Sustainability of a Clinical Decision Support Intervention for Outpatient Care for Emergency Department Patients With Acute Pulmonary Embolism

Key Points Question Were the outcomes associated with a physician champion-led, electronic health record–embedded clinical decision support intervention for risk-stratifying adults presenting at the emergency department (ED) with acute pulmonary embolism sustained 4 years after initial promotion? Findings In this cohort study of 1039 patients across 21 EDs, outpatient management increased significantly compared with prior practices in former control EDs. Former intervention sites continued to outperform former controls in managing acute pulmonary embolism among patients with low risk. Meaning These findings suggest that this champion-led, clinical decision support intervention was associated with sustained practice change in identifying and safely discharging patients with low-risk pulmonary embolism.


eAppendix 1. Anticoagulant Recommendations During the SUS-EFX Study Period
Throughout the study period, guidelines from our health system recommended rivaroxaban, a direct oral anticoagulant, for the initial treatment of acute PE in patients without severe renal disease. Rivaroxaban, though not on the system's formulary, was preferred over dabigatran, as rivaroxaban avoided the need for a lead-in period of low molecular weight heparin. Dabigatran, however, was on formulary and hence less expensive for patients. For patients with active cancer, a low molecular weight heparin or a direct oral anticoagulant was recommended, depending on the location and nature of the malignancy as well as the patient's risk of bleeding. For patients with severe renal disease, warfarin was recommended with a concomitant bridge of either unfractionated or low molecular weight heparin, depending on the degree of renal insufficiency. eAppendix 2. Importance of Ready Access to Pharmacotherapy and Close Follow-up Ready access to pharmacotherapy and close follow-up are both prerequisites for outpatient care according to the American College of Chest Physicians (CHEST) guidelines: "In patients with low-risk PE we recommend outpatient treatment over hospitalization provided access to medications, ability to access outpatient care, and home circumstances are adequate." 1 Health systems without these structures have had to create them before implementing outpatient PE treatment pathways. 2 Every week, the principal investigator of the eSPEED trial received a list of patients at the 10 study EDs who were probable candidates for RISTRA-PE (using criteria described previously). 13,14 He reviewed each case, confirmed eligibility, and identified among the eligible those whose physicians had used RISTRA-PE and those who had not. He sent each site champion a weekly list of their facility-specific results, which they used to provide feedback to their department.

eAppendix 5. Altered Mental Status Variable of the Pulmonary Embolism Severity Index
In prior ED PE studies in our setting, approximately 5% of patients had documented altered mental status, the variable with the highest score in the PE Severity Index (see eTable 1 above). 13,14 There were, however, no statistically significant differences in altered mental status between intervention and control EDs in the eSPEED trial. 14 The assumption that all our SUS-EFX patients lacked altered mental status will have miscategorized some actual higher-risk patients (who are unlikely to be managed as outpatients) into the lower-risk strata (just based on lower scores). This may have slightly reduced the proportion of patients categorized as lower-risk that received outpatient care. The true proportion may have been higher.

eAppendix 6. Selective Manual Electronic Health Record Review
We performed manual chart review of 109 cases. We found that computerized evaluation of study eligibility (108/109) and site-of-care categorization (106/109) were accurate. The 3 corrected site-of-care cases were found during a focused review of patients with prolonged ED length-of-stay (>12 hours). These 3 had been misclassified as ED-only cases, but in reality, had been transferred from the ED to an outpatient observation unit for a brief stay (<12 hours) prior to discharge home. We corrected their categorization.

eAppendix 7. Defining 7-day Pulmonary Embolism-Related Hospitalization for Those Managed as Outpatients
Our primary safety outcome for those managed as outpatients was 7-day hospitalization for PE-related signs, symptoms, or interventions, defined a priori and used in prior studies in our setting. 13,14 These include complaints of dyspnea, chest pain, syncope or pre-syncope, limb pain or swelling, bleeding or findings of pleural effusion, elevated liver enzyme levels, new anemia or hemorrhage, or new or worsening deep venous thrombosis or PE, or interventions including assistance with medication administration, respiratory support (nonrebreather mask, noninvasive ventilation, intubation or mechanical ventilation), parenteral vasopressor administration, inferior vena cava filter placement or removal, or cardiopulmonary resuscitation. This near-term temporal horizon (7 days) is of concern for treating emergency physicians 15  The patient presented to an intervention ED with a chief complaint of tachycardia and mild dysuria for 1 week. Her past medical history was notable for psychosis but without risk factors for venous thromboembolism. Her initial vital signs were normal except for tachycardia, which improved after intravenous fluid treatment, but did not fully resolve. The rest of her physical examination was normal. Urinalysis was negative. CTPA was positive for PE without right ventricular dilatation. She was started on rivaroxaban in the ED and discharged home with a next-day appointment with her primary care physician and a consult to Anticoagulation Management Services. She returned to the ED less than 24 hours later complaining of abdominal pain and chest pain and was admitted to the hospital. She had delusional parasitosis and agitation, requiring intravenous antipsychotics. She remained hemodynamically stable and did not require supplemental oxygen or respiratory support. She was discharged 5 days later to a psychiatric facility and continued rivaroxaban for 6 months without consequence.

