Mirrors.sunsite.dk

Title Sensitivity Analysis of a Meta-analysis with Unpublished but Author Noory Y. Kim. Advisors: Shrikant I. Bangdiwala, Gerald Gartlehner.
Maintainer Noory Y. Kim <[email protected]> Description This package contains R functions to gauge the impact of unpublished stud- ies upon the meta-analytic summary effect of a set of published studies. (Credits: The re-search leading to these results has received funding from the European Union’s Seventh Frame-work Programme (FP7/2007-2013) under grant agreement no. 282574.) BHHR2009p92 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Fleiss1993 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
forestsens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
funnelplot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
greentea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Hpylori . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A fictional meta-analytic data set with 6 published studies and 2 unpublished studies. The binaryoutcome event is not desired.
A data frame with 8 observations on the following 8 variables.
Denotes whether a study is unpublished, and if so, what outlook it has.
The sample size of the experimental arm.
The number of (undesired) events within the control arm.
The number of (undesired) events within the experimental arm.
The outlook of a study can be one of the following: published, very positive, positive,negative, very negative, current effect, no effect, very positive CL, positive CL, negative CL,or very negative CL.
Since the outcome event is undesired, when using the function forestsens(), specify the optionhigher.is.better=FALSE. Since this is the default setting for forestsens(), this does not needto be specified explicitly.
Borenstein, Hedges, Higgins, and Rothstein. Introduction to Meta-analysis. Wiley, 2009, page 92.
forestsens(table=BHHR2009p92, binary=TRUE, higher.is.better=FALSE) # To assign all unpublished studies to each of ten outlooks, one at a time,# and then return a table of summary effects, their 95% confidence interval, # and tau-squared.
summtab <- forestsens(table=BHHR2009p92, binary=TRUE, higher.is.better=FALSE, all.outlooks=TRUE)summtab A meta-analytic data set that includes 7 published placebo-controlled randomized studies of the ef-fect of aspirin in preventing death after myocardial infarction. The data set also includes 2 (fictional)unpublished studies.
The defined binary outcome event is death, and is undesired. When using the function forestsens(),specify the option higher.is.better=FALSE.
A data frame with 9 observations on the following 8 variables.
Denotes whether a study is unpublished, and if so, what outlook it has.
The sample size of the experimental arm.
The number of (undesired) events within the control arm.
The number of (undesired) events within the experimental arm.
The outlook of a study can be one of the following: published, very positive, positive,negative, very negative, current effect, no effect, very positive CL, positive CL, negative CL,or very negative CL.
Since the outcome event is undesired, when using the function forestsens(), specify the optionhigher.is.better=FALSE.
Fleiss, JL. (1993) "The statistical basis of meta-analysis." Stat Methods Med Res. 2(2):121-45.
forestsens(table=Fleiss1993, binary=TRUE, higher.is.better=FALSE) # To assign all unpublished studies to each of ten outlooks, one at a time,# and then return a table of summary effects, their 95% confidence interval,# and tau-squared.
summtab <- forestsens(table=Fleiss1993, binary=TRUE, higher.is.better=FALSE, all.outlooks=TRUE)summtab This function imputes missing effect sizes for unpublished studies and creates a forest plot. A setof forest plots can be generated for multiple imputations.
binary = TRUE, mean.sd = FALSE,higher.is.better = FALSE,outlook = NA, all.outlooks = FALSE,rr.vpos = NA, rr.pos = NA, rr.neg = NA, rr.vneg = NA,smd.vpos = NA, smd.pos = NA, smd.neg = NA, smd.vneg = NA,level = 95,binary.measure = "RR", continuous.measure="SMD",summary.measure="SMD", method = "DL",random.number.seed = NA, sims = 10, smd.noise = 0.01,plot.title = "", scale = 1, digits = 3) The name of the table containing the meta-analysis data.
TRUE if the outcomes are binary events; FALSE if the outcome data is continu-ous.
TRUE if the data set includes the mean and standard deviation of the both thecontrol and experimental arms of studies with continuous outcomes; FALSE oth-erwise.
TRUE if higher counts of binary events or higher continuous outcomes are de-sired; FALSE otherwise. For continuous outcomes, set as FALSE if a loweroutcome (eg. a more negative number) is desired.
If you want all unpublished studies to be assigned the same outcome, set this pa-rameter to one of the following values: "very positive", "positive", "current effect","negative", "very negative", "no effect", "very positive CL", "positive CL","negative CL", "very negative CL".
If TRUE, then a forest plot will be generated for each possible outlook.
The user-defined relative risk for binary outcomes in unpublished studies with a"very positive" outlook.
The user-defined relative risk for binary outcomes in unpublished studies with a"positive" outlook.
The user-defined relative risk for binary outcomes in unpublished studies with a"negative" outlook.
The user-defined relative risk for binary outcomes in unpublished studies with a"very negative" outlook.
The user-defined standardized mean difference for continuous outcomes in un-published studies with a "very positive" outlook.
The user-defined standardized mean difference for continuous outcomes in un-published studies with a "positive" outlook.
