A reviewer wants 95% CIs. Should I report mean, SD and 95% CI, or drop the SD?

I-Chen Tsai, MD, PhDAdapted from InnovaRad's public Q&A (Taiwan)Original published Dec 27, 2022

The question. The paper I asked you about earlier has been provisionally accepted with minor revision. A reviewer says continuous results must be reported as mean and 95% CI. I already gave mean and SD, and adding the 95% CI would make the table look crowded. Should I report mean, SD and 95% CI, or delete the SD and keep only mean and 95% CI?

Either works. Personally, I would report mean, SD and 95% CI. Besides satisfying the reviewer, anyone who later wants to include this paper in a meta-analysis will not have to convert back to the SD themselves.

In fact, the SD and the 95% CI of a group mean can be converted into each other. The 95% CI is the mean ± 1.96 × SD ÷ √n, so going the other way, SD = √n × (upper limit − lower limit) ÷ 3.92. That holds for large groups. When a group is small (the Cochrane Handbook suggests fewer than about 60 participants), the CI should be calculated with a value from the t distribution with n − 1 degrees of freedom instead of 1.96: for n = 25, for example, the 3.92 becomes 4.128. Your statistics software will give you the CI for each group directly; if you use an online calculator, check that it uses the t distribution for small samples. The formulas and a worked example are at the link below.

Cochrane Handbook, Section 6.5.2.2: Obtaining standard deviations from standard errors and confidence intervals for group means (Chapter 6, last updated August 2023; version 6.5, 2024): https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-06#section-6-5-2-2

這些回答是經驗與建議，不是規定。你的稿件以目標期刊的規定為準，你的研究以所屬機構的倫理審查為準。