Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Sample size calculation


From   Neil Shephard <nshephard@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Sample size calculation
Date   Wed, 26 Oct 2005 15:15:51 +0100

On 10/26/05, slaowattana@pol.net <slaowattana@pol.net> 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.
>
> 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.)

Probably not the answer/help your looking for, but....

Sample size is dependent on a number of factors.  The significance
level at which you wish to detect differences (termed alpha), the
power you wish your study to have (termed beta), and the size of
effect you are expecting to see.  The type of test you are to perform
also has an influence, and from the mention of mean and sd, its a fair
guess your likely to use a t-test (providing the assumptions are met).

You have not mentioned any of these quantities, and only you can
determine appropriate alpha and beta levels (and tests) that are
acceptable to your study design.  You will have to decide on these
prior to performing any sample size calculations.

You do however indicate that to your knowledge there is no literature
regarding the measurements you wish to take.  I'm not an expert on MS,
but since it is an autoimmune disease I believe cytokine activity is
one facet that is commonly investigated.

A search on PubMed indicates that there are 96 articles with the both
the terms "cytokine" and "multiple sclerosis" in the title.  Obviously
these may not be the cytokines your intersted in but they should help
provide some information.

(To duplicate this seach got to http://wwwncbi.nlm.nih.gov/, select
PubMed as the database to search, and enter the search term 'cytokine
multiple sclerosis' and hit return.  Now click on the 'Limits' tab and
in the first pull-down box select 'Title' to restrict the search to
artciles with the terms in their titles).

You may also benefit from reading the OMIM (Online Mendelian
Inherithance in Man) entry for MS which can also be accessed via NCBI

However, the question still remains as to how much difference the
treatment is likely to make to the mean, and that is presumably why
you are carrying out the experiment.  You don't state the organism,
but I'll guess that your working with humans.  Is the treatment with a
drug?  If so it is highly likely to have been tested in animal models
before being trialed in humans, and these may inform you about the
expected difference.  What sort of effects have been reported for
other treatment effects/cytokines in MS?

If there is absolutely no prior information to inform on the expected
size of effect then you can not possibly determine an appropriate
sample size, and your best approach would be to determine sample sizes
for a range of effects, see how these compare to your available
resources (money).  That way you could say to the IRB (Independent
Review Board?) that with a sample size of N, you would have X% power
to detect an effect of size Y at a significance level of alpha.  You
could even show them figures of your calculations which show that
larger sample sizes would allow you to detect smaller differences in
effect.

HTH's

Neil

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index