Diagnostic Pathways in Pancreatic Cancer
Project Details
The aim of this study was to examine the utility of linked primary care data and clinical cancer registry data to determine variation in diagnostic pathways for patients diagnosed with pancreatic cancer.
Patron ID: PAT081
Project Lead:
A detailed review of the baseline demographics and referral pathways for patients within the PURPLE Translational Registry has not previously been undertaken. This project examined the utility of linked primary care data and clinical cancer registry data to determine variation, including time to diagnosis and intervals in diagnostic pathways for patients diagnosed with pancreatic cancer
Research Outcomes
Project Outcome
Information on symptoms in the PATRON dataset is included as free text in the ‘reason for encounter’ field. This can include multiple different ways of describing the same symptom. The reason for encounter data therefore needs to be cleaned before it can be used so that these symptoms can be identified. We carried out a process called ‘symptom mapping’ where a list of search terms is generated which describe the same symptom.
These search terms are then used to identify all the instances of the same symptom in the PATRON dataset. This was done for 20 different symptoms. Sometimes blood test results can also be used to identify symptoms or diagnoses, for example ‘jaundice’, ‘anaemia’ or diabetes. We therefore also searched the PATRON blood test data to find any additional cases of symptoms or diagnoses. Additional information on family history, medical history and smoking history were also extracted from the PATRON dataset. All of these relevant variables were then combined into one dataset and linked to hospital data and cancer registry data for use in research projects.
One such research project was to use the primary care data to identify patients with symptoms of possible pancreatic cancer. For each symptom/diagnosis, we calculated the risk of the patient being diagnosed with pancreatic cancer in the next 12-24 months. We then applied three existing primary care tools for calculating the risk of pancreatic cancer to compare which of the tools was the best at picking up pancreatic cancer in our patient population. This project has been completed and an article manuscript outlining the results has been written.
Research Publications
The sensitivity of decision support tools for identifying patients with pancreatic cancer https://bjgpopen.org/content/early/2025/11/14/BJGPO.2025.0142
Schrader S, Rafiq M, Martinez-Gutierrez J, Waterhouse M, Lee B, Neale RE, Emery J. Decision support tools for pancreatic cancer detection: external validation in Australian primary care — a retrospective cohort study. BJGP February 2026; DOI: https://doi.org/10.3399/BJGP.2025.0328
Neale RE, Jordan SJ, Thompson B, Bernardes CM, Adams J, Baggoley C, Barreto SG, Baxter CM, Croagh D, Devereaux B, Emery J, Collins LG, Gilhotra R, Grogan P, Hourigan L, Martinez-Gutierrez J, Metz AJ, Philcox S, Rafiq M, Rhee J, Schrader S, Stewart M, Windsor J, Zalcberg J, Waterhouse M. The sensitivity of decision support tools for identifying patients with pancreatic cancer. BJGP Open November 2025; DOI: https://doi.org/10.3399/BJGPO.2025.0142