Who Should Be Treated? Interferon-β for Multiple Sclerosis

Presentation to Information Workshop on Disease Moderating Agents, sponsored by the Multiple Sclerosis Society of New Zealand: Thursday 17 June 1999.

Keywords Health

The Public Policy Context

Public policy is about problem solving. The specific problem I shall look at today is who with multiple sclerosis should be able to obtain access to Interferon-beta therapy? But public policy answers to such questions have to take place from a wider perspective. Resolving the public spending problem on multiple sclerosis has to take place mindful of the tradeoffs with other public spending (including on other diseases) and the raising of the revenue (typically via taxation) to fund them. There has to be a consistency between the specific answers, for otherwise the various decisions – the resolutions to the problems – becomes a matter of arbitrariness, creating ongoing tensions (and further problems) because of inconsistencies.

Each democracy solves its public policy questions differently in terms of the institutions it uses, and technologies it uses for evaluation. There is however, considerable convergence between different countries’ practices. The New Zealand arrangements I am going to describe today would be recognized as similar to those in most western democracies – the US, with its emphasis on commercial principles excepted although they may be at different stages of implementing the systematic practices.

At this point, I should state that I am neither a neurologist nor a pharmacologist, so I know very little about these aspects of the treatments considered here. My contribution is to say something about the resource consequences of the use. I shall be using published papers, each typically by a team which includes a neurologist, a pharmacologist and an economist. If my task were to do an original evaluation, I would work with such a team too.

The registration of pharmaceuticals, is medical-pharmacological matter, which among other matters considers safety aspects of the drug. Once a drug is registered, it is available for a treatment. Registration provides no guarantees of universal accessibility if the drug is expensive. Those who can afford the drug may purchase it. Aspirin is available from a chemist. However many drugs – especially new drugs – may be far too expensive for the typical sufferer to afford. At which point arises the public policy problem of whether it should be funded in whole, or part, by the public purse.

The Funding of Public Health Care

A simplified version of the public policy towards the funding of health care goes like this. At the top of the system the government decides how much it is willing to spend on all its activities. That decision depends largely on the amount of taxation revenue it raises, and on a judgement of the effects of various tax rates and the willingness of the community to pay those taxes.

There is no value-free formulae which tells a government how much it should tax or spend. In a democracy that decision is made by the political judgements of the government in power. Once every three years we have the opportunity to vote for a parliament to select a government. One of the factors which influences our voting is our personal preference for more or less government spending and less or more taxation. Within the period between elections, the government is continually judging the extent to which the public in general, and its supporters and potential supporters in particular, want to spend and tax more or less.

Having decided on total spending, the government allocates it between the various general areas, including health. For instance, in last month’s budget the government announced total spending of about $36.4 billion for this fiscal year, of which $6.8 billion it allocated to health spending. Again there is no rigorous formula which says how much the government should allocate to health or any other spending. That is a matter of political judgement.

I have stressed these political judgements at upper levels, because as the money moves down through the system, technical considerations become increasingly important, and politicians become increasingly less involved. That is how it should be in a democracy. We do not want politicians making the decisions in the clinic or surgery. Even so, technical decisions almost always value judgements lurking near them.

Reflecting the changing balance of political and technical, parliament decides the amount of the health vote which goes to the Health Funding Authority which funds (in the jargon “purchases”) the various health and disability services the public health sector provides. This year the amount is about $6.2 billion. The HFA allocates about $700 million to Pharmac or the Pharmaceutical Management Agency, which is charged with administering the public subsidies on about 3000 pharmaceuticals.

Pharmac’s Problem

The policy problem which Pharmac faces is how to use its funds effectively. Some of its actions are fairly obvious. It buys drugs as cheaply as possible. It tries to make sure the drugs it pays for actually work. It tries to make sure they are not wasted. Even when such sensible things are done, the demand for medication far exceeds the funds that Pharmac has available.

