Background
Pulmonary embolism (PE) is a potentially life-threatening condition caused by a blood clot traveling to the arteries of the lungs. In the emergency department, clinicians must quickly estimate which patients are at higher risk of deterioration, need intensive therapy, or are likely to die within days to weeks. Accurate risk stratification helps guide decisions about hospital admission, monitoring level, thrombolysis, catheter-based intervention, or safe discharge.
Several tools are commonly used for this purpose. The simplified Pulmonary Embolism Severity Index (sPESI) is widely known and easy to apply. The European Society of Cardiology (ESC) classification combines clinical findings, biomarkers, and imaging features to estimate risk. The Composite Pulmonary Embolism Shock (CPES) score is a newer tool designed to identify patients with hemodynamic strain or shock-related features by combining several bedside and diagnostic variables.
This study compared how well CPES performs against sPESI and ESC classifications for predicting short-term outcomes in emergency department patients with PE.
Study Design and Methods
This was a retrospective analysis using prospectively collected registry data from six academic emergency departments between 2016 and 2020. The investigators reviewed patients diagnosed with pulmonary embolism and assessed risk scores in a standardized way.
The CPES score was calculated only when at least 5 of its 6 components were available, ensuring the score was based on enough information to be meaningful. The researchers then compared three main tools: CPES, sPESI, and ESC risk categories.
The primary outcome was in-hospital death or clinical deterioration. Clinical deterioration generally refers to worsening that requires escalation of care, such as vasopressors, thrombolysis, ICU admission, or urgent intervention. Secondary outcomes included advanced intervention and 30-day mortality.
To measure association with outcomes, the study used logistic regression. To measure classification performance, it examined the area under the receiver operating characteristic curve (AUC), which reflects how well a tool separates patients with and without an outcome. It also calculated likelihood ratios, which show how much a test result changes the probability of an outcome.
Key Findings
A total of 1,731 patients were included. The primary outcome occurred in 193 patients (11.1%). Advanced intervention was required in 123 patients (7.1%), and 30-day mortality occurred in 124 patients (7.2%).
For the primary outcome, a CPES score of 3 or higher had a sensitivity of 72% and specificity of 57%, with an AUC of 0.68. This means the score had moderate ability to identify patients who would worsen in the hospital, but it was not clearly superior to sPESI or ESC classification. In fact, its likelihood ratios were similar to those of the other tools.
For advanced intervention, CPES performed better. Its AUC was 0.78, indicating stronger discrimination than for the primary outcome. The positive likelihood ratio was 2.07 and the negative likelihood ratio was 0.20, suggesting CPES was reasonably useful for identifying patients who were more likely to need advanced treatment and for ruling out that need when the score was low.
For 30-day mortality, the three tools performed similarly. None clearly outperformed the others, and their AUCs and likelihood ratios were broadly comparable.
An important additional finding was that central thrombus was independently associated with the need for advanced intervention, with an odds ratio of 3.46. However, central thrombus was not independently associated with the primary outcome of in-hospital death or clinical deterioration. Concomitant deep venous thrombosis was not associated with either outcome.
What the Results Mean in Practice
These findings suggest that CPES is a useful emergency department risk tool, but not necessarily better than the established options for all outcomes. For short-term death or general deterioration, CPES, sPESI, and ESC appear to offer similar clinical value. That is important because sPESI is simple and fast, while ESC classification is more comprehensive but may require more data.
Where CPES may stand out is in identifying patients who will go on to need advanced intervention. This could help emergency clinicians recognize higher-risk patients earlier and consider closer monitoring, specialist consultation, or transfer to a higher level of care.
The association between central thrombus and advanced intervention also makes clinical sense. A central clot burden often reflects more extensive pulmonary vascular obstruction and may be more likely to cause hemodynamic compromise or prompt invasive therapy. Still, clot location alone should not replace a broader risk assessment.
Clinical Context
Risk stratification in PE is not just about predicting death. Many patients survive but still experience serious complications such as hypotension, worsening oxygenation, right ventricular strain, or escalating treatment needs. That is why tools that predict intervention or deterioration can be especially valuable in emergency medicine.
In current practice, clinicians typically combine several elements when deciding management:
Patient symptoms and vital signs, such as blood pressure, oxygen saturation, and heart rate
Evidence of right ventricular dysfunction on imaging or echocardiography
Biomarkers such as troponin or brain natriuretic peptide
Overall comorbidity burden and functional status
Clot burden and location on imaging
The CPES score appears to add another structured way to summarize some of these risks, especially when the goal is to identify patients who may need more aggressive treatment.
Strengths and Limitations
A major strength of this study is that it used a multicenter registry from academic emergency departments, which improves real-world relevance. The sample size was also substantial, and the investigators evaluated several important short-term outcomes rather than focusing only on mortality.
However, some limitations should be kept in mind. The study was retrospective, so it can show association but not prove causation. The CPES score was only calculated when most components were available, which may limit generalizability in settings with incomplete documentation. In addition, treatment decisions were made in real clinical practice and may have influenced outcomes, especially advanced intervention, which is partly dependent on clinician judgment and local resources.
Another important point is that risk tools should support, not replace, bedside assessment. A patient with apparently low risk may still deteriorate if they have borderline blood pressure, poor oxygenation, or limited physiologic reserve.
Take-Home Message
In this multicenter emergency department study, the Composite Pulmonary Embolism Shock score performed similarly to sPESI and ESC classification for predicting short-term adverse outcomes overall. Its main advantage was better separation for predicting advanced intervention use.
For emergency clinicians, the practical message is that CPES may be especially helpful when deciding which PE patients need closer observation or early escalation planning. Still, the classic tools remain valuable, and the best approach is to integrate any score with clinical judgment, imaging findings, and laboratory data.
Conclusion
Among patients with pulmonary embolism evaluated in the emergency department, CPES, sPESI, and ESC risk classifications showed similar performance for short-term in-hospital adverse outcomes and 30-day mortality. CPES demonstrated stronger ability to distinguish patients who would require advanced intervention. Central thrombus was linked to advanced intervention but not to overall in-hospital deterioration, while concomitant deep vein thrombosis was not associated with outcomes.
Overall, CPES appears to be a promising adjunctive risk stratification tool, particularly for anticipating escalation of care in acute PE.
