September 25th, 2013

Is It Safe to Treat Acute Pulmonary Embolism in an Outpatient Setting?

CardioExchange’s John Ryan interviews Christopher Kabrhel about his research group’s study of the factors associated with clinical deterioration and the need for hospital-based interventions after an acute pulmonary embolism (PE). The study was presented in June at the 24th congress of the International Society on Thrombosis and Haemostasis.


From October 2008 through December 2011, researchers prospectively enrolled 298 consecutive patients with radiographically confirmed PE at the emergency department (ED) of a U.S. tertiary care hospital. One third of patients experienced serious clinical deterioration or required a major hospital-based intervention (the primary outcome) within the first 5 days after PE diagnosis. The median length of hospital stay was 3 days. The mortality rate was 1% within 5 days and 4% within 30 days.

In multivariable analyses, factors significantly associated with the primary outcome were abnormal vital signs, coronary artery disease, cerebrovascular disease, deep-vein thrombosis (DVT), and right-heart strain on echocardiogram; factors significantly associated with 30-day all-cause mortality were malignancy and chronic lung disease.


Ryan: How do these findings change your practice?

Kabrhel: In the United States, nearly all patients with PE are hospitalized for anticoagulation and monitoring. We know that anticoagulation can be initiated and coordinated without hospitalization. This is the current standard for DVT, and novel anticoagulants make this even easier. The main justification for admission is monitoring.

In our study, 67% of patients diagnosed with PE in the ED suffered no clinical deterioration and required no hospital-based intervention during hospitalization. We defined clinical deterioration very broadly to capture a wide spectrum of events that might justify hospitalization. For example, simply requiring more than 2 liters of nasal cannula oxygen was considered deterioration. Similarly, hypotension was considered deterioration even if it did not require treatment. Given that the majority of patients with PE did not clinically deteriorate and did not require any hospital-based intervention, hospitalizing patients with low-risk PE for monitoring seems unnecessary.

Our data support the current trend toward treating low-risk patients with PE as outpatients and show that the proportion who might be appropriate for outpatient treatment is quite large. However, we also found that existing clinical prediction instruments were only moderately sensitive for identifying patients who clinically deteriorate. Clinicians should exercise caution in using these rules as the basis for disposition decisions.

Ryan: What do you advocate including in the PE guidelines — or what further research needs to be done first?

Kabrhel: Outpatient treatment of PE is reasonable and may be appropriate for a large number of patients. However, further research is needed before we establish clinical guidelines. Clinical decision instruments should be designed and validated using outcomes that reflect the decision they are designed to support. Logically, a decision rule designed to help physicians decide which patients with PE benefit from hospitalization (or conversely, which patients can be treated as outpatients) should be studied relative to events that occur during a typical hospitalization for PE. We took that approach in our study.

Ryan: As you say, we need better rules to decide which patients with PE require hospitalization. What’s the best way to currently identify who will do well?

Kabrhel: We found that certain clinical factors (abnormal vital signs, right-heart strain on echocardiogram, residual DVT, and a history of cardiovascular disease) were associated with clinical deterioration in the 5 days after an acute PE. These factors can help clinicians decide which patients require hospitalization. In settings where an echocardiogram is difficult to obtain, biomarkers (e.g., troponin, brain-type natriuretic peptide) or ECG changes may be useful surrogates. Of course, clinicians should also consider other factors, such as the social circumstances that affect a patient’s ability to adhere to outpatient therapy.


Does this study influence your judgment about whether a substantial percentage of patients with acute PE can be treated safely in an outpatient setting?

4 Responses to “Is It Safe to Treat Acute Pulmonary Embolism in an Outpatient Setting?”

  1. Esteban Gandara, MD says:

    Very interesting study, in our practice we treat close to 60% of the PEs diagnosed at the ER completely as outpatients (J Thromb Haemost. 2010 Nov;8(11):2412-7). A question for the author would be, did you use a pre-specified cut-off for troponins and pro-Bnp or they were included in the model as continuous variables? Did you analyze the results excluding massive PE?

  2. Judith Andersen, AB, MD says:

    No. Dr Kabrhel’s caveats are wise. Patients with PE, however “trivial, need short term inpatient care. Any PE that brings a patient to the ED and results in a hospital evaluation should be a reason for a short term admission. It’s a cinch that all the identified risk factors for deterioration should be evaluated and often can not all be obtained withing 23 hours. And stable patients can be sent home within 24-48 hours. I appreciate Dr Kabrhel’s judicious assessment, but a 23 hr success rate is not necessarily predictive of success and the availabiity of excellent oral treatment strategies does not abrogate the necessity for short-term clinical judgment -particularly for elderly patients and those with unclear support strategies or uncertain outpatient prescription support.

  3. Christopher Kabrhel, MD MPH says:

    Thank you for your comments.

    Dr. Gandara, I have heard similar comments from my other Canadian colleagues. The practice patterns in Canada, the US and Europe do vary widely – and are likley to continue to evolve. With regards to your questions about troponin and NT-proBNP, we did use specific cutoffs. For troponin, we used 0.1, which is typical in this literature. For NT-proBNP, the standard cutoff is not well established. We used 200, which had the most discriminatory power in our analysis. We did perform a sub-analysis excluding any patient with a shock index (SBP/HR)<1, and results were similar.

  4. Esteban Gandara, MD says:

    Thank you for the answer. Talking about low risk patients, how are you treating non cancer patients with Subsegmental PE and no DVT?