Median to mean

Estimate each arm's mean and SD from the sample size and the median with the interquartile range, the range, or both. Mean and SD appear as soon as enough cells are filled. Methods and skewness test as in median2mean by Tong et al. (opens in a new tab)

Studies

Study (optional) Intervention Control Remove
nMedianQ1Q3MinMaxMeanSDnMedianQ1Q3MinMaxMeanSD

Needs n and the median, plus Q1 and Q3 (scenario 2), the minimum and maximum (scenario 1), or all of them (scenario 3). Mean by Luo et al. (2018); SD by Wan et al. (2014) in scenarios 1 and 2 and Shi et al. (2020) in scenario 3; skewness by Shi et al. (2023). Hover a result to see the method used.

Questions

Conversions, not data. Your table stays in the browser.

When should I use this?

When a trial reports a median rather than a mean, and your meta-analysis needs a mean and a standard deviation. It covers the three scenarios the literature defines: median with the minimum and maximum, median with the first and third quartiles, and both together.

Which methods does it use?

The mean follows Luo and colleagues (2018). The standard deviation follows Wan and colleagues (2014) in scenarios 1 and 2 and Shi and colleagues (2020) in scenario 3. The skewness test is Shi and colleagues (2023). Hovering a result shows which method produced it, because the scenario decides that, not you.

What does the skewness flag mean?

These conversions assume the underlying data are roughly normal. The test checks whether the reported summary is compatible with that assumption; when it is not, the row is marked and the normal-based mean and SD should not be trusted. The usual answers are to leave the study out of the pooled estimate, to use a method meant for skewed data, or to report it narratively.

Does anything get uploaded?

No. The table lives in the page, the calculations run as you type, and the .xlsx file is written in the browser. Nothing is sent anywhere and nothing is kept after the tab closes.