A massive multi-center study reveals that the new 2026 AHA/ACC pulmonary embolism framework adds clinical complexity without improving risk prediction for the vast majority of patients.

Why do medical consensus panels keep adding clinical subcategories if they do not help doctors make better decisions? The 2026 AHA/ACC pulmonary embolism framework replaced a simple four-tier system with five categories and multiple subcategories. The goal was precision. Instead, a new analysis shows this added complexity fails where it is needed most: the messy middle-risk patient.

This finding challenges the assumption that more granular guidelines equal better care. For high-risk patients, the system works well. But for the intermediate-risk patients who make up the bulk of clinical dilemmas, the new subcategories create noise rather than clarity.

Researchers analyzed 17,223 patients with confirmed pulmonary embolism across three major health systems. This cohort included 12,992 patients from Duke, 3,870 from Stanford, and 361 from Beth Israel Deaconess. They mapped these patients to the new categories (B through E) and tracked 30-day mortality.

The broad categories did show a clear trend. Pooled 30-day mortality rose from 1.5% in category B, to 8.9% in category C, 15.5% in category D, and 31.9% in category E. But the precision broke down entirely in the subcategories.

The Messy Middle

For intermediate patients in subcategories C1 through D1, mortality rates did not follow a logical order. The risk bounced from 9.2%, to 10.8%, to 8.1%, and then 10.9%. Adding subcategories to category C failed to improve the predictive power at Duke, holding flat at a C-index of 0.699.

This lack of predictive power in mid-risk cohorts is a recurring problem in cardiovascular medicine. Similar struggles to find clear prognostic signals in intermediate cohorts have been documented in other risk models, such as the simplified PESI score.

Furthermore, 12.7% of category C patients lacked both echocardiography and biomarker testing. Yet this untested group had a mortality rate of 10.4%, which actually exceeded their fully tested peers. This suggests that in real-world clinics, missing data is often a marker of unmeasured risk.

What to Rethink

When compared to the older 2019 European Society of Cardiology (ESC) model, the new framework offered little practical advantage. It reclassified 5.7% of ESC intermediate-risk patients into the higher category D. While these reclassified patients had a slightly higher mortality rate than those left in category C (10.8% versus 8.9%), the difference was not statistically significant.

This raises a critical question about the utility of consensus-driven guidelines. When expert panels design complex systems without real-world testing, they risk creating administrative burdens that do not translate to better triage. Clinicians need simple, actionable tools, like those used to manage acute pulmonary embolism risk-stratification, rather than hyper-segmented categories that fail to differentiate risk in the clinic.

Key Findings

• Overall mortality rose predictably across main categories, from 1.5% in B to 31.9% in E.
• Subcategory risk stratification failed in the middle, with mortality fluctuating non-monotonically between 8.1% and 10.9%.
• Untested category C patients faced a 10.4% mortality rate, highlighting a major gap in real-world diagnostic compliance.

This study is limited by its retrospective design and its reliance on electronic health record data, which may contain missing clinical variables.

Read the full preprint on medRxiv.