When calculating the sample size is not necessary

#1
Hi everyone.
I am planning to do an observational study in which it is intended to analyze the dispensation in pharmacies of a certain drug.
I will have access to the entire dispensing database for this drug, is it necessary to determine the sample size?
If calculation is not necessary, how should I mention in the study protocol that this calculation is not necessary?
I cannot find any bibliographic reference that can support this non-need.
Can anyone help me?
Thank you so much in advance
Adriana
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
Will you also have every year's worth of data. If so, you would have the full population and statistics with precision estimates may not be necessary, unless you are predicting future events.

As for the sample size calculation, it doesn't hurt to do one, since it forces you to write out your assumptions a priori. What is your actual study question?
 
#3
Thank you so much for your answer.
It is a retrospective study (observing past information) and we need to determine the % of the persistence of a drug use at a one-year since the first dispensation.
I already determine the sample size, but I was asked if it is really necessary, considering the fact that it is possible to observe all drug dispensations.
I start in a new company as a statiscian and I have little experience in this matter.
If it is correct to determine the sample size, than it is already done, but if it is unnessary, I don't know how to justify. References, for instance.
Could you give me some lights? I'm a little bit lost...
Adriana
 

hlsmith

Less is more. Stay pure. Stay poor.
#4
Welcome to the forum and congrats on your new job.

No reference, but sample size calculations help protect against two things in particular. First, that you did your due diligence to speculate that there will be sufficient data to answer the question. If not, you just wasted everyone's time and potentially compromised individuals' confidentiality and privacy by accessing their data. And secondly, if forces you to formulize the question and hypothesis to prevent from disseminating accidental or false discovery. Many journals may want to see you have done this to curtail reproducibility concerns in research.
 
Last edited: