hartung knapp metafor

Author: GNtem2

Clinical-Trials/RCT/RCT.Rmd

@@ -15,7 +15,7 @@ This is an R Markdown document. Markdown is a simple formatting syntax for autho
 
 When you click the **Knit** button a document will be generated that includes both content as well as the output of any embedded R code chunks within the document. You can embed an R code chunk like this:
 
-The data is from Jama Cardiology on Associations of Omega-3 Fatty Acid Supplement Use With Cardiovascular Disease Risks Meta-analysis of 10 Trials Involving 77917 Individuals
+The data is from Jama Cardiology on Associations of Omega-3 Fatty Acid Supplement Use With Cardiovascular Disease Risks Meta-analysis of 10 Trials Involving 77917 Individuals. 
 
 ```{r metafor}
 library(tidyverse)
@@ -41,20 +41,67 @@ result<-predict(res, transf=exp, digits=2)
 forest(res, targs=list(study=rct$study), 
 main="RCT of Omega3 fatty acid for cardiovascular disease")
 # funnel plot 
+png(file="funnelplot.png")
 funnel(res, refline=0, level=c(90, 95, 99), shade=c("white", "gray", "darkgray"))
+dev.off()
 
 ```
 
 ```{r random effect}
 dat <- escalc(measure = "OR",ai=ai, n1i=n1i, ci=ci, n2i=n2i,data=rct)
 res.DL<-rma(yi,vi, method = "DL",data=dat)
 forest(res.DL,main="RCT of Omega3 fatty acid for cardiovascular disease")
+
+#create subgroups
+rct1<-rct[1:5,]
+rct2<-rct[6:10,]
+rct1$group[rct1$year>=1999]<-"wide"
+rct2$group[rct2$year>=1999]<-"narrow"
+rct3<-rbind(rct1,rct2)
+dat$group<-rct3$group
+
+#plot subgroups
+res.group <- rma(yi, vi, mods = ~ group, data=dat)
+res.wide<- rma(yi, vi, subset=(group=="wide"), data=dat)
+res.wide
+res.narrow<- rma(yi, vi, subset=(group=="narrow"), data=dat)
+res.narrow
+#https://stackoverflow.com/questions/39392706/using-the-r-forestplot-package-is-there-a-way-to-assign-variable-colors-to-boxe
+#confidence interval
+rmeta_conf <- 
+  structure(list(
+    mean  = c(NA, NA, exp(-0.0676), exp(-0.0266),  NA, exp(-0.0352)), 
+    lower = c(NA, NA, exp(-0.1873), exp(-0.0715),  NA, exp(-0.0753)),
+    upper = c(NA, NA, exp(0.0520), exp(0.0183),  NA, exp(0.0049))),
+    .Names = c("mean", "lower", "upper"), 
+    row.names = c(NA, -6L), 
+    class = "data.frame")
+
+
+#table data
+tabletext<-cbind(c("","Trials","wide","narrow",NA,"Summary"),
+    c("Events","(Drugs)",sum(filter(dat,group=="wide")$ai),sum(filter(dat,group=="narrow")$ai),NA,NA),
+    c("Events","(Control)",sum(filter(dat,group=="wide")$ci),sum(filter(dat,group=="narrow")$ci),NA,NA),     
+    c("","OR","0.934","0.974",NA,"0.965")
+                 )
+
+#use forestplot
+library(forestplot)
+ forestplot(tabletext, 
+       rmeta_conf,new_page = TRUE,
+       is.summary=c(TRUE,TRUE,rep(FALSE,8),TRUE),
+       clip=c(0.1,2.5), 
+       xlog=TRUE, 
+       col=fpColors(box="royalblue",line="darkblue", summary="royalblue"))
+
 ```
 ```{r baujat}
 # adjust margins so the space is better used
+png(file="baujat.png")
 par(mar=c(5,4,2,2))
 # create Baujat plot to explore source of heterogeneity
 baujat(res.DL, xlim=c(0,20), ylim=c(0,0.2))
+dev.off()
 ```
 ```{r gosh}
 ### fit FE model to all possible subsets
@@ -63,7 +110,10 @@ sav <- gosh(res.DL)
 ### create GOSH plot
 ### red points for subsets that include and blue points
 ### for subsets that exclude study 16 (the ISIS-4 trial)
+png(file="gosh.png")
 plot(sav, out=dim(rct)[1], breaks=100)
+dev.off()
 ```
 
+
 Note that the `echo = FALSE` parameter was added to the code chunk to prevent printing of the R code that generated the plot.

