A digital quality measure for emergency presentation of pancreatic cancer
Menée aux Etats-Unis à partir de données portant sur 4 415 patients atteints d'un cancer, cette étude évalue la performance d'un algorithme pour détecter des cancers du pancréas à la suite de soins d'urgence liés à la maladie
Background : Over half of patients with pancreatic cancer experience an emergency cancer diagnosis, but tools for large-scale study are lacking. Towards this end, we developed a digital quality measure (dQM) to automate the detection of pancreatic cancer emergency presentations (EPs).
Methods : A dQM for pancreatic cancer EPs was developed within the U.S. Veterans Affairs health care system. Multivariable regression models were used to study the associations between EPs and cancer outcomes. Records of EP cases were manually reviewed to identify missed opportunities in diagnosis.
Results : The dQM had a positive predictive value of 86.4% (95% CI 80.0–92.8) for accurately identifying EPs among patients with pancreatic cancer. Among 4415 pancreatic cancer patients, 60.9% were identified as EPs by the measure. Patients with EPs had more advanced-stage disease (adjusted OR 1.38; 95% CI 1.20–1.59) and higher mortality (adjusted HR 1.64; 95% CI 1.51–1.77). Nearly one in five EP cases had missed opportunities in diagnosis.
Conclusions : Our dQM had strong performance characteristics for identifying pancreatic cancer EPs, which were independently associated with worse patient outcomes. A notable subset of cases was potentially avoidable. The dQM is a promising strategy for health care systems to identify and measure EPs for quality improvement initiatives.
British Journal of Cancer , article en libre accès, 2026