Metanalisis AGT
Ismael Calandri
5/6/2021
###data tidy####
data$`Recurrence as a main outcome`<-factor(data$`Recurrence as a main outcome`,
levels = c(0,1),
labels = c("No", "Yes"))
library(expss)##
## Use 'expss_output_rnotebook()' to display tables inside R Notebooks.
## To return to the console output, use 'expss_output_default()'.##
## Attaching package: 'expss'## The following objects are masked from 'package:dplyr':
##
## between, compute, contains, first, last, na_if, recode, varsdata$recurrence_prevalence<-data$n_recurrency/data$n_total
data<-apply_labels(data,
recurrence_prevalence= "Prevalence of recurrency (%)",
n_recurrency = "Recurrency (n)",
n_total = "Total (n)")Prevalence metanalysis
data_prevalence<-data%>% select(n_recurrency,n_total,Study, TGA_definition, `Recurrence as a main outcome`)
data_prevalence<-na.omit(data_prevalence)
meta_prevalence<- metaprop(data_prevalence$n_recurrency, data_prevalence$n_total, studlab=data_prevalence$Study, sm="PFT", data=data_prevalence, method="Inverse", method.tau="DL")
summary(meta_prevalence)## Number of studies combined: k = 14
##
## proportion 95%-CI
## Fixed effect model 0.1165 [0.1044; 0.1291]
## Random effects model 0.1337 [0.1044; 0.1658]
##
## Quantifying heterogeneity:
## tau^2 = 0.0043 [0.0018; 0.0199]; tau = 0.0653 [0.0423; 0.1411]
## I^2 = 74.6% [57.0%; 85.0%]; H = 1.98 [1.53; 2.58]
##
## Test of heterogeneity:
## Q d.f. p-value
## 51.15 13 < 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
## - Freeman-Tukey double arcsine transformationforest.meta(meta_prevalence,
comb.r=T,
comb.f=F,
prediction = T,
leftcols = c("Study","n_recurrency","n_total", "TGA_definition", "Recurrence as a main outcome"),
leftlabs = c("Author", "Events", "Total", "TGA criteria", "Recurrency outcome"),
xlab="Prevalence of recurrency")
forest.meta(meta_prevalence,
comb.r=T,
comb.f=F,
prediction = T,
xlab="Prevalence of recurrency")
Outliers and influential analysis
find.outliers(meta_prevalence)## Identified outliers (fixed-effect model)
## ----------------------------------------
## "Romoli, 2020", "Oliveira, 2020", "Moon, 2016"
##
## Results with outliers removed
## -----------------------------
## proportion 95%-CI %W(fixed) %W(random) exclude
## Dong Ah Lee, 2021 0.1250 [0.0641; 0.2127] 4.1 7.7
## Romoli, 2020 0.0743 [0.0534; 0.1002] 0.0 0.0 *
## Tynas, 2020 0.1613 [0.0932; 0.2520] 4.3 8.0
## Morris, 2020 0.1370 [0.1167; 0.1593] 48.0 18.7
## Oliveira, 2020 0.2714 [0.1720; 0.3910] 0.0 0.0 *
## Alessandro, 2019 0.0788 [0.0457; 0.1248] 9.4 12.1
## Himeno, 2017 0.0667 [0.0337; 0.1162] 7.6 11.0
## Arena, 2017 0.1403 [0.0973; 0.1932] 10.2 12.5
## Moon, 2016 0.3333 [0.1459; 0.5697] 0.0 0.0 *
## Kwon, 2014 0.1176 [0.0712; 0.1795] 7.1 10.6
## Auyeung, 2011 0.1852 [0.0630; 0.3808] 1.3 3.2
## Lampl, 2004 0.1875 [0.0405; 0.4565] 0.8 2.0
## Agosti, 2006 0.1412 [0.0751; 0.2336] 3.9 7.5
## Hinge, 1986 0.2162 [0.1289; 0.3272] 3.4 6.9
##
## Number of studies combined: k = 11
##
## proportion 95%-CI
## Fixed effect model 0.1236 [0.1096; 0.1382]
## Random effects model 0.1241 [0.1005; 0.1496]
##
## Quantifying heterogeneity:
## tau^2 = 0.0016 [0.0000; 0.0105]; tau = 0.0396 [0.0000; 0.1026]
## I^2 = 50.0% [0.3%; 75.0%]; H = 1.41 [1.00; 2.00]
##
## Test of heterogeneity:
## Q d.f. p-value
## 20.01 10 0.0292
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau
## - Freeman-Tukey double arcsine transformation
## - Clopper-Pearson confidence interval for individual studies
##
## Identified outliers (random-effects model)
## ------------------------------------------
## "Romoli, 2020", "Oliveira, 2020"
##
## Results with outliers removed
## -----------------------------
## proportion 95%-CI %W(fixed) %W(random) exclude
## Dong Ah Lee, 2021 0.1250 [0.0641; 0.2127] 4.0 7.9
## Romoli, 2020 0.0743 [0.0534; 0.1002] 0.0 0.0 *
## Tynas, 2020 0.1613 [0.0932; 0.2520] 4.3 8.1
## Morris, 2020 0.1370 [0.1167; 0.1593] 47.6 16.1
## Oliveira, 2020 0.2714 [0.1720; 0.3910] 0.0 0.0 *
## Alessandro, 2019 0.0788 [0.0457; 0.1248] 9.3 11.5
## Himeno, 2017 0.0667 [0.0337; 0.1162] 7.5 10.6
## Arena, 2017 0.1403 [0.0973; 0.1932] 10.1 11.8
## Moon, 2016 0.3333 [0.1459; 0.5697] 1.0 2.9
## Kwon, 2014 0.1176 [0.0712; 0.1795] 7.0 10.3
## Auyeung, 2011 0.1852 [0.0630; 0.3808] 1.3 3.5
## Lampl, 2004 0.1875 [0.0405; 0.4565] 0.8 2.3
## Agosti, 2006 0.1412 [0.0751; 0.2336] 3.9 7.7
## Hinge, 1986 0.2162 [0.1289; 0.3272] 3.4 7.1
##
## Number of studies combined: k = 12
##
## proportion 95%-CI
## Fixed effect model 0.1253 [0.1113; 0.1400]
## Random effects model 0.1302 [0.1040; 0.1586]
##
## Quantifying heterogeneity:
## tau^2 = 0.0022 [0.0002; 0.0160]; tau = 0.0474 [0.0135; 0.1266]
## I^2 = 56.9% [18.0%; 77.4%]; H = 1.52 [1.10; 2.10]
##
## Test of heterogeneity:
## Q d.f. p-value
## 25.55 11 0.0076
##
## Details on meta-analytical method:
## - Inverse variance method
## - DerSimonian-Laird estimator for tau^2
## - Jackson method for confidence interval of tau^2 and tau
## - Freeman-Tukey double arcsine transformation
## - Clopper-Pearson confidence interval for individual studiesinf.analysis <- InfluenceAnalysis(x =meta_prevalence,
random = TRUE)## [===========================================================================] DONEplot(inf.analysis, "influence")
# Produce funnel plot
funnel.meta(meta_prevalence,
studlab = TRUE)
title("Funnel Plot (TGA recurrency prevalence)")
col.contour = c("gray75", "gray85", "gray95")
# Generate funnel plot (we do not include study labels here)
funnel.meta(meta_prevalence,
contour = c(0.9, 0.95, 0.99),
col.contour = col.contour)
# Add a legend
legend(x = 0.55, y = 0.02,
legend = c("p < 0.1", "p < 0.05", "p < 0.01"),
fill = col.contour)
# Add a title
title("Contour-Enhanced Funnel Plot (TGA recurrency prevalence)")
eggers.test(meta_prevalence)## Eggers' test of the intercept
## =============================
##
## intercept 95% CI t p
## 1.598 -0.28 - 3.47 1.67 0.120869
##
## Eggers' test does not indicate the presence of funnel plot asymmetry.Risk factors prevalence
Female sex
m.sex <- metabin(data$`sex_fem_eventos en exp`,
data$`sex_fem_numero de expuestos`,
data$`sex_fem_eventos en no exp`,
data$sex_fem_n_no_exp,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.sex## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 0.3931 [0.1077; 1.4352] 8.9
## Romoli, 2020 NA 0.0
## Tynas, 2020 1.3333 [0.4404; 4.0368] 10.8
## Morris, 2020 1.0252 [0.7193; 1.4612] 22.8
## Oliveira, 2020 5.0469 [1.0486; 24.2897] 6.8
## Alessandro, 2019 0.9278 [0.3342; 2.5759] 11.9
## Himeno, 2017 1.4400 [0.3668; 5.6531] 8.3
## Arena, 2017 0.7896 [0.3684; 1.6924] 15.6
## Moon, 2016 1.0000 [0.0748; 13.3670] 3.0
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 1.2500 [0.0885; 17.6531] 2.9
## Agosti, 2006 0.4359 [0.1206; 1.5761] 9.0
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 10
