QResearch Logo

Decision risk tool for early diagnosis of pancreatic ductal adenocarcinoma (PDAC) among new-onset diabetes




Development and validation of a decision risk tool for early diagnosis of pancreatic ductal adenocarcinoma (PDAC) among new-onset diabetes in UK primary care

What is the aim of the study and why is it important?

Despite improvements in treatment and care, pancreatic cancer remains a very deadly disease. This is mainly due to patients presenting at late stages of disease when there are fewer treatment options. Patients with pancreatic cancer diagnosed at earlier stages are more likely to be treated successfully and survive longer. Unfortunately, pancreatic cancer is challenging to detect early as it mostly shows no or vague symptoms in early stages. However, many patients who develop pancreatic cancer also have been shown to have diabetes diagnosed in the months/years before a pancreatic cancer diagnosis. A new diagnosis of diabetes is therefore a potential indicator of pancreatic cancer which can be used for early detection. However as diabetes is a relatively common condition in the population, it remains impracticable to refer everyone for pancreatic cancer scans. Hence, GPs will need a more accurate tool to better identify patients who will be more likely to have pancreatic cancer for further tests.

In this study, we will perform the following
1. Develop and validate a risk-prediction tool for early diagnosis of PDAC among new-onset type-2 diabetes mellitus (T2DM) in primary care

2. Examine the scope for machine learning approaches to develop a risk prediction tool for PDAC in people with new-onset T2DM, and compare these approaches with traditional statistical modelling

How is the research being done?

This study will build a risk prediction tool to identify patients who are more likely to have or develop pancreatic cancer among those with new-onset diabetes. use information from QResearch – a database containing anonymised GP records of patients in England, taken over a period of over 30 years. This will be linked to hospital records, cancer registry and death records to search for individual new-onset diabetes patient characteristics (e.g. age, gender, lifestyle) that are associated with a later diagnosis of pancreatic cancer. We will compare the accuracy of risk prediction using computer algorithms and evaluate if a risk-based tool used in GP settings will be effective to detect pancreatic cancer at earlier stages among new-onset diabetes patients.

Chief Investigator

Dr Pui San Tan, University of Oxford and Prof Julia Hippisley-Cox, University of Oxford

Lead Applicant Organisation Name


University of Oxford

Location of research

University of Oxford

Date on which research approved


Project reference ID


Generic ethics approval reference


Are all data accessed are in anonymised form?


Brief summary of the dataset to be released (including any sensitive data)

EMIS, HES, ONS, Cancer Registry

Implications and Impact

The risk prediction tool developed in this study will provide a very useful clinical decision tool for use in primary care to better assess pancreatic cancer risk and guide further referrals/scans to aid early detection of pancreatic cancer.

Funding Source

Pancreatic Cancer UK

Public Benefit Statement

Research Team

Dr Ashley Clift, University of Oxford       

Dr Shivan Sivakumar, University of Oxford                        

Dr Martina Patone, University of Oxford                           

Mr Weiqi Liao, University of Oxford                     

Miss Winnie Xue Mei, University of Oxford                                                                 

Dr Rachael Bashford-Rogers, University of Oxford           

Prof Carol Coupland, University of Oxford & University of Nottingham

Prof Steve Pereira, University College London    

Prof Richard Clifton, University of Oxford                          


  • Predicting risk of pancreatic cancer in individuals with new-onset type-2 diabetes in primary care: protocol for the development and validation of a clinical prediction model (QPancreasD)
    Authors: Pui San Tan, Ashley Kieran Clift, Weiqi Liao, Martina Patone, Carol Coupland, Rachael Bashford-Rogers, Shivan Sivakumar, David Clifton, Stephen P Pereira, Julia Hippisley-Cox

Press Releases

Access Type

Trusted Research Environment (TRE)

Share this