Making a case for life-saving data

The NHS holds a huge amount of data that could help us to better diagnose and treat patients and even cure them, but that data is not always joined up.

Data saves lives. That’s the title of a recent strategy paper published by the Department of Health and Social Care, and it couldn’t be truer.

The NHS is swimming in data that could help us to better diagnose and treat patients and even cure them. However, that data is not always joined up. We may not be able to see what it is telling us. And many medical practitioners are reluctant to share useful patient records in the first place.

So, why does this matter, and what can we do to get the most out of this incredible resource?

Millions of records are hard to tame

The NHS puts the UK in an almost unique position of having, in theory, a unified healthcare record for every citizen. Anyone who has been treated recently won’t be surprised to learn that medical records in the NHS are not that joined up.

The records are huge and grow at an amazing rate with around 1 million GP consultations and 300,000 diagnostic tests every single day. Add in X-rays, scans and routine procedures and it’s clear that health records are a mammoth beast to tame.

Completing the data jigsaw is hard

Huge efforts are being made to bring this disparate data together but knitting a single record for an individual patient is far from easy.

Data must be validated. Some errors, like a patient with a body mass index of over 2,000, may be easy to spot and explained by a decimal point in the wrong place. Others are harder to see.

Fortunately, the UK has developed many high-quality datasets. However, it can still be incredibly hard for researchers to find who holds which bits of information.

If a researcher wants to review patients with asthma on a specific combination of medicines, how do they even begin to find the right information?

Until recently, the only way to find if a database held the required information would be to request access and then work through it. You could waste time trawling through records only to find that key bits of information were missing or were never collected.

A gateway to accessing data

These challenges drove UK Research and Innovation to fund creation of the Health Data Research Innovation Gateway by Health Data Research UK. The gateway is, at the simplest level, a catalogue that lists the different sets of health data.

Hardly groundbreaking. The clever bit is that the catalogue also captures supporting information. Access to these extra bits of information or metadata means it is possible to understand exactly what type of information is held.

It describes all the different bits of data recorded, for example the ages, smoking history and medicines prescribed. So when researchers ask to see the real data, they will already be confident that it can meet their needs.

Data is key to better medicine

While researchers might want to work with small discrete repositories of health data, what about the data that comes flooding out from the NHS every single day?

The biggest potential may be in addressing the efficacy gap between how medicines should work as predicted in clinical trials and how well they work in the real world.

Estimates suggest less than 50% of the medicine given to patients makes patients better. Around 5% of admissions to hospital are the result of adverse drug reactions.

Doctors face an almost impossible task picking from more than almost 2,000 different medicines, each with their own unique dosages.

While prescribing guidance covers the most common drug interactions, giving guidance for every permutation would be almost impossible. More and more patients take complex combinations of drugs. At least 2 million UK patients routinely take more than 7 prescribed medicines.

The solution is to use real data – and lots of it. To capture enough cases of each different permutation to provide a meaningful result means combining records from thousands if not millions of patients.

It is possible to get a flavour of the kind of transformational insights data might bring from a paper published this year. Using data held by UK Biobank researchers have been able to relate changes in the blood vessels in the back of the eye to the risk of heart disease. While it will require significantly more work before opticians can routinely make assessments of this kind it illustrates the important and perhaps surprising ways that health data will change the way care is delivered.

We must win hearts and minds

This large-scale data mining is critical to improved prescribing, but only if the analysis is readily accessible to those making the prescribing decisions.

If data really can save lives, why have the medical professionals been some of the loudest raising concerns about the sharing of patient records? This suggests that the strategic approach to health data has failed to win the hearts and minds of the clinical community.

Doctors are rightly protective of their patients and their privacy. However, their reluctance can be explained in other ways. Could data, for example, be used to monitor performance of individual doctors or used to restrict the freedom to use clinical judgement to assess which medicines are best for their patients?

An emphasis on collecting increasingly detailed information could simply add to the workload of hard-pressed clinicians.

Value lies in rich and complex data

We must demonstrate the value of this data to doctors.

The first step is getting that data. The real benefit will come when those choosing medicines have relevant evidence at their fingertips to know how a specific patient is likely to respond to a given medicine based on experience.

Ironically, much of the anxiety around sharing data is about privacy of the individual. This misses the point. The true value of the data does not lie in any single record, but in the richness and complexity of data when the records are considered together.

Reference: Diaz-Pinto, A., Ravikumar, N., Attar, R. et al. Predicting myocardial infarction through retinal scans and minimal personal information. Nature Machine Intelligence 4, 55 to 61 (2022).

Rudnicka AR, Welikala R, Barman S, et al Artificial intelligence-enabled retinal vasculometry for prediction of circulatory mortality, myocardial infarction and stroke. British Journal of Ophthalmology, published online first: 4 October 2022. DOI: 10.1136/bjo-2022-321842

Top image:  Credit: UK Research and Innovation

This is the website for UKRI: our seven research councils, Research England and Innovate UK. Let us know if you have feedback or would like to help improve our online products and services.