d_to_logRR: Convert Cohen's d to log risk ratio

In MetaUtility: Utility Functions for Conducting and Interpreting Meta-Analyses

Description

Converts Cohen's d (computed with a binary X) to a log risk ratio for use in meta-analysis. Under the assumption that Y is approximately normal, the resulting log risk ratio represents a dichotomization of Y that is near its median and otherwise will tend to be conservative.

Usage

d_to_logRR(d, se = NA)

Arguments

d	Cohen's d
se	Standard error of d

Details

Internally, this function first converts d to a log odds ratio using Chinn's (2000) conversion. The resulting log odds ratio approximates the value that would be obtained if Y were dichotomized; if Y is approximately normal, the log odds ratio is approximately invariant to the choice of threshold at which Y is dichotomized (Chinn, 2000). Then, the function converts the log odds ratio to a log risk ratio using VanderWeele's (2020) square-root conversion. That conversion is conservative in that it allows for the possibility that the dichotomized Y is not rare (i.e., the point of dichotomization is not at an extreme value of Y).

References

VanderWeele, TJ (2020). Optimal approximate conversions of odds ratios and hazard ratios to risk ratios. Biometrics.

Chinn S (2000). A simple method for converting an odds ratio to effect size for use in meta-analysis. Statistics in Medicine.

Examples

d_to_logRR( d = c(0.5, -0.2, .1),
        se = c(0.21, NA, 0.3) )

MetaUtility documentation built on Oct. 30, 2021, 5:07 p.m.