Foundation

When the respected economist Kenneth Arrow completed his paper on the challenge of managing uncertainty in health care, he would have had no expectation that it would set the course for the creation of a whole new academic discipline.

Health economics owes its heritage to that paper, published in 1963, which set out a range of fundamental questions around the social contract between individuals and healthcare systems — which are still pertinent today.

The related issues of agency and insurance are key concepts in the paper, called ‘Uncertainty and the Welfare Economics of Medical Care’, explains Professor Adrian Towse, Director Emeritus of the Office of Health Economics and its former director for 25 years.

“Rather like employing a builder to do your kitchen extension, when it comes to needing health care you face a lot of uncertainty. You don’t know how competent the key people carrying out the work on your behalf are or whether they will make the right decisions – so you have this problem of agency — they have knowledge that you don’t have.

“You also don’t know when you’re going to get sick or what you’re going to get wrong with you, so that’s the other big uncertainty, but you do know that you want someone to look after you when it happens. That’s why we need some form of insurance to deal with that uncertainty, in just the same way we’d have insurance to cover our kitchen gadgets.

“That insurance is what the NHS provides, as a publicly funded payer. In other systems that insurance risk is covered by other means. The critical point that Arrow addressed was to what extent should that uncertainty be fully covered?”

At a basic level, Arrow’s paper asks whether insurers should provide for any procedure or intervention that works, along the lines of “if I get sick and there’s something the doctor (my agent and the insurer’s agent) says can help me and it costs £10 million, should my insurer, in our case the NHS, pay for it?”

The answer is “no”, argues Arrow, because the system only works if the services and treatments provided have health gains that exceed the costs of providing them. The question then becomes ‘how do you measure the health gains?’

“You’ve got this problem of insurance and agency,” explains Towse, “because at the point you pay your taxes or pay for your insurance, you have one set of priorities, but when you get sick, you have a different set, along the lines of ‘I want everything that can help me’. Doctors, as agents both for the patient and for the system, are stuck in the middle. So, what Arrow says is that the system should provide, and you should pay for, through your insurance premiums or your taxes depending on whether it’s a public or private system, is enough — sufficient to provide procedures where the health benefits exceed the cost of providing them.”

Arrow’s paper is credited with influencing US president Lyndon Johnson’s decision to create the US Medicare and Medicaid health insurance systems for retired people and poorer people in mid-1960s. But more than that, it is regarded as a scene setter for the discipline, Towse argues. “Because it’s still relevant today, it still has currency — we’re still trying to answer the questions that it posed.”

During the 1960s, when Arrow wrote his paper, health care presented a lot of economic unknowns. Concepts such as disease burden or hospital costs were not fully understood. Nobody had calculated the full cost of treatments or the benefits of tackling particular diseases.

A young research economist called Marty Feldstein began the process of challenging this and, would as a result, go on to contribute significantly to the creation of the discipline also. Feldstein got his Oxford PhD in 1967 with an economic doctoral thesis on using econometric methods to measure NHS hospital efficiency and the potential for reducing hospital costs.

Both Arrow and Feldstein are recognised as giants in the field of economics, who chose to dabble in the field of health economics during the 1960s and thereby “kick started it all,” says Towse.

“They were both examining health care and coming up with very important but complementary strands of economic thinking and application,” says Towse.  “The two together are scene-setting for the development of health economics in subsequent decades.”

In some ways, it is no surprise that health care became a focus for economic interrogation during this period. In the UK, the unanticipated high costs of running the NHS, which had been created in 1948 with a promise to provide free health care to all at the point of need, began to attract political attention in the 1950s. As part of a drive to cut public spending, including the cost of the NHS, the Conservative government set up a review to examine healthcare costs, called the Guillebaud Committee[i], advised by LSE economist Brian Abel-Smith (who was a member of OHE’s first advisory group) and LSE social policy academic Richard Titmuss. The Committee found that rather than being too expensive, the NHS had been underfunded as its share of the overall GDP had fallen from 3.75% to 3.25% during the period of review (1948-54).

This backdrop of political interest in the cost of health care – and a desire to understand and measure its benefits — led to the growth of health economics as a discipline in the 1960s.

[i] https://blogs.lse.ac.uk/lsehistory/2018/05/02/the-guillebaud-report-the-nhs-and-lse

The desire to find a way to measure the benefits of health treatments or to somehow pin down their value would go on to dominate health economic research and activity during the next decade.  It would result in the creation of the ‘QALY’ — or ‘quality adjusted life year’ — which has now become a key tool for the economic evaluation of medical interventions.

Although the UK is regarded as the de-facto global leader in health economics[i], the creation of the QALY arose from parallel activity in several countries.

“The origins of the QALY cannot be tracked in a straight line,” says Towse. “As always, with these sorts of great ideas, there are parallel strands of activity and, in this case, there were four of them — in the US, Canada, the UK and Sweden. The US led the way (Bush, J.W; Fanshel, S; Chen, M (1972); Zeckhauser, R; Shephard, D (1976) and, most famously, Weinstein, M. C and Stason, W. B (1977).

“Professor George Torrance at McMaster University in Canada was also developing an approach, and in the UK work by Rachel Rosser, Head of Psychiatry at UCL, and the economist Professor Alan Williams at the University of York led firstly to a quality-of-life index and then to the QALY.

“Other key economists at York working alongside Alan Williams were Alan Maynard and Tony Culyer. Later, Mike Drummond joined them from the University of Birmingham. Finally, Bengt Johnson, a Swedish economist, visited Alan Williams and switched his PhD to the application of cost-benefit analysis in health, publishing in 1976. Bengt then set up the Swedish Institute of Health Economics (IHE) in 1979.

“Many people were working on this question of how you can identify indexes of benefit across completely different disease areas to get some sort of generic measure or descriptor of benefits. The QALY was the result.”

A final piece of the health economics jigsaw appeared in this decade with the seminal contribution of Michael Grossman’s ‘The Demand for Health: A Theoretical and Empirical Investigation.’ This book by the (now) professor emeritus at the City University of New York Graduate Center (CUNY), was the first to model the demand for health services. Health was seen as an (inherited) durable capital good which depreciates, interpreted as the natural deterioration of health over time. Health care is an investment in health. This led to a whole stream of thinking and literature on the determinants of the health status of populations, only one if which was medical care. Uncertainty was added to the Grossman model by Chuck Phelps, bringing us back full circle to Arrow’s examination of the case for health insurance.

[i] See other article in series: UK influence on health economics

If we characterise the 1960s as driven by theoretical thinking around the social contract between individuals and healthcare systems and the practicalities of analysing data to measure costs and efficiency; and the 1970s as concerned with finding practical ways to measure healthcare benefits and to model the demand for health care, the 1980s would see the emergence of cost-effectiveness and, ultimately, its use as an efficient tool for rationing. ‘How can you use a health benefit measure to decide what treatments should be provided?’ became the key question.

Here, the UK led the field.  Professor Alan Williams, University of York, wrote a paper on coronary bypass grafting[i] which is now widely regarded as a seminal publication, reporting the cost-per-QALY of this intervention.

“The application of the QALY in terms of its use in public policy, to think about answering that Kenneth Arrow question about what we should we be paying for and not paying for, was very much driven by UK economists,” says Towse.

“This was fuelled by successive governments who wanted an appropriate way of managing spending on the NHS, other than having regular political bust-ups about how much to spend on health care and newspaper stories about treatments not being available. It was about finding a more rational way of deciding how the NHS budget should be spent, and therefore also having a more rational discussion about whether to spend more (or less) on the NHS.”

In practical terms, this meant continuing to improve a generic measure of benefit which could be used to compared treatments — a QALY— with the EuroQol group being set up in 1987, but, importantly, moving to a tool that also included costs. This enabled decisions as to which ones delivered better value for money and which, therefore, should be prioritised for NHS funding — the cost per QALY.

It would, however, be another decade before health economists would start to see advantages of collecting quality of life and cost evidence alongside clinical trials to enable them to carry out a benefit analysis using this ‘cost per QALY’ approach.

Any summary of this decade would not be complete without referring to the RAND health insurance experiment — a randomized study in the US where different groups of people were put on to different types of health insurance plans to see if their demand for health care changed, depending on how much they had to pay towards the costs of treatment.

Researchers were keen to know whether payers would be less inclined to seek a doctor’s advice for trivial complaints. The study had two very important results. First, researchers found that making patients pay affected their utilisation of services — applying charges reduced their demand for health care. But the second important finding was that users of services were not able to distinguish between important health problems and less important ones.

“This brings us right back to the Arrow’s “agency problem”,” says Towse. “The reason we go to doctor is because we don’t have medical knowledge and we’re not good at self-diagnosis.  The RAND study showed that if you charge people, their utilisation of services will go down, but they will not just stop coming with trivial things, they’ll stop coming with serious things, because they don’t know the difference.”

Findings from the eight-year study, which ended in 1981 made a very significant contribution towards health economic research, one that is still referenced today in debates about top-up payments in health care or to applying hotel costs in hospitals.

[i] Economics of Coronary artery bypass grafting’. BMJ. 1985

Work continued apace in the 1990s on generating ‘cost per QALY’ evidence. In 1996, a breakthrough research paper[i] demonstrated just how useful ‘cost per QALY’ evidence could be. The paper, by Bengt Jonsson and colleagues at the Stockholm School of Economics, reported findings from a survival study which showed that a cholesterol lowering statin was not only effective at saving lives, but it was also highly cost effective — by reducing the number of heart attacks it saved lives, gained QALYs and reduced hospital costs.

This approach to thinking about costs and benefit started to shape decision-making and health policy. Various public bodies were set up to oversee the application of this methodology to practical decision making on new treatments – particularly in assessing the value of new medicines, as cost-effectiveness became an important component of health technology assessment (HTA).

Australia was first. It established a Pharmaceutical Benefits Advisory Committee which issued its first guidelines in 1990. In 1993, in a global first, these became mandatory for consideration of new medicine approvals.

Close behind was Canada. It set up an oversight organisation in 1989, now called the Canadian Agency for Drugs and Technologies in Health (CADTH), which first issued guidelines on conducting economic evaluations in 1994. The UK followed, establishing NICE in 1999.

These developments were accompanied by a parallel strand of work concerned with how best to model uncertainty and present it in a way that decision makers could understand cost-effectiveness. Indeed, this was the subject of much intellectual debate both in the UK and the Netherlands, a country that had also established an HTA body. Andrew Briggs at Brunel University developed statistical methods for measuring uncertainty in cost-effectiveness analysis and Ben van Hout, now at the University of Sheffield, but who spent his early career in the Netherlands, introduced the cost-effectiveness acceptability curve as a way to present the results of uncertainty.

[i] Cost effectiveness of cholesterol lowering. The European Heart Journal. 1996. Jonsson et al.

By the early 2000s there had been a notable downturn in the number of breakthrough drug treatments launched to market. There was concern that the pharmaceutical industry was losing its scientific edge — instead of competing to develop breakthrough new innovations in new disease areas, they were focused on launching so called “me-too” new products in the same class of medicines where competitors had already launched.

The industry’s reliance on developing ‘me too’ products was understandable – they meant lower R+D costs but not necessarily lower profits.  And, of course, the nature of the emerging model of academic – industry collaboration meant that scientific breakthroughs in our understanding of disease led to companies competing to develop drugs, such that several drugs in a new therapy area could come to market within months of each other. However, it became clear that the pharmaceutical marketplace needed fixing to incentivise breakthrough innovation.

“The response of policymakers, particularly the regulatory bodies, was to try and find ways of increasing the returns from breakthroughs by prioritizing them through the regulatory system,” says Towse.

The European Medicines Agency (EMA)n agency of the European Union charged with evaluating and supervising medicinal products, launched two initiatives to address the problem — the EMA Conditional Marketing Authorisation (2006) and PRIME (2016).

Similarly, the US equivalent, the Food and Drug Administration (FDA) launched a number of fast track progammes for critical or breakthrough therapies from 1997 to 2007.

Work by health economists Fabio Pammolli and Laura Magazzini[i] explored the issue further at the end of the decade and found that the pipeline problem of low numbers of new drugs wasn’t driven by “me-toos”, but the reverse. R+D had shifted to much more difficult therapy areas where breakthroughs were needed, the science was more difficult, and the failure rates were, therefore, higher.

“They found fewer new drugs were being launched but it wasn’t due to lower underlying productivity or that the industry was finding it more difficult to find a new treatment for heart disease, it was that we already had treatments for heart disease and the industry was now shifting into oncology, which is a more challenging area,” says Towse. “So, success rates were going down, but it wasn’t, it wasn’t because the industry was less successful in each individual disease area, it was because the disease priorities had changed.”