I am now going to try to explain the policy framework which Pharmac is increasingly using to decide which drugs to fund, given the budget constraint it faces. The underlying theory is complex, although I have done my best to keep the exposition as simple as I dare. Some readers may find it convenient to jump three pages, accepting that there is a criteria, and the ratio of the Net Cost to Quality Adjusted Life Year (or QALY) is a rational way of making the decision whether to fund a drug or not. However others will value the opportunity to obtain some idea as to the underlying analysis behind the theory.

The (over) simple example is set out in the Table 1. It assumes there are just 10 drugs (Column 1) to be funded, which in total cost $55m (Column 2). But suppose there is only a total of $20m available for funding. How is the $20m to be spent given that the demand for funds far exceeds those funds available?

There are many possible rules, some of which have developed in an ad hoc way over the years. For instance, the “grandfather rule” is that if the treatment had applied in the past it should continue into the future. Whatever its merits – much of today’s treatments are grandfathered rather than using any systematic approach – the rule does not tell us how to decide between new therapies.

Another approach might be called the “shout principle”: those who make the most noise get the treatment. There are a number of dangers: given all the shouting, no-one may be heard, however justified their need; it is an encouragement for political interference in technical decisions; and it can be very wasteful of resources.

To illustrate, suppose that therapies are chosen in order to have the maximum number of people treated – since loudness is in numbers. That is equivalent to allocating the funds by cost per patient (or shout). Column 3 shows the cost per patient for each therapy. Table 2 ranks the drugs by cost per patient. Drug A costs $10 a patient, cheaper than any other, so the rule would fund the 1,000,000. Drugs E and G are the next to lowest in cost per patient terms, so they are funded. The three cost $20m, and exhaust all the available funds. On the shout criteria, the drugs A, E, and G would be funded.

There is absolutely no allowance in this assessment for the benefits of the treatment. For instance, drug A may be something like an aspirin, which is of small benefit to a lot of people. In contrast drug H might be a matter of life and death – if the patient does not get it the patient dies. Intuitively one feels that drug H should be given more significance than drug A, but the cost per patient criteria does not take that into consideration.

QALYS and Decision Making

In order to get around the problem of treatments having different benefits, economists have evolved the QALY, or “quality adjusted life year”. QALYs work like this. A person living a normal life, in normal health, has a QALY of 1. A person who is dead has a QALY of zero. People whose health is between have a QALY level between 1 and 0. If a person’s health changes, then the QALY level also changes. For instance a person who is saved from certain death has a QALY level of 1 rather than 0, and a treatment that saves is treated as adding one QALY a year to the person’s life. Calculating QALYs is complicated and riddled with problems. Let us take it here that they are a reasonable way of resolving the impossible problem of comparing everybody’s health. We know no better one.(1)

Historically, a common means of measuring the effects of Multiple Sclerosis has been by the Kurtzke expanded disability status scale (EDSS) of neurological impairment on a 0 (normal) to 10 (dead) scale, based on responses to 20 questions. The EDSS scale can be mapped onto a QALY one, which applies to all health statuses.(2)

Column 4 of Table 1 shows the QALY gained from the treatment for each patient. The aspirin of drug A gives only a small gain of .0001 QALY per patient, but the life saving drug H treatment gives a gain of 1 QALY. The gains for the other drugs are in the range from .1 to .4 QALYs. Column 5 gives the total QALYs gained from each treatment, and Column 6 gives the cost of each QALY gained. They range from $6000 for drug H to $500,000 for Drug I. If we want to maximize the QALYs to be gained from treatment, we will choose those which are cheapest in terms of the cost per QALY. This is summarised in Table 3, where you see that drugs D,E,G and H are the cheapest and there total cost is $20m. So if we choose our therapies by the criteria of QALY gained, we replace the low return drug A (the aspirin) with drugs D and life saving H which seems to be a more sensible outcome.

There is one further major modification to the Pharmac evaluation method. Some therapies save resources elsewhere in the health system. For instance, a course of Interferon-β may mean fewer relapses, and less use of medical and hospital resources. Column 8 of Table 1 lists the cost savings to the health system for the various therapies, Column 9 shows the net total cost to the health system when the savings per patient are deducted. It makes sense to include these cost savings in the calculations. For instance, without them the method could rule out a better and cheaper drug therapy for a particular condition, because it would ignore the savings from abandoning the old therapy. So instead of using gross cost per QALY, we would use net cost per QALY, which is shown in Column 10.

