfill_leak.fraction<-meta::trimfill(mcont_leak.fraction)
trimfill_leak.fraction## MD 95%-CI %W(random)
## Abdel-Ghaffar 2022 4.2900 [-0.2797; 8.8597] 13.1
## Carron 2012 6.4000 [ 3.2042; 9.5958] 16.4
## Lai 2017 0.7000 [-0.7933; 2.1933] 20.2
## Maharhan 2013 -0.1600 [-1.2745; 0.9545] 20.8
## Filled: Abdel-Ghaffar 2022 -3.9947 [-8.5644; 0.5750] 13.1
## Filled: Carron 2012 -6.1047 [-9.3005; -2.9089] 16.4
##
## Number of studies combined: k = 6 (with 2 added studies)
##
## MD 95%-CI z p-value
## Random effects model 0.1953 [-2.4413; 2.8318] 0.15 0.8846
## Prediction interval [-8.6660; 9.0565]
##
## Quantifying heterogeneity:
## tau^2 = 8.3767 [4.0071; 102.4653]; tau = 2.8942 [2.0018; 10.1225];
## I^2 = 86.3% [72.4%; 93.2%]; H = 2.70 [1.90; 3.84]
##
## Test of heterogeneity:
## Q d.f. p-value
## 36.54 5 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau
## - Trim-and-fill method to adjust for funnel plot asymmetryGastric insufflation
gastric.insuf<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Insuflation.csv")
length(gastric.insuf$gastric.insuf.e1)## [1] 10#Number of comparisons with zero gastric insufflation in both arms
gastric.insuf_zeros<-dplyr::filter(gastric.insuf,gastric.insuf$gastric.insuf.e1==0 & gastric.insuf$gastric.insuf.e2==0)
length(gastric.insuf_zeros$gastric.insuf.e1)## [1] 7#Table for Meta-analysis of gastric insufflation
gastric.insuf_analysis<-dplyr::filter(gastric.insuf,gastric.insuf$gastric.insuf.e1>0 | gastric.insuf$gastric.insuf.e2>0)
#Number of comparisons and patients meta-analized for gastric insufflation
length(gastric.insuf_analysis$gastric.insuf.e1)## [1] 3sum(gastric.insuf_analysis$gastric.insuf.t1,gastric.insuf_analysis$gastric.insuf.t2)## [1] 202#Meta-analysis for gastric insufflation
mbin_gastric.insuf_random<-meta::metabin(gastric.insuf.e1,gastric.insuf.t1,gastric.insuf.e2,gastric.insuf.t2,data = gastric.insuf_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_gastric.insuf_random## RR 95%-CI %W(random)
## Baik 2003 1.1647 [0.3932; 3.4499] 56.0
## Carron 2012 1.4605 [0.2587; 8.2464] 30.9
## Saraswat 2011 7.0000 [0.3774; 129.8374] 13.1
##
## Number of studies combined: k = 3
##
## RR 95%-CI t p-value
## Random effects model 1.5795 [0.2653; 9.4045] 1.10 0.3852
## Prediction interval [0.0003; 7821.8350]
##
## Quantifying heterogeneity:
## tau^2 = 0.2764 [0.0000; 36.0648]; tau = 0.5257 [0.0000; 6.0054];
## I^2 = 0.0% [0.0%; 83.7%]; H = 1.00 [1.00; 2.47]
##
## Test of heterogeneity:
## Q d.f. p-value
## 1.27 2 0.5292
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Estimated probability of gastric insufflation with endotracheal tubes
meta::metaprop(event = gastric.insuf.e2,n = gastric.insuf.t2 ,studlab = paste(study),data = gastric.insuf,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Baik 2003 0.1515 [0.0511; 0.3190]
## Bhushan 2022 0.0000 [0.0000; 0.0544]
## Carron 2012 0.0541 [0.0066; 0.1819]
## Gulec 2012 0.0000 [0.0000; 0.1122]
## Hartmann 2001 0.0000 [0.0000; 0.0698]
## Ibrahim 2011 0.0000 [0.0000; 0.1157]
## Kang 2019 0.0000 [0.0000; 0.1234]
## Khan 2020 0.0000 [0.0000; 0.1372]
## Lim 2007 0.0000 [0.0000; 0.0402]
## Saraswat 2011 0.0000 [0.0000; 0.1157]
##
## Number of studies combined: k = 10
##
## proportion 95%-CI
## Random effects model 0.0012 [0.0000; 0.1017]
##
## Quantifying heterogeneity:
## tau^2 = 8.3874; tau = 2.8961; I^2 = 87.8%; H = 2.87
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 1.70 9 0.9954 Wald-type
## 27.60 9 0.0011 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)#Forest plot for gastric insufflation
meta::forest(mbin_gastric.insuf_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for gastric insufflation
dmetar::find.outliers(mbin_gastric.insuf_random)## No outliers detected (random-effects model).#Meta-regression for gastric insufflation
#Controling for operator experience
meta::metareg(mbin_gastric.insuf_random, intervention.experience)##
## Mixed-Effects Model (k = 3; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.5384 (SE = 0.8532)
## tau (square root of estimated tau^2 value): 0.7338
## I^2 (residual heterogeneity / unaccounted variability): 29.88%
## H^2 (unaccounted variability / sampling variability): 1.43
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 1) = 1.2728, p-val = 0.2592
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 1) = 0.0215, p-val = 0.9072
##
## Model Results:
##
## estimate se tval pval ci.lb
## intrcpt 0.5732 0.7599 0.7542 0.5886 -9.0827
## intervention.experienceexperienced -0.1944 1.3244 -0.1467 0.9072 -17.0230
## ci.ub
## intrcpt 10.2290
## intervention.experienceexperienced 16.6343
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mbin_gastric.insuf_random,population)## Warning in rma.uni(yi = TE[!exclude], sei = seTE[!exclude], data = dataset, :
