Governments make policy decisions involving trade-offs between different outcomes that vary in their mortality risks. These trade-offs have become more pronounced with Covid-19, which also highlights limitations of some of the metrics used, such as the value of a statistical life.
The valuation of life is required for analyses that aim to compare the costs and benefits of alternative policies that affect mortality. This is standard not only in healthcare but also in other areas of public policy, such as environment and transport, where there is a risk of mortality to citizens. To conduct such analyses, policy-makers employ a number of metrics, including the value of a statistical life, the value of a statistical life year and quality-adjusted life years (QALYs).
During the current pandemic, there are many policy decisions that will influence mortality from Covid-19 but also morbidity and mortality from non-Covid-19 causes. These include policies relating to reductions in diagnosis and treatment of other conditions such as cancer. They also include policies that will affect long-term morbidity and mortality arising from impacts on jobs and livelihoods, as well as educational impacts on children and young people.
Providing monetary estimates for life, alongside these other impacts, enables policy-makers to choose the option that provides most benefit (or, as is more likely for decision-making during the pandemic, the least loss). One notable example in the UK was whether children should return to school in September even if that increases the risk of more people dying from Covid-19.
The decision to put an explicit monetary value on life and the methods for doing so are contentious. Monetary values are available to place on reduced mortality risks, but it would be more appropriate in this context to focus on the effect on years of life adjusted for quality of life impacts.
The usual approach in health is perhaps too narrow for the pandemic context given that the impacts on quality of life go beyond those related to health. While putting a monetary value on life, it is also important to keep in mind that these approaches say nothing about who benefits. Additional information is needed to give adequate consideration to the distributional impact of policies, although that is not considered here.
What sorts of lives should we value?
Assigning a monetary value to reduced risk of mortality involves ethical, methodological and practical judgements. Ethically, should values be based on ‘known’ individuals or ‘statistical’ individuals?
Once specific individuals are known to be in peril, vast sums of money are sometimes spent on preventing them from coming to harm. Examples from recent years of this ‘rule of rescue’ (Jonsen, 1986) coming into play are the rescue of the Chilean miners in 2010 and the rescue of a Thai soccer team from the Tham Luang cave in 2018. The implied value of life from such rescues is huge. Yet implied values of life from other policy decisions, where individuals are not identifiable, are much lower.
Should we value these known individuals (sometimes called ‘ex-post’ valuation) or should we focus on the values associated with saving the lives of unknown individuals (sometimes called ‘ex-ante’ valuation)? The usual approach in economics is not to accept the values associated with identifiable individuals (Richardson, 2003) and instead to focus on the valuation of lives of unknown or ‘statistical’ individuals; essentially the valuation is of the reduction in the risk of dying (OECD, 2011).
This is appropriate for decisions about policies designed to reduce exposure to Covid-19, given that neither those who die from Covid-19 nor (for the most part) those who die as a result of policies intended to reduce exposure to Covid-19 (for example, individuals whose diagnosis or treatment of cancer is delayed, those who take their own lives following job losses or those whose life chances are reduced and who die before they would otherwise have done so) are identifiable in advance.
How can we value a statistical life?
The concept of a statistical life is used in several sectors to assess government policies that pose a mortality risk and determine the benefit of such a policy (OECD, 2011). The value of a statistical life represents the monetary amount that society is willing to pay to reduce the risk of death of one unknown or statistical life. Several methods have been used to obtain these values (Kniesner and Viscusi, 2019):
Human capital approach
The human capital approach relies on estimating the value of lost productivity (lost earnings). It is seen as problematic as it ignores those who are not part of the labour market, and also ignores a standard economic judgement that individuals themselves are the best judges of their own welfare (Robinson, 1993).
Revealed preferences are based on estimates that indirectly ‘reveal’ the implicit monetary value based on behaviour and choices. These are most often made with respect to the trade-offs between the premium that is paid to workers in jobs where the risk of death is increased and the level of that risk (Viscusi, 2018) – although it can also be related to health and safety features of goods purchased.
This approach also focuses on those who are part of the labour market; it further assumes that those who take these jobs understand the risk trade-offs that they are making.
Approaches based on stated preferences use survey questions that directly establish the trade-off that individuals are willing to make by asking how much they would be willing to pay to reduce (or be willing to accept for an increase in) the risk of death in particular scenarios.
