Safety of outpatient treatment in acute pulmonary embolism


Marc Carrier, Department of Medicine, Ottawa Hospital General Campus, 501 Smyth Road, Box 201, Ottawa, Ontario K1H 8L6, Canada.
Tel.: +1 613 737 8899 ext. 73034; fax: +1 613 739 6266.


See also Baglin T. Fifty per cent of patients with pulmonary embolism can be treated as outpatients. This issue, pp 2404–5; Kovacs MJ, Hawel JD, Rekman JF, Lazo-Langner A. Ambulatory management of pulmonary embolism: a pragmatic evaluation. This issue, pp 2406–11.

Summary. Introduction: Data regarding outpatient treatment of pulmonary embolism (PE) is scarce. This study evaluates the safety of outpatient management of acute PE. Methods: This is a retrospective cohort study of consecutive patients presenting at the Ottawa Hospital with acute PE diagnosed between 1 January 2007 and 31 December 2008. PE was defined as an arterial filling defect on CTPA or a high probability V/Q scan. Patients were managed as outpatients if they were hemodynamically stable, did not require supplemental oxygenation and did not have contraindications to low-molecular-weight heparin therapy. Results: In this cohort of 473 patients with acute PE, 260 (55.0%) were treated as outpatients and 213 (45.0%) were admitted to the hospital. The majority of the patients were admitted because of severe comorbidities (45.5%) or hypoxia (22.1%). No outpatient died of fatal PE during the 3-month follow-up period. At the end of follow-up, the overall mortality was 5.0% (95% CI, 2.7–8.4%). The rates of recurrent venous thromboembolism (VTE) in outpatients were 0.4% (95% CI, 0.0–2.1%) and 3.8% (95% CI, 1.9–7.0%) within 14 days and 3 months, respectively. The rates of major bleeding episodes were 0% (95% CI, 0–1.4%) and 1.5% (95% CI, 0.4–3.9%) within 14 days and 3 months, respectively. Four (1.5%) outpatients were admitted to the hospital within 14 days. Conclusions: A majority of patients with acute PE can be managed as outpatients with a low risk of mortality, recurrent VTE and major bleeding episodes.


Patients with deep vein thrombosis (DVT) can be safely managed and treated as outpatients using subcutaneous low-molecular-weight heparins (LMWH) [1,2]. However, the role of outpatient therapy in patients with pulmonary embolism (PE) is unclear and has been a matter of debate in the literature.

A large majority of patients with PE are still systematically admitted to the hospital to avoid potential complications such as death, progressive right ventricular (RV) failure and major bleeding. Although the risk of fatal recurrent PE is higher in patients treated for PE than the risk of fatal PE in patients treated for DVT [3,4], the literature suggests that only 4.5% of PE-patients develop serious complications during the first 10 days of treatment [5].

Small cohort studies and subgroup analyses of randomized controlled trials have investigated the outcomes of outpatient treatment for PE. Two systematic reviews, which include observational studies of patients diagnosed with symptomatic PE who were treated completely as outpatients or had early discharge (within 3 days), concluded that outpatient therapy is safe in hemodynamically stable patients without hypoxia [6,7]. More recently, a randomized controlled trial was stopped early due to an increased mortality in the group of patients treated as outpatients [8].

Outpatient treatment in patients with PE may be important to reduce hospitalizations and medical costs and hence save scarce health care resources. In order to counsel hemodynamically stable PE patients on the risks and benefits of outpatient treatment, clinicians require estimates of the rates for overall mortality, fatal PE, recurrent venous thromboembolism (VTE) and major bleeding at 14 days and 3 months. To address these knowledge gaps and assess the safety and feasibility of outpatient treatment of PE, we performed a retrospective study of patients diagnosed with PE and treated as outpatients at the Ottawa Hospital.


Study population

A retrospective cohort study of consecutive patients with suspected acute PE that underwent computed tomographic pulmonary angiography (CTPA) or lung ventilation/perfusion (V/Q) scan was conducted at the Ottawa Hospital Thrombosis Program, Ottawa, Canada. All reports of CTPA or V/Q scans conducted from 1 January 2007 to 31 December 2008 at the Ottawa Hospital, a large tertiary care center serving a catchment area with a population of greater than 1 million, were reviewed. PE was defined as a pulmonary artery filling defect on CTPA or as a high probability V/Q scan.

Patients diagnosed with PE during hospitalization, patients with chronic PE and patients in whom anticoagulation treatment was not initiated (e.g. palliative care patients and those with small clinical non-significant PE) were excluded from the analyses. Patients were also excluded if they received treatment at another hospital or were followed-up by a health care professional out of the Ottawa Hospital. All included patients were followed for 3 months.

At our institution, the following patients are managed, as per hospital protocol, as outpatients: patients who do not have a systolic blood pressure of 100 mmHg or less, who have an oxygen saturation of 92% or more and do not require supplemental oxygen, who do not have contraindications for the use of LMWH (such as a high bleeding risk or renal failure) and who do not need hospitalization for other comorbidities (such as severe cancer, heart failure, chronic obstructive pulmonary disease, coronary artery disease, cerebrovascular events, etc.). Outpatients with PE are treated with daily injections of LMWH for a minimum of 5 days with concomitant oral vitamin K antagonist (VKA). LWMH is discontinued if the International Normalized Ratio (INR) is above 2.0 on two consecutive readings. Outpatients are discharged home from the emergency department and are followed-up in the Outpatient Thrombosis Assessment and Treatment Unit within 24–48 h of diagnosis, then again at 7 days and 3 months.


The following variables were specified prior to data collection: (i) patient demographics; (ii) reasons for hospitalization; (iii) recurrent VTE and bleeding episodes; and (iv) death (date and cause). Data were extracted from hospital discharge reports and consultation notes from the Thrombosis Assessment and Treatment Unit. All outcomes were reviewed independently by two investigators (E. Gandara, P. M. G. Erkens). Disagreements on information were resolved by consensus or retrieving further information from other medical records.

The primary outcomes of this study were overall mortality and fatal recurrent PE. The cause of death was identified by a review of hospital records including death summaries. Secondary outcomes included recurrent VTE, major bleeding and VTE-related hospitalization at 14 days and 3 months of follow-up. Recurrent PE was defined as a new arterial filling defect on CTPA or a new mismatched area on a high probability V/Q-scan, and recurrent DVT was defined as a new non-compressible venous segment on an ultrasound of the extremities. Major bleeding events were defined as: (i) fatal bleeding; (ii) symptomatic bleeding in a critical area or organ such as intracranial, intraspinal, intraocular, retroperitoneal, intra-articular or pericardial, or intramuscular with compartment syndrome; and/or (iii) bleeding causing a fall in hemoglobin level of 20 g L−1 (1.24 mm) or more, or leading to transfusion of two or more units of whole blood or red cells [9].

Statistical analysis

Descriptive statistics were used to describe the patient characteristics and outcomes. Ninety-five per cent confidence intervals (95% CI) and P-values were calculated for each event rate by using Fisher’s exact test. A subgroup analysis was performed for patients without active cancer. The statistical analyses were performed using the Statistical Package for the Social Sciences software (version 16; SPSS, Chicago, IL, USA).


Over a 2-year period, 4410 CTPA and 740 V/Q scans were carried out for suspected PE at the Ottawa Hospital. A total of 724 patients had a positive CTPA for PE and 115 patients had a high probability V/Q scan. A flow diagram including the reasons for a patient’s exclusion is depicted in Fig. 1. A total of 473 patients presenting at the Ottawa Hospital with confirmed PE were included in the study. Two hundred and sixty (55.0%; 95% CI, 50.4–59.5) patients were managed entirely as outpatients and 213 (45.0%; 95% CI, 40.5–49.6) patients were admitted to the hospital.

Figure 1.

 Flow diagram of the study.

Baseline characteristics of patients treated as outpatients and of those admitted to the hospital (inpatients) are shown in Table 1. Admitted patients were more likely to have had a previous diagnosis of heart failure whereas outpatients were more likely to have presented with chest pain. All outpatients were initially treated with LMWH. Most (70.4%) of the inpatients were also treated with LMWH while 25.8% received UFH (Table 1).

Table 1.   Baseline characteristics of the cohort
 Outpatient treatment (n = 260)Inpatient treatment (n = 213) 
  1. VTE, venous thromboembolism; COPD, chronic obstructive pulmonary disease; CAD, coronary artery disease; DVT, deep vein thrombosis; SBP, systolic blood pressure; LMWH, low molecular weight heparin; UFH, unfractionated heparin. *Saturation assessed with and without the administration of supplemental oxygen. Other initial treatment: five patients received an IVC filter, two patients received thrombolysis, one patient was treated with argatroban because of heparin induced thrombocytopenia.

 Female, n (%)128 (49.2%)106 (49.8%) 
 Age, mean years (SD) 54.7 (17.1) 62.0 (17.5) 
 n (%; 95% CI)n (%; 95% CI)P-value
Risk factors for VTE
 Immobilization for at least 3 days40 (15.4%; 11.2–20.4)35 (16.4%; 11.7–22.1)0.801
 Surgery in the previous 4 weeks41 (15.8%; 11.6–20.6)28 (13.1%; 8.9–18.4)0.436
 Previous objectively diagnosed VTE53 (20.4%; 15.9– 25.8)27 (12.7%; 8.5–17.9)0.027
 Active cancer83 (31.9%; 26.3–38.0)86 (40.4%; 33.7–47.3)0.067
 Heart failure4 (1.5%; 0.4–3.9)15 (7.0%; 4.0–11.4)0.004
 COPD9 (3.5%; 1.6–6.5)18 (8.5%; 5.1–13.0)0.027
 CAD22 (8.5%; 5.4–12.5)27 (12.0%; 8.5–17.9)0.172
 Stroke8 (3.1%; 1.3–6.0)14 (6.6%; 3.6–10.8)0.082
Clinical presentation
 Dyspnea191 (73.5%; 67.7–78.7)176 (82.6%; 76.9–87.5)0.020
 Chest pain167 (64.2%; 58.1–70.1)95 (44.6%; 37.8–51.6)0.000
 Hemoptysis11 (4.2%; 2.1–7.4)16 (7.5%; 4.4–11.9)0.163
 Syncope2 (0.8%; 0.1–2.8)13 (6.1%; 3.3–10.2)0.001
 Clinical symptoms of DVT82 (31.5%; 25.9–37.6)48 (22.5%; 17.1–28.8)0.030
 Heart rate > 100 bpm33 (12.7%; 8.9–17.4)61 (28.6%; 22.7–35.2)0.000
 SBP < 100 mmHg0 (0%; 0–1.4)15 (7.0%; 4.0–11.5)0.000
 Arterial oxygen saturation < 90%*3 (1.2%; 0.2–3.3)45 (21.1%; 15.9–27.3)0.000
Initial treatment
 LWMH260 (100%; 98.6–100.0)150 (70.4%; 63.8–76.4)0.000
 UFH0 (0%; 0.0–1.4)55 (25.8%; 20.1–32.3)0.000
 Other initial treatment0 (0%; 0.0–1.4)8 (3.8%; 1.6–7.3)0.002

For the patients admitted to the hospital, the median length of admission was 6 days (range 1–209 days). Almost 75% of the inpatients were admitted for 10 days or less. The majority of these patients were admitted because of severe comorbidities (45.5%), hypoxia (22.1%), extensive PE (8.5%), hypotension (7.0%), renal dysfunction (6.6%), pain management (4.7%) or syncope (2.3%). The reason for admission was unclear in seven (3.3%) patients.

Overall mortality and fatal PE

Only one outpatient (0.4%; 95% CI, 0.0–2.1%) died within the first few weeks after the index PE (Table 2). This patient was initially sent home from the emergency room to be treated as an outpatient but was admitted within 24 h of discharge. The patient was transferred to a palliative care facility after 10 days of hospitalization and died from progression of her metastatic cancer. After 3 months of follow-up, the overall mortality in the outpatient group was 5.0% (95% CI, 2.7–8.4%). No patient died of recurrent fatal PE (Table 2).

Table 2.   Main outcomes
 Outpatient treatment (n = 260)
n (%; 95% CI)
Inpatient treatment (n = 213)
n (%; 95% CI)
  1. PE, pulmonary embolism; VTE, venous thromboembolism. *This outpatient received palliative care 10 days after PE. Five of these inpatients received palliative care at 14 days follow-up. Major hemorrhage defined according to the Control of Anticoagulation Subcommitte (ISTH SSC 2004): (i) fatal bleeding, and/or (ii) symptomatic bleeding in a critical area or organ, such as intracranial, intraspinal, intraocular, retroperitoneal, intra-articular or pericardial, or intramuscular with compartment syndrome, and/or (iii) bleeding causing a fall in hemoglobin level of 20 g L−1 (1.24 mm) or more, or leading to transfusion of two or more units of whole blood or red cells. §Three of these outpatients received palliative care at 3 months follow-up. No patients with palliative care. **Nine patients with one readmission and two patients with two readmissions.

