How much is your life worth? Your health? Can you put a dollar value on it? Healthcare providers need to, governments need to, and although you may not realise it, you do too. You put a price on your life and health when you chose one insurance plan over another, or when you decide that gym membership is too expensive. What about someone else’s life? How much is the life of your mother, your friend, someone who lives down the road that you’ve never met, or a starving child who lives overseas worth? Are they all worth the same? Healthcare is composed of finite resources that need to be rationed like any others; from a technical point of view medical decisions often need to be made through balancing length of life with quality of life.
Both in Australia and globally, healthcare costs are rising in real terms and as a proportion of GDP. The discussion around this in the media and policy centres on this fact, and solutions are proposed to change how much we spend; yet less effort is expended on analysing how we decide to spend it. We propose that health economics is flawed both in its implementation and its theory.
What is the endpoint of health spending? It is to improve peoples’ length and quality of life. It is important to note that, contrary to how we talk, it is not possible to ‘save a life’. Everyone must die – it’s as immutable as taxes. What we can do is extend the time until death. The first outcome of interest is the length of life (Life-years, LYs), and the second is the ‘quality of life’ (QoL). These two components are multiplied to give the end point which is measured – quality-adjusted life years (QALYs), and the absolute majority of health interventions are measured in $/QALY. It seems to make intuitive sense, then, that the aim of healthcare services is to maximise the amount and quality of life lived with minimal costs.
There is an oft-repeated maxim that “all lives have equal value”, but this is demonstrably false. Think about your life as it is now, in its entirety. Now imagine living your life, but having severe migraines every second day. Do these two lives of your have equal value to you, whether measured monetarily or otherwise? The only way for this is to be true is for you to have no preference which of those two lives you wake up in tomorrow – and we’re sure that you have a clear favourite out of the two options. These two lives not only have different values to you, but they also have different values to your loved ones (do they have to care for you every day?), to society, and to the government (how productive are you?). If there are potential differences in the valuations of your own life, then not only are some lives worth comparatively more, they are worth absolutely more. It therefore logically stands that there will be differences in value between the lives of individuals. This is what a QALY takes into account, and by definition healthy peoples’ lives give more QALYs than others’ lives do. That is, extending the life years of an otherwise healthy person with good QoL should have greater benefit than extending the life years of someone with poor QoL by the same amount.
For international context, according to the World Health Organisation (WHO) guidelines countries should aim to spend between one and three times their Gross Domestic Product (GDP) per capita for one of their citizens per QALY gained from the treatment. This puts the value of a life in Luxembourg between $111,162 – $333,486, the value in Mexico between $10,307 – $30,921 (the lowest GDP/capita in the OECD), and between $226 – $678/QALY in the Democratic Republic of Congo. (All figures current $US for 2013 from the empirical World Bank data; claim about Mexico based on OECD website for 2012 GDP/capita). What impact do figures like this have on constructs such as the right to health? Is it only a right if your disease is cheap to treat or if you live in a wealthy country? And how do we measure the ‘health’ you have a right to from the perspective of QALYs? Is it 65 years at perfect health or some equivalent QALY that we should be aiming for, or is it simply maximising the QALYs for any individual within the price limits of the local Government?
Think back to your answer at the beginning of the article: how much is your life worth in monetary value; how much do you think your government should pay for your life (or anyone else’s)? Or, if your government was failing in this, how much should other Governments be willing to pay for your health? How did your answer compare to the above figures? These figures are WHO guidelines and significantly differ from real spending in different countries. The WHO guidelines for Australia calculate a recommended price of $72,620 – $217,900 per QALY (figures in 2014 $AU). Real spending is significantly less, with available data on decisions made by the Australian Pharmaceutical Benefits Advisory Committee (PBAC – the committee that recommends which drugs are to be listed as governmentally subsidised medications). A study by Harris and colleagues using 1994 – 2004 data showed that the mean $/QALY for drugs that received a positive recommendation for PBS listing was $46,000 (in 2004 dollars, $66,905 adjusted for Health Price Index to 2014) per QALY gained, with a 6% reduction in approval chances per $10,000 increment. This is substantially below the lower bound of the WHO recommendations of $72,620. Of note, using other methods to evaluate the value of a human life in Australia gives values ranging from $42,000 to $881,000 (2013 $AU)(3).
The dazzling flaw in the practical implementation of health economics is the consistently inconsistent application of the guidelines as the ideals are diluted by political influence and lobbying groups. Although the PBAC makes recommendations for funding, final decisions are made by cabinet. The guidelines centre on $/QALY, but also take into account other important factors such as total budget impact. It is unsustainable to treat common diseases expensively; paradoxically, an expensive treatment for a disease is less likely to be implemented if the disease is common than if it’s rare due to the multiplicative effect on cost. Decisions regarding funding are significantly impacted by both visibility and political point scoring ability. Powerful patient representative groups bear significant responsibility for this situation, and are often funded by pharmaceutical companies. Organisations with more funding have improved lobbying ability7, and as such are more likely to affect funding decisions relating to specific illnesses. To be clear, these patient groups are simply looking after their patients and need funding to survive, but it is well known that drug companies’ charitable donations are more extensive to patient groups who need expensive drugs(2). Examples of these drugs include ‘Interferon beta’ for multiple sclerosis ($126,400/QALY in 2008) and trastuzumab for breast cancer ($68,610/QALY in 2008). Both multiple sclerosis and breast cancer have significant lobby groups and public campaigns (World MS Day, Pink Ribbon Fundraisers). Furthermore, politically boring but effective areas are ripe for the cutting – there is data suggesting that the entire Australian National Preventive Health Agency (ANPHA), recently cut by the Government, is much cheaper per QALY saved than many of the drugs we pay for. For example, one program which was funded as part of ANPHA included the national tobacco campaign, estimated to be between ‘cost saving’ and $100/QALY(4). There won’t be lobby groups demanding that the government does more to prevent disease; the passionate lobbying is by people who already are, or were, sick. These inefficiencies in the apparently technocratic costing process arise before taking into account that the data used to determine cost effectiveness is compiled and calculated by pharmaceutical companies, which is a topic beyond the scope of this article.
