I am thinking about cultures. Can you do a pooled sensitivity and specificity estimate on the performance of a culture technique on different kinds of samples? I mean, put together sputums, cerebrospinal fluid, whole blood, etc. if it is cultured with the same technique or would that be, from a methodologically point of view, very incorrect?
Mathematically you can do that, but coceptually it is not OK. The best way is to perform meta-analysis on each culture separately and then compare the results with each other. Combining all samples does not add much to what you are going to convey by the meta-analysis
Mathematically you can do that, but coceptually it is not OK. The best way is to perform meta-analysis on each culture separately and then compare the results with each other. Combining all samples does not add much to what you are going to convey by the meta-analysis
Hi Philip
I think first you should research if the type of sample used is affecting the sensitivity and specificity performance of the technique. For example the ability to detect the presence of a bacteria in whole blood might be different from the ability to detect them in cerebrospinal fluid. If this is the case then I don't think you can pool the results.
One way to go around the use of different samples if the technique is the same is to actually do subgroup analysis instead of pool them all together if the evidence so that the sensitivity and specificity is affected by the sample you use.
The other thing that Patrick has mentioned is how standardized is the method across different institute. I think a lot of this techniques especially in a hospital setting will need to fulfill certain standards and be calibrated. If this is the case maybe you can ask from the relevant institutes to provide that information by contacting the microbiology department for example.
- Badges
- Science method