## Redundant predictors dropped from the model.##
## Mixed-Effects Model (k = 3; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 0.5384 (SE = 0.8532)
## tau (square root of estimated tau^2 value): 0.7338
## I^2 (residual heterogeneity / unaccounted variability): 29.88%
## H^2 (unaccounted variability / sampling variability): 1.43
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 1) = 1.2728, p-val = 0.2592
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 1) = 0.0215, p-val = 0.9072
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt 0.3788 1.0847 0.3492 0.7861 -13.4041 14.1617
## populationgeneral 0.1944 1.3244 0.1467 0.9072 -16.6343 17.0230
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Influence Analysis for gastric insufflation
inf_analysis_gastric.insuf<-dmetar::InfluenceAnalysis(mbin_gastric.insuf_random,random = TRUE)## [===========================================================================] DONEplot(inf_analysis_gastric.insuf,"baujat")
#Funnel Plot for gastric insufflation
meta::funnel(mbin_gastric.insuf_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::trimfill(mbin_gastric.insuf_random)## RR 95%-CI %W(random)
## Baik 2003 1.1647 [0.3932; 3.4499] 33.7
## Carron 2012 1.4605 [0.2587; 8.2464] 22.2
## Saraswat 2011 7.0000 [0.3774; 129.8374] 10.9
## Filled: Carron 2012 0.9288 [0.1645; 5.2442] 22.2
## Filled: Saraswat 2011 0.1938 [0.0104; 3.5945] 10.9
##
## Number of studies combined: k = 5 (with 2 added studies)
##
## RR 95%-CI t p-value
## Random effects model 1.1647 [0.3570; 3.8004] 0.36 0.7385
## Prediction interval [0.0684; 19.8303]
##
## Quantifying heterogeneity:
## tau^2 = 0.6120 [0.0000; 11.2961]; tau = 0.7823 [0.0000; 3.3610];
## I^2 = 0.0% [0.0%; 72.5%]; H = 1.00 [1.00; 1.91]
##
## Test of heterogeneity:
## Q d.f. p-value
## 3.03 4 0.5530
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Trim-and-fill method to adjust for funnel plot asymmetryVentilation inadequacy
inadequacy<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Inadequacy.csv")
length(inadequacy$inadequacy.e1)## [1] 18#Number of comparisons with zero inadequate ventilations in both arms
inadequacy_zeros<-dplyr::filter(inadequacy,inadequacy$inadequacy.e1==0 & inadequacy$inadequacy.e2==0)
length(inadequacy_zeros$inadequacy.e1)## [1] 14#Table for Meta-analysis of ventilation inadequacy
inadequacy_analysis<-dplyr::filter(inadequacy,inadequacy$inadequacy.e1>0 | inadequacy$inadequacy.e2>0)
#Number of comparisons and patients meta-analized for ventilation inadequacy
length(inadequacy_analysis$inadequacy.e1)## [1] 4sum(inadequacy_analysis$inadequacy.t1,inadequacy_analysis$inadequacy.t2)## [1] 305#Meta-analysis for ventilation inadequacy
mbin_inadequacy_random<-meta::metabin(inadequacy.e1,inadequacy.t1,inadequacy.e2,inadequacy.t2,data = inadequacy_analysis,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,method.tau = "SJ",hakn = TRUE,prediction = TRUE,incr = 0.5,sm = "RR")
mbin_inadequacy_random## RR 95%-CI %W(random)
## Abdel-Ghaffar 2022 1.0000 [0.2191; 4.5639] 36.3
## Dunnebier 2017 24.4737 [1.5468; 387.2154] 21.6
## Kim 2021 1.0000 [0.0646; 15.4775] 21.8
## Maltby 2002 9.1651 [0.5054; 166.1919] 20.4
##
## Number of studies combined: k = 4
##
## RR 95%-CI t p-value
## Random effects model 3.1296 [0.2481; 39.4781] 1.43 0.2475
## Prediction interval [0.0065; 1510.8749]
##
## Quantifying heterogeneity:
## tau^2 = 1.4283 [0.0000; 34.5143]; tau = 1.1951 [0.0000; 5.8749];
## I^2 = 42.3% [0.0%; 80.6%]; H = 1.32 [1.00; 2.27]
##
## Test of heterogeneity:
## Q d.f. p-value
## 5.20 3 0.1575
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.5 in studies with zero cell frequencies#Estimated probability of inadequate ventilation with second generation SGAs
meta::metaprop(event = inadequacy.e1,n = inadequacy.t1 ,studlab = paste(study),data = inadequacy,method = "GLMM",sm = "PLOGIT",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## proportion 95%-CI
## Abdi 2010 0.0000 [0.0000; 0.0521]
## Abdel-Ghaffar 2022 0.1000 [0.0211; 0.2653]
## Bhushan 2022 0.0000 [0.0000; 0.0544]
## Biswas 2015 0.0000 [0.0000; 0.1089]
## Dunnebier 2017 0.5357 [0.3387; 0.7249]
## Griffiths 2013 0.0000 [0.0000; 0.0606]
## Gulec 2012 0.0000 [0.0000; 0.1089]
## Hartmann 2001 0.0000 [0.0000; 0.0725]
## Ibrahim 2011 0.0000 [0.0000; 0.1157]
## Kang 2019 0.0000 [0.0000; 0.1234]
## Khan 2020 0.0000 [0.0000; 0.1372]
## Kim 2021 0.0233 [0.0006; 0.1229]
## Kuvaki 2019 0.0000 [0.0000; 0.0711]
## Lim 2007 0.0000 [0.0000; 0.0402]
## Maltby 2002 0.0741 [0.0206; 0.1789]
## Nagahisa 2017 0.0000 [0.0000; 0.0499]
## Saraswat 2011 0.0000 [0.0000; 0.1157]
## Ahmed 2015 0.0000 [0.0000; 0.0881]
##
## Number of studies combined: k = 18
##
## proportion 95%-CI
## Random effects model 0.0009 [0.0000; 0.0306]
##
## Quantifying heterogeneity:
## tau^2 = 11.2749; tau = 3.3578; I^2 = 92.2%; H = 3.58
##
## Test of heterogeneity:
## Q d.f. p-value Test
## 28.01 17 0.0448 Wald-type
## 113.95 17 < 0.0001 Likelihood-Ratio
##
## Details on meta-analytical method:
## - Random intercept logistic regression model
## - Maximum-likelihood estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Logit transformation
## - Clopper-Pearson confidence interval for individual studies
## - Continuity correction of 0.5 in studies with zero cell frequencies
## (only used to calculate individual study results)#Forest plot for ventilation inadequacy
meta::forest(mbin_inadequacy_random,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for ventilation inadequacy
dmetar::find.outliers(mbin_inadequacy_random)## No outliers detected (random-effects model).#Influence Analysis for ventilation inadequacy
inf_analysis_inadequacy<-dmetar::InfluenceAnalysis(mbin_inadequacy_random,random = TRUE)## [===========================================================================] DONEplot(inf_analysis_inadequacy,"baujat")
#Meta-regression for ventilation inadequacy
#Controling for operator experience
meta::metareg(mbin_inadequacy_random, intervention.experience)##
## Mixed-Effects Model (k = 4; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 2.1334 (SE = 2.0505)
## tau (square root of estimated tau^2 value): 1.4606
## I^2 (residual heterogeneity / unaccounted variability): 57.28%
## H^2 (unaccounted variability / sampling variability): 2.34
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 2) = 5.1411, p-val = 0.0765
##
## Test of Moderators (coefficient 2):
## F(df1 = 1, df2 = 2) = 0.0101, p-val = 0.9290
##
## Model Results:
##
## estimate se tval pval ci.lb
## intrcpt 1.2756 1.3059 0.9768 0.4317 -4.3434
## intervention.experienceexperienced -0.1985 1.9711 -0.1007 0.9290 -8.6796
## ci.ub
## intrcpt 6.8946
## intervention.experienceexperienced 8.2825
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mbin_inadequacy_random,population)## Warning in rma.uni(yi = TE[!exclude], sei = seTE[!exclude], data = dataset, :
## Redundant predictors dropped from the model.##
## Mixed-Effects Model (k = 4; tau^2 estimator: SJ)
##
## tau^2 (estimated amount of residual heterogeneity): 1.1911 (SE = 2.2382)
## tau (square root of estimated tau^2 value): 1.0914
## I^2 (residual heterogeneity / unaccounted variability): 36.53%
## H^2 (unaccounted variability / sampling variability): 1.58
## R^2 (amount of heterogeneity accounted for): 16.61%
##
## Test for Residual Heterogeneity:
## QE(df = 1) = 1.1857, p-val = 0.2762
##
## Test of Moderators (coefficients 2:3):
## F(df1 = 2, df2 = 1) = 1.3693, p-val = 0.5172
##
## Model Results:
##
## estimate se tval pval ci.lb ci.ub
## intrcpt 0.0000 1.1610 0.0000 1.0000 -14.7520 14.7520
## populationgeneral 1.0682 1.6042 0.6659 0.6260 -19.3148 21.4513
## populationmale 3.1976 1.9334 1.6538 0.3462 -21.3692 27.7644
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Funnel Plot ventilation inadequcy
meta::funnel(mbin_inadequacy_random,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::trimfill(mbin_inadequacy_random)## RR 95%-CI %W(random)
## Abdel-Ghaffar 2022 1.0000 [0.2191; 4.5639] 25.6
## Dunnebier 2017 24.4737 [1.5468; 387.2154] 18.7
## Kim 2021 1.0000 [0.0646; 15.4775] 18.9
## Maltby 2002 9.1651 [0.5054; 166.1919] 18.0
## Filled: Dunnebier 2017 0.0882 [0.0056; 1.3948] 18.7
##
## Number of studies combined: k = 5 (with 1 added studies)
##
## RR 95%-CI t p-value
## Random effects model 1.7226 [0.1231; 24.0958] 0.57 0.5977
## Prediction interval [0.0028; 1066.8706]
##
## Quantifying heterogeneity:
## tau^2 = 3.1777 [0.0000; 37.7522]; tau = 1.7826 [0.0000; 6.1443];
## I^2 = 59.3% [0.0%; 84.8%]; H = 1.57 [1.00; 2.57]
##
## Test of heterogeneity:
## Q d.f. p-value
## 9.83 4 0.0434
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Trim-and-fill method to adjust for funnel plot asymmetryHeart rate
hr<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA HR.csv", header = TRUE)
#Number of comparisons and patients meta-analized for heart rate
length(hr$hr.m1)## [1] 10sum(hr$hr.f1,hr$hr.f2)## [1] 893#Meta-analysis for heart rate
mcont_hr<-meta::metacont(hr.f1,hr.m1,hr.sd1,hr.f2,hr.m2,hr.sd2,data=hr,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,prediction = TRUE,sm="MD")
mcont_hr## MD 95%-CI %W(random)
## Bhushan 2022 -11.4300 [-15.9153; -6.9447] 10.3
## Biswas 2015 -19.2300 [-25.3234; -13.1366] 9.4
## Carron 2012 -6.4000 [-11.4243; -1.3757] 10.0
## Gulec 2012 1.7300 [ -4.0073; 7.4673] 9.6
## Hartmann 2001 -4.0000 [ -9.1042; 1.1042] 9.9
## Koo 2003 -9.5000 [-15.6539; -3.3461] 9.4
## Lim 2007 -3.0000 [ -6.8095; 0.8095] 10.6
## Parikh 2017 -2.7100 [ -5.9089; 0.4889] 10.8
## Tosh 2021 -16.2000 [-20.8552; -11.5448] 10.2
## Ye 2020 -19.3200 [-24.4915; -14.1485] 9.9
##
## Number of studies combined: k = 10
##
## MD 95%-CI z p-value
## Random effects model -8.9103 [-13.2889; -4.5316] -3.99 < 0.0001
## Prediction interval [-24.9580; 7.1374]
##
## Quantifying heterogeneity:
## tau^2 = 43.4379 [18.3169; 169.1026]; tau = 6.5907 [4.2798; 13.0039];
## I^2 = 88.3% [80.5%; 92.9%]; H = 2.92 [2.26; 3.76]
##
## Test of heterogeneity:
## Q d.f. p-value
## 76.74 9 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau#Forest plot for heart rate
meta::forest(mcont_hr,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for heart rate
