What the trial found
Cancer patients are among the most likely to be sent for a CT scan when pulmonary embolism (PE) is suspected — they have a higher baseline clot risk, and their D-dimer levels are often elevated for reasons unrelated to PE, which pushes clinicians toward imaging. The result is a lot of CT pulmonary angiography (CTPA) that comes back negative.
The Hydra trial, published in JAMA on July 12, 2026, tested whether a structured decision rule could safely cut that volume. Investigators led by Bram Akerboom, MD, of Leiden University Medical Center randomized 698 patients with active cancer and suspected acute PE across 21 hospitals in the Netherlands, Italy, Switzerland, Belgium, France, and Spain. Roughly half were managed with the YEARS algorithm; the other half went straight to CTPA.
The headline result: the YEARS strategy avoided a CT scan in 22% of patients — 77 of the 352 assigned to the algorithm — while symptomatic venous thromboembolism or PE-related death during follow-up was 1.8% in the YEARS group versus 5.5% with CTPA-only, meeting the trial's prespecified noninferiority threshold. In the authors' words, "the YEARS algorithm is as safe as a CTPA-only approach to exclude PE in patients with active cancer and avoids CTPA scans."
How the algorithm decides who skips the scan
The YEARS algorithm is a clinical decision rule, not an imaging product. It combines three simple items with a D-dimer blood test:
Clinical signs of deep vein thrombosis
Objective signs of a clot in the leg — swelling, tenderness along the deep veins.
Hemoptysis
Coughing up blood, a classic symptom associated with pulmonary embolism.
PE as the most likely diagnosis
The clinician's judgment that PE is the leading explanation for the presentation.
The D-dimer threshold flexes with those items. If none are present, PE can be ruled out without imaging when D-dimer is below 1000 ng/mL; if one or more are present, the cutoff tightens to 500 ng/mL. Everyone else still gets the scan. The point isn't to scan fewer high-risk patients — it's to stop scanning the low-risk ones who were never going to have a positive result.
What "skip the scan" actually buys
For the roughly one in five patients who avoid CTPA, the benefits stack up: no radiation dose, no iodinated contrast (a real consideration in cancer patients, many of whom have compromised kidney function), a shorter emergency-department stay, and lower cost. As the authors put it, implementing the algorithm "will reduce health care costs and the growing workload for radiology departments" while sparing patients "less radiation" and "less exposure to contrast material."
This fits a broader pattern. Appropriate-use rules and AI triage are increasingly deciding which imaging studies get ordered in the first place — a counterweight to the long-running trend of rising imaging volume. Fewer low-yield scans is good medicine and good economics. But it changes the shape of the problem rather than eliminating it.
The scans that remain deserve better reads
Here's the part that's easy to miss. Even in the YEARS arm, most patients — 78% of them — still needed the CT. And every one of those studies still has to be interpreted correctly, reported clearly, and turned around fast enough to guide treatment. Reducing unnecessary scans doesn't reduce the stakes on the necessary ones; if anything, it raises them, because the remaining pretest probability is higher and the findings matter more.
So the two levers of imaging efficiency work together, not against each other:
| Appropriate use (YEARS-type rules) | Reporting quality (the reads that happen) | |
|---|---|---|
| Goal | Order fewer low-yield scans | Get more value from each scan performed |
| Lever | Clinical criteria + D-dimer | Accurate, structured reports; fast turnaround |
| Effect on volume | Fewer studies enter the queue | Studies clear the queue faster |
| Who stays accountable | Ordering clinician | The reading radiologist who signs the report |
Where AI CT reporting fits
The YEARS algorithm decides whether a scan happens. AI CT reporting works on the reads that do — it produces a structured, comprehensive draft for a chest CT or other study, xAID's in-house radiologist reviews every preliminary, and the report is delivered ready-to-sign so your reading radiologist signs the final. On a workload already trimmed by appropriate-use rules, faster and more consistent reporting is where the remaining time is won — see how that translates into turnaround-time benchmarks and what the accuracy evidence shows. Fewer scans and better reads aren't competing strategies; they're the same efficiency story told from both ends.
Frequently asked questions
Can cancer patients safely skip a CT scan for pulmonary embolism?
In some cases, yes. In the Hydra randomized trial published in JAMA in July 2026, a diagnostic strategy using the YEARS algorithm was as safe as scanning everyone with CT pulmonary angiography (CTPA), and it avoided a CT scan in 22% of patients with active cancer who had suspected pulmonary embolism. The decision rests on clinical criteria and a D-dimer blood test, not on the scan alone.
What is the YEARS algorithm for pulmonary embolism?
The YEARS algorithm is a clinical decision rule that combines three items — clinical signs of deep vein thrombosis, hemoptysis, and whether pulmonary embolism is the most likely diagnosis — with a D-dimer blood test. Pulmonary embolism can be ruled out without imaging if no items are present and D-dimer is below 1000 ng/mL, or if at least one item is present and D-dimer is below 500 ng/mL. It is a clinical protocol, not an imaging AI product.
How many CT scans did the YEARS algorithm avoid in the trial?
In the Hydra trial of 698 cancer patients across 21 hospitals in six countries, the YEARS algorithm obviated CTPA in 22% of patients — 77 of the 352 assigned to the algorithm group. Symptomatic venous thromboembolism or PE-related death during follow-up was 1.8% in the YEARS group versus 5.5% in the CTPA-only group, meeting the trial's noninferiority safety threshold.
What does the YEARS trial mean for radiology departments and AI reporting?
Risk-stratification tools like YEARS reduce the number of low-yield CT scans, easing radiology workload and sparing patients radiation and contrast. But they do not touch the scans that are still ordered. For the CTPAs and other CTs that remain clinically warranted, accurate, structured reporting and fast turnaround matter more, not less — appropriate use and reporting quality are two sides of the same efficiency story.
Source: Akerboom B, et al. "YEARS Algorithm for Diagnosis of Suspected Pulmonary Embolism in Patients With Cancer: A Randomized Clinical Trial" (the Hydra study), JAMA, July 12, 2026, doi.org/10.1001/jama.2026.10676; as reported by AuntMinnie and Radiology Business. YEARS criteria and D-dimer thresholds per the external validation study. Figures are rounded as reported.