Quantifying how many frail elderly patients there are could revolutionise their care – and the project may be the first of many to exploit big NHS data. By Jennifer Trueland

Finding new ways to care for the increasingly elderly population is big business for the NHS. Ministerial announcements, reports from learned think tanks and serious academics all come down to one thing: we’ve got to get better at looking after older people.

There is a moral imperative to do this and a strong economic argument too. With a growing elderly population comes greater demands on healthcare systems and hence more pressure on tightening healthcare budgets. One of the great difficulties, however, is knowing which older people are likely to need the extra resources and, crucially, finding them in time to do something about it.

‘A simple use could be to lengthen consultation time for frail, older people to better address their complex needs’

Ministers want action on this: in September, health secretary Jeremy Hunt called for a more proactive approach to looking after older people, to keep them out of hospital where possible, and said all older people should have a named doctor, or “accountable clinician”.

But how do we shift the focus of care from a secondary care reaction to crisis to a primary care focus on prevention? One answer perhaps is to treat frailty as a long term condition.

Changing focus

According to older people’s medicine specialist Dr Andrew Clegg, while most of us would recognise older people who are very frail, defining frailty and classifying it is a different story. That’s why Dr Clegg, clinical senior lecturer at Leeds University, and honorary consultant geriatrician at Bradford Teaching Hospitals, has been working on an index to do just that.

It has been built on the intelligence gathered from some 5m NHS patient records held in TPP’s SystmOne clinical system. Together, TPP, the University of Leeds and the UK government’s Technology Strategy Board have anonymised clinical and administrative data and developed ResearchOne, a not for profit health and care research database.

According to TPP head of informatics and analytics Chris Bates, this has the potential to become one of the largest healthcare research databases in the world, containing up to 30 million records.

Developing it took two years of hard but worthwhile work by analysts from TPP and the University of Leeds, and now researchers such as Dr Clegg are beginning to reap the benefits.

As a geriatrician, in his day-to-day practice he finds that simple observation provides one of the best indications of how frail a patient is. In the clinic, he watches patients as they make their way from waiting room to consulting room to see how easily they get up from a chair and how quickly they walk across the room.

That’s an instinctive judgment based on years of experience – but how do you make it quantifiable? In essence, the frailty index is a numerical way of measuring just how frail a patient is based on a number of factors.

Frailty indicators

“We have used a cumulative deficit model for our frailty index,” he says. “Deficits are things that can indicate frailty and can be clinical symptoms, signs, diseases and results of laboratory tests.”

‘The more deficits people have, the more likely they are to be frail, so counting up the number of deficits from head to toe enables clinicians to calculate the frailty index’

So if a person had hearing problems that would be a symptom; if they had a tremor that would be a clinical sign; if they had osteoporosis that would fall under disease, while an abnormal blood or urine result would also be part of the picture. The more deficits people have, the more likely they are to be frail, so counting up the number of deficits from head to toe enables clinicians to calculate the frailty index.

All of these factors are part of the jigsaw which helps determine just how frail an elderly person is, and just what their care needs are at a given point.
In order to build up the index, Dr Clegg and colleagues have studied hundreds of thousands of anonymised patient records and categorised around 3,000 Read Codes into approximately 50 deficits.

“It’s mathematically derived but it’s fairly simple,” says Dr Clegg, explaining that people get a frailty “score” depending on how many factors they display.

There would be many uses for an accurate frailty index. For example, a simple use could be to lengthen consultation time for frail older people to better address their complex needs and multiple health problems. A frailty index could also be used to design improved, evidence-based pathways of care for frail older people, including delivery of interventions and medication reviews.

“From an end of life care perspective, if someone had a frailty score of, say, 0.7, then we would know they were likely to be in the terminal phase of life and would be able to plan accordingly. It would help with decisions around advanced care planning, and even palliative care,” he adds.

The real beauty of the process is that the results of the research based on ResearchOne data can be fed right back into the clinical systems of those who have provided the basic information, explains Dr Bates. “The ability to build results into the clinical system is very exciting. Plus, we’re committed to publishing all results in an open-source format.”

The data could be useful in many areas: public health researchers are also starting to use it. “We’ve had requests from the academic community and also from national bodies,” says Dr Bates.

One project under discussion would look at patient record data and compare this with data from the Met Office, to see the impact of weather on pressures on maternity services in hospital and at home. Another project, being run by the London School of Hygiene and Tropical Medicine, is considering whether text reminders are an effective way of getting at-risk under-65s to attend flu vaccination appointments.

“It’s still very new, but the number of requests is growing rapidly,” says Dr Bates. “We have the data from 5m records at the moment – about a sixth of the records in SystmOne. But more and more providers are opting in to the project and allowing us to use anonymised data all the time.”

Yes, they need reassurance on the information governance – which includes permission from the National Research Ethics Committee and the Health Research Authority’s confidentiality advisory group. “But as soon as we explain to practices how useful a resource this is, and when they are reassured about governance and we explain that it’s not commercial, and we’re not going to sell the data, then they are keen to sign up.”

ResearchOne gives the opportunity to do important research that would otherwise be prohibitively costly, he says. More than that, he adds: “What ResearchOne does is combine data mining with biology and clinical knowledge - that’s a powerful combination, and it has huge potential.”

Professor John Young and Dr Martin McShane on a long term condition

Frailty is an enigma: it surrounds us in health and social care and we recognise it when we see it – but it is, paradoxically, invisible because we do not regard it as a diagnosis or formally record it. Indeed, we tend to use the term as an adjective – “the frail elderly” – rather than considering it more properly as an abnormal health state, like a long term condition.This is important because frailty absolutely behaves like a long term condition: it has a high prevalence; it is progressive; it impacts adversely on life experience; it is expensive; and it has acute exacerbations.

Unfortunately, because we have not been in a position to manage frailty as a long term condition, the acute exacerbations are the predominant presentation of frailty into health and social care.

The nature of frailty exacerbations is typically dramatic and hyperacute with sudden immobility, delirium (acute confusion) and falls. So people with frailty tend to present late and in crisis and there is an over-reliance on secondary care responses. For example, there are over 650,000 fall presentations for people aged over 65 to hospital emergency departments each year.

The electronic frailty index (eFI) has the potential to address these issues and relocate frailty in the guise of a long term condition in primary care. The eFI is important because it is evidence-based (underpinned by the internationally validated cumulative deficit model of frailty), it is simple (entirely derived from the existing health care record) and it will separate older people into categories of fitness. Indeed, it might be better regarded as a fitness index.

At individual or practice level, the primary care team will be able to identify people with various grades of frailty. This opens up frailty management to strategies used for other long term conditions: case finding; individualised care planning; and multidisciplinary team reviews.

The care content should be around promoting exercise, medication reviews, social networks, home adaptations, carer support and nutritional support. These care components, targeted at modifiable aspects of frailty, are currently delivered sporadically rather than systematically. We will move from a reactive approach to frailty to proactive primary care-based service model.

Professor John Young is national clinical director, integration and frail elderly; Dr Martin McShane is director, domain 2 (long term conditions), NHS England

Primary care supplement: The big sleep problem