Capacity and Access Payment (CAP)

• PCNs can qualify for a Capacity and Access Improvement Payment if they can demonstrate improvement (or high achievement) in three areas:

  • Patient experience
  • Ease of access
  • Accuracy of appointment records

• They must undertake “trend analysis” of GP Patient Survey (GPPS) data over the last five years to qualify

• Some PCNs are struggling to do this, for several reasons

Access to GPPS & FFT Data

• GP Patient Survey and Friends & Family Test (FFT) data is publicly available but difficult to work with
• This makes it difficult (and time-consuming) for PCNs to access the data, let alone analyse it
• Wrangling Excel spreadsheets can be painful - this is a problem across the entire NHS!

Limited Analytics Capabilities

• Not all PCNs have the capabilities to carry out the required analysis, especially given the difficulty accessing the data
• This is especially the case if they do not have their own data analysts
• They may be missing out on vital funding
• How do PCNs make best use of what is available?

Making the Data & Analysis Accessible

• We created an app that makes it easier for PCNs to access the GPPS & FFT data
• The app includes visualisations of trends, with simple, user-friendly customisation
• This significantly reduces the effort required to turn this very useful but inaccessible data into real value for PCNs

CAIP App UI

Visualising GPPS Data

Visualising FFT Data

Enabling PCNs Beyond the App

• The app itself is a useful snapshot of national, regional, ICB, PCN, and GP practice performance, but making this data as valuable as possible involves enabling PCNs to use it outside the app too

• Everything shown to the user is downloadable, and the data is formatted such that it should significantly reduce PCN workloads

• Everything is open-source - an increasing priority in the NHS (Goldacre and Morley 2022)

Analysis Without Comparison

• The CAIP app is a good start, but it lacks meaningful comparisons
• How do you know a GP practice is performing well (or not) without comparing against other GP practices?
• Comparing against national averages, or other GP practices in the local area, is limited at best

Comparisons Using Public Data

• We sought to make comparisons between practices based on factors that were likely to impact their performance and the challenges they face:

• While some of the features were engineered, all of the data is publicly available

Clustering GP Practices

• Using K-Means clustering, we were able to group GP practices into three clusters:
  • Urban - High Deprivation
  • Urban - Mid to Low Deprivation
  • Rural
• Evaluation metrics & validation against GPPS data suggest these are distinct, separable clusters
• Clusters are also consistent with existing academic research (Booth et al. 2021)

Incorporating Clusters into the CAIP App

• We will enhance the CAIP app using the GP Practice Similarities Model:
  • Visualising GP practice performance with comparisons against the average performance of their cluster
  • Identifying the \(n\) most similar practices using distances within clusters
  • Detailing what the clusters mean and how they can better illustrate practice performance

Extending/Improving the Model

• The model is in the relative early stages of development - it can definitely be improved in future iterations
• There is a lot of public data that we are not making full use of, whether it be as features in the model or for validation purposes
  • We plan to explore the use of QOF data and ONS Area Classifications data
• There are a wide range of potential use cases in Population Health Analytics and across SCW

Expanding Access to GPPS & FFT Data

• Public data should be easier for analysts to access
• We plan to build an API to make access to the GPPS & FFT data easier
• It will follow tidy data principles, to ensure it is immediately valuable to analysts
• In the future, we would like to build R and Python packages for accessing the API