orbbound function

Sensitivity Analysis for Outcome Reporting Bias (ORB)

Implementation of the method by Copas & Jackson (2004) to evaluate outcome reporting bias in meta-analysis. An upper bound for outcome reporting bias is estimated for a given number of studies suspected with outcome reporting bias.

orbbound(x, k.suspect = 1, tau = x$tau, left = NULL, backtransf = x$backtransf)

Arguments

• x: An object of class meta.
• k.suspect: Number of studies with suspected outcome reporting bias.
• tau: Square-root of between-study variance tau-squared.
• left: A logical indicating whether the cause of any selection bias is due to missing studies on the left or right of the funnel plot: left hand side if left=TRUE, right hand side if left=FALSE. If not set, the linear regression test for funnel plot asymmetry (i.e., function metabias(..., meth="linreg")) is used to determine whether studies are missing on the left or right hand side.
• backtransf: A logical indicating whether results should be back transformed in printouts and plots. If backtransf=TRUE (default), results for sm="OR" are printed as odds ratios rather than log odds ratio, for example.

Returns

An object of class c("orbbound") with corresponding print and forest function. The object is a list containing the following components: - k.suspect, tau: As defined above.

• maxbias: Maximum bias for given values of k.suspect.

• common: Adjusted treatment estimates and corresponding quantities for common effect model (a list with elements TE, seTE, lower, upper, z, p, level, df).

• random: Adjusted treatment estimates and corresponding quantities for random effects model (a list with elements TE, seTE, lower, upper, z, p, level, df).

• left: Whether selection bias expected on left or right

• x: Meta-analysis object (i.e. argument x from function call).

• call: Function call.

• version: Version of R package metasens used to create object.

Details

This function provides the method by Copas and Jackson (2004) to estimate an upper bound for bias for a given number of studies with suspected outcome reporting bias.

Based on the upper bound of outcome reporting bias, treatment estimates and confidence limits adjusted for bias are calculated.

For comparison, the original meta-analysis is always considered in the sensitivity analysis (i.e. value 0 is always added to k.suspect).

Examples

data(Fleiss1993bin, package = "meta")

m1 <- metabin(d.asp, n.asp, d.plac, n.plac, data = Fleiss1993bin, sm = "OR")

orb1 <- orbbound(m1, k.suspect = 1:5)
print(orb1, digits = 2)
forest(orb1, xlim = c(0.75, 1.5))

# Same result
#
orb2 <- orbbound(m1, k.suspect = 1:5, left = FALSE)
print(orb2, digits = 2)

# Assuming bias in other direction
#
orb3 <- orbbound(m1, k.suspect = 1:5, left = TRUE)
print(orb3, digits = 2)

References

Copas J, Jackson D (2004): A bound for publication bias based on the fraction of unpublished studies. Biometrics, 60 , 146--53

See Also

forest.orbbound, print.orbbound

Author(s)

Guido Schwarzer guido.schwarzer@uniklinik-freiburg.de