14 days follow-up
 Overall mortality1 (0.4%; 0.0–2.1)*27 (12.7%; 8.5–17.9)0.000
 PE-specific mortality0 (0%; 0–1.4)5 (2.3%; 0.8–5.4)0.018
 Recurrence VTE1 (0.4%; 0.0–2.1)4 (1.9%; 0.5–4.7)0.180
 Major hemorrhages0 (0%; 0.0–1.4)13 (6.1%; 3.3–10.2)0.000
 Readmissions VTE/treatment related4 (1.5%; 0.4–3.9)4 (1.9%; 0.5–4.7)1.000
3 months follow-up
 Overall mortality13 (5%; 2.7–8.4)§57 (26.7%; 20.9–33.2)0.000
 PE-specific mortality0 (0%; 0–1.4)5 (2.3%; 0.8–5.4)0.018
 Recurrence VTE10 (3.8%; 1.9–7.0)10 (4.7%; 2.3–8.5)0.654
 Major hemorrhages4 (1.5%; 0.4–3.9)17 (8.0%; 4.7–12.5)0.001
 Readmissions VTE/treatment related6 (2.3%; 0.9–5.0)11 (5.2%; 2.6–9.1)**0.135

In the inpatient group, five (2.3%; 95% CI, 0.8–5.4) patients died because of fatal PE. All of these patients died within 7 days of the index PE diagnosis. The overall mortality was 12.7% (95% CI, 8.5–17.9) within 14 days and 26.7% (95% CI, 20.9–33.2) within 3 months.

The majority (76.4%) of patients died from cancer. In a subgroup analysis excluding patients with active cancer, the 3-month overall mortality in the outpatient and inpatient group was 0% (95% CI, 0–2.1) and 9.4% (95% CI, 5.0–15.9), respectively.

Recurrent VTE and major bleeding episodes

One (0.4%; 95% CI, 0.0–2.1) outpatient had recurrent PE within 14 days. At 3 months follow-up 10 outpatients (3.8%; 95% CI, 1.9–7.0) developed recurrent VTE (Table 2). Four patients had DVT and six patients had recurrent PE. The incidence of recurrent VTE in hospitalized patients was similar at 14 days and 3 months (= 0.180 and 0.654, respectively).

None of the outpatients suffered from major bleeding during the first 2 weeks. Four (1.5%; 95% CI, 0.4–3.9) outpatients, however, had a major bleed during the 3 months follow-up (Table 2). The percentage of major bleeding episodes was higher in the inpatient group. Thirteen (6.1%; 95% CI, 3.3–10.2) inpatients suffered from major bleeding episodes within 14 days after the index PE, and at 3 months follow-up 17 (8.0%; 95% CI, 4.7–12.5) inpatients had suffered a major bleeding episode (Table 2).

Hospitalization during follow-up

Four (1.5%; 95% CI, 0.4–3.9) outpatients were admitted to the hospital within 14 days of their index PE diagnosis. Outpatients were admitted because of progressive shortness of breath (n = 2), presyncopal episode (n = 1) or heparin-induced thrombocytopenia (n = 1). During the 3 months follow-up two more patients were admitted (one recurrent VTE; one major bleeding episode). Therefore, the VTE-related admission rate of outpatients at 3 months follow-up was 2.3% (95% CI, 0.9–5.0) (Table 2). No differences were seen in the readmission rates between patients originally admitted and outpatients at the end of follow-up (= 0.135).


Our study suggests that outpatient PE treatment is feasible and safe in a majority of patients. To our knowledge, this is one of the largest cohort studies to evaluate the safety of outpatient treatment in low-risk patients with acute PE. More than 50% of the patients were treated as outpatients with a low subsequent risk of death, recurrent VTE and major bleeding episodes in the first 14 days. None of the outpatients died from PE and there is only one (0.4%) death from metastatic cancer within the first few weeks.

