All steps of the diagnostic process are undertaken based on our prior knowledge of the disease that we would like to diagnose or rule out. A key prior knowledge is represented by the expected probability of the disease, in this case pulmonary embolism (PE), across different groups of subjects. This means that the performance of a diagnostic algorithm are not immutable, but largely depends on the probability of PE before having performed any diagnostic test. The main clinical consequence of that is that a diagnostic algorithm could be safe to rule out PE in a low-risk population, but should be used carefully if the expected prevalence of PE is high.
In this work, the authors studied the performance of available PE diagnostic strategies to rule out PE across patient groups and settings.
The results are straightforward. The probability of having PE was highest among hospitalized or nursing home patients, followed by referred secondary care, and primary healthcare. The authors analysed and discussed the safety and efficiency of the most studied diagnostic strategies in the different settings. This study has a potential direct clinical implication, as it permits to tailor the use of the most appropriate rule out strategy in each healthcare setting optimizing safety and, at the same time, minimizing unnecessary imaging tests