OpenPathology: supporting better diagnostic testing

We are a team based at the DataLab at the University of Oxford. We're developing free, open tools for analysing pathology data in the NHS. These will give clinicians simple and actionable insights, to improve patient care and reduce costs. Get involved

Why does this matter?

Better audit and feedback about test requests in primary care has the potenital to improve patient safety and reduce costs significantly:

  • England sees 1.12 billion tests per year, costing £2.2 billion, 35-45% of which are requested from primary care (source).
  • Testing rates are increasing rapidly: patients in 2001 had on average 1.5 tests per year; five years later they were having 5 tests per year (source).
  • There is huge, often unwarranted variation in tests (source), which can lead to patient harm and unecessary costs.
  • Quality improvement studies have already demonstrated the potential to reduce costs and patient harm without impacting outcomes (source).

What will it look like?

Our prototype system will focus on primary care. Practices will be able view a range of test orders and results, in the context of their peers. As well as charts, all data will downloadable for further analysis. We will write a series of papers exploring trends and variation across the data.

Data will be presented on this website, in dashboards similar to the ones we developed for OpenPrescribing (example). As an example, here's how we might show potassium test ordering and results for a single practice:

Number of Potassium tests per person
Chart showing potassium ordering
This practice requests relatively more than other practices in the CCG.
Proportion of potassium tests that are high
Another chart showing potassium ordering
You can see big winter peaks across the region, but this practice has reduced the seasonal fluctuation in high potassiums, and reduced the baseline across the year.

How can I get involved?

Anyone with an interest in pathology data should get in touch. At this stage, we're particularly interested in finding partners who want to include their data in the project. Get involved