Everyone clearly hates PESIT. Ever since its publication in the New England Journal, the rage has been virtually palpable – both in the editorial throwaways and in the academic correspondence. Then, its publication has inspired multiple follow-up refutations of the generally observed prevalence of pulmonary embolism in patients with syncope.
So, this is one more of those. This is a retrospective analysis of large administrative databases from Canada, Denmark, Italy, and the United States. These databases totaled 1,671,944 patients administratively labeled as having an Emergency Department diagnosis of syncope. Their primary outcome was patients discharged with a diagnosis of PE, with sensitivity analyses including any diagnosis of PE within 90 days as well as a wider definition of venous thromboembolism.
Their range of answers, of course, is far below the 17% found in PESIT. The top of the range on initial ED presentation, using the most generous definitions for PE prevalence, was 1.37%, but the most likely prevalence is around 0.25-0.5%. With respect to hospitalized patients, the top end of the estimate was 3.86%, with the more likely prevalence around 1%.
There are, of course, limitations associated with coding and using administrative databases – particularly as compared to a prospective study like PESIT. However, all these other snapshots of data are consistent in finding vastly different results more consistent with our anecdotal experience.
“Prevalence of Pulmonary Embolism in Patients With Syncope”