Estimating sample means and standard deviations from the log-normal distribution using medians and quartiles: evaluating reporting requirements for primary and secondary endpoints of meta-analyses in anesthesiology
Authors' conclusion
All evaluated methods of estimating means and standard deviations from quartiles of log-normal distributed data result in confidence interval coverages below the expected 95%.
Study context
- Population
- Simulated log-normal data with zero log-scale means (n = 15, 27, 51; CV = 0.15, 0.3, 0.5), representative of anesthesiology studiessimulated samples
- Comparison
- Methods estimating means and standard deviations (SDs) from medians and quartiles: Luo/Wan vs McGrath quantile estimation vs McGrath Box-Cox vs Cai maximum likelihood
- Outcome
- Coverage of 95% generalized CIs for the ratio of means and ratio of SDs
Methodological detailslog-normal data simulation
- Log-normal data simulation
Main finding
As reported in the abstract
95% CI coverage for ratio of means and of SDs
| Group | Mean ratio coverage | SD coverage |
|---|---|---|
| Luo et al. and Wan et al. | 92.4% to 93.6% | 79.2 to 89.6 |
| McGrath quantile estimation | 88.5% and 91.5% | 78.0 and 82.7 |
| McGrath Box-Cox | 86.6% to 94.4% | 67.1 to 83.1 |
| Cai maximum likelihood | 78.9% to 86.4% | 67.6 to 78.0 |
Mean ratio coverage
Luo et al. and Wan et al.92.4% to 93.6%
McGrath quantile estimation88.5% and 91.5%
McGrath Box-Cox86.6% to 94.4%
Cai maximum likelihood78.9% to 86.4%
SD coverage
Luo et al. and Wan et al.79.2 to 89.6
McGrath quantile estimation78.0 and 82.7
McGrath Box-Cox67.1 to 83.1
Cai maximum likelihood67.6 to 78.0
Original publication
Estimating sample means and standard deviations from the log-normal distribution using medians and quartiles: evaluating reporting requirements for primary and secondary endpoints of meta-analyses in anesthesiology. Can J Anaesth. Published 11 April 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →