Derivation and Validation of a Novel Prediction Model to Identify Low-Risk Patients With Acute Pulmonary Embolism
Subramanian, M. et al. The American Journal of Cardiology 2017. [paper]
Why I chose this study
The topic of discharging patients with pulmonary embolism (PE) came up in conversation recently. I started asking fellow residents if they have ever discharged a low-risk PE from the ED or if they have ever heard of the PESI prognostic model. I only found one person who had heard of PESI and had used it. Even so, after they determined the patient was low-risk they admitted the patient anyway.
This is consistent with my experience in the ED, which is, we have gotten comfortable discharging patients with deep vein thrombosis (DVT) on a direct oral anticoagulant (DOAC), but it isn’t the same for PE. On the medicine side I think most of us have seen this too, hemodynamically stable patients who are admitted for a day or two, then started on either a DOAC or Lovenox® and sent home. Maybe that admission didn’t need to happen and the anticoagulant could have been started from the ED.
The following is another scoring system for establishing low-risk (low mortality/complication) groups that could be potentially discharged from the ED. Spoiler Alert: I don’t think this ‘not yet externally validated’ prediction model offers anything new (we have PESI and Hestia already) but I do think it reinforces the idea that there is a group of patients with PE who do not need hospital admission.
To derive and validate a prediction model to effectively identify patients with PE at low-risk of short-term mortality, right ventricular dysfunction (RVD), and other nonfatal outcomes.
Single center retrospective data analysis at a tertiary care hospital in Chennai, India that included all patients with PE diagnosed between January 2011 – March 2015. 400 patients were included, 300 were used to extrapolate a risk model and 100 were used to internally validate the model. After this they tested the model on data from 82 patients at “another tertiary care hospital”.
Age > 18 years and acute PE confirmed by CT.
All-cause 30-day mortality after diagnosis of acute PE.
- In-patient mortality
- RVD by echocardiography (within 24 hours of presentation)
- Nonfatal cardiogenic shock
- Nonfatal cardiorespiratory arrest
Variables that were readily accessible by an emergency physician and previously shown to be associated with short-term mortality in acute PE were selected. Candidate models were analyzed in an attempt to find models that identified a low-risk group with a membership of at least 20% of the total derivation sample and a 30-day mortality of <1%.
- Univariate analysis to select predictor variables for the multivariate model, using a cutoff of p <0.20.
- Continuous variables that were statistically significant were then categorized, choosing the most discriminative cut-off points.
- Variables that were statistically significant were included in a logistic regression model.
- A regression coefficient for each significant variable in the final model was calculated.
The criteria: Systolic blood pressure, diastolic blood pressure, heart rate, PaO2, and modified electrocardiographic score (EKG changes converted to a point score: tachycardia (2), incomplete RBBB (1), RBBB (3), T wave inversions in v1 – v3 (4), or S1Q3T3 (4)). If you are wondering how they got HOPPE as the name it stands for the components of the score: HR, PaO2, SBP, DBP, and ECG. Each variable was divided into three categories and assigned point values of 1-3 points.
They then took the model they came up with and compared it against the Pulmonary Artery Obstruction Index as well as the PESI score.
Based on this model they were able to roughly divide the groups by thirds: 1/3 low-risk, 1/3 intermediate-risk, and 1/3 high-risk with strong correlation of primary outcome with group assignment based on prediction score. The low-risk group had <1% 30-day mortality in the derivation and two validation groups. With most of the secondary outcomes there was low morbidity in the low-risk group with <1% in-hospital mortality, cardiogenic shock and cardiopulmonary arrest. However, the secondary outcome of RVD, while lower incidence, was still seen in 21% of low-risk patients.
The low-risk group had a high sensitivity (96-99%) and high negative predictive value (95-96%) for 30-day mortality. The authors also claim the HOPPE score has a higher accuracy and better discriminatory power than the previously validated PESI score (sensitivity 86.2% vs 98.7%, p = 0.036, specificity 37.2% vs 48.3%, p = 0.043) and discriminative power (AUC 0.74 vs 0.85, p = 0.033).
As I mentioned at the beginning, I think this article is useful in terms of promoting discussion of which patients we can send home with a PE. But, I don’t think that it provides me with a new tool (yet) due to some noticeable limitations.
- It was a small, single center trial in India that has not been validated (can I generalize these data, probably not).
- There was a significant presence of RVD in their low-risk group and I don’t think it was adequately addressed.
- There was no breakdown of treatments received or complications of treatment.
The authors acknowledge the high rate of RVD in their discussion and attribute it to a lack of uniform criteria for the diagnosis. They suggest further study of the short-term risk in the subset of patients who are low-risk but have RVD by echo.
There is also a question of when in the course of hospitalization the echocardiogram was performed and does the timing affect the results of the echo. The authors suggest that delayed echo could have missed acute RVD that resolved. To me this doesn’t seem to be as important as delayed RVD developing over days to months. One long term complication of PE that was not addressed in this study, but used as an reason to given tPA for ‘submassive’ PE, is the risk of future pulmonary hypertension and right heart failure.
At this point I think the PESI is a better or at least a better validated tool, although perhaps future validation of this study will show the HOPPE score to be as good as the authors claim. Even with the handful of relatively new scores for identifying low-risk PE there are still institutional and socio-economic barriers to discharging patients from the ED. I’m interested to hear about your practice with regards to low-risk PE patients, do you use a scoring system to aid your clinical judgement and do you send them home?
Dr. Emily Wheelis is a EM/IM resident, class of 2021.
EM/IM Sessions are reviewed in journal club style by the current attendings and residents, as well as alumni of the UIC IM/EM program prior to publication. This post was specifically reviewed by Sukhi Bains, MD, IM/EM Assistant Professor. Elspeth Pearce, MD Editor.