Syndicate content

We just learned a whole lot more about achieving Universal Health Coverage

Adam Wagstaff's picture

Subsidized health insurance is unlikely to lead to Universal Health Coverage (UHC); insurance coverage doesn’t always improve financial protection and when it does, doesn’t necessarily eliminate financial protection concerns; and tackling provider incentives may be just as – if not more – important in the UHC agenda as demand-side initiatives. These are the three big and somewhat counterintuitive conclusions of the Health Equity and Financial Protection in Asia (HEFPA) research project that I jointly coordinated with Eddy van Doorslaer and Owen O’Donnell.

As we all now know, UHC is all about ensuring that everyone – irrespective of their ability to pay – can access the health services they need without suffering undue financial hardship in the process. The HEFPA project set out to explore the effectiveness of a number of UHC strategies in a region of the world that has seen a lot of UHC initiatives: East Asia. The project pooled the skills of researchers from six Asian countries (Cambodia, China, Indonesia, the Philippines, Thailand and Vietnam), several European universities and the World Bank.

We have you covered – or do we?

The lever that policymakers have tended to reach for first in their quest to achieve UHC is the financial protection scheme. In many countries, this has meant bringing people not covered by a health insurance scheme into an existing scheme, or creating a new scheme for these hitherto ‘uncovered’ groups. The poor often have their contributions paid by the tax-payer, while families above the poverty line are eligible for partial subsidies if they enroll voluntarily. The problem is that in many countries they don’t. Coverage rates among these families is often very low, leading to a ‘missing middle’ in coverage: poor families are covered at the tax-payer’s expense, and the better off are covered by schemes for formal-sector workers or by schemes for civil servants, but families in the middle remain uncovered.

One of the questions the HEFPA project asked was: How far would coverage rates rise if those in the missing middle were given fuller information about the scheme, and/or given a more generous subsidy? To answer this, the HEFPA teams in the Philippines and Vietnam – two countries where the missing middle problem is most pronounced – set up randomized control trials (RCTs).

In the Philippines a ‘treatment’ group was offered an information package (including enrollment forms) and a 50% subsidy via a voucher, and a control group was given neither. In Vietnam one treatment group was given information, another was given a 25% subsidy, and a third was given both. In neither country, did the treatment groups come anywhere close to 100% enrollment: in the Philippines, the intervention increased enrollment from 10% to just 15% around a year later; in Vietnam the most effective intervention increased enrollment from 6% to just 7% 12 months later.

A neat feature of the Philippines study was the sub-experiment implemented towards the end of the 12 months. The team tracked enrollment in the administrative data and could see before its end that the experiment was having a very limited impact. So the team decided to do an ‘experiment within an experiment’. Ten months into the project, it divided the ‘non-compliers’ (the households who had been given the subsidy voucher but hadn’t used it) randomly into two groups. One group was given the information package all over again. The other got a lot of hand-holding in the enrolment process: during the endline survey, the enumerator offered to help the family complete its application form, took the completed form and the family’s subsidy voucher and payment to the health insurance agency, and had the ID card mailed to the family. This intervention worked much better: it got the enrollment rate up to 40%, compared to 3% in the non-complier control group. But even this much more costly intervention left 60% of families choosing not to enroll – hardly a reassuring tale for policymakers planning to achieve 100% coverage through subsidized voluntary enrolment.

We have you covered – now what?

In fact, the countries that have successfully covered large sections of the population who are not civil servants or formal sector workers have mostly done so by financing coverage through general government revenues, or through donor funding, as in Cambodia where ‘health equity funds’ have been set up to pay for free care at (some) government facilities, and vouchers have been used to give pregnant women free maternal services. (In Asia the big counterexample is China whose government subsidizes the insurance scheme for rural residents but leaves people free to make the decision whether to enroll – most do.)

For these countries and subpopulations, the interesting question is not how to cover people but rather whether the coverage works. Does it push the country towards the twin goals of UHC? Does it raise use of needed services including among the poor? And does it reduce the financial hardship associated with usage of health services?

Here rather than implement RCTs, the HEFPA teams used quasi-experimental methods, taking advantage of quirks in the implementation of the coverage expansion programs such as a phased rollout across the country, or finding a group whose coverage wasn’t affected by the rollout.

First the good news: the teams found extensive evidence that when insurance coverage was acquired utilization increased, at least among some groups. This was seen in Indonesia and Thailand, as well as in Cambodia among poor women who were given vouchers to cover the cost of maternal care. In China, too, more generous coverage led to higher utilization.

Now the less good news: the HEFPA teams found mixed evidence on the degree to which coverage limited how much people paid for health services. In Indonesia, there was no evidence of any negative impact on out-of-pocket spending; in fact, among urban residents just the opposite. In China, the project confirmed the finding of some of the previous studies on the topic, namely that China’s rural health insurance scheme has not improved financial protection. More encouraging were HEFPA’s findings in Cambodia, where health equity funds reduced out-of-pocket spending by 26% (by even more among the poor), and in Thailand where the so-called universal coverage (UC) scheme reduced out-of-pocket spending on average by one-third, the probability of a family spending more than 10% of the household budget on health care (‘catastrophic’ spending) fell by 35%, and large out-of-pocket expenditures (expenditures at the 95th percentile) dropped by one-half.

These findings from Cambodia and Thailand are encouraging in that they show that the schemes did achieve a reduction in out-of-pocket spending. But it’s important to keep in mind that neither eliminated out-of-pocket spending; nor is it clear that they even reduced out-of-pocket payments to a level that eliminated financial protection concerns. In Cambodia, in areas acquiring a health equity fund, 44% of families still incurred out-of-pocket spending after the funds arrived, 4% still had health-related debt, and out-of-pocket spending as a share of total consumption was still 6%; these statistics reflect the fact that at the time the funds subsidized care only at public hospitals.  Thailand achieved (approximately) 33% reductions in out-of-pocket spending as a share of consumption and in the risk of catastrophic spending. Thailand didn’t achieve 100% reductions. Among the UC scheme target subpopulation, even after the scheme was rolled out, 67% still reported out-of-pocket spending, the share of consumption absorbed by out-of-pocket health spending was 2%, and catastrophic spending was 4.5%.

It’s not clear then that either country can be said to have reduced out-of-pocket payments to a level where people can use health services without suffering undue financial hardship. None of this is to belittle the achievements of the Cambodian and Thai initiatives, or to say that the government and donor funding involved wasn’t money well spent. Rather we need to be clear that a UHC initiative can be successful in terms of pushing a country toward the goals of UHC, without necessarily getting it all the way there. Which prompts the question:  What complementary levers might policymakers reach for to help their country get further down the road to UHC?

“I want to go into medicine and help people who can pay out-of-pocket”

Out-of-pocket payments are a cost to a family, but a source of income to a health provider. Such payments persist even after health insurance coverage expansions – or perhaps even partly because of health insurance coverage expansions – because providers rely on them for their income. Where providers are paid fee-for-service, there’s a strong temptation to focus on treating more patients, doing more tests, prescribing more – and more expensive – drugs, and so on. Shifting from fee-for-service toward payment methods such as capitation and salaries, and combining these with incentives for delivering good quality care, may be a more effective approach to reducing out-of-pocket spending. It may also help curb unnecessary care, thus helping a country get toward the first of the UHC twin goals – making sure that everyone gets the care they need.

To explore this idea, the HEFPA project set up yet more RCTs, this time in China – the country in E Asia where provider incentives are most skewed towards the delivery of costly and unnecessary care. In two provinces – Shandong and Ningxia – the HEFPA team helped local government officials shift from fee-for-service to capitation, but also randomly assigned some township health centers to a payment regime where facilities earned points according to the quality of the care they delivered. Given the problem of overprescribing in China, many of the indicators focused on prescribing patterns – use of antibiotics, intravenous drugs, steroids, etc. The points were used to calculate how much of a facility’s capitation budget that was withheld at the start of the monitoring period would be ‘returned’ to it to at the end of the monitoring period. In the case of Shandong, performance was compared to pre-announced targets, and the maximum a facility could earn was 100% of its capitation budget. In Ningxia, a facility’s performance was compared to average performance in the county, and above-average performers in effect earned a supplement to their capitation budget.

