When a new medical drug comes on the market patients want to be given immediate access to it, especially if their condition threatens their lives. However, healthcare systems around the world are coming under increasing pressure due to rising costs. There is a difficult balance to strike: with a finite budget, the NHS cannot afford everything, but every decision to fund a treatment for one patient group may come at the expense of others. It could mean a baby's life in an area with high infant mortality, no decent palliative care for other terminally ill patients, or no new diagnostic equipment for a hospital.

Making these difficult decisions is one of the roles of the National Institute for Health and Clinical Excellence (NICE). NICE aims to ensure that the NHS's scarce budget is spent on care that delivers the greatest possible health benefits. In order to decide whether something represents value for money, NICE has to compare it to existing care in the NHS. NICE asks the question "how much extra health is this new treatment giving us compared to standard NHS care and how much extra are we being asked to pay for those additional benefits?". It does this using health economics.

Comparing like with like

When a new treatment or pathway of care becomes available, and it does more or less the same thing and produces similar results as the existing NHS care, the choice is straightforward. All things being equal it makes sense to choose the cheapest alternative — the economists call this type of analysis cost–minimisation analysis.

However, more often than not, the cost of treatments, as well as the health benefits they provide will differ. If the treatments you're comparing treat the same condition, you can compare the cost of producing the same outcome for each treatment (the health gain as health economists like to put it). For example, if the treatments are designed to lower blood pressure, you compare the cost per reduction in unit of blood pressure. You also need to take into account things like side-effects, which may reduce benefits and can be costly to treat, and nursing time to administer treatments. This process of comparison is called cost-effectiveness analysis.

Comparing like with unlike

Things get more complicated if you are trying to get the best possible value for money in healthcare across all different treatments. You need a way of comparing a treatment for, say, high blood pressure with a treatment that prevents recurrence of breast cancer. In this instance the measurements of value for money are expressed in different ways. Not only can they not be compared; they can't even be used to decide which represents the better value for money and is therefore the "better buy".

To compare treatments for different conditions, you need a common measuring instrument, a yardstick. The yardstick used by NICE is the quality adjusted life year (the QALY).

The QALY takes into account both the increase in life expectancy from an intervention and any change in quality of life. This reflects the value judgment that living longer, in itself, is an insufficient measure of success; and that the quality of life also needs to be considered. Of course many treatments (such as hip replacement surgery) may not directly increase longevity but give a better quality of life during a person's remaining years. Therefore, it is important to take both into account.

One QALY is equivalent to one year of life in perfect health. If a new treatment on offer gives you an additional QALY, then this could correspond to one year of life in perfect health, two years of life in 50% health, or any other combination. In the health economic approach used by NICE, you compare how many extra QALYs the new treatment provides compared to the old treatment, and also how much those QALYs cost to provide. In other words you are comparing the cost per QALY for these treatments. This process of comparison is called cost-utility analysis.

Measuring the extra length of life provided by a treatment is easy, but it is more difficult to measure quality of life.

How is quality of life measured?

Traditional methods of evaluating disease and response to treatment examined the process of health care from the point of the physician, focusing on symptoms, cure and whether or not the patient died. In the past few decades, however, the focus has moved more towards the point of view of the patient and the impact a treatment has on their life and how they live it, in other words the impact on the patient's quality of life (QoL).

QoL assessments are made based on the attitudes of the general public, as well as people suffering a condition.

QoL can be affected by many factors and is referred to using many terms. To look at healthcare, we have to focus on the aspects of people's lives that are affected by their health. There are a large number of different potential aspects and illness will affect each person differently. You cannot measure all of them, therefore in order to get some consistency in measurement of quality of life, scientists have tried to identify those aspects (or dimensions) that are affected most often and have the most impact on someone's overall quality of life.

Several large surveys of different people across Europe, including those who were well and those who suffered from a wide variety of diseases, found five aspects people consistently identified as having the biggest impact on health-related quality of life: whether a patient was mobile, whether they could look after themselves, to what extent they could do the things they usually did, how much pain or discomfort they were in, and how much anxiety or depression they were suffering from.

Based on these results, scientists then devised a questionnaire, known as the EuroQol–5D (or EQ–5D), giving people the choice of whether they had (1) no problem, (2) some problems, and (3) major problems. It has now been widely tested in many different patient populations.

There are a total of 3⁵=243 possible different combinations of the five responses to the questionnaire. Each different combination of options is termed a health state (see table 3). There are an additional two possible health states, unconscious and dead, so we get a total of 245 health states.

In order to use the results of the questionnaire in QALY calculations, you have to convert each combination of options into a number which reflects overall QoL. The numbers between 0 to 1 (see table 3) were chosen and they are called health utilities. The value 0 is equivalent to being dead and 1 represents the best possible health state. However, some health states are regarded as being worse than 0 and are given a negative value. For example, some people may consider being in a permanent vegetative state worse than death and so would give such a health state a negative utility.

But how do we know which of the 243 combinations of the different health states (in the left hand column of table 3) result in a particular utility (in the right hand column of table 3). For example, how do we know that that 22222 (Some problems walking about; some problems washing or dressing self; some problems with performing usual activities; moderate pain or discomfort; moderately anxious or depressed) corresponds to a utility of around 0.5 or 50% quality of life? To find out, the scientists undertook a survey of the views, attitudes and preferences of 3,000 members of the UK general public, asking them detailed questions about how much length of life they would hypothetically sacrifice to gain greater quality of life. This allowed them to mathematically "value" a sample of the different combinations of health states.

The results also indicated that the five dimensions are not considered equal. The British public considers being confined to bed a much greater negative impact on their QoL than being unable to perform their usual activities, and would be willing to trade off more years to improve this aspect of their quality of life. As a result of this survey, the scientists were able to create a formula that converts each health state into a utility value between 0 and 1.

It has been suggested that this formula should have been derived from surveys of people with different diseases rather than surveys of the general public. However, this would be impractical. For every conceivable disease, the experiment would need to be repeated in a large sample of patients. Moreover, there is also the argument that in a publicly funded health care system — such as the NHS — it is right that the public should determine the values given to individual elements of QoL. Because the NHS has a fixed budget, each decision impacts on all NHS users, not just those with a given condition. Finally there is also the argument that patients adapt to their condition; people get used to their circumstances — after one year things might not seem so bad as they did at the start of the year.

The EQ5D is the simplest of many different QoL questionnaires. Most other questionnaires contain more than five aspects. They are therefore more time consuming to fill in and there are many more combinations of answers. As a result, the experiments to derive the necessary formulas have not been carried out. For these reasons, the other questionnaires are more difficult to use in health economic studies.

NICE is often unfairly criticised because people think that its calculations are based on responses from the public, rather than from people suffering a condition. Because the calculations are quite complicated and rarely explained, people don't realise that there are two steps involved in the process of calculating QALYs. The first is to capture the actual health states associated to a treatment, which are assessed using the responses to the EQ-5D questionnaire from people who actually suffer the relevant condition and have been given the treatment. Secondly, there is the conversion of health states to numerical utility values, which is based on the preferences of members of the public, who do not necessarily suffer from that condition.

So we now have a way of "weighting" each year a patient is in a particular health state by the quality of life experienced in that year: the number of QALYs of the health state is defined as the utility of that state multiplied by the number of years an average patient is in that state. So one year in the best possible health state is 1 year × 1 QoL, which equals 1 QALY. Whereas if it's only 50% of perfect health, it's 1 year × 0.5 QoL, which equals 0.5 QALYs.

What dimension do you value most?

To compare two different treatments, you carry out a randomised controlled trial of the treatments and include a measurement of QoL to find out how many QALYs each produces. You ask the groups of people receiving the different treatments to each fill out the EQ5D prior to treatment and then after they have had the treatment. You then convert the EQ5D health states to utilities and use them to calculate the QALYs. This gives a comparison of how many QALYs individuals in each group gained on average as a result of the treatment, and you can statistically compare them to see if there are any differences. Since the treatments are commonly undertaken in a randomised controlled trial, any change in either quality or length of life is assumed to be a direct result of that treatment.

When health economists calculate QALYs they do not take any account of how many QALYs a person has already had in the past. For example, if a woman is successfully treated for breast cancer and gains 30 QALYs, it makes no difference to the calculation of the number of QALYs she gets when her hip is replaced 15 years later.

Cost per QALY

The final step of the cost-utility analysis is to compare the cost of the QALY gains in each treatment group. You can either add up the costs associated with each step of the trials in which the treatments have been tested, or estimate them based on agreed average national cost tables.

Using the hip replacement example, if the total cost of a hip replacement with artificial hip of type A is £5,000, then the cost per QALY is

5,000/1.5=£3,333 per QALY.

If the cost for the initial operation for a hip of type B is also £5,000, then type B costs £5,000/0.5=£10,000 per QALY: over three times the cost per QALY even without the costs and loss of quality of life associated with further hip replacements.

Is it cost-effective? The ICER

Because the NHS has a fixed budget, any money spent on a new intervention is not available to spend on other things. As a result, something (an opportunity) will have to be given up either for this patient group or other patient groups. This opportunity that has been forgone is termed the opportunity cost and it can be valued in both monetary terms (the money that has been spent) and in health benefit terms (the health measured in QALYs that you have given up because you didn't choose that option).

Is the treatment worth the extra cost?

Bearing this in mind, it makes sense that a new intervention, as a minimum, should have the equivalent benefits (in QALY terms) as the things you are going to be giving up. This could then be considered an efficient (cost-effective) use of NHS resources. Ideally, the new intervention should have greater benefits than what you are giving up. However, as is often the case, the new intervention will have greater benefits, but at a greater cost.

To assess this extra cost of a new intervention compared to existing treatments, NICE uses the incremental cost-effectiveness ratio, or ICER, which is worked out as follows. First, compute the difference between the cost c₁ per patient of the new intervention and the cost c₂ per patient of the standard existing intervention. Then work out the difference between the new intervention's gain in QALYs q₁ and the existing intervention's gain in QALYs q₂. Now compute the ratio between the two differences:

(c₁-c₂)/(q₁-q₂).

This measures the additional cost per QALY of the new intervention compared to the existing intervention.

For example, suppose that a new intervention A provides a gain of 0.8 QALYs while existing intervention B provides a gain of 0.25 QALYs. If intervention A costs £10,000 per patient and intervention B costs £2,000 per patient, then the cost per QALY is £10,000/0.8=£12,500 for intervention A and £2,000/0.25=£8,000 for intervention B. The ICER of the new treatment compared to the existing is

(10,000-2,000)/(0.8-0.25)=14,545.45.

The threshold

We obviously cannot measure the cost-effectiveness of every alternative use of the same money. Also we often don't know what will be given up in favour of a new treatment because decisions are made locally and NICE does not directly substitute new interventions for old. We have to decide how much extra money we're maximally willing to spend per QALY for a new intervention, and therefore whether it can be deemed to be cost-effective or not. Economists call this maximum a threshold, and NICE's lies between £20,000 and £30,000 per QALY. NICE uses the threshold as a comparison tool: an intervention with an ICER below the threshold is generally regarded as cost-effective. If its ICER lies above the threshold, then the intervention may be deemed not cost-effective for the benefits it offers compared to existing treatments, and it may not be made available on the NHS. In addition to economics, NICE's independent advisory bodies consider many other factors when they are making decisions, and have input from all stakeholders including patients. The factors include ethical principles, legal requirements, the need to avoid discrimination, and the promotion of equality.

On the face of it, cost-effectiveness may seem like a very mathematical way of dealing with issues of personal health. But the fact of the matter is that the health care pie is finite and decisions for one and decisions or one patient group affect other patients with different diseases. The health economical approach used by NICE is an attempt to divide it up as fairly as possible. It's not a matter of mechanically plugging made-up numbers into equations, but of carefully assessing the effects of treatments and comparing them using measures that are as objective as possible.

About this article

Dr Sarah Garner is Associate Director for Research and Development at NICE.