Table 4 shows the rankings. Now drug J is the cheapest, and that with drugs H, D, G, and F would be the recommended ones. However, while their gross cost is $20m, their net cost (after we have deducted the savings to the health sector) comes to only $12.2m. We can now also afford drug E, and still have a bit over for some spending on drug C.(3)

Table 5 summarises the outcomes of different choice of drugs within the budget constraint. By changing the selection criteria we change the set of drugs which the agency would provide free. As the criteria moves from gross medical cost per patient to cost per QALY, drug A gets dropped in favour of D and H. Shifting to net medical cost enables drugs F, G, and part C to be added.

What is unavoidable however, is that given the budget constraint a selection has to be made, and that some selections are better than others.

Some Problems with the Method

I have simplified the method for expository purposes. However there are some issues which are still being debated and need to be briefly mentioned:

– Should other net gains to government revenue (such as savings in social security benefits and additional tax payments from working) be included? But we need to be careful about the implications. For instance, incorporating the gains from employment could mean that the potentially employed were given preference to the unemployable.

– Should non-government gains (such as the extra earnings of a well person) be include? We have to be careful here not to double count, since the quality of life includes the consumption which goes with it. Moreover, too wide an approach can lead to ethical issues: if some illnesses cause marital breakups, which treatment prevents, does that mean the treatment should preferentially be given to the married?

– Are all QALYs are the same? Some surveys suggest that people think some lives have characteristics (such as age) which are more worthy.

– How are the effects of the treatment on the quality of life of others to be included? For instance there is a case for a higher QALY score of parents with young children, because it improves the young’s quality of life too (or adding in a QALY for the better life of a child without a sick parent).

– Should low probability health events (such as MS) be treated differently from high probability ones (headaches)? Part of the issue here is what costs can be reasonably attributed to the person. We dont give higher social security benefits to people who need to more to eat.

– What about moral hazard, the technical term for situations which affect one’s health or entitlement to treatment? Most people would be more reluctant to provide aspirins for a headache precipitated by a hangover, which the patient’s drinking brought it on.

– How events that take place over time to be treated?. The technical notion is “discounting”: there is some dispute over what is the correct discount factor. Different ones may affect the relative ranking of wasting diseases such as multiple sclerosis compared to one-off gains from surgery.

These are proper matters for debate. Very often they will reduce the cost per treatment but not change the rankings. Whatever the outcome of these debates, I doubt they will have much influence of the question of the funding of Interferon-β therapy. And they certainly do not invalidate the need for an approach which involves a systematic decision rule.

In Defence of Pharmac

After presentations such as this, economists are often accused of being calculating and hard hearted. Calculating we are, just like engineers who design bridges you can trust. In the economists’ jargon the calculations are to get “as big a bang for the buck as possible.” Without them there will be a waste of resources, and less people would be effectively treated.

But economists are not necessarily hard-hearted. What we are aware is that where there is a budget constraint, being warm hearted to one means being cold hearted to others. It is a sad truism that people allege cold heartedness when they are turned down, but dont mention a warm heart if they are given a “yes”. Economists dont push into health policy because they are intelligent, beautiful, witty, wise, or power hungry. They may (or may not) be some (or all) of these. The reason economists get involved because of the funding constraint. Were there no economists, the government’s health agencies would still face the problem of what to do with limited funds. Eventually someone would recommend a prioritisation not unlike that which economists recommend.

My impression of the economists and others who work for Pharmac is that they would love to be able to say “yes” to everyone. They, too, have friends and family who suffer illness but are unable to get the maximum treatment for conditions they suffer, because of the budget constraint. Of course they are tough. They have to be, when they say “no” to one treatment because the funds could be better spent on another. And they have to be tough when the front up to a pharmaceutical company to get the best possible price, knowing the money saved can be spent on other drugs.

The Threshold

In practice, Pharmac does not do the elaborate exercise I have just described, for that would mean evaluating every drug therapy every time a new drug appears. Rather they calculate the net cost to QALY ratio of each new drug, and compare the ratio with that for other drugs. This is equivalent to a “threshold” approach where if the cost per QALY of a new drug is below the base threshold it is subsidised. Usually there is a grey area, where drugs with ratios above the base threshold up to a second higher threshold, where other factors not included in the Cost per QALY consideration are taken into account.

Looking at Table 4, the lower threshold might be $10,000. If the ratio comes below that, and Pharmac has the money, they are likely to recommend to the provision of the drug without charge to the patient. In the grey area, say up to $20,000, they are likely to say no, unless there are some other relevant considerations.

Pharmac has not said what its thresholds are. Essentially they are the consequence of the amount that parliament provides for health care. The more money voted, the higher the thresholds, and the more drugs and other treatments that can be funded by the public. However, we can guess the Pharmac thresholds must be near $20,000 a QALY. That is about the average (public and private) consumption per New Zealander. If, for instance, the threshold was $200,000 a year, then ten people in the community would have to give up all their consumption in order to gain one QALY. It seems unlikely that New Zealanders would be willing to accept the taxation consequences (and their personal consumption loss) of such a high threshold. Thus the technical threshold for evaluating the drug ratio reflects a political judgement at a higher level.

(Even so, I am of the view that there is a case for spending about an extra $500m a year on the health sector, in terms of what New Zealanders are willing to sacrifice, and the substantial health improvements which would arise as a result.)

The Net Cost to QALY Ratio for Interferon-β

Although there has been no New Zealand specific evaluation of the costs per QALY for Interferon-β treatment for multiple sclerosis, there are four foreign reports in the public domain, which give us a strong indication of what a New Zealand analysis would look like. They identify two key benefits from the treatment, where it is applicable (noting that there is no evidence those with primary MS benefit from the drug). Among those with its relapse-remission from it appears that as a result of using Interferon-β:
– there are (perhaps a third) fewer relapses, which saves medical costs and gives a better quality of life;
– the progression of the disease is retarded. However, not a lot is know about the long run retardation, mainly because the systematic evaluation of the treatment is recent, so there has not been enough time to observe its effects in the long run. Trials have been at most three years, and there may never be long term trials because of the ethical issues involved.

Here is a brief summary of each, focusing on the key efficiency finding:

A Therapeutic and Economic Assessment of Betaseron@ in Multiple Sclerosis. (Dalhousie MS Research Unit, Halifax, Nova Scotia, Canada: July 1996)
The net cost/QALY of the Betaseron® therapy amounts to around $CAN160,000 (say $NZ200,000) for women and $CAN136,000 ($NZ170,000) for men.

Interferon Beta-1a in Relapsing-Remitting Multiple Sclerosis. (The Wessex Institute for Health Research and Development: December 1997)
The net cost/QALY of the Interferon Beta-1a therapy amounts to between £740,000 and £5.5m ($NZ2.1m and $NZ16.5m). The report concludes “not proven”, the lowest level of its five recommendations.

Comparison of Drug Treatment for Multiple Sclerosis. (Canadian Coordinating Office for Health Technology Assessment: November 1998)
The net cost/QALY of the Rebfin® therapy amounts to at least $CAN406,400 ($NZ.500,000).

A Cost-Utility Analysis of Interferon Beta for Multiple Sclerosis. (University of Newcastle: December 1997)
The net cost/QALY of the Interferon Beta therapy amounts to between £228,300 and £809.900 ($NZ680,000 to $NZ2.4m). This study, the most comprehensive, looks at various retardation rates and reports “[t]he most optimistic estimate from the ten year model, incorporating some extreme assumptions about natural history and the impact of therapy was £74.500 [$NZ220,000] per QALY gained.”

These are all very high figures – the lowest is around $200,000 per QALY, well beyond any likely threshold that Pharmac (that means “us”) can afford. I was asked to look at doing a New Zealand study, which would have applied New Zealand parameters to these studies. My advice to the MS Society was that they would be wasting their money, because whatever changes were made, the ratio would remain well out of line with any Pharmac threshold.