53-year-old F PESI: 50 (Class I)
The patient presented to a control ED with 1 day of right-sided chest pain that radiated to the right shoulder. She denied dyspnea. History was notable for right knee replacement 10 days prior with twice-daily aspirin for venous thromboembolic prophylaxis. Vital signs were unremarkable, and no leg swelling was noted. CTPA was positive for PE without right ventricular dilatation. Her serum troponin and 12-lead electrocardiogram were negative. Rivaroxaban was prescribed with strict return precautions to the ED. She returned the next day for new-onset dyspnea. Serial vital signs were normal. She felt better and was discharged home with next-day with follow-up with her primary care physician. At that visit she complained of worsening dyspnea, severe pleuritic chest pain, fever, and cough. Her vital signs were normal, except for an elevated temperature. A portable chest radiograph showed a new right pleural effusion and adjacent focal opacities. Her physician consulted an emergency physician, then had the nurse transport the patient by wheelchair on nasal cannula oxygen to the ED on the same campus. A bedside echocardiogram revealed a normal heart and pulmonary findings consistent with pneumonia. She was started on intravenous piperacillin-tazobactam and hospitalized. She improved in house and was discharged on the 3rd hospital day on amoxicillin-clavulanate. She completed 3 months of rivaroxaban without incident.

65-year-old F PESI: 115 (Class IV)
The patient presented to an intervention ED with sudden-onset dyspnea on exertion 1 week prior associated with cough. History was notable for metastatic breast cancer, neutropenia from chemotherapy, and prior pleural effusions. Her initial set of vital signs were normal except for mild tachycardia that resolved after intravenous fluid administration. CTPA revealed multiple bilateral PE in the segmental and subsegmental branches, as well as moderate right pleural effusion. There was no right ventricular dilatation. Though a DOAC was offered, she preferred to start enoxaparin in the ED. She was discharged home with enoxaparin and warfarin and returned the next day to medicine clinic for instruction on enoxaparin administration. She also spoke by telephone with Anticoagulation Management Service. She presented to the ED 5 days later for worsening dyspnea on exertion and several days of fever. Her laboratory results showed therapeutic international normalized ratio as well as influenza B and elevated lactate. She was hospitalized for intravenous cefepime and oseltamivir for severe sepsis. She remained hemodynamically stable throughout her hospital course. She was discharged home on the 4th hospital day on cefpodoxime, doxycycline, and oseltamivir and indefinite warfarin management without incident.
66-year-old M PESI: 76 (Class  II) The patient presented to an intervention ED with 1 week of right-sided pleuritic chest pain and cough. History was notable for hypertension and remote kidney transplant. A D-dimer was found to be elevated and a CTPA was positive for 2 subsegmental, right-sided PEs with a small, right-sided pleural effusion. He was discharged home on rivaroxaban and followed up with Anticoagulation Management Services by telephone the following day. He complained of persistent right-sided chest pain and chills at his primary care physician's office 3 days later. An outpatient chest radiograph showed a new focal consolidation and worsening right-sided pleural effusion. He was sent to the ED, where he was diagnosed with pneumonia and admitted for intravenous antibiotics. During hospitalization, the patient required no supplemental oxygen and was discharged home on the third hospital day on oral antibiotics. He has continued his rivaroxaban indefinitely without sequelae.
30-day all-cause mortality 98-year-old M PESI: 178 (Class V) The patient presented to an intervention ED with 2 days of dyspnea and right calf pain. History was notable for hypertension, melanoma, atrial fibrillation, cerebrovascular disease, protein calorie malnutrition and deep vein thrombosis. He was not taking anticoagulation medication at the time due to previous upper gastrointestinal bleeding. He did not want resuscitation. His vital signs were normal, except a chronically low systolic blood pressure. He was cachectic on examination and had no leg swelling or tenderness. D-dimer was found to be elevated and a CTPA revealed 2 subsegmental right-sided PEs. No deep vein thrombosis was identified on compression ultrasonography. In a shared decisionmaking conversation with the patient and his family, they agreed to restart anticoagulation. Follow-up was arranged with his primary care physician and Anticoagulation Management Services. He was discharged home on enoxaparin and warfarin and spoke with a pharmacist by telephone the next day for anticoagulation management. He presented to the ED 2 weeks later for failure to thrive and was subsequently discharged home with hospice care, where he died 3 days later.
58-year-old M PESI: 98 (Class III) The patient presented to a control ED with 2 weeks of progressively worsening fever, dyspnea and cough. History was notable for glioblastoma multiforme status post palliative resection and chemotherapy. He was hypoxic in the ED and was placed on supplemental oxygen. A D-dimer returned elevated. CTPA was positive for multiple scattered emboli in both lower lobes and a right lower lobe pulmonary infarct. He declined anticoagulation therapy due to increased bleeding risk and elected palliative measures instead. He was discharged home with hospice care and died 2 days later.

eAppendix 8. Thirty-Day All-Cause Mortality of Emergency Department Patients With Acute Pulmonary Embolism
Thirty-day all-cause mortality varied by site of care: it was lower among those discharged home than those hospitalized: 0.7% (n=2) vs 5.7% (n=43) (P<.001). We describe the characteristic and course of the 2 outpatients who died within 30 days in eTable 3 above.
Thirty-day all-cause mortality also varied by risk class. We report mortality stratified by PE Severity Index classification in eTable 4 below. Thirty-day mortality was similar between ED groups: 4.7% (n=26) in the intervention group and 3.9% (n=19) in the control group. Among patients with complete mortality data, the 30-day mortality was 3.8% and was lower among outpatients than their hospitalized counterparts (0.9% vs 4.8%).  16 This attests to changing practice patterns in some circles, but the results from this study cannot be compared with our own. The initial site-of-care (clinic vs ED) was not reported, and we know that clinic patients with PE are far lower risk than those diagnosed in the ED. 17,18 Also, sicker patients were excluded from their study (e.g., those treated with low molecular weight heparin, like most patients with active cancer, and those with contraindications to anticoagulation). Excluding these higher-risk patients from the study may have inflated the proportion of patients undergoing outpatient management.