In October 2016 the Emergency Medicine community was faced with the publication of the now infamous PESIT trial, and the symbolic wrench its authors carelessly tossed into the already indecisive diagnostic work up of patients presenting to the Emergency Department following a syncopal event1. Published in the NEJM, Prandoni et al enrolled patients who were admitted to the hospital following a syncopal event and engaged in what can only be described as diagnostic insanity. The authors employed a strategy in which all patients underwent an evaluation for PE. Patients were risk stratified, and then underwent serum screening using a d-dimer assay or radiographic imaging. The authors reported implementation of this diagnostic pathway, led to 17.3% of the cohort being diagnosed with a pulmonary embolism. If these results were true it would mean that either the underlying epidemiology of syncope had swiftly and drastically changed, or we have been missing an unacceptably high number of PEs.
A far more likely explanation for these authors unbelievable results is that they are simply not to be believed. In truth, they represented the natural consequence of the empiric use of a d-dimer assay in an elderly population and the overdiagnosis which follows. Despite the PESIT results being an extreme outlier from all the previous data examining syncope and its underlying causes, there has been a lack of high-quality evidence directly disputing PESIT’s claims. But with a recent publication in Annals of Emergency Medicine by Thiruganasambandamoorthy et al we finally have evidence directly contradicting the findings presented by the PESIT authors2.
Performing a secondary analysis of two large prospective cohorts, the authors sought to quantify the proportion of patients presenting to the Emergency Department with syncope, in which a PE was ultimately responsible for their transient loss of consciousness. The first was a large trial performed in 17 Canadian Emergency Departments enrolling patients older than 16 presenting to the ED following a syncopal event. The second was an equally large study performed in multiple US Emergency Departments enrolling patients greater than 60 presenting with either syncope or a near syncopal episode. The authors collected demographics, medical history, serological and radiological investigations for PE that occurred during the initial syncope presentation and 30-day outcome data.
Overall the authors enrolled 9,374 patients over a 6-year period. 273 patients (2.9%) had incomplete follow up, while an additional 10 (0.1%) withdrew prior to completion of the study. This left 9,091 for the final analysis. Of the entire cohort 6% underwent investigations for pulmonary embolism, only 56 patients (0.6%) were determined to have a PE as the cause of their syncopal event. 4 deaths could be attributed to PEs and 52 (0.6%) of patients died within 30-days due to an unknown cause.
These results fall in stark contrast to the 17.3% observed in the PESIT cohort. How do we reconcile these differences? The first reason could be that we are comparing two very different cohorts. Thiruganasambandamoorthy et al enrolled all patients who presented to the ED with syncope. In contrast, PESIT only enrolled patients that were being admitted to the hospital because of their syncopal event. Of the 2584 who presented to one of the participating ED for syncope during the enrollment of the PESIT cohort, only 560 (21.6%) were enrolled in the study. Do the results of these two cohorts diverge from one each other so drastically because they represent two very different patient populations?
Even if you examine the subset of patients in the Thiruganasambandamoorthy et al study that were admitted to the hospital (39.1% of the entire cohort) for their syncopal event, only 1.2% were diagnosed with a PE.
The PESIT cohort was an older population at higher risk of PE. Was this the reason for the contrasting rates of PE? The mean age in the PESIT cohort was 76. By comparison the Thiruganasambandamoorthy et al cohort had a mean age of 66. But if you examined the subset of patients enrolled in the US study where the enrolment criteria demand patients be older than 60, the mean age was 71. In this subset of patients PE was determined to be the cause of their initial presentation in 1.3%.
Clearly the differences in the rates of pulmonary embolism between these two cohorts are not simply due to patient selection. Rather, the most likely reason the PESIT authors found a much higher incidence of PE was simply because they looked harder. The question becomes how much of this represents true clinically meaningful disease, and how much is simply overdiagnosis? In order to make the claim that the large majority of the additional PEs identified in the PESIT cohort were clinically important disease, then one has to assume that if the large majority of these PE had gone unidentified, and in turn, untreated in the Thiruganasambandamoorthy et al, there would be clinical consequences for missing these diagnoses. This is not the case. Of the entire Thiruganasambandamoorthy et al cohort, only 6% underwent testing for PE. Of the patients who did not undergo initial testing for PE in the ED, who were admitted to the hospital, only 0.2% went on to be diagnosed with a PE as an in-patient. Of the patient who did not undergo testing for PE during their initial presentation, none were subsequently diagnosed with PE and only 0.5% died due to an unknown cause within in the 30-day follow up period. If 17% of these patients were secretly concealing pulmonary emboli that went undiagnosed on their initial presentation, one would have to imagine the clinical consequences of missing such diagnoses would be more apparent than what these authors observed.
Following the publication of the Thiruganasambandamoorthy et al cohort, it is apparent that the results of the PESIT trial were immensely inflated. The vast majority of the PEs diagnosed in the PESIT cohort fell into one of two categories. They were either clinically important PEs, identifiable as such, as they presented with signs and symptoms consistent with PE, or radiological diagnoses of no clinical import, simply the natural consequence of a diagnostic strategy which divorces itself from clinical reasoning.