The user-defined standardized mean difference for continuous outcomes in un-published studies with a "negative" outlook.
The user-defined standardized mean difference for continuous outcomes in un-published studies with a "very negative" outlook.
binary.measure The effect size measure used for binary outcomes. "RR" for relative risk; "OR" The effect size measure used for continuous outcomes. "SMD" for standardizedmean difference (with the assumption of equal variances).
The measure used for summary effect sizes.
The same parameter in the escalc() function of the metafor package. "DL" forthe DerSimonian-Laird method.
Leave as NA if results are to be randomized each time. Set this value to a integerbetween 0 and 255 if results are to be consistent (for purposes of testing andcomparison).
The number of simulations to run per study when imputing unpublished studieswith binary outcomes.
The standard deviation of Gaussian random noise to be added to standardizedmean differences when imputing unpublished studies with continuous outcomes.
Changes the scaling of fonts in the forest plot.
The number of significant digits (decimal places) to appear in the table of sum-mary results which appears if all.outlooks=TRUE.
For unpublished studies with binary outcomes, random numbers are generated from binomial dis-tributions to impute the number of events in the experimental arms of experimental studies. Theparameter of these distributions depends out the outlook of the unpublished study and the rate ofevents in the control arms of published studies. By default, 10 simulations are run and their averageis used to impute the number of events in the experimental arm.
For unpublished studies with continuous outcomes, a ’very good’ approximator mentioned byBorenstein is used to impute the variance of the standardized mean difference. See Borensteinet al, 2009, pages 27-28.
The function employs functions in the metafor package: escalc() and forest().
Borenstein M, Hedges LV, Higgins JPT, and Rothstein HR (2009). Introduction to Meta-Analysis.
Chichester UK: Wiley.
Cooper HC, Hedges LV, & Valentine JC, eds. (2009). The handbook of research synthesis andmeta-analysis (2nd ed.). New York: Russell Sage Foundation.
DerSimonian R and Laird N (1986). "Meta-analysis in clinical trials." Controlled Clinical Trials7:177-188 (1986).
Viechtbauer W (2010). Conducting meta-analyses in R with the metafor package. Journal of Sta-tistical Software, 36(3), 1–48. data(Hpylori)forestsens(Hpylori, binary=TRUE, higher.is.better=FALSE)forestsens(Hpylori, binary=TRUE, higher.is.better=FALSE, plot.title="Test")forestsens(Hpylori, binary=TRUE, higher.is.better=FALSE, random.number.seed=52)forestsens(Hpylori, binary=TRUE, higher.is.better=FALSE, outlook="negative")forestsens(Hpylori, binary=TRUE, higher.is.better=FALSE, all.outlooks=TRUE) data(greentea)forestsens(greentea, binary=FALSE, mean.sd=TRUE, higher.is.better=FALSE)forestsens(greentea, binary=FALSE, mean.sd=TRUE, higher.is.better=FALSE, forestsens(greentea, binary=FALSE, mean.sd=TRUE, higher.is.better=FALSE, outlook="negative", smd.noise=0.3) This function (1) imputes data for a meta-analytic data set with unpublished studies, then (2) gen-erates a funnel plot.
binary=TRUE, mean.sd=TRUE,higher.is.better=NA,outlook=NA,vpos=NA, pos=NA, neg=NA, vneg=NA,level=95,binary.measure="RR", continuous.measure="SMD",summary.measure="SMD", method="DL",random.number.seed=NA, sims=1, smd.noise=0.01,title="", pch.pub=19, pch.unpub=0) The name of the table containing the meta-analysis data.
TRUE if the outcomes are binary events; FALSE if the outcome data is continu-ous.
TRUE if the data set includes the mean and standard deviation of the both thecontrol and experimental arms of studies with continuous outcomes; FALSE oth-erwise.
TRUE if higher counts of binary events or higher continuous outcomes are de-sired; FALSE otherwise. For continuous outcomes, set as FALSE if a loweroutcome (eg. a more negative number) is desired.
If you want all unpublished studies to be assigned the same outcome, set this pa-rameter to one of the following values: "very positive", "positive", "current effect","negative", "very negative", "no effect", "very positive CL", "positive CL","negative CL", "very negative CL".
The user-defined effect size for unpublished studies with a "very positive"outlook.
The user-defined effect size for unpublished studies with a "positive" outlook.
The user-defined effect size for unpublished studies with a "negative" outlook.
The user-defined effect size for unpublished studies with a "very negative"outlook.
binary.measure The effect size measure used for binary outcomes. "RR" for relative risk; "OR" The effect size measure used for continuous outcomes. "SMD" for standardizedmean difference (with the assumption of equal variances).
The measure used for summary effect sizes.
The same parameter in the escalc() function of the metafor package. "DL" forthe DerSimonian-Laird method.
Leave as NA if results are to be randomized each time. Set this value to a integerbetween 0 and 255 if results are to be consistent (for purposes of testing andcomparison).
The number of simulations to run per study when imputing unpublished studieswith binary outcomes.
The standard deviation of Gaussian random noise to be added to standardizedmean differences when imputing unpublished studies with continuous outcomes.