In Ningxia, pay-for-performance (P4P) led to improvements in prescribing behavior (e.g. fewer antibiotics, and fewer injected antibiotics). In Shandong, P4P improved the quality of care in the first of the two study counties, but not in the second. The reason for the difference is linked to the fact that payments in the Shandong experiment were based on performance relative to targets. By the time the study started most facilities in the second county had already achieved their targets; by contrast, those in the first had not and thus had an incentive to continue to improve their prescribing quality indicators.

On the second of the UHC twin goals – financial protection – neither experiment produced the hoped-for result; only in village posts in Ningxia did P4P reduce the amount that a patient paid out-of-pocket during a visit, and even then the reduction was just 3%.

Where does this leave us?

So the HEFPA project has given us some important insights into achieving UHC. Subsidized health insurance doesn’t look like it’s the answer to UHC’s ‘missing middle’ problem. In any case, expanding insurance coverage doesn’t seem to be the silver bullet many seem to think it is: insurance coverage doesn’t always improve financial protection, and when it does, doesn’t necessarily eliminate financial protection concerns. That points us toward grappling with provider incentives, and the conclusion that addressing them may be just as – if not more – important in the UHC agenda than working on demand-side interventions.

But the project also makes clear that we also have some work still to do. Yes, P4P holds some promise as a potential UHC policy instrument. But the China HEFPA studies suggest caution is warranted. We haven’t yet got the formula for a P4P scheme (for rural China, at least) that will help us toward both of the UHC goals. So like all good research projects, HEFPA generated some important actionable insights but also highlighted areas where more research is needed.



Submitted by Bakhuti Shengelia on

Excellent summary of the research evidence. It is so true that policy makers limit themselves to only one "control knob" in their afforts to achieve universal coverage, leaving out issues related to supply-side readiness, availability of human resources to delivery care, quality, etc. Policy makers need to understand that UHC does not mean an universal enrollment into a risk pool. Perhaps promoting a new measure of UHC which takes into account supply-side dimensions more prominently will contribute to the shift of focus from "universal enrollement" to "universal coverage".

Thanks for the nice comments, Bakhuti. I agree with you that when we're monitoring progress toward UHC we shouldn't be looking at the fraction of the population covered by a financial protection scheme. That's why colleagues from the World Bank and the World Health Organization and I proposed tracking (a) the rates of (effective, ideally) coverage among the population and the poorest for certain key indicators, and (b) a measure of financial protection like catastrophic payments or impoverishment. That allows for the possibility that a country may get everyone into a financial protection scheme without getting close to UHC. I'd be interested to hear your thoughts on this idea. Cheers, Adam

Submitted by Joe Kutzin on

Congratulations on this excellent summary and to the whole HEFPA team for this work. The findings give weight to the broad conclusion - perhaps overused and overly simple - that there are no magic bullets in health reform. But of course the work goes far beyond this, helping to consolidate some important messages on financing for UHC, for example

- getting people into an insurance scheme is not enough, you have to think about all aspects of financing (arrangements for collection, pooling, purchasing and benefit design) in a coherent way
- voluntary prepayment, even when heavily subsidized, is not a viable pathway to sustained coverage expansion. Adverse selection is part of the "physics" of volutary health insurance (VHI), and even putting the word "community" in the label is not a sufficient measure to mitigate this

- just throwing more public money at the system is similarly not sufficient to sustain real coverage improvement (i.e. better financial access and financial protection). Have to think about using the money effectively to both manage system expenditure growth and ensure that people are protected against financial risk. Here the studies show well the contrast between the reforms in Thailand and China: both greatly expanded population affiliation to a scheme, but with very different results in terms of financial protection due to critical differences in the combination of provider payment methods and benefit design.

What next? Many further areas for research, of course, but at the moment I'm more concerned with something else: how to communicate these findings, align with other similar conclusions, and then do our best in a health financing policy "harm reduction" effort? Of course we don't know everything that needs to be done, but at the same time we do know many things about what NOT to do, as well as the kinds of issues that countries need to be thinking about as they proceed with reforms. Yet we see many countries repeating initiatives that both theory and practice tell us will fail (e.g. expecting high levels of voluntary enrollment by the informal sector, or failure to focus adequately on strengthening provider payment arrangements). So how can we do a better job in dissuading countries (and those advising them!) from these pitfalls? We have the evidence, but something in our dissemination/communication seems to be lacking. Will keep trying nevertheless...

Congratulations again to you and the HEFPA team on this important contribution.


Thanks, Joe, for the kind comments on the post and the project. I agree that dissemination is key, but with the voluntary health insurance issue, I'm wondering whether part of the challenge is that people like you haven't up to now had clean easy-to-understand results to point to. It's one thing to talk about principles like adverse selection; it's another to point to real-life examples of experiments that tried out the idea but the idea didn't work. But I'm sure that's only part of the story! Cheers, Adam

Submitted by Vincent Okungu on

Great piece by Adam as always!
A few observations and questions:
(1) It is difficult to eliminate OOP spending; we can only limit such expenditures so the aim ideally is not elimination...but great if this could happen.
(2) OOP spending in Thailand has been on the downward trend from government funding (talk of proper targeting of public spending?)
(1) It is not crystal clear for what reasons coverage has failed to reduce concerns over financial protection in the countries studied? Is it about availability of services or aligning population health needs with available services? Is the benefits package too limited?
(2) How was P4P structured on the demand side or was it a supply- side issue?


Vincent, thanks for the nice remarks. On (1) I agree -- I think the emphasis is on making sure OOPs don't cause undue financial hardship, at the very minimim not keeping a family in poverty and not pushing families into poverty. On (2) yes Thailand's trend was downward even before UC, but the results suggest UC did put downard pressure on OOPs of the target population beyond the reduction that would have beene xpected on the basis of the downward trend. On your 2nd (1) I think the answers vary by country: in Cambodia, as the post says, the focus of the scheme was on fees in public hospitals; in China it's partly copayments and reimbursement ceilings, but even more important is the use of FFS to pay providers. On your 2nd (2) P4P was a provider payment scheme. Cheers, Adam

Submitted by Bistra Zheleva on

Fascinating report and findings. I have a question - why are people not enrolling? I have a fresh example from Vietnam where we encountered many cases of families supposedly not knowing about the health insurance. Clearly there are some inefficiencies in how the enrollment happens and how it's promoted but in the first interventions HEFPA did there people still did not enroll. Why is that? What is preventing them? Is it trust in the system? I'm sorry if there is an obvious answer and I don't see it but to me is baffling to have this resource at your fingertips and not use it, so there must be a reason I am not seeing...

Thanks, Bistra, for sharing your thoughts. The experiment was geared to lessening two barriers commonly claimed to be the barriers preventing higher enrollment in Vietnam: lack of information about the scheme; and the unaffordability of health insurance. Both showed up actually as reasons for non-enrollment in the baseline survey results when respondents were asked why the weren't insured. The experiment, I think, shows pretty convincingly that information isn't the issue. On the affordability, we tested only a 25% subsidy, and it's possible a larger subsidy would have produced a bigger effect. But frankly with such a low baseline level, even a more generous one would have done well to have got enrollment up to even 25%. The post doesn't talk about the issue of adverse selection but the paper does: we found larger effects among the less healthy, which is consistent with the fact that many in the baseline said they weren't enrolled because they were healthy and didn't neeed insurance. So like most voluntary schemes, this one looks like it suffers from adverse selection.  Cheers, Adam

Submitted by Owen on

Adam - this is great stuff and I would fully subscribe to the main messages. Deng Xiaoping famously likened Chinese reforms to "crossing the river by feeling the stones". Would you say that the stones you looked at are not worth stepping on, or possibly useful incremental steps to get to the other side? No one is crossing the UHC river overnight. Your reply can be as philosophical as my question!

