You can certainly calculate the "required sample size" for finding a statistically significant difference (between one sample pre-post, or two samples, or various other designs), using retrospective data.

However, you need to provide more information as to what you plan to do with that information?

I would envision that you'd use the retrospective data to (1) help determine the sample size needed for a future prospective study, or (2)  you have conducted a study already and found that it did not achieve statistical significance. Thus you want to determine what sample size would be required for this result to be statistically significant.

The more information you can provider, the greater likelihood that you'll get helpful responses.

I echo Ariel's answer.

Once you determined your study design, the next step should be figuring out the sample size needed to adequately answer your scientific question (and thus accept or reject your null hypothesis).

In order for you to complete the above, you need to figure out (or at least estimate) the magnitude of the difference you want to demonstrate (e.g., "on average, one treatment is 20% more effective than the other treatment). You then need to pick your alpha and your power. Once that's set, you should figure out the sample size for each group (e.g., experimental versus control) using one of the freely available sample size estimation software packages. One of those that will cover the most common study designs is listed below:

http://biostat.mc.vanderbilt.edu/wiki/Main/PowerSampleSize

You can download it from there. It's legit and it's free. It is a Windows program, so in Linux you will have to use Wine. Don't know about Macs too much.

Hope this helps!

Regards,

Stan.

Hi,

Your question: Is it possible to calculate the sample size in retrospective patients?

If you want to decide the sample size for the retrospective study (for your study) you can follow the following:

1.  You have to take all the patients data from 2009 to 2014   (or)

2.  Your sample size depending on your study title.  Suppose, your study title is "Study on spinal cord injury in between 2009 - 2014" Then you will take the whole patients data in between the above mentioned years.  This is called as "Census Method"

3.  Suppose, you have considered any prevalence (in %), Then you have to calculate the sample size according to the prevalence of spinal cord injury in your study area.  Refer some previous study (Journal papers). 

Any other help kindly contact me.

Thanks.

Best regards.

It depends what you are planing to do with the sample size.

1. If you are analyzing the current data to answer a research question, then that is the sample for the analyses.

2. If you have calculated an effect size for the outcome and it is not significant statistically but you believe that it is clinically relevant. So, you want to do the power analysis to see whether you have enough power with this sample for the statistical analysis.

3. You want to use the data  for computing the effect size to design a new study.

In the last two cases, you can carry out a power analysis using the retrospective sample data.

Hope this help

Muhammad Riaz

The issue is how much is the sample, so you do power calculation and if you get the mimimum number in your available data, then you are fine.

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.

The issue is how much is the sample, so you do power calculation and if you get the mimimum number in your available data, then you are fine.

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.