The symbol used to denote a published study.
The symbol used to denote an unpublished study.
The function employs functions in the metafor package: escalc() and forest().
data(Hpylori)funnelplot(Hpylori, binary=TRUE, higher.is.better=FALSE, data(greentea)funnelplot(greentea, binary=FALSE, higher.is.better=FALSE) The effect of green tea on weight loss.
Randomized clinical trials of at least 12 weeks duration assessing the effect of green tea consump-tion on weight loss.
A data frame with 14 observations on the following 9 variables.
Denotes whether a study is unpublished, and if so, what outlook it has.
The sample size of the experimental arm.
The mean effect within the control arm.
The mean effect within the experimental arm.
The standard deviation of the outcome within the control arm.
The standard deviation of the outcome within the experimental arm.
The outlook of a study can be one of the following: published, very positive, positive,negative, very negative, current effect, no effect, very positive CL, positive CL, negative CL,or very negative CL.
In this setting, a more negative change in outcome is desired; specify the option higher.is.better=FALSEfor the function forestsens().
Jurgens TM, Whelan AM, Killian L, Doucette S, Kirk S, Foy E. "Green tea for weight loss andweight maintenance in overweight or obese adults." Cochrane Database of Systematic Reviews2012, Issue 12. Art. No.: CD008650. DOI: 10.1002/14651858.CD008650.pub2.
Figure 6. Forest plot of comparison: 1 Primary outcomes, outcome: 1.2Weight loss studies con-ducted in/outside Japan.
forestsens(greentea, binary=FALSE, mean.sd=TRUE, higher.is.better=FALSE) # To fix the random number seed to make the results reproducible.
forestsens(greentea, binary=FALSE, mean.sd=TRUE, higher.is.better=FALSE, random.number.seed=52) # To modify the outlooks of all unpublished studies to, say, "negative".
forestsens(greentea, binary=FALSE,mean.sd=TRUE,higher.is.better=FALSE,random.number.seed=52, # To modify the outlooks of all unpublished studies to, say, "negative", and# overruling the default standardized mean difference (SMD) assigned to "negative".
# (In this case, for a negative outlook we might assign a positive SMD, which corresponds to# having weight loss under green tea treatment less than weight loss under control treatment,# i.e. the green tea treatment is less effective at achieving weight loss than control treatment.
forestsens(greentea, binary=FALSE, mean.sd=TRUE, higher.is.better=FALSE,random.number.seed=52, # To generate a forest plot for each of the ten default outlooks defined by forestsens().
forestsens(greentea, binary=FALSE, mean.sd=TRUE, higher.is.better=FALSE, random.number.seed=52, Healing of duodenal ulcers by Helicobacter pylori eradication therapy Randomized clinical trials comparing duodenal ulcer acute healing among (1) patients on ulcerhealing drug + Helicobacter pylori eradication therapy vs. (2) patients ulcer healing drug alone.
The event counts represent the numbers of patients not healed.
A data frame with 33 observations on the following 7 variables.
Denotes whether a study is unpublished, and if so, what outlook it has.
The sample size of the experimental arm.
The number of (undesired) events within the control arm.
The number of (undesired) events within the experimental arm.
The outlook of a study can be one of the following: published, very positive, positive,negative, very negative, current effect, no effect, very positive CL, positive CL, negative CL,or very negative CL.
Since the outcome event is undesired, when using the function forestsens(), specify the optionhigher.is.better=FALSE.
Ford AC, Delaney B, Forman D, Moayyedi P. "Eradication therapy for peptic ulcer disease in He-licobacter pylori positive patients." Cochrane Database of Systematic Reviews 2006, Issue 2. ArtNo.: CD003840. DOI: 10.1002/14651858.CD003840.pub4.
Figure 3. Forest plot of comparison: 1 duodenal ulcer acute healing hp eradication + ulcer healingdrug vs. ulcer healing drug alone, outcome: 1.1 Proportion not healed.
forestsens(table=Hpylori, binary=TRUE, higher.is.better=FALSE, scale=0.8) # To fix the random number seed to make the results reproducible.
forestsens(table=Hpylori, binary=TRUE, higher.is.better=FALSE, scale=0.8, # To modify the outlooks of all unpublished studies to, say, "very negative".
forestsens(table=Hpylori, binary=TRUE, higher.is.better=FALSE, scale=0.8, random.number.seed=106, outlook="very negative") # To modify the outlooks of all unpublished studies to, say, "very negative",# and overruling the default relative risk assigned to "very negative".
forestsens(table=Hpylori, binary=TRUE, higher.is.better=FALSE, scale=0.8, random.number.seed=106, outlook="very negative", rr.vneg=2.5) # To generate a forest plot for each of the ten default outlooks# defined by forestsens().
forestsens(table=Hpylori, binary=TRUE, higher.is.better=FALSE, scale=0.8, random.number.seed=106, all.outlooks=TRUE) BHHR2009p92, Fleiss1993, greentea, Hpylori,

Source: http://mirrors.sunsite.dk/cran/web/packages/SAMURAI/SAMURAI.pdf