Can we personalise urgent referrals to find cancer earlier?
Aim
To find out if using information available in GP records could personalise the process of urgent cancer referral.
Why?
The earlier patients are diagnosed with cancer, the more likely they are to be cured. One way to find cancers earlier is by improving the process of urgent referrals from GPs to hospitals when a patient has a symptom that might be cancer.
What questions will I ask?
1. Will using extra information from medical notes help GPs to make better decisions about who to test for cancer?
We want to find out if we can help GPs make better decisions by using more of the information that they already have in your medical records.
2. What impact can this have on patients?
In particular:
• Will it change the number of patients being referred?
• Will cancers be diagnosed earlier?
• Will cancer patients live longer?
What impact might this have?
What we find out in this study could become part of future guidelines for doctors assessing whether patients might have cancer. This has the potential to help diagnose patients earlier and save lives. This not only helps patients and their families but also the NHS as it can save money and improve resource allocation.
Background
A key route for cancer diagnosis is through urgent referral from primary to secondary care; by optimising this pathway it should be possible to increase the numbers of patients diagnosed at an early stage. Current urgent cancer referral criteria include only symptoms, signs and abnormal test results (“features” from here on) and limited demographics. However, GP records contain a wealth of other information could be used to inform the need for referral, including widened demographic and clinical data.
Aims
1. Develop a personalised risk score of undiagnosed cancer for patients presenting to their GP with features of colorectal, gastro-oesophageal or pancreatic cancers, using routine data including demographics, test results, and comorbidities.
2. Model the potential impact on patient outcomes of the personalised risk score for urgent cancer referrals in terms of numbers eligible for referral and change in 10 year survival.
Methods
Personalised risk scoring systems will be developed using logistic regression on data from CPRD Aurum. NCRAS linked data will be used to define the primary outcome of diagnosis with cancer and secondary outcome of stage.
Impacts on patient outcomes of the new scoring system will be estimated and compared to outcomes using current NICE guidance. The best performing risk prediction models will be applied to estimate the difference in numbers of patients eligible for referral and date of eligibility for referral.
Anticipated impact
This has significant potential to influence public health policy; it provides critical evidence for the early detection and diagnosis of cancer roadmap, will be reported to the policy research unit on cancer awareness, screening and early diagnosis and could be of critical use in the next refinement of national guidance.
All outcomes refer to cancers of the pancreas, colon, rectum, stomach or oesophagus
Primary
Diagnosis of cancer;
Personalised risk score of undiagnosed cancer;
Secondary
Stage at diagnosis;
We will also model the impact on the outcomes below but these models will be predictions based on other existing findings, rather than derived from CPRD data:
Estimated numbers of urgent referrals;
Date that a patient might become eligible for urgent referral;
Stage at cancer diagnosis;
Patient 10 year survival following cancer diagnosis.
Sarah Moore - Chief Investigator - University of Exeter
Sarah Moore - Corresponding Applicant - University of Exeter
Fiona Walter - Collaborator - Queen Mary University of London
Gary Abel - Collaborator - University of Exeter
Richard Neal - Collaborator - University of Exeter
Sarah Price - Collaborator - University of Exeter
William Hamilton - Collaborator - University of Exeter
NCRAS Cancer Registration Data;No additional NCRAS data required;Patient Level Index of Multiple Deprivation;Practice Level Index of Multiple Deprivation