A challenge to all health systems is how best to manage “high need, high cost patients,” a term developed by the Commonwealth Fund. Its president, David Blumenthal, discussed the best response in last week’s annual health lecture of Imperial College’s Institute of Global Health Innovation.

The first reason for caring about such patients is that they could be our family members or even us one day. But the reason that means they get special attention is that they account for high costs: in the US 5% of patients account for 49% of costs, and 1% of patients for 22% of costs. Blumenthal couldn’t find directly comparable figures for Britain, but in one British teaching hospital the sickest 3% of patients accounted for 45% of costs. It is, he believes, a pattern seen in all industrialised countries.

These patients are, Blumenthal emphasised, a heterogenous group, but they are likely to be over 65; have multiple conditions, functional disability (not being able to dress, feed, or bathe themselves), and behavioural problems; “face material hardship” (a polite way of saying poor); and be near the end of life. The elderly are increasing as a proportion across the industrialised world, and those over 65 will increase from 18% to 24% between 2016 and 2040 in Britain. (I’ll be one of them.)

Having multiple conditions and functional disabilities is a strong driver of cost. In the US the average annual cost of a patient is $4800, and there are 231 million of them. There are 79 million patients with at least three conditions, and they cost on average $7526 a year, but the 11.8 million who in addition have one functional disability cost $21 021. (I think here of my mother, who has no physical health problems and takes no drugs, but because of her dementia needs residential and nursing care at a cost of £54 000 ($70 000) a year.)

These high need, high cost patients present quality and safety problems as they are more likely than average patients to be over medicalised, have more tests, and suffer more adverse effects.

Health systems need to improve how they manage these patients, but before he described some programmes, Blumenthal took a diversion to describe the difference between macrosystems and microsystems in healthcare. Both influence the performance of the health system. Microsystems are systems that interact directly with patients. Clinicians often control microsystems, and they are the sharp end of healthcare. Macrosystems are things like government regulation, payment systems, hospitals, and accountable care organisations. The NHS and clinical commissioning groups are macrosystems.

We have, said Blumenthal, lots of evidence of microsystems that work: he quoted the examples of primary care, reminder systems, and computerised decision support. Unfortunately, macrosystems often get in the way of implementing microsystems. “We have,” he said, “failed to create macrosystems that support initiatives that work at the microsystem level. We must make it easier to do the right thing.”

We do, he continued, know something about what works for high need, high cost patients at the microsystem level. Targeted interventions and segmentation of patients are important, recognising that the patients are a heterogenous group. Close coordination of those providing care is essential, and Blumenthal emphasised the importance of face to face meetings. Electronic communication is not adequate, but strong information technology, particularly around risk prediction and continuing surveillance, is helpful. Also essential is patient and caregiver engagement.

Blumenthal described several programmes that have been developed using these features and which have managed to improve care and reduce costs. One feature of several is that they have been able to spend money on transport, housing, and food—things that it’s often not possible to pay for within health budgets. “The key,” said somebody in the audience, “is to give patients what they want,” not, the implication was, what the health system thinks they need or is willing to provide.

The programmes Blumenthal described have managed only tens of thousands of patients when tens of millions need improved care. Macrosystems are getting in the way. American doctors are still mainly paid by fee for service, which means that reducing the services they provide would lead to reduced income. In England hospitals are paid for admitting patients, not keeping them out of hospital. Blumenthal advocated payment systems that pay for value not activity. The biggest macrosystem block, however, may be “culture,” getting health workers to work in new ways.

Blumenthal—who was a professor of medicine at Harvard and a manager within Partners Healthcare System, which manages the big hospitals in Boston—described, however, how the whole system, which has been traditional and slow to change, “pivoted in months” once it became an accountable care organisation and responsible for the costs of all of the patients it managed. The leaders of the organisation became very interested in high need, high cost patients.

The Commonwealth Fund has formed an international working group to scan international programmes for high need, high cost patients and make recommendations to health ministers. Managing these patients effectively and efficiently will be crucial in keeping health systems affordable.

Richard Smith was the editor of The BMJ until 2004. 

Competing interest: RS is an unpaid adjunct professor in the Institute of Global Health Innovation.