PESIT Investigators: the prevalence of PE in those hospitalized following first syncope
“The only way to get rid of temptation is to yield to it.”
A 76 year old woman without known medical comorbidity is ambulating along 7th avenue, rounding the corner where St. Vincent’s Hospital once operated. It is an exceptionally humid August afternoon and she has not been eating well because of the recent death of her beloved dachshund, Archie. She calls out that she feels faint, reaches out for a nearby brick wall and is overcome by a feeling of warmth. She then stumbles to the ground, breaking her left humerus. A passerby rushes towards her and finds her transiently unresponsive; she regains consciousness quickly and emergency medical services are activated.
The transient loss of consciousness with spontaneous resolution due to cerebral hypoperfusion – syncope – has a plethora of possible causes. Amongst the long lists of syncope etiologies lies pulmonary embolus [PE]. The mechanism is certainly straightforward; a large, abrupt increase in right ventricular [RV] afterload transiently impairs RV forward flow, the transmural pressure of the pulmonary artery rises and RV volume increases. The engorged RV stiffens the interventricular septum and left ventricular [LV] filling is impaired. Thus, LV diastole is doubly embarrassed – there is diminished preload from falling perfusion through the pulmonary vascular tree, and the LV filling characteristics are stiffened by its encroaching RV neighbor; pulsus paradoxus may be seen.
Consequently, LV stroke volume falls as a function of contracted end-diastolic volume. To the extent that hypoxemia, or underlying myocardial dysfunction spoil LV contractility, LV end-systolic volume will simultaneously rise – LV stroke volume plummets, syncope ensues.
Given the aforementioned, the PESIT investigators – reported in yesterday’s New England Journal of Medicine – yielded to the temptation of pulmonary embolus as a cause of syncope; their findings have already created quite a stir.
2584 patients with first-time syncope were screened in 11 Italian emergency departments – 9 of which were non-academic; 72% of these patients were discharged home based on a clinical evaluation by a physician in the emergency department. Therefore, 717 patients were admitted to the hospital for the following reasons:
-trauma related to falls
-severe coexisting conditions
-failure to identify an explanation for the syncope
-high probability of cardiac syncope based on the EGSYS score
Of these remaining patients admitted to the hospital, another 157 were excluded for the following reasons: 118 were receiving anticoagulation therapy, 82 had atrial fibrillation, 36 had other reasons, 35 had recurrent syncope, 4 declined to participate.
Thus, the residual 560 inpatients were then all evaluated for pulmonary embolus. This followed an algorithm which utilized the Well’s Score to predict either “unlikely” [≤ 4 points] or “likely” [> 4 points] pulmonary embolus followed by a d-dimer assay for those deemed to be “unlikely” to have a PE. If the patient had a negative d-dimer as determined by the manufacturer of the assay and was deemed “unlikely” to have a PE by the Well’s Score, the patient was ruled out for PE – no further investigation was done. If the patient was either deemed “likely” by the Well’s Score to have a PE or if the d-dimer was positive, the patient received a CT angiogram [CTA] or a ventilation-perfusion study to evaluate for PE.
Of the 560 patients admitted to the hospital for first-time syncope, about 6 in 10 were ruled out for PE based on Well’s Score and a negative d-dimer assay. The remaining 4 in 10 underwent radiographic evaluation for PE. Of this group of patients who underwent radiographic evaluation, about 4 in 10 were diagnosed with PE. In totality, of all patients admitted for first time syncope, about 1 in 6 [17.3%] were found to have a PE.
Among those found to have a PE by CT angiography, 4/10 had a PE in the main pulmonary artery and another quarter of patients had PE in a lobar artery. Only 7% were found to have sub-segmental pulmonary emboli by CTA.
Of all inpatients with first-time syncope diagnosed with PE, one quarter had no clinical manifestations of the diagnosis [e.g. tachypnea, tachycardia, hypotension, or clinical signs or symptoms of deep-vein thrombosis].
One very important, and potentially overlooked, finding of this study is that a large majority [72%] of patients presenting to the emergency department with first-time syncope could be – seemingly – safely discharged home. These patients were, as expected, younger and healthier on average.
Amongst those who were evaluated in the ED, the EGSYS score was used to predict a high-probability of cardiac syncope. Typically those patients with a score of 3 points or more are considered at high risk for cardiac syncope and therefore admitted. The points are allocated by the following:
4 points for palpitations prior to syncope
3 points for heart disease, abnormal ECG or both
3 points for syncope with effort
3 points for syncope while supine
-1 point for precipitating or predisposing factors [warm, crowded, fear/pain/emotion, prolonged standing]
-1 point for autonomic prodrome [e.g. nausea, vomiting]
Note that, pathophysiologically, a pulmonary embolus irritates the heart via multiple mechanisms. For example a clot in transit may precipitate an arrhythmia, produce an abnormal ECG in response to RV strain and hypoxemia and test the heart’s physiological reserve on exertion. It shouldn’t be surprising, therefore, that in a patient population selected for a high EGSYS score that PE is over-represented.
With respect to their algorithm for detecting PE, the authors did not utilize age-adjusted d-dimer cut-offs. In theory, correcting d-dimer for age in those admitted to the hospital might spare a significant number of inpatients with syncope from additional radiographic investigation. For example, patients between the ages of 71-80 have an increase in specificity [fewer false positives] from 25% to 44% while those older than 80 have an increase in specificity from 15% to 35%. In other words, 20% fewer radiographic studies in these age groups.
Interestingly, overall clot burden has been generally unhelpful in predicting patient mortality in PE. Rather, it is clot location which provides a better method of prognostication. Notably, central and lobar PEs predict ‘death or disability’ better than overall radiographic burden. Importantly, in those patients found to have PE by CTA by the PESIT investigators, 67% had PE in these concerning locations! While an enlarged RV on CTA is also noted to help predict patient outcome, this variable received no mention by Prandoni et al. Ostensibly one would expect to see an engorged RV in clinically significant PE; it would have been interesting to note the prevalence in the PESIT study.
Regarding the radiographic diagnosis of PE, given that 9 of the 11 hospitals were ‘non-academic’ it raises the question as to whether or not dedicated chest radiologists were interpreting the angiograms. In fact, there is no mention in the article as to who was making the final diagnosis of PE [i.e. centralized, blinded interpretation versus locally-based radiologists]. This is of importance given recent data implying that upwards of 25% of pulmonary emboli on CTA were initially falsely detected when re-read by 3 separate and dedicated chest radiologists.
Lastly, it is not clear how well these results can be completely adapted to other countries as the threshold for admission to the hospital for syncope may differ. For example, in one abstract, the rate of syncope admission in Canada was 12%, while in the U.S. it was 50% [interestingly, the rate in Italy was more similar to the U.S. at 44%]; overall, however, there was wide variability. To the extent that this variability in geographical practice changes the prevalence of PE in hospitalized patients, the likelihood of detecting PE may similarly vary.
Return to the Case
The woman’s EGSYS score is found to be 2 and she is admitted for further management of her humerus fracture. The patient’s history was strongly suggestive of poor solute intake and volume depletion secondary to a typical humid Manhattan afternoon, so she receives 1 L of lactated ringers intravenously. On reaching the floor, however, the admitting house-officer sees that the ED sent a d-dimer which was elevated. She obtains consent from the patient and orders a CT angiogram. An hour later the radiologist urgently pages her. The patient has a saddle embolus with RV distention on the CTA. She is clinically and hemodynamically well, but sent to the MICU on a heparin infusion for close hemodynamic monitoring.