dmetar::find.outliers(mcont_hr)## Identified outliers (random-effects model)
## ------------------------------------------
## "Gulec 2012", "Ye 2020"
##
## Results with outliers removed
## -----------------------------
## MD 95%-CI %W(random) exclude
## Bhushan 2022 -11.4300 [-15.9153; -6.9447] 12.8
## Biswas 2015 -19.2300 [-25.3234; -13.1366] 11.4
## Carron 2012 -6.4000 [-11.4243; -1.3757] 12.4
## Gulec 2012 1.7300 [ -4.0073; 7.4673] 0.0 *
## Hartmann 2001 -4.0000 [ -9.1042; 1.1042] 12.3
## Koo 2003 -9.5000 [-15.6539; -3.3461] 11.4
## Lim 2007 -3.0000 [ -6.8095; 0.8095] 13.3
## Parikh 2017 -2.7100 [ -5.9089; 0.4889] 13.8
## Tosh 2021 -16.2000 [-20.8552; -11.5448] 12.7
## Ye 2020 -19.3200 [-24.4915; -14.1485] 0.0 *
##
## Number of studies combined: k = 8
##
## MD 95%-CI z p-value
## Random effects model -8.8449 [-13.0731; -4.6167] -4.10 < 0.0001
## Prediction interval [-23.4775; 5.7877]
##
## Quantifying heterogeneity:
## tau^2 = 31.1069 [10.6209; 152.6137]; tau = 5.5774 [3.2590; 12.3537];
## I^2 = 85.1% [72.5%; 91.9%]; H = 2.59 [1.91; 3.52]
##
## Test of heterogeneity:
## Q d.f. p-value
## 46.97 7 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau#Meta-regression for heart rate
#Controling for operator experience
meta::metareg(mcont_hr,intervention.experience)##
## Mixed-Effects Model (k = 10; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 10.7419 (SE = 8.5193)
## tau (square root of estimated tau^2 value): 3.2775
## I^2 (residual heterogeneity / unaccounted variability): 65.10%
## H^2 (unaccounted variability / sampling variability): 2.87
## R^2 (amount of heterogeneity accounted for): 75.27%
##
## Test for Residual Heterogeneity:
## QE(df = 8) = 22.9202, p-val = 0.0035
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 19.0428, p-val < .0001
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt -16.1975 2.1563 -7.5117 <.0001 -20.4238
## intervention.experienceexperienced 11.8171 2.7080 4.3638 <.0001 6.5096
## ci.ub
## intrcpt -11.9712 ***
## intervention.experienceexperienced 17.1246 ***
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mcont_hr,population)##
## Mixed-Effects Model (k = 10; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 58.5361 (SE = 36.4902)
## tau (square root of estimated tau^2 value): 7.6509
## I^2 (residual heterogeneity / unaccounted variability): 90.80%
## H^2 (unaccounted variability / sampling variability): 10.87
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 7) = 76.0862, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 0.2324, p-val = 0.8903
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -10.9567 5.6509 -1.9389 0.0525 -22.0323 0.1189 .
## populationgeneral 2.2520 6.4242 0.3506 0.7259 -10.3392 14.8432
## populationobese 4.5567 9.8509 0.4626 0.6437 -14.7507 23.8641
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Subgroup analysis with studies with experienced operators
hr.experience<-dplyr::filter(hr,hr$intervention.experience=="experienced")
mcont_hr.experience<-meta::metacont(hr.f1,hr.m1,hr.sd1,hr.f2,hr.m2,hr.sd2,data=hr.experience,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,prediction = TRUE,sm="MD")
mcont_hr.experience## MD 95%-CI %W(random)
## Bhushan 2022 -11.4300 [-15.9153; -6.9447] 16.8
## Carron 2012 -6.4000 [-11.4243; -1.3757] 15.5
## Gulec 2012 1.7300 [ -4.0073; 7.4673] 13.9
## Hartmann 2001 -4.0000 [ -9.1042; 1.1042] 15.3
## Lim 2007 -3.0000 [ -6.8095; 0.8095] 18.5
## Parikh 2017 -2.7100 [ -5.9089; 0.4889] 20.0
##
## Number of studies combined: k = 6
##
## MD 95%-CI z p-value
## Random effects model -4.3798 [ -7.6173; -1.1423] -2.65 0.0080
## Prediction interval [-14.6703; 5.9106]
##
## Quantifying heterogeneity:
## tau^2 = 11.0084 [1.5136; 101.2982]; tau = 3.3179 [1.2303; 10.0647];
## I^2 = 68.8% [26.5%; 86.8%]; H = 1.79 [1.17; 2.75]
##
## Test of heterogeneity:
## Q d.f. p-value
## 16.03 5 0.0068
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau#Estimating the mean heart rate for endotracheal tubes
meta::metamean(n = hr$hr.f2,mean= hr$hr.m2, sd=hr$hr.sd2,studlab = study,data = hr,method.tau = "SJ",sm = "MRAW",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## mean 95%-CI %W(random)
