Hey Sarah,

here are some ideas:

Here is a link to an online calculator:

http://www.stat.ubc.ca/~rollin/stats/ssize/caco.html

You could use G-Power (a free tool for power calculations):

http://www.gpower.hhu.de/

And anoother link to a book for sample size estimation in health studies:

http://whqlibdoc.who.int/publications/9241544058_%28p1-p22%29.pdf

I hope this helps.

Regards,

Timo

Thank you so much Timo, very grateful for your help! 

Hi Sarah

There are many online sites for calculating sample size for all types of studies

A useful & easy to use is: www.dssresearch.com

You should define the following: alpha error, bets error (or study power), the expected proportion ( or mean & SD) of the outcome in both cases & controls. Note that sample size calculated is per group

The sample size for a case control study is based on the reduction od the odds ratio considered clinically relevant. In addition, you have to fix the statistical significance (alpfa = 0.05, two tailed) and the power (at least, 0.80) and also on the number of controls for each case.

The above link (http://www.stat.ubc.ca/~rollin/stats/ssize/caco.html), uses the baseline proportion (p0) and the relative risk instead of the odds ratio and it can be correct in the case of unmatched case - control, as it is correctly reported.

Thank you Mr Mario!

Thank you Mr Gilany!

Sample size calculation for case-control studies depends on the hypothesis to be tested. The main feature of your task is to consider as "controls" individuals that are mostly equal in every feature to your "cases", except for the probability of exposure you will use as the main or primary hypothetical category to be studied. Once you have this clear, you will need to have an "under the null hypothesis" the (reasonably) estimated rate of exposure for controls. From here on, the sample size calculations will depend on whether you want to carry out a 1:1 rate of cases and controls, as well as whether you will do it as a "paired" or "matched" design.

As you have been told, there is an array of programs for “mechanically” making the calculation, some expensive, some other for free. What is most important is to having clear the issues mentioned above, otherwise you might get lost among the not always straightforward information the programs use to require from you to fill in the “forms”.

I would add -to your better understanding and success in your effort- to have in mind the number of analysis you have in mind to carry out on the data calculated for your main hypothesis, something that might advise you to enlarge your sample size, so as avoid to falling short during a secondary or tertiary analysis looking for confounders. This is achieved more easily if your expected odds ratio is small (not over 2 or 2.5, unless you have nice reasons to expect it high). The temptation of selecting small, easily achievable sample sizes makes likely to leading you to incomplete or biased results and –worse- interpretations.

Thank you Mr Herrera! 

Very grateful Mr Sahin! 

Dear Sarah, there are many statistical packages for mathematical calculations. Yet, before using any of them, remember that a working hypothesis and expected Ors are critically needed  to properly interpretation of the program output.

Sincerely,

to do power analysis to estimate your sample size, you have to write your hypothesis, and based on that you decide what statistical test you will use. It should be one of the inferential statistics. so you need to determine the following: alpha {standard to be .05}, power [standard to be .80], effect size {small, moderate, or large, each test has its own value, you can find these values in the net}. Then download free programs to calculate the sample size such as G. power.

For a  case-control study sample size calculations you will need: a) the exposure rate in controls (an estimate, sometimes hard to get), b) the size of the difference between cases and control exposure rates (a given OR size may be selected, but beware:  the bigger the OR, the smaller the sample size and higher  the likelihood of  of the lack of statistical). If you will calculate the sample size using an OR, my advice is to pick OR=2 or  2.5 as the most. Then you need the standar alpha value (ussually p= 0.05 two tailed, which is Z =1.96. and the beta error, usually  size 0.80. If you intend to carry out subgroup analysis or test more than one hypothesis, you must have it in mind at the time of sample size calculations.

Dear Sarah, Mr Herrera had told you a good method, but a) the exposure rate in controls (an estimate, sometimes hard to get) is a big matter to calculate sample size for case and control study, I am sending a article, may be it help you, At the end of this article , some free web sites and links , kindly see and use it,

i think , EpiTools epidemiological calculators and open epi is best

thanks

I had similar question and the responses have helped me.

Thank you

Thank you.

The answers were helpful.