An interesting side note on implementation and spending: research by the philosopher Peter Singer (Princeton University) suggests that a child’s life in equatorial Africa can be ‘saved’ (saving a child = a lot of QALYs) for every $500 spent on preventive bed nets. Compare this to the $45,000 we spend per single QALY here in Australia. Or, for a personal perspective, the next time you buy a shiny new TV, consider whether it’s worth as much as two children’s lives9.
Even more concerning to us than the flaws in implementation are the flaws in the theory and calculation of $/QALY. There are myriad papers written on the topic so we are going to focus on two core issues here. The first is the modelling methods, and the second is the calculation of QALYs. The health outcome modelling used for calculating $/QALY assumes an otherwise healthy patient with the disease of interest, and compares the outcomes in two theoretical situations – on the treatment you’re testing, versus best current treatment. The majority of individuals with health issues are over 65, and they nearly all have more than one health issue; the modelling doesn’t apply to them but it’s what we use to make decisions for them. Taking co-morbid (that is co-existent) disease into account would notch up the complexity of the model significantly, but let us give you a clear example of where we are without it. It is common to use a ten year time limit for a model: pretend for some data that those on the new therapy have an extra 200 people alive at ten years’ time. But every single one of them will die of something – they will all get sick again, and there is no ‘terminal value’ in these models. Furthermore, several of those 200 ‘extra’ people alive at the end would have developed another disease in that time period (they’re now 75 years old), costing significantly more than the equivalent ‘dead’ 200 people in the group not utilising the new intervention. The outcome from this is almost invariably going to be an underestimation of the gross cost impact of the new therapy. This is central to the rising healthcare costs but nobody seems to want to talk about it: every time we prolong someone’s life and avoid severe illness, we are guaranteeing a future cost to the system; they will become sick again, and they will die. As an aside, death costs the health system around $5000 depending on the cause and location. This is not built into our costing, which takes a one disease at a time approach, and it is worth reiterating: rising healthcare costs are a natural conclusion of a better healthcare system, because if people survive they will need to be treated again – this will continue indefinitely until their health is beyond our ability to treat. We need to begin using the totality of the complex data available in healthcare to build better models.
The other issue in our approach to health economics is how we measure QALYs. Years lived are pretty simple, but the other half, utility, not so much. How exactly do you measure a year of life compared to other possible years of life? Technically, utilities are scored from negative infinity to one where one is perfect health, zero is death, and negative values are life states worse than death; however in practice researchers only give patients the option to score their QoL between zero and one. There are three direct measures used to measure utility: the standard gamble (most theoretically accurate but complex), the time trade off (middle ground), and the visual analogue scale. The method these authors prefer is the “time trade-off”; you ask the patient how many life years of perfect health they would be willing to trade for an arbitrary amount of time with disease impacted health (often an arbitrary ten years, sometimes the life-expectancy of the disease). These are then put into a ratio. For example, if you are willing to trade ten years of migraines for seven years of perfect health, your utility is 0.7. The issues with this (and the other approaches) are numerous and amongst them are: risk neutrality, anchoring bias, and the availability heuristic. The availability heuristic is particularly damning because when a doctor asks the patient this question they are explicitly thinking about their disease. It is their focus, and there are many unpleasant health states which individuals do not enjoy thinking about (e.g. stoma; Google it if you aren’t eating) but which actually have minimal impact on their day-to-day life. We know these patients self-report very low utilities when asked by a doctor compared to their response when asked at random intervals by psychologists in non-medical settings(5). It is likely in these cases that patients’ answers do not represent their true experiences as they go about their day-to-day life. Worse than this though, the medical and medical economic literature is rife with utilities (and therefore QALYs) which are essentially made up by well-intentioned researchers who believe they understand their patients’ lives and preferences. In one author’s own research on a health topic, we found that the utility estimated by the researchers for a subset of patients was 0.63, but their utilities calculated elsewhere by getting them to do a time trade off was 0.81. This means that patients valued their life 29% higher than researchers assumed they would; calculating this accurately would therefore generate 29% more QALYs in this population over a given time and yield a potential 29% more funding for their treatment. Further to the issues explored above, one may also question what impact artificially binding utility between 0 and 1 has on our calculations; philosophically it is unlikely that it is truly impossible for one to have utilities less than zero or greater than one.
There are two distinct and significant problems for health economics to overcome in the future. The first is modelling accuracy to avoid resource and budget blowouts; if we don’t use appropriately complex and detailed models we are likely to continue to underestimate the costs of treating various diseases. Secondly, we need to start to address philosophical concerns. All resources are finite, and at some point we as a society will need to have a conversation about quality versus quantity of life for primacy of health budgeting, and it’s likely a good idea to have this discussion before we are forced into making rash decisions due to spiralling unsustainable costs.
If we continue at the current arc, we’ll continue down a path of ever increasing healthcare costs for what may very well be different outcomes to those that patients actually want.
1. Harris, A.H., et al., The role of value for money in public insurance coverage decisions for drugs in Australia: a retrospective analysis 1994-2004
2. Bad Pharma, Ben Goldacre
4. A colleague’s unpublished research
3. Strategic Review of Health and Medical Research, Department of Health and Aging, February 2013
5. Thinking, Fast and Slow, Daniel Kahneman.