Sample size estimation for evaluation of diagnostic tests

#1
I am studying 2 blood tests (A and B) to detect infection following exploratory laparotomy undertaken to treat various surgical lesions. In other words, the population is somewhat heterogenous. In a similar previous study (51 patients), other authors reported on test A. The blood will be sampled daily for 7 days.

The incidence of infection following exploratory laparotomy is approximately 10-20% in multiple reports. About 4 other types of complication may also occur which may influence the test results and these complications have similar incidence (i.e. 10-20%) following this surgery.

My hypothesis is that the novel test (B) will have greater accuracy than test A to detect infection. In the previous study, test A could not distinguish between complications at early time points but could discriminate infection with 50% positive predictive value at the final time point.
My 2 question are, please.
  • Is this accurately described as a cohort study?
  • How do I calculate my ideal sample size based on the information I have from the previous study of test A given an expected infection incidence of 10%?