##
## OR 95%-CI t p-value
## Random effects model 0.9799 [0.6279; 1.5293] -0.10 0.9201
## Prediction interval [0.2988; 3.2138]
##
## Quantifying heterogeneity:
## tau^2 = 0.2266 [0.0000; 1.1175]; tau = 0.4760 [0.0000; 1.0571]
## I^2 = 0.0% [0.0%; 60.8%]; H = 1.00 [1.00; 1.60]
##
## Test of heterogeneity:
## Q d.f. p-value
## 8.63 9 0.4720
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects modelforest(m.sex,
xlab="Female sex")
Hypertension
m.hta <- metabin(data$`HTA_eventos en exp`,
data$`HTA_numero de expuestos`,
data$`HTA_eventos en noexp`,
data$`HTA_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.hta## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 0.7896 [0.2609; 2.3900] 14.9
## Morris, 2020 NA 0.0
## Oliveira, 2020 2.0870 [0.7065; 6.1645] 15.1
## Alessandro, 2019 1.6846 [0.5884; 4.8233] 15.4
## Himeno, 2017 0.3560 [0.0910; 1.3927] 12.9
## Arena, 2017 1.0148 [0.4701; 2.1906] 17.6
## Moon, 2016 0.8889 [0.1252; 6.3103] 9.1
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 49.2353 [0.0823; 29450.1833] 1.4
## Agosti, 2006 0.3295 [0.0908; 1.1954] 13.5
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 8
##
## OR 95%-CI t p-value
## Random effects model 0.9224 [0.4504; 1.8893] -0.27 0.7977
## Prediction interval [0.0977; 8.7071]
##
## Quantifying heterogeneity:
## tau^2 = 0.7498 [0.0000; 4.3050]; tau = 0.8659 [0.0000; 2.0748]
## I^2 = 25.1% [0.0%; 66.0%]; H = 1.16 [1.00; 1.72]
##
## Test of heterogeneity:
## Q d.f. p-value
## 9.35 7 0.2284
##
## 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 0.1 in studies with zero cell frequenciesforest(m.hta,
xlab="Hypertension")
## Dyslipidemia
m.dlp <- metabin(data$`DLP_eventos en exp`,
data$`DLP_numero de expuestos`,
data$`DLP_eventos en noexp`,
data$`DLP_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.dlp ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 1.4041 [0.4636; 4.2527] 19.2
## Morris, 2020 NA 0.0
## Oliveira, 2020 NA 0.0
## Alessandro, 2019 1.5136 [0.5286; 4.3341] 21.0
## Himeno, 2017 0.9458 [0.2654; 3.3712] 15.0
## Arena, 2017 1.1455 [0.4945; 2.6535] 30.7
## Moon, 2016 1.0000 [0.0748; 13.3670] 3.9
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 0.1705 [0.0003; 109.7893] 0.6
## Agosti, 2006 0.9231 [0.1804; 4.7221] 9.5
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 7
##
## OR 95%-CI t p-value
## Random effects model 1.1815 [0.9304; 1.5004] 1.71 0.1384
## Prediction interval [0.6510; 2.1445]
##
## Quantifying heterogeneity:
## tau^2 = 0.0442; tau = 0.2103; I^2 = 0.0%; H = 1.00
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.88 6 0.9899
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.1 in studies with zero cell frequenciesforest(m.dlp,
xlab="Dyslipidemia")
Smoking
m.tbq <- metabin(data$`TBQ_eventos en exp`,
data$`TBQ_numero de expuestos`,
data$`TBQ_eventos en noexp`,
data$`TBQ_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.tbq ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 0.8929 [0.2774; 2.8738] 21.8
## Morris, 2020 NA 0.0
## Oliveira, 2020 1.4333 [0.3767; 5.4543] 16.7
## Alessandro, 2019 1.0500 [0.3656; 3.0158] 26.7
## Himeno, 2017 0.8625 [0.1035; 7.1841] 6.7
## Arena, 2017 0.8824 [0.3161; 2.4628] 28.2
## Moon, 2016 NA 0.0
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 NA 0.0
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 5
##
## OR 95%-CI t p-value
## Random effects model 1.0033 [0.7860; 1.2808] 0.04 0.9715
## Prediction interval [0.7178; 1.4026]
##
## Quantifying heterogeneity:
## tau^2 = 0.0033; tau = 0.0578; I^2 = 0.0%; H = 1.00
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.40 4 0.9826
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Hartung-Knapp adjustment for random effects modelforest(m.tbq,
xlab="Smoking")
Diabetes
m.dbt <- metabin(data$`DBT_eventos en exp`,
data$`DBT_numero de expuestos`,
data$`DBT_eventos en noexp`,
data$`DBT_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.dbt ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 1.0429 [0.1131; 9.6200] 17.7
## Morris, 2020 NA 0.0
## Oliveira, 2020 0.4824 [0.0955; 2.4375] 29.2
## Alessandro, 2019 0.0825 [0.0002; 42.4399] 2.6
## Himeno, 2017 1.6111 [0.1852; 14.0133] 18.5
## Arena, 2017 1.5690 [0.3173; 7.7577] 29.7
## Moon, 2016 NA 0.0
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 0.3548 [0.0005; 258.9388] 2.3
## Agosti, 2006 NA 0.0
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 6
##
## OR 95%-CI t p-value
## Random effects model 0.9304 [0.4342; 1.9936] -0.24 0.8174
## Prediction interval [0.1879; 4.6064]
##
## Quantifying heterogeneity:
## tau^2 = 0.2440 [0.0000; 2.4619]; tau = 0.4940 [0.0000; 1.5690]
## I^2 = 0.0% [0.0%; 35.3%]; H = 1.00 [1.00; 1.24]
##
## Test of heterogeneity:
## Q d.f. p-value
## 1.96 5 0.8546
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.1 in studies with zero cell frequenciesforest(m.dbt,
xlab="Diabetes")
Stroke
m.acv <- metabin(data$`ACV_eventos en exp`,
data$`ACV_numero de expuestos`,
data$`ACV_eventos en noexp`,
data$`ACV_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.acv ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 1.3750 [0.3368; 5.6136] 38.0
## Morris, 2020 NA 0.0
## Oliveira, 2020 1.3824 [0.2318; 8.2442] 29.8
## Alessandro, 2019 0.8923 [0.1091; 7.2956] 24.5
## Himeno, 2017 NA 0.0
## Arena, 2017 0.0348 [0.0001; 17.6350] 4.1
## Moon, 2016 0.1677 [0.0002; 115.8494] 3.7
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 NA 0.0
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 5
##
## OR 95%-CI t p-value
## Random effects model 0.9874 [0.3229; 3.0194] -0.03 0.9764
## Prediction interval [0.0581; 16.7715]
##
## Quantifying heterogeneity:
## tau^2 = 0.6300 [0.0000; 14.7908]; tau = 0.7937 [0.0000; 3.8459]
## I^2 = 0.0% [0.0%; 51.0%]; H = 1.00 [1.00; 1.43]
##
## Test of heterogeneity:
## Q d.f. p-value
## 1.70 4 0.7913
##
## 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 0.1 in studies with zero cell frequenciesforest(m.acv,
xlab="Stroke history")
Coronary artery disease
m.coro <- metabin(data$`coro_eventos en exp`,
data$`coro_numero de expuestos`,
data$`coro_eventos en noexp`,
data$`coro_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.coro ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 1.3462 [0.2563; 7.0706] 28.6
## Morris, 2020 NA 0.0
## Oliveira, 2020 NA 0.0
## Alessandro, 2019 0.7644 [0.0944; 6.1921] 20.8
## Himeno, 2017 NA 0.0
## Arena, 2017 0.6000 [0.0741; 4.8590] 20.9
## Moon, 2016 0.1677 [0.0002; 115.8494] 2.8
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 6.0000 [0.2571; 140.0446] 10.8
## Agosti, 2006 3.2273 [0.2694; 38.6553] 16.1
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 6
##
## OR 95%-CI t p-value
## Random effects model 1.2896 [0.4970; 3.3467] 0.69 0.5234
## Prediction interval [0.1614; 10.3060]
##
## Quantifying heterogeneity:
## tau^2 = 0.4227 [0.0000; 5.0601]; tau = 0.6502 [0.0000; 2.2495]
## I^2 = 0.0% [0.0%; 50.6%]; H = 1.00 [1.00; 1.42]
##
## Test of heterogeneity:
## Q d.f. p-value
## 2.57 5 0.7662
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.1 in studies with zero cell frequenciesforest(m.coro,
