In case you hadn’t noticed, there’s a growing clamor for a global commitment to universal health coverage (UHC). You might have seen the recent special issue of the Lancet on “the struggle for UHC”. Inevitably, accompanying this clamor, there’s been a lot of wracking of brains on how to measure progress toward UHC. With the excitement of a new political agenda, there’s understandably a desire to carve out a new measurement agenda too. While not wanting dampen people’s enthusiasm for the UHC cause, I would like us to reflect whether on the measurement agenda we’re building enough on what’s been done before.
So what’s been done already on UHC measurement?
At the core of UHC is the idea that people should be able to get the health care they need without experiencing financial hardship as a result. This isn’t a new idea, of course. It underpins the British National Health Service (NHS), as well as other welfare states all around the OECD and beyond. It would be a little surprising if someone somewhere hadn’t thought before now of how to measure these countries’ success in achieving their shared UHC objective. And indeed they have.
One part of the literature, starting with Julian Le Grand’s famous 1978 Economica article, looks for evidence that people’s use of health care – relative to their need – depends on whether they are poor or well off. If it does, we can infer that the limited resources of the poor somehow left them receiving less care than they ought to have received given their needs, and given what better-off people with similar needs received. Such a situation goes counter to the idea of UHC, and is a violation of the UHC principle of equal treatment for equal need irrespective of ability to pay. Le Grand found that the British NHS failed to deliver on its commitment to UHC – the poor were using services more than the rich, but their higher use fell short of what would be expected given their higher needs and the level of use among the better off with similar needs.
Le Grand’s measurement ideas led a group of us in the late 1980s to start a project trying to compare countries in terms of how close they came to achieving ‘equal treatment for equal need irrespective of ability to pay’. We built on Le Grand’s ideas (here and here) to come up with some indices with which we could quantify inequity in a country, and then ranked a subset of the OECD countries including the US. We updated these results in 2000.
My co-conspirator, Eddy van Doorslaer, and a colleague of his recently extended the analysis to all the OECD countries. They found a large variation in the degree of inequity in the utilization of physician and hospital care across countries, but interestingly, as I reported in an earlier blog post, these differences weren’t systematically related to whether the country was strongly committed to UHC. Ann Mills, Di Mcintrye and colleagues have used these methods to look at equity in the use of care in several African countries, and Rachel Lu and colleagues have used them on selected Asian countries. Several papers have studied one or a few countries at a time – I’m probably missing studies but the countries studied include at least Argentina, Australia, Brazil, Chile, China, Finland, Guinea, the Ivory Coast, Mali, New Zealand, Senegal, Spain, Uruguay, and the US.
When the individuals being compared are all in equal need for a specific service – pregnant women, for example, in the case of antenatal care – there’s no need to control for differences across income groups in need, and the exercise becomes one of simply looking across income groups for systematic differences in use of care. Inequity in this case can be measured simply by the ‘concentration index’. Several studies of inequalities across ‘wealth’ quintiles have been done using data from the Demographic and Health Survey (DHS) and Multiple Indicator Cluster Survey (MICS), one particularly large study being the one just completed by the World Bank. Of course, such work gives us a very partial picture of ‘equal treatment for equal need’. It may well be the case that there is a high degree of equity among MCH interventions, and that it is the treatment of conditions requiring hospitalization that we see a high degree of inequity.
It’s important to be clear that these measures of inequity do not try to measure the average level of utilization. Rather they try to measure the degree of inequality in utilization (adjusting for need) across income or wealth groups. That, after all, is what UHC is all about: ensuring that needed services are available to everyone irrespective of their ability-to-pay. A lot of the methods and empirical work being bandied around right now in the UHC debate don’t get at this core idea at all.
What about financial protection?
Imagine an authoritarian but unequal country where everyone gets all the health care they need but everyone pays the same price for each procedure. From the point of view of the principle of ‘equal treatment for equal need, irrespective of ability-to-pay’, this would look a very equitable country. But it might well be one in which the poor face financial hardship as a result of using health care.
A supplementary measure is required therefore to capture financial protection. In a 2003 paper, Eddy van Doorslaer and I suggested a couple of measures. One captures the incidence of ‘catastrophic’ out-of-pocket spending – spending that exceeds a specific percentage of income or consumption. This doesn’t get directly at the question of hardship, however. So we proposed a second set of measures that capture the ‘impoverishment’ associated with out-of-pocket payments – one captures whether a household is pushed below the poverty line because of their out-of-pocket spending, and another captures the degree to which households end up below the poverty line because of out-of-pocket spending on health.
There have been a number of studies using these methods. Several WHO studies have reported the incidence of catastrophic spending in a large number of countries. A study by Eddy van Doorslaer and others looked at the incidence of catastrophic and impoverishing spending in 10 Asian territories. A World Bank study looked at the degree of impoverishment due to out-of-pocket spending in 10 countries in Eastern Europe and the Former Soviet Union. A number of studies explore how policies and institutions impact on the incidence of catastrophic and/or impoverishing health spending, either by comparing across countries, or by looking at the impacts of reforms on the incidence of catastrophic and/or impoverishing spending in specific countries; these include China, Estonia, India, Mexico, Thailand, Uganda, Vietnam, and Zambia.
Making UHC measurement tools accessible to everyone
These methods aren’t that complicated to implement, but they aren’t that easy either. To make them more accessible, three colleagues and I have written a free ‘how-to’ handbook (the ‘blue book’), complete with worked examples and editable computer code for Stata. Customizable ‘do’ files are available for download as well.
Not everyone likes or feels confident playing with Stata code. For these researchers, we have developed a free software program known as ADePT, which ‘cans’ most of the methods in the ‘blue book’, including the measures of inequity and financial protection. In the former, users load their dataset, specify what their utilization variables are, what their need and non-need variables are, and hit ‘go’, whereupon ADePT will do the relevant computations and spit out the results in nicely formatted tables in an Excel file. In the financial protection module, users specify their health spending variable, the variable capturing non-medical spending, and the poverty line – ADePT takes care of the rest. There’s also a free manual for the ADePT health module that walks the reader though the concepts and methods in plain English, and shows her how to get accurate results quickly using ADePT. It was certainly our intention – and Kara Hanson and Di McIntyre said as much in their remarks on the cover – that ADePT should help countries assess whether their UHC reforms are getting them toward the ultimate goal – ensuring that everyone can get the care they need without experiencing financial hardship as a result.
The ideas keep coming
Like all flourishing research programs, this one is still in development. A paper published just this year by Ellen Van de Poel and others in the Journal of Health Economics extends the ‘old’ analysis by allowing for the possibility that there are differences across income groups in the way that people with unequal needs are treated. There may be a more favorable relationship among the better off, and ignoring this results in an underestimate of the degree to which the better off receive favorable treatment given their needs. The magnitude isn’t always that large, but the bias is statistically significant in several countries. Gabriela Flores and Owen O’Donnell have recently offered a new way of measuring financial protection that is forward-looking and captures ex ante exposure to risk. These and other improvements in methods will likely be added to the ADePT software in due course.
My sense is that we already have a rich toolkit of methods that get at both how far people in equal need get treated the same irrespective of their ability-to-pay, and the degree to this goal is achieved without people suffering financially through the use of needed health care. Moreover, these methods are accessible, even to researchers without extensive statistical training and knowledge of statistical software, and are being extensively used, not least to track progress toward UHC. All in all, this a much better place to start than a blank sheet of paper!
Join the Conversation