## Bhushan 2022 88.1500 [84.8038; 91.4962] 10.3
## Biswas 2015 98.4600 [94.1048; 102.8152] 9.8
## Carron 2012 91.7000 [88.1556; 95.2444] 10.2
## Gulec 2012 81.7700 [76.8136; 86.7264] 9.5
## Hartmann 2001 82.0000 [78.1577; 85.8423] 10.1
## Koo 2003 81.9000 [77.0353; 86.7647] 9.5
## Lim 2007 80.0000 [77.5208; 82.4792] 10.7
## Parikh 2017 87.5000 [84.7339; 90.2661] 10.6
## Tosh 2021 95.9000 [91.4995; 100.3005] 9.8
## Ye 2020 85.5700 [80.9401; 90.1999] 9.6
##
## Number of studies combined: k = 10
##
## mean 95%-CI
## Random effects model 87.2759 [82.7546; 91.7972]
##
## Quantifying heterogeneity:
## tau^2 = 35.8862 [14.6642; 129.8275]; tau = 5.9905 [3.8294; 11.3942];
## I^2 = 90.5% [84.7%; 94.1%]; H = 3.25 [2.56; 4.13]
##
## Test of heterogeneity:
## Q d.f. p-value
## 94.96 9 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Untransformed (raw) means#Funnel Plot for heart rate
meta::funnel(mcont_hr,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::metabias(mcont_hr, method.bias = "linreg")##
## Linear regression test of funnel plot asymmetry
##
## data: mcont_hr
## t = -1.4194, df = 8, p-value = 0.1935
## alternative hypothesis: asymmetry in funnel plot
## sample estimates:
## bias se.bias intercept
## -6.013312 4.236388 6.101627trimfill_hr<-meta::trimfill(mcont_hr)
trimfill_hr## MD 95%-CI %W(random)
## Bhushan 2022 -11.4300 [-15.9153; -6.9447] 8.5
## Biswas 2015 -19.2300 [-25.3234; -13.1366] 8.0
## Carron 2012 -6.4000 [-11.4243; -1.3757] 8.4
## Gulec 2012 1.7300 [ -4.0073; 7.4673] 8.1
## Hartmann 2001 -4.0000 [ -9.1042; 1.1042] 8.3
## Koo 2003 -9.5000 [-15.6539; -3.3461] 8.0
## Lim 2007 -3.0000 [ -6.8095; 0.8095] 8.7
## Parikh 2017 -2.7100 [ -5.9089; 0.4889] 8.9
## Tosh 2021 -16.2000 [-20.8552; -11.5448] 8.5
## Ye 2020 -19.3200 [-24.4915; -14.1485] 8.3
## Filled: Biswas 2015 7.1846 [ 1.0912; 13.2780] 8.0
## Filled: Ye 2020 7.2746 [ 2.1031; 12.4461] 8.3
##
## Number of studies combined: k = 12 (with 2 added studies)
##
## MD 95%-CI z p-value
## Random effects model -6.2998 [-11.0405; -1.5590] -2.60 0.0092
## Prediction interval [-24.8438; 12.2443]
##
## Quantifying heterogeneity:
## tau^2 = 63.4159 [32.0400; 218.6151]; tau = 7.9634 [5.6604; 14.7856];
## I^2 = 91.3% [86.7%; 94.3%]; H = 3.39 [2.74; 4.19]
##
## Test of heterogeneity:
## Q d.f. p-value
## 126.24 11 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau
## - Trim-and-fill method to adjust for funnel plot asymmetryMean arterial pressure
has<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA HAS.csv", header = TRUE)
#Number of comparisons and patients meta-analized for mean arterial pressure
length(has$has.m1)## [1] 6sum(has$has.f1,has$has.f2)## [1] 442#Meta-analysis for mean arterial pressure
mcont_has<-meta::metacont(has.f1,has.m1,has.sd1,has.f2,has.m2,has.sd2,data=has,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,prediction = TRUE,sm="MD")
mcont_has## MD 95%-CI %W(random)
## Biswas 2015 -1.1700 [-14.6350; 12.2950] 8.0
## Carron 2012 -7.7000 [-12.2336; -3.1664] 19.4
## Gulec 2012 0.7400 [ -7.0356; 8.5156] 14.3
## Hartmann 2001 -1.0000 [ -4.7592; 2.7592] 20.6
## Parikh 2017 -2.0700 [ -7.6342; 3.4942] 17.8
## Tosh 2021 -12.6000 [-16.8456; -8.3544] 19.9
##
## Number of studies combined: k = 6
##
## MD 95%-CI z p-value
## Random effects model -4.5588 [ -9.2424; 0.1248] -1.91 0.0564
## Prediction interval [-19.7080; 10.5904]
##
## Quantifying heterogeneity:
## tau^2 = 24.0612 [0.8276; 174.6723]; tau = 4.9052 [0.9097; 13.2164];
## I^2 = 76.7% [47.9%; 89.6%]; H = 2.07 [1.38; 3.10]
##
## Test of heterogeneity:
## Q d.f. p-value
## 21.43 5 0.0007
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau#Forest plot for mean arterial pressure
meta::forest(mcont_has,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for mean arterial pressure
dmetar::find.outliers(mcont_has)## No outliers detected (random-effects model).#Meta-regression for mean arterial pressure
#Controling for operator experience
meta::metareg(mcont_has,intervention.experience)##
## Mixed-Effects Model (k = 6; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 9.7218 (SE = 13.3303)
## tau (square root of estimated tau^2 value): 3.1180
## I^2 (residual heterogeneity / unaccounted variability): 54.18%
## H^2 (unaccounted variability / sampling variability): 2.18
## R^2 (amount of heterogeneity accounted for): 59.60%
##
## Test for Residual Heterogeneity:
## QE(df = 4) = 8.7292, p-val = 0.0682
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 3.5239, p-val = 0.0605
##
## Model Results:
##
## estimate se zval pval ci.lb
## intrcpt -10.2904 3.3914 -3.0343 0.0024 -16.9374
## intervention.experienceexperienced 7.4452 3.9661 1.8772 0.0605 -0.3282
## ci.ub
## intrcpt -3.6434 **
## intervention.experienceexperienced 15.2186 .
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mcont_has,population)##
## Mixed-Effects Model (k = 6; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 32.4829 (SE = 32.4742)
## tau (square root of estimated tau^2 value): 5.6994
## I^2 (residual heterogeneity / unaccounted variability): 80.02%
## H^2 (unaccounted variability / sampling variability): 5.00
## R^2 (amount of heterogeneity accounted for): 0.00%
##
## Test for Residual Heterogeneity:
## QE(df = 4) = 20.0178, p-val = 0.0005
##
## Test of Moderators (coefficient 2):
## QM(df = 1) = 0.3414, p-val = 0.5590
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -3.7063 2.9813 -1.2432 0.2138 -9.5496 2.1370
## populationobese -3.9937 6.8353 -0.5843 0.5590 -17.3907 9.4033
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Estimating the mean MAP for endotracheal tubes
meta::metamean(n = has$has.f2,mean= has$has.m2, sd=has$has.sd2,studlab = study,data = has,method.tau = "SJ",sm = "MRAW",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## mean 95%-CI %W(random)
## Biswas 2015 90.8500 [ 80.0296; 101.6704] 14.5
## Carron 2012 82.8000 [ 79.8034; 85.7966] 17.3
## Gulec 2012 100.2600 [ 95.4444; 105.0756] 16.9
## Hartmann 2001 85.0000 [ 81.9811; 88.0189] 17.3
## Parikh 2017 78.3000 [ 73.1972; 83.4028] 16.8
## Tosh 2021 111.0000 [108.2729; 113.7271] 17.3
##
## Number of studies combined: k = 6
##
## mean 95%-CI
## Random effects model 91.4204 [78.3920; 104.4488]
##
## Quantifying heterogeneity:
## tau^2 = 145.7983 [54.9323; 898.1387]; tau = 12.0747 [7.4116; 29.9690];
## I^2 = 98.2% [97.4%; 98.8%]; H = 7.48 [6.18; 9.05]
##
## Test of heterogeneity:
## Q d.f. p-value
## 279.44 5 < 0.0001
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Untransformed (raw) means#Funnel Plot for mean arterial pressure
meta::funnel(mcont_has,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
trimfill_has<-meta::trimfill(mcont_has)
trimfill_has## MD 95%-CI %W(random)
## Biswas 2015 -1.1700 [-14.6350; 12.2950] 7.0
## Carron 2012 -7.7000 [-12.2336; -3.1664] 17.0
## Gulec 2012 0.7400 [ -7.0356; 8.5156] 12.6
## Hartmann 2001 -1.0000 [ -4.7592; 2.7592] 18.0
## Parikh 2017 -2.0700 [ -7.6342; 3.4942] 15.5
## Tosh 2021 -12.6000 [-16.8456; -8.3544] 17.4
## Filled: Gulec 2012 -12.1897 [-19.9653; -4.4141] 12.6
##
## Number of studies combined: k = 7 (with 1 added studies)
##
## MD 95%-CI z p-value
## Random effects model -5.5154 [ -9.9042; -1.1266] -2.46 0.0138
## Prediction interval [-19.4082; 8.3774]
##
## Quantifying heterogeneity:
## tau^2 = 24.1949 [2.1828; 148.5776]; tau = 4.9188 [1.4774; 12.1892];
## I^2 = 75.3% [47.7%; 88.3%]; H = 2.01 [1.38; 2.93]
##
## Test of heterogeneity:
## Q d.f. p-value
## 24.28 6 0.0005
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau
## - Trim-and-fill method to adjust for funnel plot asymmetryTime for insertion
time<-read.csv2("~/Desktop/Systematic Reviews/SR SGA vs TT/Tables for Analyses/SGA Time.csv", header = TRUE)
#Number of comparisons and patients meta-analized for time for insertion
length(time$time.m1)## [1] 20sum(time$time.f1,time$time.f2)## [1] 2734#Meta-analysis for time for insertion
mcont_time<-meta::metacont(time.f1,time.m1,time.sd1,time.f2,time.m2,time.sd2,data=time,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,prediction = TRUE,sm="MD")
mcont_time## MD 95%-CI %W(random)
## Abdi 2010 -96.0000 [-109.0522; -82.9478] 3.7
## Badheka 2015 -3.0500 [ -4.2315; -1.8685] 5.2
## Abdel-Ghaffar 2022 -4.7700 [ -6.5735; -2.9665] 5.2
## Bhushan 2022 -6.6000 [ -8.9502; -4.2498] 5.2
## Carron 2012 5.9000 [ 3.0999; 8.7001] 5.1
## Dunnebier 2017 -7.5100 [ -20.2380; 5.2180] 3.8
## Gombar 2012 3.6000 [ 1.3261; 5.8739] 5.2
## Hartmann 2001 3.0000 [ -0.6654; 6.6654] 5.1
## Kuvaki 2019 -5.0800 [ -7.1615; -2.9985] 5.2
## Lim 2007 -17.0000 [ -17.7449; -16.2551] 5.2
## Lorenz 2009 -3.8000 [ -4.9449; -2.6551] 5.2
## Ng 2021 -20.6000 [ -25.1558; -16.0442] 5.0
## Oczenski 1999 0.0000 [ -6.3474; 6.3474] 4.8
## Ozbilgin 2021 -7.7600 [ -10.3792; -5.1408] 5.2
## Panneer 2017 -2.2000 [ -3.4396; -0.9604] 5.2
## Sabuncu 2018 -2.6300 [ -5.3599; 0.0999] 5.1
## Saraswat 2011 -1.1600 [ -2.9629; 0.6429] 5.2
## Tosh 2021 21.0000 [ 16.1834; 25.8166] 5.0
## Yao 2019 -23.0000 [ -23.7483; -22.2517] 5.2
## Ahmed 2015 -0.6000 [ -1.8266; 0.6266] 5.2
##
## Number of studies combined: k = 20
##
## MD 95%-CI z p-value
## Random effects model -7.2342 [-11.8880; -2.5803] -3.05 0.0023
## Prediction interval [-29.5909; 15.1226]
##
## Quantifying heterogeneity:
## tau^2 = 107.6009 [80.1658; 375.1458]; tau = 10.3731 [8.9535; 19.3687];
## I^2 = 99.4% [99.3%; 99.4%]; H = 12.51 [11.66; 13.41]
##
## Test of heterogeneity:
## Q d.f. p-value
## 2972.23 19 0
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau#Forest plot for time for insertion
meta::forest(mcont_time,sortvar=TE,lab.e="SGAs",lab.c="ET",col.study="black",col.square="black",col.diamond="blue")
#Detecting Outliers for time for insertion
dmetar::find.outliers(mcont_time)## Identified outliers (random-effects model)
