Patient data use case studies
Questions about these case studies or how patient data is handled?
During our Review of Informed Choice for Cancer Registration, patients clearly told us they would like to know what their data are being used for. We agreed that as a charity, we would highlight where we are using patient-data for research and analysis. Relevant examples of work by the Cancer Intelligence team at Cancer Research UK will be available here, and we will add to this as we embark on new projects.
Lung cancer diagnostic pathways
Identifying common lung cancer diagnostic pathways using linked datasets.
Small proportions of patients in 2013-2015 meet the timings of the optimal pathway. Diagnostic pathway length varies by source of imaging request.
Variation in ovarian cancer treatment rates
Investigating variation in treatment for ovarian cancer patients.
We expect to establish if there is variation between Cancer Alliances/Trusts, identify the odds ratios/ regression co-efficients for predictor variables, and if it's the affect of healthcare system level factors.
Variation in lung cancer treatment rates
Investigating variation in treatment for non-small cell lung cancer patients.
We expect to establish if there is significant variation between Cancer Alliances/Trusts (geography still to TBC) and identify the odds ratios/regression co-effecients associated with each predictor variable.
Bowel cancer screening campaign evaluation
Evaluating the bowel cancer screening regional pilot campaign in the North West of England.
Analysis of interim results following the campaign are promising. We are currently working to finalise the analysis and results will follow in due course.
Colorectal cancer diagnostic pathways
Coming June 2018
The earlier diagnosis of lung cancer will save lives but also puts additional pressure on diagnostic services. It is therefore important that lung cancer pathways are organised to be as effective and efficient as possible and ensure patients are given their diagnosis as soon as possible.
The data used was lung cancer registrations (2013-2015) from National Cancer Registration Analysis Service (NCRAS), diagnostic imaging data (DID) and cancer waiting times (CWT) data from NHS England and involved data from over 100,000 patients.
We hoped to improve understanding of pre-diagnostic events, intervals and patterns for lung cancer patients on a national scale, and also to benchmark to the timings in the newly adopted National Optimal Lung Cancer Pathway (NOLCP).
Following data linkage, time intervals between events were calculated, different scenarios of events were investigated and timings of events were compared with those from the NOLCP.
Many different diagnostic scenarios exist, from simple to complex, which varies by CCG. Time intervals from imaging to diagnosis differed by source of image referral, with those ordered by GP direct access imaging having longer pathways. Benchmarking to the NOLCP timings showed small proportions (less than 6%) of patients meeting timings, this also varied widely by CCG. The benchmarking exercise can be repeated to show degree of compliance to optimal pathway timings.
Partners: ACE Programme (funded by CRUK, Macmillan, NHS England), CRUK-PHE Partnership, NCRAS
This analysis will establish a detailed picture of ovarian cancer treatments and outcomes and is intended to provide initial insights into a planned national ovarian cancer audit benchmarking pilot, with the ultimate aim to act as a catalyst to reduce variation and drive improvements to clinical practice.
Data used will be: patients diagnosed with ovarian cancer (1st July 2014 to 31st March 2015) who were eligible for SACT (from NCRAS), associated patient-level data and tumour characteristics (from HES, COSD, RTDS and Cancer Registry), and treatments patients received (from SACT).
Survival for ovarian cancer in the UK is lower than other comparable high-income countries and there is substantial regional variation within the UK. Access to and/or quality of treatment may be contributing factors.
The linked data will be used to compare treatment access rates between Cancer Alliances and/ or Trust of MDT/diagnosis, to determine the extent to which this can be explained by regional differences in patient demographics, such as age and socioeconomic status (case-mix), and we will investigate the influence of healthcare system levels factors on access to treatment, such as provider and type, and consultant volume and specialisation.
Findings are due Winter 2018, but we expect to find how much geographical variation in access to treatment for ovarian cancer there is in England, and to what extent is this affected by patient demographics and healthcare system level factors.
The purpose of this service evaluation is to establish; factors that affect access to lung cancer treatments for non-small cell lung cancer (NSCLC) patients, whether there is variation between Cancer Alliances/ hospital trusts in access to a range of treatment pathways, and the impact of patient, tumour and provider characteristics on access to treatments.
Data used will be: patients diagnosed with lung cancer (1st April 2014 to 31st March 2015) who were eligible for SACT (from NCRAS), associated patient-level data and tumour characteristics (from HES, COSD, RTDS and Cancer Registry), and treatments patients received (from SACT).
Understanding variation in accessing treatments will inform policy makers and commissioners regarding where efforts should be focused to ensure equitable access to effective treatments for NSCLC, and improve patient outcomes.
The linked data will be used to calculate access rates for different treatments, and then statistical tests will measure the overall geographic variation and significance. Finally, logistic regression will identify which factors are predictive of whether people receive various treatments (e.g. age, deprivation, ethnicity).
Findings are due Winter 2018, but we expect to establish the presence of variation at a geographic level, to understand the statistical significance of any variation, and to identify the odds ratios/regression co-effecients associated with each predictor variable.
In England, less than 60% of eligible bowel cancer screening participants take part in the programme. A regional Be Clear on Cancer pilot campaign was carried out across the North West of England to encourage participation.
The campaign ran in 33 CCGs and consisted of advertising (including TV) and direct mail. Advertising ran alone for 6 weeks and was later combined for 6 weeks with direct mail in 22 CCGs to first-timers and non-responders. The remaining 11 CCGs continued to receive advertising only. The campaign was aimed at people aged 55-74 in the lower socioeconomic groups (C2DE) and skewed towards men; this was based on pilots previously run by Cancer Research UK.
We evaluated whether the campaign advertising, or advertising paired with sending a CRUK-endorsement letter with the kit increased uptake. Differences have also be compared across different socioeconomic groups.
The data we used to do this were extracted from the national bowel cancer screening database. We used age, location, date of test kit received and date kit sent back. For the direct mail element of the campaign, we knew who had received a letter and who hadn’t. We did not receive any names, addresses, or NHS numbers, so no individuals could be identified.
Analysis of interim results following the campaign are promising. We are currently working to finalise the analysis and results will follow in due course. If successful, the pilot could be extended to cover all eligible bowel cancer screening participants.
Coming June 2018