For example, one study asked how much a person would be willing to pay to reduce their risk of death by a total of five in 1,000 over the next 10 years. This value represents willingness to pay for a risk of death of five in 10,000 each year; multiplying by 10,000 and dividing by five gives the value of a statistical life (Alberini et al, 2006).
The key study that informs UK monetary values was conducted in 1997 in the context of road transport accidents (Carthy et al, 1998); this study is now quite old, and the value has also been criticised because of the small sample size used in estimating the value (Thomas, 2020). More generally, stated preference studies can be criticised because of the hypothetical nature of the questions that are asked.
The UK uses a combination of approaches to value a prevented fatality including productivity elements, healthcare costs and human costs (which relate to the stated preference study above, adjusted for growth in GDP) (HM Treasury, 2018). A May 2020 document from the Department for Transport gives the value of a statistical life as £1.53 million in 2010 prices.
Other countries also mix approaches. Canada, for example, combines values from revealed and stated preference studies (OECD, 2011); and France has based values just on stated preferences, but across a number of studies (Téhard et al, 2020; Lindhjem et al, 2011).
Would it be better to value life years than lives?
The next question to consider is whether the focus for policy analysis related to Covid-19 should solely be on the loss of statistical lives. This may be useful, for example, in estimating the value of prevented fatalities in road traffic accidents where risks are largely similar across the population.
Policies relating to Covid-19 may, however, have differential effects across age groups. The infection itself is known to result in disproportionate deaths of older people (Oke et al, 2020). Policies such as lockdowns may, however, have impact on lives and life expectancy more broadly across the population.
Where government policies produce different impacts on years of life expectancy, the use of a value of a statistical life year, rather than a statistical life, can help to differentiate between policies. The value of statistical life years is stated as being calculated using similar methods to the value of a statistical life and the current UK estimate is £60,000 (HM Treasury, 2018; Glover and Henderson, 2010), with similar values in the European Union more generally (OECD, 2011).
Using the value of a statistical life year, rather than the value of a statistical life, to estimate the overall impacts of different policy options may give a more nuanced understanding of effects. In practice, it may give less weight to policies to prevent deaths from Covid-19 and more weight to policies that reduce the impact on other health conditions and people’s livelihoods more generally.
Should all life years be valued in the same way?
The valuation of life has been of particular interest in the health setting, but this introduces another complication. The business of healthcare and public health is both lengthening life and reducing morbidity. A common criticism of the value of a statistical life or the value of a statistical life year is that they do not take into consideration the quality of the remaining life (Ashenfelter, 2006; Karimi and Brazier, 2016).
This is pertinent for Covid-19 where there is increasing evidence that some survivors are left in poor health, but also where some individuals lose time in better health – for example, through missing out on cancer treatment or experiencing poor mental health as a result of lockdown policies.
Quality-adjusted life years (QALYs) were largely developed as a way of measuring outcomes that apparently avoided the difficult question of the monetary valuation of life while capturing morbidity. They focus on combining information about quality – morbidity – and quantity of life – mortality (Torrance et al, 1982; Williams, 1985).
To obtain the quality adjustment in QALYs, numerous measures of quality of life have been developed, which each include different health state descriptive systems with a range of dimensions that collectively summarise health-related quality of life (Herdman et al, 2011; Brazier et al, 2002; McCabe et al, 2005). The combination of quality of life and length of life results in a single measure that can be used to compare the outcomes of different interventions across many disease types and this sort of analysis is starting to be applied for Covid-19 (Briggs, 2020).
But the need to provide some sort of monetary valuation even of the QALY has proved inescapable, in order to make decisions about whether to pay for new healthcare interventions. The current value per QALY gained is accepted for the UK as £60,000, based on the value of a statistical life, discounted and quality-adjusted (HM Treasury, 2018; Glover and Henderson, 2010). But its application in the budget-constrained NHS uses a value between £20,000 and £30,000 per QALY gained.
This NHS threshold range is not based on empirical research but instead relates to early evaluation decisions made by the National Institute for Health and Clinical Excellence, NICE (Culyer et al, 2007; McCabe et al, 2008); other settings may peg values to GDP. Increasing criticism of threshold values without a clear evidence base has sparked empirical research based on two alternative approaches outlined below and designed to provide a stronger evidence-base for the values used in decision-making.