Our results are consistent with previously published smaller studies assessing outpatient management or an early discharge strategy (< 3 days of hospitalization) in patients with hemodynamically stable acute PE [8,10–20]. No PE-related deaths and only one death due to major bleeding were reported in these studies. The overall short-term mortality in our study was 0.4% (95% CI, 0–2.1%). Although this is consistent with previously reported cohort studies [10,12,13,15–18], a recent randomized controlled trial reported an overall mortality of 2.8% within the first 10 days after the index PE. This study was stopped early due to this increased mortality in the outpatient group [8]. The reasons behind this discrepancy are unclear but might be explained by the difference in follow-up management between studies (in-person vs. telephone follow-up). Short-term recurrent VTE in the literature ranged from 0% to 3.2% and major bleeding from 0% to 2.3% [8,10,12,13,15–18]. Similarly, our results are consistent with prior studies assessing long-term outcomes in patients with PE treated as outpatients. The 3 months overall mortality in these studies ranged from 0% to 43.5%. The percentage of recurrent VTE and major bleeding ranged from 0% to 9.3% and 0% to 3.7%, respectively [8,10–20].

In our cohort study, the decision regarding whether patients with PE could be managed as outpatients or required hospitalization was based on the clinical judgment of the doctor at the emergency department. All hemodynamically stable patients that did not require supplemental oxygenation and had no contraindications to LMWH or significant comorbidities were considered for outpatient management. Our rate of serious adverse events is higher in patients who were admitted to the hospital immediately after their index PE (Table 2), suggesting that these simple criteria are able to discriminate between low and high-risk patients.

Several studies investigated risk stratification in PE and developed risk-prediction models to identify patients with acute PE who are at low risk for adverse outcomes and could be treated as outpatients [21]. The most extensively validated models are the Geneva Prognostic Score (GPS) and the Pulmonary Embolism Severity Index (PESI) [21–23]. Other studies have shown that patients with acute PE who have elevated cardiac enzymes (such as troponin, brain natriuretic peptide (BNP) and creatinine kinase), or evidence of right-heart dysfunction on either echocardiography or CTPA, have a worse short-term survival rate than those without these findings [21,23–26]. There are also studies pointing to the role of plasmin degradation products, such as D-Dimer concentration, in the risk stratification of patients with PE [10,27]. More recently, a prospective management study showed that out of hospital treatment is safe in hemodynamically stable PE-patients with low NT-proBNP level (< 500 pg mL−1) [10]. Future prospective management studies are required to assess if the use of biomarkers can improve stratification of patients with PE and optimize the number of patients with PE that can be safely treated as outpatients.

In a recent publication, Aujeskey et al. [27] showed that emergency physicians in the United States are reluctant to discharge patients with PE for outpatient treatment due to insufficient data supporting the effectiveness and safety of the outpatient management of PE. However, safe outpatient management of patients with PE may lead to: (i) a decrease in unnecessary hospitalizations; (ii) a decrease in hospital-acquired infections and death; (iii) improvement in health-related quality of life; and (iv) reduction in health care costs [27–30]. Our results suggest that outpatient treatment is feasible and safe in selected patients with PE.

It is important to note the limitations of our study. First, this is a retrospective cohort study and as such, our findings may be subject to bias, incomplete information or misdiagnosis. We tried to minimize selection bias by including all consecutive patients with confirmed PE on CTPA or V/Q scan over a defined time period and by adjudicating all outcome events (two independent reviewers). Complete information about follow-up was only missing in 5.1% of the sample. We performed a worst-case-scenario analysis for all patients who were lost to follow-up. Assuming that all of these patients died at their end of follow-up, the overall mortality for the outpatients would have been 8 (2.9%) within the first 14 days and 25 (9.1%) at 3 months follow-up. Second, as previously stated, the decision to admit or discharge the patient was left to the discretion of the treating physician. Hence, it is possible that some admitted patients could have been safely managed as outpatients. Nonetheless, our results are conservative as they suggest already that a majority of the patients can be treated as outpatients with low risk of adverse events. Future prospective management studies are required to validate this approach. Finally, all patients with PE treated as outpatients were seen within 24–48 h in our thrombosis clinic and followed up after 7 days. Close follow-up of patients managed as outpatients in specialized thrombosis clinics might not be possible in all settings and therefore rates of major bleeding and recurrent VTE may differ if patients are followed-up in another health setting.

In conclusion, a majority of patients with acute PE at the Ottawa Hospital were treated as outpatients with a low risk of mortality, recurrent VTE and major bleeding episodes. Out of hospital treatment for PE should be considered as a feasible and safe treatment in uncomplicated patients and can contribute to a substantial reduction in health care costs.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.