New score predicts thrombosis risk in hospitalized pancreatitis patients

Key messages

• Among 647 patients, 114 (17.6%) had at least one thromboembolic event diagnosed during hospitalization.
• A six-parameter score separated patients into low-, intermediate-, and high-risk groups with observed event rates of 7.6%, 22.0%, and 38.5%.
• The model showed moderate discrimination but was assessed only in the cohort used to develop it; prospective external validation is required.

Why thrombosis matters in pancreatitis

Evidence for screening and treatment remains limited. Symptoms may overlap with pancreatitis itself, and imaging practices vary. The study therefore examined both abdominal and extra-abdominal thromboembolic events and asked whether information available at admission could identify patients at higher risk of having thrombosis detected during the same hospital stay.

What the researchers did

The investigators retrospectively analyzed patients who had been prospectively enrolled in an institutional database between January 2018 and June 2023. The cohort included 365 patients with acute pancreatitis and 282 with chronic pancreatitis. The primary outcome was any diagnosed deep vein thrombosis, pulmonary embolism, portal vein thrombosis, or other splanchnic vein thrombosis during hospitalization. Categories could overlap within one patient.

What the study found

The point score produced an area under the receiver operating characteristic curve of 0.709 using data-derived thresholds. A version using clinically selected thresholds had an area of 0.683; the difference was not statistically significant. These values indicate moderate separation between patients with and without detected thrombosis, not diagnostic certainty.

Meaning, limitations, and next steps

Because the study was observational, it cannot establish whether thrombosis caused poorer outcomes or mainly marked more severe illness. Imaging was not standardized, so some events may have been missed. The single-center score also had no internal resampling or independent external validation, meaning its reported performance may be optimistic.

Future multicenter studies should test calibration, diagnostic yield, unnecessary imaging, and clinical net benefit. Prospective trials are also needed to determine whether score-triggered vascular imaging leads to earlier diagnosis and better outcomes without excess testing or bleeding from subsequent treatment.

Source:

First Hospital of Jilin University

Journal reference: