A recently published paper and policy briefing highlight the challenge of disease risk prediction tools rarely reaching patients and not being evaluated if they do, and what we can do to overcome the technical and structural barriers to their implementation.
Why does risk prediction matter?
Primary care – particularly GP surgeries – are the first point of contact with the health system for most patients. Risk prediction tools, which are digital models that use a range of patient information to predict risk of future disease, have the potential to transform healthcare: primary care can identify people at risk of disease earlier and start managing this risk, and patients benefit from improved health outcomes.
Where are the risk tools?
It would therefore be expected for tools and resources that support risk prediction in primary care to be numerous and well integrated and for there to be extensive follow up data available on their performance, however this is not the case. The research found that while plenty of good quality risk prediction tools are being made, very few make it into clinical practice, and of those that do, they are not systematically evaluated.
The project team carried out literature reviews, a workshop and consensus meetings with key stakeholders from Medicines and Healthcare products Regulatory Agency (MHRA), NHS England, Electronic Healthcare Record (EHR) vendors, academia and industry. The findings showed that the problems are not down to the tools being developed, rather the issues are primarily technical and structural.
Breaking the deadlock
The workshop and consensus meetings developed fifteen recommendations to break this deadlock. In the briefing, we incorporate the recommendations into four structural conditions which need to move in parallel to make a difference:
- Early co-production and cross-sector alignment. This means involving risk prediction tool and EHR developers from the start of the technology development process. This needs early and consistent collaboration between academia, industry and clinicians.
- A shared deployment infrastructure. Improved NHS digital infrastructure to support tool delivery will be essential. This can include a libraries of tools linked with EHR software through standard interfaces, enabling GPs to access the tools they need easily from their work computers.
- A fit for purpose regulatory pathway. Ongoing work by the MHRA on regulation of software as a medical device will help smooth the path to tool implementation. Key to this are evidence standards for real-world evaluation – clarity around what types of information are needed to assess how tools are performing in the clinic.
- Shared liability and sustained funding. We need processes that allow the risk of developing tools to be shared, and not fully taken on by manufacturers, and clear pathways of when and how developers are paid for the work that they do.
None of these conditions can be fixed by any one organisation and extensive cooperation and coordination will be essential to make them happen. However, given that prevention is now a key pillar of health policy, including the 10 year health plan for England, there is now ample opportunity to act and support implementation and use of these tools. In future blogs we will explore these conditions in more detail and focus on how – and who – can make them happen.
This research was funded by the National Institute for Health and Care Research (NIHR) Policy Research Programme, conducted through the Policy Research Unit on Cancer Awareness, Screening and Early Diagnosis (NIHR206132).