xlab="Coronary artery disease")
Atrial fibrillation
m.fa <- metabin(data$`FA_eventos en exp`,
data$`FA_numero de expuestos`,
data$`FA_eventos en noexp`,
data$`FA_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.fa ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 5.8462 [0.7554; 45.2464] 61.9
## Morris, 2020 NA 0.0
## Oliveira, 2020 NA 0.0
## Alessandro, 2019 0.1373 [0.0003; 71.6350] 19.0
## Himeno, 2017 NA 0.0
## Arena, 2017 0.0573 [0.0001; 29.4107] 19.1
## Moon, 2016 NA 0.0
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 NA 0.0
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 3
##
## OR 95%-CI t p-value
## Random effects model 1.1847 [0.0023; 609.6083] 0.12 0.9177
## Prediction interval [0.0000; 2992499958917.7930]
##
## Quantifying heterogeneity:
## tau^2 = 2.9459 [0.0000; >100.0000]; tau = 1.7164 [0.0000; >10.0000]
## I^2 = 30.6% [0.0%; 92.8%]; H = 1.20 [1.00; 3.72]
##
## Test of heterogeneity:
## Q d.f. p-value
## 2.88 2 0.2368
##
## 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 0.1 in studies with zero cell frequenciesforest(m.fa,
xlab="Atrial fibrillation")
m.mig <- metabin(data$`Mig_eventos en exp`,
data$`Mig_numero de expuestos`,
data$`Mig_eventos en noexp`,
data$`Mig_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.mig## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 1.1264 [0.3637; 3.4887] 12.7
## Morris, 2020 2.2889 [1.5687; 3.3398] 32.4
## Oliveira, 2020 2.6786 [0.7086; 10.1255] 10.1
## Alessandro, 2019 3.8880 [1.2991; 11.6365] 13.2
## Himeno, 2017 NA 0.0
## Arena, 2017 1.2917 [0.5156; 3.2358] 16.6
## Moon, 2016 1.0000 [0.0748; 13.3670] 3.2
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 0.3548 [0.0005; 258.9388] 0.5
## Agosti, 2006 NA 0.0
## Hinge, 1986 3.2667 [0.9484; 11.2520] 11.2
##
## Number of studies combined: k = 8
##
## OR 95%-CI t p-value
## Random effects model 2.0795 [1.3892; 3.1128] 4.29 0.0036
## Prediction interval [0.7314; 5.9126]
##
## Quantifying heterogeneity:
## tau^2 = 0.1533 [0.0000; 0.7264]; tau = 0.3915 [0.0000; 0.8523]
## I^2 = 0.0% [0.0%; 53.0%]; H = 1.00 [1.00; 1.46]
##
## Test of heterogeneity:
## Q d.f. p-value
## 4.83 7 0.6806
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.1 in studies with zero cell frequenciesforest(m.mig,
xlab="Migraine")
m.dep <- metabin(data$`Depresion_eventos en exp`,
data$`Depresion_numero de expuestos`,
data$`Depresion_eventos en noexp`,
data$`Depresion_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.dep ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 4.8125 [1.4695; 15.7605] 48.6
## Morris, 2020 NA 0.0
## Oliveira, 2020 4.2000 [1.3260; 13.3034] 51.4
## Alessandro, 2019 NA 0.0
## Himeno, 2017 NA 0.0
## Arena, 2017 NA 0.0
## Moon, 2016 NA 0.0
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 NA 0.0
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 2
##
## OR 95%-CI t p-value
## Random effects model 4.4871 [1.8902; 10.6517] 22.06 0.0288
##
## Quantifying heterogeneity:
## tau^2 = 0.0001; tau = 0.0109; I^2 = 0.0%; H = 1.00
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.03 1 0.8719
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Hartung-Knapp adjustment for random effects modelforest(m.dep,
xlab="Depression history")