Thanks, Owen. Great question. My sense is having people onto the big stone marked "covered by financial protection scheme" is essential to getting across the river to UHC, but that it's not just a short jump from there to the UHC bank. Steps marked "subsidized voluntary enrollment" may look like they lead to the "covered by financial protection scheme" stone, but the HEFPA studies cast doubt on that idea. By contrast, the steps marked "mandatory cover financed out of taxes" might appear to be hard to reach and a bit slippery, but with some creativity and hard work on tax reform, they may be more reachable and not as slippery as they look. Getting from the "covered by financial protection scheme" to the UHC side of the river means passing across some steps to do with benefit package breadth and depth, but also some steps to do with provider accountability and incentives, as well as steps to do with things like geographic resource allocation. I suspect it's worth watching others cross the river and learning. Turning back and starting over might not be as straightforward as it might appear—for example, it's one thing to introduce P4P in a setting where budgets and salaries are leading to activity levels that are too low; quite another when FFS has led you to have too much activity or the wrong mix. Cheers, Adam  

Submitted by Karl Waterson on

I think the ongoing crisis in West Africa with 1 million people needing food and water relief supplies and almost a billion USD economic losses just in 5 months demonstrates absence of catastrophic coverage for entire populations. Where were health policy makers that they did not establish core public health functions for prevention of contagion outbreak from spreading? What were health sector policy makers thinking that they did not stop the outbreak of Ebola in March?

This vast tragedy is evidence that indeed UHC is a low priority in developing countries. Improving health status of the population is the highest priority. UHC is not -- definitely Not -- the best way to improve health of poor people.

Watch panel four of the World Bank conference in April 2014. You will see the high-level panel casting serious doubt on the very notion of UHC. Neither prof. Larry Summers nor Mayor Bloomberg spoke up for UHC. They said many times that public health measures (reducing smoking, eliminating indoor air polution) and investment in education of girls are the highest priority . Did we learn anything about the merits of UHC from that panel?

Byinvesting in sanitation and water many more lives can be saved and disease burden lowered than by spending the same money on access tocare centers. The UHC campaign never mentions prevention, seemingly does not care about the daily carnage on roads. Or the huge setbacks from preventable infectious disease outbreaks (think Ebola in West Africa, cholera in Haiti, cholera in Ghana, cholera in any number of other places, avian flu in E Asia that already cost $3 billion in just 2 years). Preventing premature deaths and disease is demonstrably not the goal of UHC activists because if it were, they would advocate for tobacco taxes, environmental risk reduction like pollution of air, water, aflatoxin contaminated food and livestock, sugar, tobacco, indoor coal cooking. Itis these things that generate the disease burden on the poor. Not absence of UHC.

Before asking which poor country has UHC, you need to ask which poor country has good capacity for early detection and fast early control of spreading infectious diseases? Health sector policy makers are responsible for the huge damage done by public health policies in Guinea, Liberia, and Sierra Leone. perhaps this is the most important factor for achieving UHC there and in other poor countries.

Besides, if policy makers choose not to prevent disease, no developing country will be able to afford UHC . It is far more kind to preventdisease than to have policies that lead to people being ill and then offering them "universal health care". Prevention of disease has to be the utmost priority if poor people are to have better health in 2030. UHC favor patients today over investments (in sanitation, road safety etc) in the health of our children's generation. That is not fair toward our children. It is also a fiscal recklessness.

The health sector policy makers are responsible for unsafe roads, poluted water, microbes in food, spreading disease outbreaks, low tobacco taxes, sugared softdrinks everywhere in rural Africa, etc. They are responsible for policies that impact the health of the population. Unsafe water, dangerous roads, obesity, TB, malaria do not fall from the sky but their severity is caused by government policy (or lack of it, which is also a policy) that the Health Minister is responsible for. If they do not work for health (and instead only cry for UHC, the finance minister will be sceptical - listen to the Nigeria minister in the World Bank high level panel). They can barely be heard on public health, and they do even less, and instead wantmore money for equipment, HCW, vehicles for UHC, that is their own ministries and medical colleagues and friends. The cure pays off for them far better than prevention. For the poor, prevention is far better than the cure. If health policy makers who push UHC cared about health, they would be focused on preventing disease.

As Mayor Bloomberg said at the event, all it takes is leadership. UHC as a goal for poor countries is perhaps evidence of lack of leadership, with a hefty dose of populism.

Add new comment