Why is Interferon-β so cost ineffective? An examination of the studies suggests that:
1. Interferon is expensive – treatment costs about $20,000 a year;
2. It is ongoing treatment (rather that a one-off therapy – like surgery);
3. The average reductions in other medical costs are relatively small. For instance, while there are medical savings from fewer relapses, one study estimated that the saving was only 31 percent of the cost of the drug.
4. The therapy is ineffective for many people, and it does not eliminate all relapses. The studies assume that it is not possible to discriminate between those who will benefit and those who wont, and so – as it were – much of the expensive shot is wasted.
5. The health (quality of life) gains are on average relatively small per dollar spent.

It appears that in our current state of knowledge, the gains from reducing relapses are insufficient to justify Pharmac subsidizing the medication. Ultimately the justification is likely to be from better targeting on those with a high relapse rate (because that may reduce high hospitalisation and other medical costs), but the big gains arise if the progressivity is greatly retarded, especially with better targeting. Only further trials will tell if we have been too pessimistic.

In the longer run, we can expect neurologists and pharmacologists to refine their administration of the therapy to get improved patient benefits, while the drugs themselves will improve, and their costs reduce. The net cost to QALY ratio is likely to lower steadily overtime. Unless someone finds an better drug, Interferon-β may eventually be made widely available without charge by Pharmac or its successor. But that date is some time away.

A Strategy for Accessing Interferon-β for Multiple Sclerosis

What is clear is that Pharmac is not going to be funding the drug to all potential beneficiaries (no matter how small the benefit) in the near future. What is to be done in the interim? I am assuming that the MS Society will not want to do nothing, nor would it waste it time by trying to bust the policy framework.

It appears that in most countries where there is MS, make Interferon-β available for some patients without charge. But given the greatly varying utilisation rates, they must have quite different rules about its use. This suggests that New Zealand will also settle on a selective rather than universal entitlement.

The criteria for selection needs to be designed primarily by neurological and pharmacological specialists, but with a clear view of the implications of the policy framework analysis which I have outlined above. Pharmac will be most responsive to a criteria which targets those:
– who are most likely to respond to the treatment; and
– who have highest relapse rates or otherwise incur relatively high health sector expenditure; and
– who are likely to experience high retardation of the progressive deterioration; and
– where there are rigorous and effective criteria for identify those who are not responding to treatment, and should have the treatment withdrawn. (Alternatively a trial followed by selection as to those who are successful may be an option.)
Basically we should be searching for practices which cut therapeutic costs (of both the drug and of other therapies), and/or which increase the efficacy of the therapy in terms of some concept as the EDSS or QALYs.

There may also be personal factors which may affect eligibility for treatment (as occurs for the surgical waiting lists). One which may be especially relevant to a MS, with its high initial incidence among young women, is that parenthood should be taken into account, since success will also benefit the quality of life of young children.

Ultimately the criteria could even end up with a points system, like in surgery eligibility.

However, I would not wish to raise the expectations of the MS Society and its members that there will be a lot of people eligible under this criteria. Because funds are so scarce, Pharmac is more likely to look favourably on a proposition involving 25 people than 450.

If I may end on a personal note. I usually enjoy my consultancy work. This case is an exception. It gives me no pleasure to have to be so pessimistic about the short term opportunities for moderating the condition of people suffering a debilitating disease like multiple sclerosis. But that is how the material available to me appears. I would have even less enjoyment if I had used my expertise to tell the sufferers an untruth.

1. There are alternative proposals, such as Disability Adjusted Life Years (or DALYs), but they are not as nearly convincing as QALYs. They would not change the story greatly. Pharmac, rightly in my opinion, uses QALYs.
2. One contribution at the seminar argued that QALYs are not yet well defined for neurological problems.
3. I am not sure whether the HFA directly gives over to Pharmac the savings to the rest of the health sector from the use of pharmaceuticals. However, what we can be sure is that insofar as Pharmac increases its contribution to the rest of the health sectors savings, the HFA will increase Pharmac’s budget in due course.