As large number of articles evidence, a panel of biomarkers can be used for cancer diagnosis or outcome prediction. In such situation, How to design the study and How to estimate the requried sample size and optimal ratio of the case to control?
Its given in research methodology books. the adequate sample size was depends on the parameter you studied. its depends on location ,territory ,population,prevalence,suspectibility, endemicity, and main importance --availability,resources and funding if any......
..but for routine dissertation ,sample size of 100s to 500 is enough and for ph.d thesis upto 1000 is good enough arbitrarily.......
Hi,
Your question is: How to design the study and How to estimate the requried sample size and optimal ratio of the case to control.
Study Design:
a. If you want to do the study in community, then its called community based cross sectional study.
b. If you do the research based on record (Secondary data) then its called as retrosective study.
c. If you want to do a study or research in your or your near by hospital then its called hospital based observational study.
d. If you take, case and control then its called case control study.
e. Cohart study, and survellience study, there are so many studies are there.
Sample size: you can find this, by online sample size calculator in google. or otherwise see some biostatistics book.
If your study based on prevalence, then you can calculate the required sample size from the following formula:
Sample Size = n = 4 p q / (dxd)
Here, p = prevalence (in %), q = 100 - p and d = admissible error = 10% of prevalence.
Then, you do the research as per you protocol and do the analysis and use the appropriate statistical analysis and find the results and write the interpretations based on the existing results.
Hi! As soon as you have reliable data on the test performance of your biomarker assays and -even more important- the variablility of the biomarker in your target population, you will have to decide which differences are diagnostically relevant (which may vary for different diagnostic tasks, e.g. screening vs. confirmational tests). With a given study power (again dependent on the requirements) you will get a rough estimation of the sample size with one of the (freely available) sample size calculators
The study design and sample size calculation depend on your research question. And you haven't stated what that is. Are you looking at the association between one or more known biomarkers and diagnosis or prognosis of cancer in a particular population? Or do you want to identify new ones?
However, since you mentioned that there is a lot of literature on biomarker diagnosis and prediction studies in cancer research, I would start there for methodology solutions. Usually, the authors describe in detail how and why they chose a particular study design, and how they calculated sample size and power.
There are online calculators, but I don't recommend them for complex situations. It's best to contact a biostatistician, or just to carefully and thoroughly study the references cited for sample size and power calculations.
I noticed that biologists studying pollution would sometimes start from an effect size that was meaningful in theory. That is a certain % enlargement of fish livers might be meaningful in terms of pollution. This can be converted into an effect size and then power is calculated from there. This approach requires good theory to suggest what is meaningful.
Usually, when you report a panel of diangostic biomarkers or diagnostic prediction model, we need give the sensitivity, specificity, accuracy and AUC of the prediction model. Therefore, I need to estimate the sample size and power for Sen, Spe, Acc and AUC.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size