[i] Pammolli, F., Magazzini, L. and Riccaboni, M. (2011) The productivity crisis in pharmaceutical R&D. Nature Reviews Drug Discovery. 10, 428-438

Health economists began to focus less on cost effectiveness per se, but more on its link to affordability and the impact on healthcare budgets, and on the challenges posed by curative treatments – with large but uncertain benefits.  They began to consider how payment mechanisms needed to adapt and also how to use competition as a useful lever for lower prices.

This shift arose because, during the teens, science started to deliver — a rising number of breakthrough drugs were launched to market spanning cancer treatments – including immunotherapy for multiple indications, orphan drugs, and curative treatments, including gene therapies.

Pammolli and colleagues[i] re-examined pharmaceutical output again, publishing a paper in 2019 which showed that investment in science and additional incentives had paid off and there had been an upsurge of R&D productivity by the industry.

Not only that, but pharma was focusing on the high-risk/high-reward areas. The science was next generation, based on novel mechanisms and delivering advanced therapy medicinal products (ATMPs) for use in genes, tissues or cells.

Governments began faced a new problem — it was less that pharmaceutical companies were ‘not producing enough of what we need’ but more they were ‘producing so much and how can we pay for it?’.

Breakthrough treatments delivering long-term benefits became a particular challenge. “If you are delivering very long-term benefits, which would be true with something like a gene therapy, you’re effectively producing a lifetime of benefit but if you’re paying for it upfront in one shot the sticker price is very high,” explains Towse.

Gene therapy launches didn’t “break the bank” because they were launched for very small patient groups, but budget challenges became more obvious with new classes of treatments for diseases affecting much larger patient group sizes.

Hepatitis C is a case in point. When Gilead launched Solvaldi[ii], an antiviral treatment for Hepatitis C to the US market in 2013, it proved to be “cost-effective but unaffordable” due to the high numbers of people affected. In the UK, it was rationed so only those at immediate risk of live failure were able to access it. In theory, the benchmark for being “cost-effective” at the margin has to be related to the budget.  In practice this was not happening because the budget impacts were so large.

Interestingly, the market found ways to solve these affordability challenges because the science was picked up by competing companies.   Says Towse: “Lots of competing products for Hepatitis C entered the market and prices started to come down. Now the NHS is looking to eradicate Hepatitis C. It’s asking manufacturers to for proposals to eradicate it in the UK…competition has meant that the price of treatment has come down a lot – well below the prices at which NICE regarded Sovaldi to be cost-effective.”

Health economists began to propose innovative payment systems. “If a gene therapy gives you benefits over 60 years, then it’s silly to try to pay for it all in the first year and then discover that you can’t afford it. It’s far more appropriate to have the health system paying a certain amount each year that the patient is still alive and still benefiting from that treatment. That was the thinking behind innovation around managing payment mechanisms.”   Phased payments also provided a mechanism to address uncertainty. If the treatments stopped working, the payments could stop.

Work also began revisiting the benefits of health care, with discussion of “Value Frameworks” exploring other elements of benefit that are not necessarily reflected in the QALY or in the impact on health system costs. This work is an ongoing health economics research agenda.

[i] The Endless Frontier? The Recent Upsurge of R&D Productivity in Pharmaceuticals. Pammolli, F et al. May 2019.

[ii] https://www.gilead.com/news-and-press/press-room/press-releases/2013/12/us-food-and-drug-administration-approves-gileads-sovaldi-sofosbuvir-for-the-treatment-of-chronic-hepatitis-c

The 2020s have been characterised by global health threats, notably the COVID-19 pandemic requiring rapid development of new vaccines and therapeutics, and the emerging challenge of Anti-Microbial Resistance, requiring new “Netflix” style subscription models.

Two research agendas have emerged. Firstly, the need to explore new payment models to meet these challenges. This includes, importantly, a renewed focus on mechanisms to pay for vaccines and treatments to be available in low- and middle-income countries. And, secondly, as health systems struggle with the continued impact of COVID-19 and with the impact that economic slowdown has had on their finances, a renewed focus on efficiency and on ensuring benefits exceed costs.