## ------------------------------------------
## "Abdi 2010", "Carron 2012", "Gombar 2012", "Hartmann 2001", "Lim 2007", "Ng 2021", "Tosh 2021", "Yao 2019", "Ahmed 2015"
##
## Results with outliers removed
## -----------------------------
## MD 95%-CI %W(random) exclude
## Abdi 2010 -96.0000 [-109.0522; -82.9478] 0.0 *
## Badheka 2015 -3.0500 [ -4.2315; -1.8685] 13.1
## Abdel-Ghaffar 2022 -4.7700 [ -6.5735; -2.9665] 11.0
## Bhushan 2022 -6.6000 [ -8.9502; -4.2498] 9.2
## Carron 2012 5.9000 [ 3.0999; 8.7001] 0.0 *
## Dunnebier 2017 -7.5100 [ -20.2380; 5.2180] 0.7
## Gombar 2012 3.6000 [ 1.3261; 5.8739] 0.0 *
## Hartmann 2001 3.0000 [ -0.6654; 6.6654] 0.0 *
## Kuvaki 2019 -5.0800 [ -7.1615; -2.9985] 10.0
## Lim 2007 -17.0000 [ -17.7449; -16.2551] 0.0 *
## Lorenz 2009 -3.8000 [ -4.9449; -2.6551] 13.2
## Ng 2021 -20.6000 [ -25.1558; -16.0442] 0.0 *
## Oczenski 1999 0.0000 [ -6.3474; 6.3474] 2.6
## Ozbilgin 2021 -7.7600 [ -10.3792; -5.1408] 8.4
## Panneer 2017 -2.2000 [ -3.4396; -0.9604] 12.9
## Sabuncu 2018 -2.6300 [ -5.3599; 0.0999] 8.0
## Saraswat 2011 -1.1600 [ -2.9629; 0.6429] 11.0
## Tosh 2021 21.0000 [ 16.1834; 25.8166] 0.0 *
## Yao 2019 -23.0000 [ -23.7483; -22.2517] 0.0 *
## Ahmed 2015 -0.6000 [ -1.8266; 0.6266] 0.0 *
##
## Number of studies combined: k = 11
##
## MD 95%-CI z p-value
## Random effects model -3.8631 [-4.9889; -2.7374] -6.73 < 0.0001
## Prediction interval [-7.4340; -0.2922]
##
## Quantifying heterogeneity:
## tau^2 = 2.1619 [0.0000; 11.8590]; tau = 1.4703 [0.0000; 3.4437];
## I^2 = 70.9% [46.2%; 84.3%]; H = 1.85 [1.36; 2.52]
##
## Test of heterogeneity:
## Q d.f. p-value
## 34.41 10 0.0002
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau#Meta-regression for time for insertion
#Controling for operator experience
meta::metareg(mcont_time,intervention.experience)##
## Mixed-Effects Model (k = 20; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 77.6842 (SE = 44.6996)
## tau (square root of estimated tau^2 value): 8.8139
## I^2 (residual heterogeneity / unaccounted variability): 98.99%
## H^2 (unaccounted variability / sampling variability): 98.78
## R^2 (amount of heterogeneity accounted for): 27.80%
##
## Test for Residual Heterogeneity:
## QE(df = 17) = 1679.2269, p-val < .0001
##
## Test of Moderators (coefficients 2:3):
## QM(df = 2) = 6.5633, p-val = 0.0376
##
## Model Results:
##
## estimate se zval pval
## intrcpt -2.1143 2.8724 -0.7361 0.4617
## intervention.experienceexperienced -10.6045 4.1821 -2.5357 0.0112
## intervention.experienceinexperienced -1.6857 9.2885 -0.1815 0.8560
## ci.lb ci.ub
## intrcpt -7.7442 3.5155
## intervention.experienceexperienced -18.8013 -2.4078 *
## intervention.experienceinexperienced -19.8908 16.5195
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for population
meta::metareg(mcont_time,population)##
## Mixed-Effects Model (k = 20; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 44.9329 (SE = 26.3368)
## tau (square root of estimated tau^2 value): 6.7032
## I^2 (residual heterogeneity / unaccounted variability): 98.40%
## H^2 (unaccounted variability / sampling variability): 62.67
## R^2 (amount of heterogeneity accounted for): 58.24%
##
## Test for Residual Heterogeneity:
## QE(df = 15) = 940.0448, p-val < .0001
##
## Test of Moderators (coefficients 2:5):
## QM(df = 4) = 30.6667, p-val < .0001
##
## Model Results:
##
## estimate se zval pval ci.lb ci.ub
## intrcpt -19.4412 2.8804 -6.7494 <.0001 -25.0868 -13.7957 ***
## populationgeneral 18.3268 3.6178 5.0657 <.0001 11.2360 25.4176 ***
## populationmale 11.9312 9.7674 1.2215 0.2219 -7.2125 31.0750
## populationobese 25.3412 7.4344 3.4086 0.0007 10.7700 39.9124 ***
## populationpregnant 18.0413 5.5643 3.2423 0.0012 7.1355 28.9471 **
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Controling for both operator experience and population
meta::metareg(mcont_time,intervention.experience+population)##
## Mixed-Effects Model (k = 20; tau^2 estimator: DL)
##
## tau^2 (estimated amount of residual heterogeneity): 42.6808 (SE = 28.7871)
## tau (square root of estimated tau^2 value): 6.5331
## I^2 (residual heterogeneity / unaccounted variability): 98.20%
## H^2 (unaccounted variability / sampling variability): 55.52
## R^2 (amount of heterogeneity accounted for): 60.33%
##
## Test for Residual Heterogeneity:
## QE(df = 13) = 721.7587, p-val < .0001
##
## Test of Moderators (coefficients 2:7):
## QM(df = 6) = 37.7448, p-val < .0001
##
## Model Results:
##
## estimate se zval pval
## intrcpt -13.6547 3.6343 -3.7571 0.0002
## intervention.experienceexperienced -8.7648 3.5947 -2.4383 0.0148
## intervention.experienceinexperienced -6.8659 7.1165 -0.9648 0.3347
## populationgeneral 16.7206 3.6781 4.5460 <.0001
## populationmale 6.1447 9.9026 0.6205 0.5349
## populationobese 28.3195 7.3687 3.8432 0.0001
## populationpregnant 12.2547 5.8946 2.0790 0.0376
## ci.lb ci.ub
## intrcpt -20.7778 -6.5315 ***
## intervention.experienceexperienced -15.8103 -1.7193 *
## intervention.experienceinexperienced -20.8139 7.0821
## populationgeneral 9.5117 23.9294 ***
## populationmale -13.2641 25.5534
## populationobese 13.8771 42.7619 ***
## populationpregnant 0.7015 23.8080 *
##
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1#Subgroup analysis with pregnant women
time.pregnant<-dplyr::filter(time,time$population=="pregnant")
mcont_time.pregnant<-meta::metacont(time.f1,time.m1,time.sd1,time.f2,time.m2,time.sd2,data=time.pregnant,studlab = paste(study),comb.fixed = FALSE,comb.random = TRUE,prediction = TRUE,sm="MD")
mcont_time.pregnant## MD 95%-CI %W(random)
## Panneer 2017 -2.2000 [-3.4396; -0.9604] 49.8
## Ahmed 2015 -0.6000 [-1.8266; 0.6266] 50.2
##
## Number of studies combined: k = 2
##
## MD 95%-CI z p-value
## Random effects model -1.3974 [-2.9654; 0.1706] -1.75 0.0807
##
## Quantifying heterogeneity:
## tau^2 = 0.8842; tau = 0.9403; I^2 = 69.1% [0.0%; 93.0%]; H = 1.80 [1.00; 3.79]
##
## Test of heterogeneity:
## Q d.f. p-value
## 3.23 1 0.0721
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2#Estimating the mean time for insertion for endotracheal tubes
meta::metamean(n = time$time.f2,mean= time$time.m2, sd=time$time.sd2,studlab = study,data = time,method.tau = "SJ",sm = "MRAW",comb.fixed = FALSE,comb.random = TRUE,hakn = TRUE)## mean 95%-CI %W(random)
## Abdi 2010 228.0000 [218.0900; 237.9100] 5.0
## Badheka 2015 14.3300 [ 13.7718; 14.8882] 5.0
## Abdel-Ghaffar 2022 27.3400 [ 26.5847; 28.0953] 5.0
## Bhushan 2022 25.2000 [ 23.1976; 27.2024] 5.0
## Carron 2012 19.9000 [ 18.1600; 21.6400] 5.0
## Dunnebier 2017 43.9400 [ 33.6730; 54.2070] 4.9
## Gombar 2012 21.8000 [ 20.1646; 23.4354] 5.0
## Hartmann 2001 13.0000 [ 9.4322; 16.5678] 5.0
## Kuvaki 2019 20.4200 [ 18.7292; 22.1108] 5.0
## Lim 2007 37.0000 [ 36.3802; 37.6198] 5.0
## Lorenz 2009 19.3000 [ 18.3984; 20.2016] 5.0
## Ng 2021 47.2000 [ 42.9775; 51.4225] 5.0
## Oczenski 1999 28.2900 [ 23.8017; 32.7783] 5.0
## Ozbilgin 2021 21.6200 [ 19.1254; 24.1146] 5.0
## Panneer 2017 12.5000 [ 11.5083; 13.4917] 5.0
## Sabuncu 2018 16.5600 [ 14.0273; 19.0927] 5.0
## Saraswat 2011 16.9300 [ 15.4736; 18.3864] 5.0
## Tosh 2021 24.3000 [ 21.4799; 27.1201] 5.0
## Yao 2019 39.1000 [ 38.4420; 39.7580] 5.0
## Ahmed 2015 9.6800 [ 8.6109; 10.7491] 5.0
##
## Number of studies combined: k = 20
##
## mean 95%-CI
## Random effects model 34.2054 [12.4117; 55.9990]
##
## Quantifying heterogeneity:
## tau^2 = 2163.6637 [1242.1081; 4644.0293]; tau = 46.5152 [35.2436; 68.1471];
## I^2 = 99.8% [99.7%; 99.8%]; H = 20.78 [19.79; 21.82]
##
## Test of heterogeneity:
## Q d.f. p-value
## 8205.67 19 0
##
## Details on meta-analytical method:
## - Inverse variance method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Untransformed (raw) means#Funnel Plot for time for insertion
meta::funnel(mcont_time,ref.triangle=TRUE,contour.levels=c(0.9,0.95,0.99),col.contour=c("darkblue","blue","lightblue"))
meta::metabias(mcont_time,method.bias="linreg")##
## Linear regression test of funnel plot asymmetry
##
## data: mcont_time
## t = 1.9864, df = 18, p-value = 0.06243
## alternative hypothesis: asymmetry in funnel plot
## sample estimates:
## bias se.bias intercept
## 9.183456 4.623063 -16.188851trimfill_time<-meta::trimfill(mcont_time)
trimfill_time## MD 95%-CI %W(random)
## Abdi 2010 -96.0000 [-109.0522; -82.9478] 3.0
## Badheka 2015 -3.0500 [ -4.2315; -1.8685] 4.0
## Abdel-Ghaffar 2022 -4.7700 [ -6.5735; -2.9665] 4.0
## Bhushan 2022 -6.6000 [ -8.9502; -4.2498] 4.0
## Carron 2012 5.9000 [ 3.0999; 8.7001] 3.9
## Dunnebier 2017 -7.5100 [ -20.2380; 5.2180] 3.0
## Gombar 2012 3.6000 [ 1.3261; 5.8739] 4.0
## Hartmann 2001 3.0000 [ -0.6654; 6.6654] 3.9
## Kuvaki 2019 -5.0800 [ -7.1615; -2.9985] 4.0
## Lim 2007 -17.0000 [ -17.7449; -16.2551] 4.0
## Lorenz 2009 -3.8000 [ -4.9449; -2.6551] 4.0
## Ng 2021 -20.6000 [ -25.1558; -16.0442] 3.8
## Oczenski 1999 0.0000 [ -6.3474; 6.3474] 3.7
## Ozbilgin 2021 -7.7600 [ -10.3792; -5.1408] 3.9
## Panneer 2017 -2.2000 [ -3.4396; -0.9604] 4.0
## Sabuncu 2018 -2.6300 [ -5.3599; 0.0999] 3.9
## Saraswat 2011 -1.1600 [ -2.9629; 0.6429] 4.0
## Tosh 2021 21.0000 [ 16.1834; 25.8166] 3.8
## Yao 2019 -23.0000 [ -23.7483; -22.2517] 4.0
## Ahmed 2015 -0.6000 [ -1.8266; 0.6266] 4.0
## Filled: Ahmed 2015 -23.5363 [ -24.7628; -22.3097] 4.0
## Filled: Oczenski 1999 -24.1363 [ -30.4837; -17.7888] 3.7
## Filled: Hartmann 2001 -27.1363 [ -30.8016; -23.4709] 3.9
## Filled: Gombar 2012 -27.7363 [ -30.0101; -25.4624] 4.0
## Filled: Carron 2012 -30.0363 [ -32.8363; -27.2362] 3.9
## Filled: Tosh 2021 -45.1363 [ -49.9528; -40.3197] 3.8
##
## Number of studies combined: k = 26 (with 6 added studies)
##
## MD 95%-CI z p-value
## Random effects model -12.5282 [-16.9056; -8.1508] -5.61 < 0.0001
## Prediction interval [-36.0355; 10.9791]
##
## Quantifying heterogeneity:
## tau^2 = 124.7387 [109.0236; 417.9317]; tau = 11.1686 [10.4414; 20.4434];
## I^2 = 99.4% [99.3%; 99.4%]; H = 12.69 [11.95; 13.48]
##
## Test of heterogeneity:
## Q d.f. p-value
## 4025.10 25 0
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau
## - Trim-and-fill method to adjust for funnel plot asymmetry