This site is intended for health professionals only

How to turn NHS data into intelligence – part 2

How to turn NHS data into intelligence – part 2
By Dr Cornelia Junghans, Dr James Crick, Dr Peter Roderick and Dr Kathrin Thomas
13 February 2023

Data analysis and intelligence-driven decisions by ICBs are a must in modern healthcare. So how can GP practice data be used to get the best outcomes? Dr Cornelia Junghans, Dr James Crick, Dr Peter Roderick and Dr Kathrin Thomas explain in this second article of a two-part series.

GP practice data is rich and large enough to provide great insights into whole populations. That’s of enormous value when making intelligence-driven decisions for an area. As the first article in this series, ‘How to make intelligent use of NHS data,’ showed, data has huge potential as long as pitfalls are taken into account.  Here we discuss how GP practice data has much to offer in terms of insights – providing the approach is right.

For example, when it comes to using GP practice data, we need to shift our thinking to improve population health. Individual interventions targeted to those at most risk may be clinically appropriate and necessary, but a whole-population approach is needed.

That means a greater focus on the following:


This is utilising low-cost, high-volume interventions such as smoking cessation services, embedded financial inclusion support in practice, or mass text message approaches using behavioural insights to increase cervical screening uptake.

Inequalities awareness

The ‘inverse care law’ [1] posits that the higher your level of need for quality healthcare, the less likely you are to receive it. Putting in remedies to address this will align with Professor Sir Michael Marmot’s ‘proportionate universalism’.[2] That is, everyone receives the universal offer of primary care, but those from deprived backgrounds – or inclusion health groups – are given increased resources, swifter access and greater help, such as GP at the Deep End.


Evidence suggests that timely good quality clinical care only contributes around 20% of our health. So turning data into intelligence thus challenges primary care to understand the social factors behind disease registers and activity data and work more closely with local public health colleagues and the voluntary and community sector in addressing the social determinants of health.

The impact of underfunding

As Julian Tudor Hart put it: ‘Early interventions, requiring little technical skill but much time and thoughtful social experience and judgement can be far more cost-effective than late crisis interventions using high technology and expertise.

‘Yet over and over again, resource-starved acute services are compelled to steal from anticipatory care budgets so as to sustain crisis care and thus hopefully avert public scandals, careers pillorying by journalists, and the ire of politicians whose careers depend on the next election.’

The trouble is, says Hart, no one can justify withholding care in a crisis because it would be more cost-effective to provide anticipatory care for those not yet in crisis. He goes on: ‘When services are grossly underfunded overall, as they were in the UK for at least last three decades of the 20th century, anticipatory care of people not yet seriously ill is forced into grossly unequal competition with heroic end-stage salvage, and loses every time.’ [3]

The limits of GP practice data

Data certainly has its uses. But just as important is the data that is missing. Abraham Wald, a statistician working for the US Military, figured this out when asked to decide where to place armour on planes to protect them from being shot down.[4] Others suggested the obvious – putting armour plating over the wings because planes that had returned from battle had wings riddled with bullet holes.  But Wald asked for the plating to go on the areas without bullet holes. He realised that they were looking only at the planes able to fly home despite the bullets. The ones that got shot down were unavailable for inspection.

We can think of data in the same way. GP practice data only includes the people who, by definition, successfully register and attend their surgery. So, when we tailor services to those deemed most worthy of intervention, we might be getting it wrong. They may not actually be the ones who are the most vulnerable and in need of our help.

While we might find ways to include some of these people and collect the data, it’s not always feasible. The most predictive factors for needing state-funded adult social care are a decline in activities of daily living, change in personal finances or informal social support. However, these factors are likely too private, too complicated or changeable to collate and maintain. And it would be even harder to act on the information in a timely manner.[5]

Thinking beyond the numbers

And then there is the intelligence beyond the data. We might know from the data that a patient is a frequent attendee at their GP practice and A&E. It will also tell us their ethnic background, gender and preferred language. But it is only when you speak to the patient that you find out that the reason they present multiple times is that they can’t read or write and don’t know how to take their medication correctly.

Data thus becomes the start of a conversation, not the be-all and end-all. Or, in fact, the start of multiple conversations. For example, when we use data, we can join up people with shared characteristics who would not otherwise know each other and enable self or better each-other care, as with group consultations and patient groups. This might be the true power of primary care data.

So, we need to continue to change our mindset. GPs are not just in the business of treating disease but also creating health. Social determinants are our business too, not just the realm of a social prescriber. Prevention conversations are not a luxury – they are essential to shift the needle on health.

QOF may have inadvertently widened inequalities by encouraging GP practices to aim for just enough with limited resources while the remaining percentages represent the hardest to reach and the easiest to ignore. Data analysis and good intelligence-driven local healthcare decisions by ICBs are no longer a nice-to-have; many people are doing it already. If done well, it reduces demands on services in the long term, hits the targets we want, and we do right by our populations.

By Dr Cornelia Junghans, Senior clinical fellow in Primary Care at Imperial College London, Dr James Crick, Consultant in public health medicine and sessional GP Hull Health and Care Partnership, Dr Peter Roderick, Consultant in public health NHS Humber and North Yorkshire ICB andDr Kathrin Thomas, GP and Consultant in public health

[1] Tudor Hart, J. The Inverse Care Law. The Lancet Vol 1971; 297, 7696: 405-412

[2] Marmot, M. Fair society, healthy lives: the Marmot Review : strategic review of health inequalities in England post-2010. (2010)

[3] Tudor Hart, J. The Political Economy of Healthcare: a Clinical Perspective. 2006: 131-132

[4] Ellenberg J. How Not to Be Wrong: The Power of Mathematical Thinking. New York: The Penguin Press; 2014.

[5] Nakubulwa M, Junghans C, Novov V, Lyons-Amos C, Lovett D, Majeed A, Aylin P, Woodcock T. Factors associated with accessing long-term adult social care in people aged 75 and over: a retrospective cohort study. Age Ageing. 2022 Mar 1;51(3):afac038. doi: 10.1093/ageing/afac038. PMID: 35231093; PMCID: PMC8887841.

Want news like this straight to your inbox?

Related articles