Introduction

When evaluating any intervention it is just as important to consider the benefits that arise as it is to consider the costs.  This raises the question of how do we measure the benefits?  In the early days of health economic evaluation, the benefits of interventions were measured in clinical outcomes. The problem with this is that it limits making comparisons between interventions for different conditions.  Then came one of the greatest contributions to health economics, the quality adjusted life year or, as many know it, the ‘QALY’.  First proposed back in the mid 1980s, it was the first type of measure of benefit which was easy to conceptually understand, (relatively) easy to calculate and, importantly, comparable across interventions.  Such comparability is often claimed to be important for purposes of resource allocation.

However, the QALY is not without controversy (more about that later); but first to help us debate the use of QALYs we need to understand them a little better.  

The QALY is a summary measure of health gain that takes into account not only length of life but also quality of life. In the calculation of QALYs, we combine the number of life years over which an individual will experience a particular condition with an assessment of their quality of life during those years. Quality of life in the calculation of QALYs is measured on a 0 to 1 scale where 0 is equated to ‘being dead’ and 1 is ‘full/normal health’.  Values between 0 and 1 are known as ‘health state utilities’.  Essentially reflecting different degrees of impairment across different dimensions of health, these utilities can be interpreted as judgements of how ‘good’ or ‘bad’ different conditions are.

To calculate the number of QALYs for any health state you simply multiply the utility value by the number of life years. For example, two years in a health state valued at 0.5 would be one QALY. If people experience multiple states over time, the respective QALYs for time spent in each state would simply be added up.

We can illustrate this calculation using a hypothetical example, as illustrated in the diagram below.  Imagine a patient with chronic renal failure. The standard treatment is dialysis, with which the patient would live for 10 years and their quality of life is measured at 0.6, so this person would have 6 QALYs.  An alternative to dialysis is to have a kidney transplant. If a patient has a transplant, let’s say this would increase their life expectancy by 10 years from 10 years to 20 years) and would return the patient to full health (i.e. a utility value of 1) – a success which we assume for purposes of exposition, recognising that many will suffer from background morbidity. A person who had a transplant would have 20 QALYs. 

The QALY gain from having a transplant over continuing dialysis is therefore 14 QALYs (shown by the shaded area).  If however, instead of quality of life on dialysis being valued at 0.6 it was 0.8 this would give a QALY gain of 12 QALYs; so, you see, the number we put on quality of life is really important. So how do we do this?

Calculating QALYs

Remember we said calculating QALYs was (relatively) easy. Well, this is where it gets more complicated.  For a start there are a number of different methods which we can use. These methods can be categorised into two main groups: generic and condition specific.  Generic measures are likely to be the ones with which you are most familiar. A good example is the EQ-5D which was devised by the EuroQol group.  The EQ-5D has 5 dimensions each of which has 3 levels (although a new five-level version has been developed which will be more sensitive to changes in health) and there is a corresponding tariff of quality of life scores for the UK which has been developed to score all potential combinations of levels. While generic QALY measures can be applied to any group of interest, condition-specific measures focus directly on the characteristics of the condition being evaluated.  For condition-specific methods we generally start by producing a description of the condition being evaluated including items like symptoms.

Whether generic or condition-specific, health state descriptions then need to be valued; the two main methods to do this are the standard gamble and the time trade off – which, by the way, can also be used to value generic QALYs.  To explain these methods it is easier to go back to our dialysis example.  Thinking about the standard gamble technique, what we are asking people to do is choose between two alternatives – ‘alternative A’ is a certain outcome of remaining in the health state as described (so in this case dialysis for the rest of his/her life), whilst ‘alternative B’ is some form of treatment that has two possible outcomes; a return to full health for remaining life with a probability p or immediate death with a corresponding probability 1-p. That is, ’alternative B’ comes in the form of a gamble. We can illustrate this in the following diagram: 

The respondent makes a series of choices in which the probabilities for ‘alternative B’ are varied until the individual is ‘indifferen’t (i.e. finds it difficult to choose) between alternatives A and B; the utility of the description (in this case ‘being on dialysis’) is then taken as p. The time trade off involves a similar iterative process but, in this case, it asks people to choose between living in the health state described for a given period of time t or in full health for a shorter period of time x.  The period of time in full health is varied until the individual is indifferent between the two and the corresponding utility score is calculated as x/t.

Contributions and controversies – the QALY and public health interventions

Returning to the beginning of this blog, the QALY has arguably been the biggest contribution of health economics to the evaluation of health policy. However, like any measure, it is not perfect and there are some limitations to its use especially in the evaluation of public health interventions. It may be obvious from the name, but QALYs measure values only for ‘health related’ quality of life, which are then combined with life years. 

In the case of public health interventions, are QALYs an appropriate measure? Yes, if health is the main outcome of a policy, such as a mass screening programme for the prevention of CVD.  However, as we have said in previous blogs, public health interventions often include other areas of the public sector beyond the health system, resulting in non-health benefits such as reducing crime or raising education levels.  How do we capture these benefits? A possible approach is to measure ‘non health QALYs’ and then combine these with our ‘health QALYs’ to create a more comprehensive generic measure of well-being. However, despite some previous attempts to do this for crime interventions this particular research area is not well advanced.

So what advice can be offered to those undertaking economic evaluations of public health interventions today? If we want to retain our focus on health then a possible approach is to attempt to map all impacts onto QALYs, where non-health outcomes are seen as a means to a final health end. We could do this either by using empirical evidence (where it exists) or by theorising the links between non-health and health outcomes. In practical terms, such exercises would be challenging in general and also more difficult for certain interventions. For example, converting housing and regeneration interventions into QALYs may be feasible, by thinking through the links from improved housing and neighbourhood conditions into physical health (e.g. reduction in respiratory conditions) and mental health (e.g. improvements in self-worth translating into healthy behaviours , and a reduction in anti-social behaviour lowering stress). For some interventions, however, making such links explicit seems more difficult. For instance, the impacts of services to improve access to contraceptives for teenagers (where some notion of ‘enhanced life opportunities’ might be the goal) may not be meaningfully translatable into QALYs – although, of course, some QALY devotees may say everything is reducible to QALY terms. Further, it is one thing to theoreise links, but quite another to estimate the strength and timing of such relationships to assess how far QALYs would change following an intervention. Perhaps you can think of other examples of public health interventions where it would be either impractical or undesirable to convert outcomes into QALYs?

However, perhaps we’re barking up the wrong tree? Whether we should map all outcomes onto QALYs really depends on the purpose of the evaluation, as discussed in previous blogs. It may not be an appropriate exercise to undertake if we wish to take a ‘decision maker’s perspective’ and provide to information policymakers need given that, for non-NHS sectors, health is probably not the main aim.   

So where do we go from here? QALYs are a useful and accepted way to measure the benefits of health care interventions. But, as we move towards evaluating public health interventions we realise that there are limitations to the narrow focus of the QALY. This leads us to think about ways in which we can measure the wider benefits of interventions in a form that is useful for a ‘decision maker perspective’ where the aims of a policy may include, but go beyond, health. In the next blog we will be discussing one potential method to do this – willingness to pay.