CV from a given Confidence interval
Calculates the CV (coefficient of variation) from a known confidence interval of a BE study.
Useful if no CV but the 90% CI was given in literature. utf-8
CVfromCI(pe, lower, upper, n, design = "2x2", alpha = 0.05, robust = FALSE) CI2CV(pe, lower, upper, n, design = "2x2", alpha = 0.05, robust = FALSE)
pe: Point estimate of the T/R ratio.
The pe may be missing. In that case it will be calculated as geometric mean
of lower and upper.
lower: Lower confidence limit of the BE ratio.
upper: Upper confidence limit of the BE ratio.
n: Total number of subjects under study if given as scalar.
Number of subjects in (sequence) groups if given as vector.
design: Character string describing the study design.
See known.designs() for designs covered in this package.
alpha: Error probability. Set it to (1-confidence)/2 (i.e. to 0.05 for the usual 90% confidence intervals).
robust: With robust=FALSE (the default) usual degrees of freedom of the designs are used.
With robust=TRUE the degrees of freedom for the so-called robust evaluation (df2 in known.designs()) will be used. This may be helpful if the CI was evaluated via a mixed model or via intra-subject contrasts (aka basic estimator).
See Helmut presentation for the algebra underlying this function.
Numeric value of the CV as ratio.
Yuan J, Tong T, Tang M-L. Sample Size Calculation for Bioequivalence Studies Assessing Drug Effect and Food Effect at the Same Time With a 3-Treatment Williams Design. Regul Sci. 2013;47(2):242--7. tools:::Rd_expr_doi("10.1177/2168479012474273")
Original by D. Labes with suggestions by H. .
Reworked and adapted to unbalanced studies by B. Lang.
The calculations are based on the assumption of evaluation via log-transformed values.
The calculations are further based on a common variance of Test and Reference treatments in replicate crossover studies or parallel group study, respectively.
In case of argument n given as n(total) and is not divisible by the number of (sequence) groups the total sample size is partitioned to the (sequence) groups to have small imbalance only. A message is given in such cases.
The estimated CV is conservative (i.e., higher than actually observed) in case of unbalancedness.
CI2CV() is simply an alias to CVfromCI().
# Given a 90% confidence interval (without point estimate) # from a classical 2x2 crossover with 22 subjects CVfromCI(lower=0.91, upper=1.15, n=22, design="2x2") # will give [1] 0.2279405, i.e a CV ~ 23% # # unbalanced 2x2 crossover study, but not reported as such CI2CV(lower=0.89, upper=1.15, n=24) # will give a CV ~ 26.3% # unbalancedness accounted for CI2CV(lower=0.89, upper=1.15, n=c(16,8)) # should give CV ~ 24.7%