Background: Results from conventional diagnostic case-control studies have been questioned, since this design can give biased estimates of diagnostic accuracy measures. Consequently, this approach has not been recommended in the recent Standards for Reporting of Diagnostic Accuracy (STARD) guideline.

Objectives: We discuss why a nested case-control design is generally more efficient than a full cohort approach and, when conducted appropriately, equally valid.

Patients & Methods: An empirical data set comprising patients suspected of deep vein thrombosis (DVT) was used for illustration purposes. Nested case-control samples were drawn from the full cohort to compare estimates of diagnostic accuracy measures for two diagnostic tests for DVT (the D-dimer and calf difference test) with estimates obtained from the full cohort.

Results: Estimates of the diagnostic accuracy measures as estimated in the nested case-control samples were very similar to those estimated in the full cohort. For example, the positive predictive value of the D-dimer test was approximately 0.30 in both the full cohort and in all four nested case-control samples and the OR of the calf difference test was approximately 1.44 in the full cohort and in the nested case-control samples.

Conclusions: Nested case-control studies should be applied more often in diagnostic research, as this design provides valid and precise estimates of diagnostic accuracy measures. The efficiency of the nested case-control approach is particularly beneficial when the test under study is costly or difficult to collect, since the number of subjects to be tested is reduced compared to the full cohort approach.