I've been blogging a bit about Universal Health Coverage (UHC) recently. In my "old wine in a new bottle" post, I argued that UHC is ultimately about ensuring that rich and poor alike get the care they need, and that nobody suffers undue financial hardship from getting the care they need. In my "Mrs Gauri" post, I used my colleague Varun Gauri's mother as a guinea pig to see whether the general public feels that UHC is a morally powerful concept and whether it could be expressed in a way that the general public would find accessible.
My sense from Ms Gauri's comment on the post, is that the answer to both questions could well be Yes. So far so good.
Some bad news—resources are finite
But before we place orders for colorful placards and huge banners with my suggested slogans "Everyone should get the care they need!" and "End impoverishment due to health spending!", we should break some bad news to Ms Gauri and the rest of the general public: resources are finite, and especially in poor countries the available resources won't allow us to get to UHC anytime soon.
In fact, to be quite frank, even in rich countries, we may not get there—there will always be some needs that scarcity will prevent us from meeting. Sorry. That's a little blunt, I know. But it’s the truth.
Which raises the question: How should countries allocate their limited resources so as to get themselves as far toward UHC as they can?
In answering this question, let's make one requirement—that whatever method we come up with has to be feasible to operationalize and implementable at the point of use. Staff at a health facility have to have clear practical rules about what to do with each patient.
Why CEA is so compelling
One popular approach is to use cost-effectiveness analysis (CEA) to come up with a list of interventions (or "benefits package") that will be financed publicly. Patients needing an intervention not in the benefit package will have to pay out of pocket, or else go without it.
CEA can be operationalized with existing data on costs and effects of different interventions. In fact, we already have cost-effectiveness estimates for a large number of interventions, and in different settings. We can use these data to rank interventions in descending order of cost-effectiveness, estimate the number of people likely to need each intervention, and then keep adding interventions to the benefit package until the budget is exhausted.
CEA is also a rule that can be implemented at the point of use. Medical staff make the diagnosis, decide the appropriate intervention, see whether it's in the benefit package, and then break the good or bad news to the patient, who either gets the care for free, or gets out his wallet or heads for home untreated.
The big attraction of the CEA approach is that, of all the possible rules, it’s the one that will produce the maximum possible aggregate health for the population. So if we keep things simple and think of people as being either well or ill (and therefore in need of health care), applying the CEA rule will get the largest number of people well again.
That's pretty compelling. Not applying CEA will mean that some people could have been made well again with the existing budget but weren't.
Why UHC and CEA aren't happy bedfellows
But is CEA an attractive allocation rule if we're trying to pursue UHC? Of course we know that with limited resources we won't be able to achieve UHC. But is at least in the spirit of UHC?
Awkwardly for CEA, the answer is No.
If we use CEA to come up with a benefit package, we'll only get those people back into good health who have been “fortunate enough” to need a very cost-effective intervention, or whose private resources are large enough to enable them to finance a relatively cost-ineffective intervention. This group will likely include a disproportionately large number of relatively well off people.
By contrast, there will be another group of people who have had—through no fault of their own—the misfortune to get an illness that requires an intervention that's not very cost-effective (surgery, say). These people will either go without treatment, with all the unpleasant consequences—pain, disability, and perhaps death. Or else they will have to pay for it out-of-pocket, quite possibly pushing themselves and their family into poverty as a result. This group will likely include a disproportionately large number of less well off people.
Put bluntly, CEA isn’t going to help us get to UHC.
Is the future random?
So what to do?
One thing we could do with limited resources is to give everyone the same chance of getting the care they need without suffering undue financial hardship. Ex post, not everyone needing care would get it, and if they do they might suffer from financial hardship. But we could try to ensure that, ex ante, everyone—rich and poor—faces the same probability of getting the care they need without suffering financial hardship as a result of receiving it.
Could a lottery approach to allocating scare health care resources be operationalized? And could it be implemented at the point of use?
With today’s information technology the answer has to be Yes. The government would need to get its actuaries to work out how many patients in an average month it could afford to give free treatment to, and what fraction of the population would require treatment. These calculations would determine the odds that any one person seeking care could be given free care. These odds could be updated more regularly than every month if necessary—for example, if new resources become available, of if an epidemic suddenly causes a spike in demand.
For each patient, the facility manager would need to submit a request for free treatment, and get a Yes or No back within minutes. The obvious way would be for this to be done electronically. It could even be a mobile phone or a handheld device that makes the request of a central computer. The facility would submit a code for the proposed treatment and the name of the patient. The central computer would provide a Yes/No reply along with an authorization number; the facility would submit this number to claim reimbursement for the treatment costs that it incurs and for which it does not ask the patient to pay. There would need to be a mechanism to prevent requests for the same patient and same treatment being made repeatedly in the hope the patient eventually gets a Yes.
Tweaking the lottery with a nationwide ID system and biometrics
If every citizen had a unique identification number, and their identity could be confirmed electronically through biometrics, repeated queries for the same patient and same treatment would be easy to tackle. India is introducing just such a system known as Aadhaar.
There are tweaks we could make, especially if we had a system like Aadhaar in place. The country could exempt the poor: individuals from an officially poor family would be identified as such once their biometric information is obtained at the clinic, e.g. through fingerprints or an iris scan. Other groups—such as children and pregnant women—could be added to the exempt category. Of course, for every person added to the exempt list, the government would have to reduce the odds of a nonexempt person getting free care.
We could also require that anyone who has started a long-term treatment for free be allowed to continue forever for free. And we could say that someone who started a long-term treatment but didn't get the care free initially could submit a second request for free care the following month, and the third month, and so on.
Another tweak would be to have randomly generated gradations of charges. So instead of generating a Yes/No answer, the national computer could randomly generate a copayment rate—the number might not be completely random, or it could be one of 0%, 25%, 50%, 75%, or 100%.
Blending randomness and CEA
Going down the lottery road would get us much further toward the UHC ideal.
But there would be a price to be paid in terms of lower aggregate health. We would make fewer people better—probably far fewer—than if we decided on who gets free care and who doesn’t on the basis of the cost-effectiveness of the interventions they need.
We could go some way to meeting this objection by giving patients needing very cost-effective procedures slightly better odds of drawing a low copayment rate than those needing cost-ineffective procedures. This would be a practical way to address the equity-efficiency tradeoff in health. It would be much more practical than the social welfare function approach I suggested nearly 25 years ago that continues to be popular (and just as impractical) today.
Building on modern information technology and blending CEA with randomization-at-the-point-of-use would allow us to balance our objectives of maximizing health and attaining the goals of UHC. Randomness can help us protect people who—through no fault of theirs—have had the misfortune to fall ill with a condition that requires expensive care.
The future may not be completely random, but in the name of fairness, shouldn’t it be a part of it?