Adverse Clinical Outcomes Among Patients With Acute Low-risk Pulmonary Embolism and Concerning Computed Tomography Imaging Findings

Key Points Question Are concerning computed tomography findings among patients seen in the emergency department with acute, low-risk pulmonary embolism (PE) (eg, saddle PE, right ventricular strain, pulmonary infarct) associated with differences in treatment and/or clinical outcomes? Findings In this cohort study of 817 patients, concerning computed tomography findings were associated with increased hospitalization and resource utilization but not short-term adverse clinical outcomes. Meaning These findings suggest that concerning computed tomography imaging findings may be a significant barrier to outpatient treatment among patients with otherwise low-risk acute PE.


Construction of the U-M acute ED-PE registry
Construction of the U-M registry began with data from the Michigan Emergency Department Improvement Collaborative (MEDIC), a state-wide quality Collaborative Quality Initiative (CQI) which reviews the charts of all patients who undergo CTPE in one of the 37 Emergency Departments within the network. Positive PE cases from our institution, as identified by MEDIC abstractors, were pooled with those identified by query of our electronic medical record. The latter was done, in part, to capture patients diagnosed on outpatient CTPE who were immediately referred to the ED. Of the 967 cases identified by MEDIC abstractors and EMR query (Supplemental Figure 1), physician reviewers excluded 63 cases in which there was no objective evidence of PE. The most common reason for exclusion was inaccuracy in MEDIC abstraction (24 cases). Reviewers also excluded 14 cases that had been inappropriately identified as PEs via query of the EMR, 3 cases in which a diagnosis of PE was made presumptively by the ED provider (based on high clinical suspicion) but CTPE imaging was ultimately negative, and 22 cases in which the initial radiology read indicated PE, but final interpretation was negative (motion artifact, 5 cases; inadequate contrast bolus, 5 cases; infiltrating tumor, 5 cases; low flow due to severe pulmonary hypertension, 2 cases; misinterpretation by a radiology trainee, 5 cases). An additional 87 cases (8.9%) were excluded due to the PE being chronic (58 cases), not clinically significant (all isolated subsegmental PEs which were not treated, 7 cases), septic emboli (2 cases), diagnosed after the patient had left the ED (5 cases), or previously diagnosed and treated at another facility (15 cases).

Chart abstraction and data verification
For each chart identified as an acute ED-PE, demographic data (age, sex, ethnicity, date of birth, date/time of ED arrival, date/time of ED or hospital discharge), vital signs, and laboratory results were electronically abstracted from the electronic medical record. Missing values were obtained via manual chart review. Additional data elements were manually abstracted by two physician reviewers, including components of the PESI score (see below), CTPE findings (see below), ultrasonography (point-of-care ultrasonography and formal transthoracic echocardiography), treatment (anticoagulation, PE Response Team [PERT] activation, use of thrombolytics and/or advanced interventions, and hospitalization), and outcomes (need for intensive care and mortality). PERT activation was analyzed only for acute ED-PEs which occurred after June 1 st , 2017, when the PERT team was formed. Disagreements were adjudicated by third physician review.
Calculation of PESI score/class from abstracted variables PESI score and class were calculated using abstracted variables -age, gender, comorbidities, mental status, and peak vital signs during the ED stay (i.e., maximum heart rate, maximum respiratory rate, minimum systolic blood pressure, and lowest O2 saturation). In assessing history of cancer (+30 points), only active cancers were counted. Patients were considered hypoxic (+20 points) if they did not have a history of chronic (i.e., home) oxygen use and required supplemental O2 of > 2L, or if they required > 2L increase over their home oxygen requirement.

Blood-based Biomarkers
While precise cut-offs for Troponin I (TnI) and high-sensitivity Troponin T (hs-TnT) have not been defined in the context of acute ED-PE, we reasoned that EM providers would be unlikely to consider any patient with an abnormal troponins (as defined by our hospital laboratory) to be low risk. With this in mind, values above the 99 th percentile for our laboratory (> 0.16 ng/mL for TnI, > 19 ng/L for hs-TnT, and > 100 pg/mL for BNP) were considered abnormal.

eTable. Acute ED-PE Patient Characteristics and ED Disposition
Acute ED-PEs had similar demographics as compared to excluded cases. However, the rate of outpatient management for acute ED-PEs was markedly different than for non-PEs (2.6% vs. 22.2%, p<0.001) or non-acute or non-significant PEs (2.6% vs. 21.8%, p<0.001). *Other indicates patients who left AMA or were sent to OB triage or Psychiatric Emergency Services. Results of parallel analysis incorporating PESI score/class and biomarkers into risk stratification. The associations of CTPE findings with clinical outcomes, hospitalization, and resource utilization in low-risk acute ED-PEs remained unchanged. A. 7-and 30-day mortality, *** -p < 0.001 vs. both low-risk groups, B. need for intensive care, *** -p < 0.001 vs. both low-risk groups, C. Rate of outpatient management (* -p = 0.01) and hospital length of stay (LOS, ** -p < 0.01), D. point of care ultrasound (POCUS), transthoracic echocardiography (TTE), and PE response team activation (PERT), *** -p < 0.001.