Should life years adjusted for health-related quality of life be valued based on population values?
One approach is to follow on directly from the valuation of statistical lives by using similar stated preference surveys to obtain the social value of a QALY. There have been many such studies conducted internationally in the last 20 years or so. Vallejo-Torres and colleagues report a wide range of values obtained: from as little as €1,000 to more than €5 million per life year (LY/QALY in 2014 prices) (Vallejo-Torres et al, 2016).
But one major study with more than 22,000 participants across nine European countries reported much lower values, between €14,430 to €26,965 (2014 prices) (Vallejo-Torres et al, 2016; Robinson et al, 2013). These studies have the same sorts of advantages and benefits as those outlined above for stated preference studies.
Should life years adjusted for health-related quality of life be valued based on health service budget constraints?
An alternative option for putting a value on a QALY gained is to focus on what is currently spent to gain a QALY within existing health systems. Essentially, this provides an implied value, related to existing constraints on health system funding; the value represents the value of the health that would be given up, to gain health elsewhere.
Such values are constrained by the decisions that have already been made about how much to spend on health. They have been noted as being concerned with affordability rather than desirability (Téhard et al, 2020). They also tend to be relatively low (Vallejo-Torres et al, 2016). Estimates in the UK, for example, are less than £15,000 per QALY gained (Claxton et al, 2015). In the context of policies that may have influences both within and outside the NHS, they may not be appropriate.
Should life years be adjusted for broader impacts on quality of life?
The use of health-related quality of life to adjust life years is clearly appropriate where the main policy impacts are on health, but one aspect of the Covid-19 pandemic is that policy responses may also significantly affect other aspects of people’s lives. Policy responses to Covid-19 have reduced both people’s capabilities (their ability to do and be things that they value – Sen, 1993) and their subjective wellbeing (life satisfaction – Dolan et al, 2012).
There are methods to allow impacts on capability wellbeing to be assessed and valued quantitatively (Al-Janabi et al, 2012; Flynn et al, 2015; Mitchell et al, 2015) and data on policy impacts are currently being collected. Subjective wellbeing metrics such as global life satisfaction or happiness are commonly used, with life satisfaction routinely collected in the UK Office for National Statistics (ONS) survey.
Either option could be used alongside current estimates of the value of a statistical life year, for example, the £60,000 UK value, to adjust the value of life for broader quality of life impacts. An example of how this might be achieved is available already for subjective wellbeing (Layard et al, 2020); this piece of work uses a considerably higher value for a wellbeing adjusted statistical life year of £750,000 based on correlations between wellbeing and income, although it is questionable how applicable it is in a system with constrained public finances.
It is possible to value statistical lives to inform pandemic policy-making, but this is unlikely to be sufficient. Covid-19 has differential impacts at different ages, suggesting that a value for years of life saved is more appropriate.
Furthermore, the disease itself and the impacts of policies to tackle its spread in society have both health- and non-health-related impacts on quality of life that should be taken into account. All of this information should be used in evaluating possible policy responses.
Such evaluations, however, can also only tell us about impacts at the level of society as a whole, and not how those impacts are shared among the population. Any analysis that looks at the total impact on society of alternative policy options also needs to consider how those impacts are distributed before decisions are made about the best alternatives going forward.
Where can I find out more?
- Taking a wellbeing-years approach to policy choice: Jan-Emmanuel De Neve, Andrew Clark, Christian Krekel, Richard Layard and Gus O’Donnell writing in the British Medical Journal (BMJ).
- Valuating health vs wealth: The effect of information and how this matters for COVID-19 policymaking: Shaun Hargreaves Heap, Christel Koop, Konstantinos Matakos, Asli Unan and Nina Weber writing at VoxEU.
- Population health, economics and ethics in the age of COVID-19: Sanjay Reddy writing in BMJ Global Health.
- Do the benefits of COVID‐19 policies exceed the costs? Exploring uncertainties in the Age-VSL relationship: Lisa Robinson and colleagues writing in Risk Analysis.
- A scoping study on the valuation of risks to life and health: the monetary Value of a Life year (VOLY): Susan Chilton and colleagues explain how the technology exists now to generate a theoretically robust, evidence-based and updated valuation of risk to human life and health. They conclude that applying such values would lead to better and more informed policy decisions and would have major implications not only for efficiency of government spending but also for equity in population wellbeing.