Background and Aims
Pancreatic cancer typically presents with non-specific symptoms, often leading to diagnostic delays. We aimed to evaluate and compare the sensitivity of three decision support tools designed to assist primary care practitioners in identifying patients who may require pancreatic imaging.
Methods
We conducted an observational study in Australia evaluating three decision support tools: the Risk Assessment Tool (RAT), QCancer®, and a qualitative tool developed through consensus at QIMR Berghofer (QPaC Tool). We applied these tools to data from people diagnosed with pancreatic ductal adenocarcinoma (n=191) who were interviewed about their signs, symptoms, and risk factors at first clinical presentation. For QCancer®, we excluded patients with jaundice (n=142 remaining). Sensitivity was calculated as the percentage of participants correctly "flagged" by each tool according to predefined criteria.
Results
The mean age at first presentation was 66 years, 53% were men, and 24% had ever been diagnosed with diabetes. At the time of interview, 29% of participants had had surgery that successfully removed their tumour. In the full cohort (n=191), 62% experienced at least one pain symptom at first presentation. Other symptoms included jaundice (22%), altered bowel habits (18%), weight loss (17%), and malaise (16%). In these patients the QPaC Tool demonstrated a higher sensitivity than the RAT (54% versus 27%). Among the subset of patients without jaundice (n=142), QCancer® had a sensitivity of 14% at a 1% probability threshold, and the sensitivities of QPaC Tool and RAT were 44% and 7%, respectively. The superior sensitivity of the QPaC Tool was due mainly to its inclusion of severe epigastric pain and newly unstable diabetes.
Conclusions
Only the QPaC Tool achieved a sensitivity of more than 50%; however, its specificity remains unknown. Including the QPaC Tool in educational resources would encourage GPs to have a heightened index of suspicion for pancreatic cancer in patients without red flag symptoms such as jaundice and weight loss. Whether this approach decreases the delay to diagnosis without over-investigation of common symptoms will require further research.