m.tig <- metabin(data$`gatillante_eventos en exp`,
data$`gatillante_numero de expuestos`,
data$`gatillante_eventos en noexp`,
data$`gatillante_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.tig ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 1.6111 [0.3954; 6.5641] 13.6
## Romoli, 2020 NA 0.0
## Tynas, 2020 NA 0.0
## Morris, 2020 1.3327 [0.9180; 1.9347] 24.5
## Oliveira, 2020 1.0390 [0.3571; 3.0226] 17.1
## Alessandro, 2019 1.2864 [0.4290; 3.8572] 16.8
## Himeno, 2017 NA 0.0
## Arena, 2017 0.8760 [0.3692; 2.0785] 19.4
## Moon, 2016 0.3000 [0.0277; 3.2499] 7.2
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 41.8606 [0.0813; 21557.6477] 1.4
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 7
##
## OR 95%-CI t p-value
## Random effects model 1.1310 [0.6284; 2.0356] 0.51 0.6265
## Prediction interval [0.1493; 8.5705]
##
## Quantifying heterogeneity:
## tau^2 = 0.5630 [0.0000; 4.4311]; tau = 0.7503 [0.0000; 2.1050]
## I^2 = 0.0% [0.0%; 51.2%]; H = 1.00 [1.00; 1.43]
##
## Test of heterogeneity:
## Q d.f. p-value
## 3.59 6 0.7323
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.1 in studies with zero cell frequenciesforest(m.tig,
xlab="Triggers presency")
m.estres <- metabin(data$`estres_eventos en exp`,
data$`estres_numero de expuestos`,
data$`estres_eventos en noexp`,
data$`estres_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.estres ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 1.0000 [0.2689; 3.7184] 17.4
## Romoli, 2020 NA 0.0
## Tynas, 2020 NA 0.0
## Morris, 2020 0.9046 [0.4683; 1.7475] 39.8
## Oliveira, 2020 NA 0.0
## Alessandro, 2019 2.1366 [0.7330; 6.2275] 23.2
## Himeno, 2017 NA 0.0
## Arena, 2017 0.1941 [0.0004; 106.5910] 1.0
## Moon, 2016 NA 0.0
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 1.6558 [0.4736; 5.7899] 18.6
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 5
##
## OR 95%-CI t p-value
## Random effects model 1.2392 [0.7040; 2.1812] 1.05 0.3517
## Prediction interval [0.3107; 4.9423]
##
## Quantifying heterogeneity:
## tau^2 = 0.1475 [0.0000; 2.1973]; tau = 0.3840 [0.0000; 1.4823]
## I^2 = 0.0% [0.0%; 66.3%]; H = 1.00 [1.00; 1.72]
##
## Test of heterogeneity:
## Q d.f. p-value
## 2.47 4 0.6508
##
## 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 0.1 in studies with zero cell frequenciesforest(m.estres,
xlab="Trigger: stress")
m.exe <- metabin(data$`exerc_eventos en exp`,
data$`exerc_numero de expuestos`,
data$`exerc_eventos en noexp`,
data$`exerc_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.exe ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 1.4889 [0.2803; 7.9096] 25.2
## Romoli, 2020 NA 0.0
## Tynas, 2020 NA 0.0
## Morris, 2020 1.3138 [0.7653; 2.2554] 49.1
## Oliveira, 2020 NA 0.0
## Alessandro, 2019 0.0708 [0.0001; 36.2834] 3.1
## Himeno, 2017 NA 0.0
## Arena, 2017 0.3625 [0.0463; 2.8359] 19.6
## Moon, 2016 0.0812 [0.0001; 49.3877] 3.0
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 NA 0.0
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 5
##
## OR 95%-CI t p-value
## Random effects model 0.8847 [0.2840; 2.7565] -0.30 0.7797
## Prediction interval [0.0537; 14.5858]
##
## Quantifying heterogeneity:
## tau^2 = 0.6080 [0.0000; 12.3600]; tau = 0.7797 [0.0000; 3.5157]
## I^2 = 0.0% [0.0%; 71.7%]; H = 1.00 [1.00; 1.88]
##
## Test of heterogeneity:
## Q d.f. p-value
## 2.94 4 0.5678
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.1 in studies with zero cell frequenciesforest(m.exe,
xlab="Trigger: physical exercise")
m.show <- metabin(data$`shower_eventos en exp`,
data$`shower_numero de expuestos`,
data$`shower_eventos en noexp`,
data$`shower_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.show ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 NA 0.0
## Morris, 2020 1.7666 [0.8269; 3.7744] 68.3
## Oliveira, 2020 NA 0.0
## Alessandro, 2019 0.5475 [0.0009; 317.9098] 3.8
## Himeno, 2017 NA 0.0
## Arena, 2017 0.0970 [0.0002; 50.7792] 3.9
## Moon, 2016 0.6111 [0.0516; 7.2402] 20.2
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 0.0769 [0.0001; 40.7210] 3.9
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 5
##
## OR 95%-CI t p-value
## Random effects model 1.0794 [0.3260; 3.5739] 0.18 0.8679
## Prediction interval [0.0843; 13.8279]
##
## Quantifying heterogeneity:
## tau^2 = 0.4562 [0.0000; 9.4091]; tau = 0.6754 [0.0000; 3.0674]
## I^2 = 0.0% [0.0%; 65.0%]; H = 1.00 [1.00; 1.69]
##
## Test of heterogeneity:
## Q d.f. p-value
## 2.38 4 0.6670
##
## 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 0.1 in studies with zero cell frequenciesforest(m.show,
xlab="Trigger: shower")
m.sexual <- metabin(data$`intercourse_eventos en exp`,
data$`intercourse_numero de expuestos`,
data$`intercourse_eventos en noexp`,
data$`intercourse_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.sexual ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 NA 0.0
## Morris, 2020 1.3838 [0.6309; 3.0351] 61.9
## Oliveira, 2020 NA 0.0
## Alessandro, 2019 1.1929 [0.2526; 5.6341] 16.4
## Himeno, 2017 NA 0.0
## Arena, 2017 2.1149 [0.4072; 10.9852] 14.6
## Moon, 2016 NA 0.0
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 2.1212 [0.2021; 22.2589] 7.2
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 4
##
## OR 95%-CI t p-value
## Random effects model 1.4814 [1.0341; 2.1222] 3.48 0.0401
## Prediction interval [0.8050; 2.7261]
##
## Quantifying heterogeneity:
## tau^2 = 0.0073 [0.0000; 0.4740]; tau = 0.0856 [0.0000; 0.6885]
## I^2 = 0.0%; H = 1.00
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.37 3 0.9458
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects modelforest(m.sexual,
xlab="Trigger: sexual intercourse")
m.vom <- metabin(data$`vomiting_eventos en exp`,
data$`vomiting_numero de expuestos`,
data$`vomiting_eventos en noexp`,
data$`vomiting_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.vom ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 NA 0.0
## Morris, 2020 NA 0.0
## Oliveira, 2020 NA 0.0
## Alessandro, 2019 0.2774 [0.0005; 150.1113] 10.9
## Himeno, 2017 NA 0.0
## Arena, 2017 1.0222 [0.1188; 8.7922] 89.1
## Moon, 2016 NA 0.0
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 NA 0.0
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 2
##
## OR 95%-CI t p-value
## Random effects model 0.8872 [0.0051; 154.1137] -0.29 0.8174
##
## Quantifying heterogeneity:
## tau^2 = 0.0585; tau = 0.2419; I^2 = 0.0%; H = 1.00
##
## Test of heterogeneity:
## Q d.f. p-value
## 0.15 1 0.7007
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.1 in studies with zero cell frequenciesforest(m.vom,
xlab="Trigger: vomits")
DWI lesions
m.difu <- metabin(data$`DWI_eventos en exp`,
data$`DWI_numero de expuestos`,
data$`DWI_eventos en noexp`,
data$`DWI_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")## Warning in metabin(data$`DWI_eventos en exp`, data$`DWI_numero de expuestos`, :
