# Re: st: Sample size calculation

 From Roger Newson <[email protected]> To [email protected] Subject Re: st: Sample size calculation Date Wed, 26 Oct 2005 17:09:21 +0100

My previous experience with cytokines is that (if measured well) they should have an approximately log-normal distribution. Therefore, to compare 2 groups of cytokine measurements, the correct approach is probably to define a confidence interval for a ratio of geometric means. To do power calculations for this, you need to know two main things, firstly the size of a clinically interesting geometric mean ratio to be detected, and secondly the likely within-group variability of the logs (which may be measured by coefficient of variation, SD of the logs, interpercentile ratio, or sometimes by geometric SD).

Power calculations for lognormally distributed variables are discussed in Newson (2004), which can be downloaded from my website (see my signature below) either using a browser or by typing within Stata

net describe powergen, from(http://www.kcl-phs.org.uk/rogernewson/papers)

and getting the ancillary file -powergen.pdf-. Geometric means and their ratios in Stata are discussed in Newson (2003), which can also be downloaded from my website either using a browser or typing within Stata

net describe gmratio, from(http://www.kcl-phs.org.uk/rogernewson/papers)

and getting the ancillary file -gmratio.pdf-. Another good source on the lognormal distribution is Stanislav Kolenikov's website at
http://www.komkon.org/~tacik/
which features a very useful reference with formulas at
http://www.komkon.org/~tacik/science/lognorm.pdf

I hope this helps.

Best wishes

Roger

References

Newson R. 2003. Stata tip 1: The eform() option of regress. The Stata Journal 3(4): 445. Also downloadable from my website at
http://www.kcl-phs.org.uk/rogernewson/

Newson R. 2004. Generalized power calculations for generalized linear models and more. The Stata Journal 4(4): 379-401. Also downloadable from my website at
http://www.kcl-phs.org.uk/rogernewson/

At 16:01 26/10/2005, Ronan wrote:

```On 26 DF�mh 2005, at 14:19, <[email protected]>
<[email protected]> wrote:

```
```We are planning to measure some cytokines in patients with MS
comparing
between a group with treatment and control, but there have not been
literatures regarding these measurements, so we do not have an
estimated
mean and standard deviation. Is there a way to estimate a sample size?
(our IRB really wants to see sample size estimate.) Thanks.

```
```You really are in the dark, then.

There are several approaches. The Resource Equation method

http://embryo.ib.amwaw.edu.pl/invittox/er/ER/ER%2029.pdf see page 9

is useful in a preliminary experiment. It's based on the idea that
the amount of information in each new piece of data diminishes as the
sample size increases.

It allows you to estimate the sample size needed to ascertain whether
a study would be useful or not. It is often used in animal research
where investigators have no idea at all of what they might find.

Ron�n Conroy
[email protected]

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```
```
--
Roger Newson
Lecturer in Medical Statistics
Department of Public Health Sciences
Division of Asthma, Allergy and Lung Biology
King's College London

5th Floor, Capital House
42 Weston Street
London SE1 3QD
United Kingdom

Tel: 020 7848 6648 International +44 20 7848 6648
Fax: 020 7848 6620 International +44 20 7848 6620
or 020 7848 6605 International +44 20 7848 6605
Email: [email protected]
Website: http://phs.kcl.ac.uk/rogernewson/

Opinions expressed are those of the author, not the institution.

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```