Question Regarding Research Design

#1
If I have a dataset with 100 unspecified cases and I know from a previous chart review that the likelihood is that 10% of these are indeed cases how do I randomly select the 10%? I have been searching for journal articles that describe a method but have been unsuccessful. I would greatly appreciate any advice or direction to a journal article. Thanks!
 

hlsmith

Omega Contributor
#2
100 random observations, don't know if the observation(s) is a case or not? The prevalence of cases is 10% in population, so if your sample is random, 10% of your observations should be cases.


I don't get the rest of your post, why would you randomly select these 10%? Seems pretty Bayesian...


Please provide more information!!
 
#3
I completed a review of a few hundred medical charts and identified that 10% of the time when an unspecified ICD code was recorded, the presentation could be attributed to a bee sting. I now want to use this result and extrapolate (I'm not sure if that is the right term) 10% of unspecified presentations from a larger national data set (3000 presentations). Ultimately I am investigating trends over time but I can't include all unspecified presentations in my analysis as it is most likely that only 10% of these are due to bee sting for example.

I would appreciate any advice. After further investigation I think I have to calculate weighted averages but I'm not sure. Thanks for your help!