## Studies with non-positive values for n.e and / or n.c get no weight in meta-
## analysis.m.difu ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 0.7250 [0.2030; 2.5899] 24.8
## Romoli, 2020 NA 0.0
## Tynas, 2020 NA 0.0
## Morris, 2020 0.8342 [0.1808; 3.8499] 23.5
## Oliveira, 2020 241.0000 [0.4291; 135362.0532] 6.2
## Alessandro, 2019 NA 0.0
## Himeno, 2017 NA 0.0
## Arena, 2017 0.2818 [0.0005; 169.8720] 6.1
## Moon, 2016 0.6818 [0.0853; 5.4475] 20.5
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 13.6000 [1.2245; 151.0446] 18.8
## Lampl, 2004 NA 0.0
## Agosti, 2006 NA 0.0
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 6
##
## OR 95%-CI t p-value
## Random effects model 1.7385 [0.2365; 12.7784] 0.71 0.5079
## Prediction interval [0.0094; 323.0415]
##
## Quantifying heterogeneity:
## tau^2 = 2.9391 [0.0000; 31.6675]; tau = 1.7144 [0.0000; 5.6274]
## I^2 = 36.6% [0.0%; 74.7%]; H = 1.26 [1.00; 1.99]
##
## Test of heterogeneity:
## Q d.f. p-value
## 7.88 5 0.1627
##
## Details on meta-analytical method:
## - Mantel-Haenszel method
## - Sidik-Jonkman estimator for tau^2
## - Q-profile method for confidence interval of tau^2 and tau
## - Hartung-Knapp adjustment for random effects model
## - Continuity correction of 0.1 in studies with zero cell frequenciesforest(m.difu,
xlab="DWI lesions")
reflux
m.reflux <- metabin(data$`reflux_eventos en exp`,
data$`reflux_numero de expuestos`,
data$`reflux_eventos en noexp`,
data$`reflux_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")
m.reflux ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 NA 0.0
## Morris, 2020 NA 0.0
## Oliveira, 2020 NA 0.0
## Alessandro, 2019 NA 0.0
## Himeno, 2017 NA 0.0
## Arena, 2017 NA 0.0
## Moon, 2016 NA 0.0
## Kwon, 2014 NA 0.0
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 1.9333 [0.5125; 7.2935] 100.0
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 1
##
## OR 95%-CI z p-value
## Random effects model 1.9333 [0.5125; 7.2935] 0.97 0.3305
##
## 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^2forest(m.reflux,
xlab="Yugular reflux")
EEG
m.eeg <- metabin(data$`EEG_eventos en exp`,
data$`EEG_numero de expuestos`,
data$`EEG_eventos en noexp`,
data$`EEG_n no exp`,
data = data,
studlab = data$Study,
comb.fixed = FALSE,
comb.random = TRUE,
method.tau = "SJ",
hakn = TRUE,
prediction = TRUE,
incr = 0.1,
sm = "OR")## Warning in metabin(data$`EEG_eventos en exp`, data$`EEG_numero de expuestos`, :
## Studies with non-positive values for n.e and / or n.c get no weight in meta-
## analysis.m.eeg ## OR 95%-CI %W(random)
## Dong Ah Lee, 2021 NA 0.0
## Romoli, 2020 NA 0.0
## Tynas, 2020 NA 0.0
## Morris, 2020 1.1053 [0.4783; 2.5541] 42.8
## Oliveira, 2020 0.5263 [0.0739; 3.7460] 10.1
## Alessandro, 2019 NA 0.0
## Himeno, 2017 NA 0.0
## Arena, 2017 1.4848 [0.3444; 6.4010] 17.3
## Moon, 2016 NA 0.0
## Kwon, 2014 1.8276 [0.6315; 5.2887] 29.8
## Uttner, 2012 NA 0.0
## Auyeung, 2011 NA 0.0
## Lampl, 2004 NA 0.0
## Agosti, 2006 NA 0.0
## Hinge, 1986 NA 0.0
##
## Number of studies combined: k = 4
##
## OR 95%-CI t p-value
## Random effects model 1.2534 [0.6457; 2.4328] 1.08 0.3579
## Prediction interval [0.2913; 5.3919]
##
## Quantifying heterogeneity:
## tau^2 = 0.0716 [0.0000; 3.3616]; tau = 0.2675 [0.0000; 1.8335]
## I^2 = 0.0% [0.0%; 66.5%]; H = 1.00 [1.00; 1.73]
##
## Test of heterogeneity:
## Q d.f. p-value
## 1.37 3 0.7119
##
## 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 modelforest(m.eeg,
xlab="Abnormal EEG")