Clinical-Trials/RCT/RCT_omega13.Rmd

@@ -0,0 +1,66 @@
+---
+title: "RCT_omega13"
+author: "GNtem2"
+date: "27/07/2020"
+output: html_document
+---
+
+```{r setup, include=FALSE}
+knitr::opts_chunk$set(echo = TRUE)
+```
+
+## R Markdown
+
+This is an R Markdown document. Markdown is a simple formatting syntax for authoring HTML, PDF, and MS Word documents. For more details on using R Markdown see <http://rmarkdown.rstudio.com>.
+
+When you click the **Knit** button a document will be generated that includes both content as well as the output of any embedded R code chunks within the document. You can embed an R code chunk like this:
+
+```{r jaha}
+library(metafor)
+library(tidyverse)
+
+library(tidyverse)
+library(metafor)
+
+Trials=c("GISSI-P","JELIS","GISSI-HF","DOIT","SU.FOL.OM3","Alpha Omega","OMEGA","ORIGIN","R&P","AREDS-2","VITAL")
+
+#
+Treatment_AMI=c(223,71,107,11,32,89,87,228,30,145,186)
+Treatment_No=c(5666,9326,3494,282,1253,2404,1919,6239,2147,12933,7740)
+Control_AMI=c(233,93,129,9,32,102,78,248,40,200,200)
+Control_No=c(5658,9319,3481,281,1248,2433,1885,6266,2056,12938,7740)
+
+rct<-data.frame(Trials,Treatment_AMI,Treatment_No,Control_AMI,Control_No) %>%  rename(ai=Treatment_AMI,n1i=Treatment_No,ci=Control_AMI,
+         n2i=Control_No,study=Trials)
+
+#peto's fixed effect method
+res <- rma.peto(ai=ai, n1i=n1i, ci=ci, n2i=n2i, data=rct)
+print(res, digits=2)
+
+result<-predict(res, transf=exp, digits=2)
+#forest plot
+forest(res, targs=list(study=rct$study), 
+main="RCT of Omega3 fatty acid for cardiovascular disease")
+# funnel plot 
+
+funnel(res, refline=0, level=c(90, 95, 99), shade=c("white", "gray", "darkgray"))
+
+#DL
+dat <- escalc(measure = "OR",ai=ai, n1i=n1i, ci=ci, n2i=n2i,data=rct)
+res.DL<-rma(yi,vi, method = "DL",data=dat)
+forest(res.DL,main="RCT of Omega3 fatty acid for cardiovascular disease-DL")
+
+#Hartung-ignore error message
+res.HK<-rma.uni(yi,vi=1/vi,method="FE",knha=TRUE,data=dat)
+forest(res.HK,main="RCT of Omega3 fatty acid for cardiovascular disease-DL")
+```
+
+## Including Plots
+
+You can also embed plots, for example:
+
+```{r pressure, echo=FALSE}
+plot(pressure)
+```
+
+Note that the `echo = FALSE` parameter was added to the code chunk to prevent printing of the R code that generated the plot.

README.md

@@ -159,7 +159,9 @@ The issue regarding combining data from RCT and observational studies in the set
 
 
 ### RCT
-Revman is an excellent tool for performing meta-analysis of trials but lacks the flexibility of R packages. Here we will use metafor package, see rct.Rmd. The following methods are demonstrated: fixed effect and random effect analyses, forest plot, funnel plot (for exploring publication bias), Baujat plot (for exploring source of heterogeneity) and GOSH plot (for exploring study heterogeneity using output of fixed effect model for all possible subsets). The data is from Jama Cardiology 2020 on Associations of Omega-3 Fatty Acid Supplement Use With Cardiovascular Disease Risks Meta-analysis of 10 Trials Involving 77917 Individuals.
+Revman is an excellent tool for performing meta-analysis of trials but lacks the flexibility of R packages. Here we will use metafor package, see RCT.Rmd. The following methods are demonstrated: fixed effect and random effect analyses, forest plot, funnel plot (for exploring publication bias), Baujat plot (for exploring source of heterogeneity) and GOSH plot (for exploring study heterogeneity using output of fixed effect model for all possible subsets). 
+The data is from Jama Cardiology 2020 on Associations of Omega-3 Fatty Acid Supplement Use With Cardiovascular Disease Risks Meta-analysis of 10 Trials Involving 77917 Individuals. Subsequently a meta-analysis in J Am Heart
+Assoc. 2019;8:e013543. DOI: 10.1161/JAHA.119.013543 reported that contrary to the earlier meta-analysis, Omega-3 lowers the risk of cardiovascular diseases with effect related to dose. The analysis using DerSimonian Laird method is provided in RCT.Rmd and Hartung Knapp method is provided in RCT_omega13.Rmd.
 
 Funnel plot of RCT on Associations of Omega-3 Fatty Acid Supplement Use With Cardiovascular Disease Risks
 ![here](./Clinical-Trials/RCT/funnelplot.png)