29th March 2020
The current COVID pandemic has brought a very thorny and difficult issue to the forefront. How much money should we, as a society, spend on keeping people healthy/alive? No-one has ever fully got to grips with this question, but it has never been more important than now.
The reason why I say this is that the US Govt has set aside two trillion dollars to deal with the crisis, in the UK it is over three hundred and fifty billion pounds, which is almost three times the current yearly budget for the entire NHS. Is this a price worth paying?
I know that some people will instantly dismiss such a question as being cold-hearted, and simply stupid. ‘You cannot put a value on a human life.’ Is an argument that I have heard many times, almost whenever health economics is discussed.
The counter argument is that – if funds are not limitless – then we should focus on doing things whereby we can do the most good (save the most lives) for the least possible amount of money. Or use the money we have, to save the most lives. In fact, this is why the National Institute for Health and Care Excellence (NICE) was established.
NICE reviews interventions and decides whether they provide value for money. The economic term for this is cost-effectiveness. This work is complex and often relies on assumptions that can be difficult to verify.
However, keeping this as simple as possible, NICE tries to compare healthcare interventions against each other by using a form of ‘currency’ called the cost per QALY. A QALY is a Quality Adjusted Life Year. One added year of the highest quality life would be one QALY.
People with conditions such as cancer, or severe heart disease, or who are suffering from chronic pain can be considered to have a quality of life less than one. For the sake of argument, we can say that their quality of life is 50%. Thus, one year of additional life gained for them, would have a value of 0.5 of a QALY.
It also needs to be borne in mind that not everything that is measured using a QALY, relates to saving, or extending, lifespan. For example, someone could have chronic hip pain, and a quality of life of 0.5. Then they have a hip replacement, and their pain goes away, their quality of life can improve from 0.5 to 1. If they live another twenty years, they will have gained 20 x 0.5 QALYs = 10 QALYs.
Obviously, things can get significantly more complicated than this, and the validity of the measured quality of life is a matter of considerable debate.
However, the fundamental question as always, comes down to the following. How much are we willing to pay for one QALY? [How much can you afford to pay for one QALY?] Not just the NHS, but the country as a whole? The current answer, in the UK, is that NICE will recommend funding medical interventions if they cost less than £30,000/QALY. Anything more than this is considered too expensive.
This figure is not set in stone and can vary depending on circumstances. Interventions for young children tend to get more spent per QALY, and powerful lobbying groups can bring pressure to bear on that figure.
However, the figure of £30,000 is generally accepted – if not widely publicised.
Which means that, if we are going to spend £350,000,000,000.00 in the UK, on managing the coronavirus, how many QALYs do we need to get back? The simple answer is to divide three hundred and fifty billion by thirty thousand. Which leaves us with slightly more than eleven and a half million (11,666,666).
To put it in more stark terms. In order to spend three hundred and fifty billion pounds, we require a return on investment of eleven point six million QALYs. If not, NICE would reject it.
[For those who think this an impossible/inhuman calculation, you always have to consider how many other lives could be saved, how much other suffering, or death, could be prevented, by spending three hundred and fifty billion pounds in another way. Because that is what you are really trying to work out].
Are we likely to achieve this level of benefit? Of course, any attempt to model this requires several assumptions to be made. However, the model we can use in this case only has four variables, two of which are (pretty much) known. The variables are:
- How many people will die?
- What is the average age of death?
- What is the average reduction in life expectancy in those who die?
- What is the average quality of life of those who die?
[In truth, average age of death is only needed to calculate the average reduction in life expectancy.]
So, for example
- 500,000 die
- Average age at death 78.5
- Average reduction in life expectancy 3 years
- Average quality of life of those who die 0.7
QALYs lost: 500,000 x 3 x 0.7 = 1,050,000
Using these figures, if we spend three hundred and fifty billion pounds – in the hope of reducing the ‘QALYs lost’ figure to zero, then each QALY will have cost £333,000. Which is more than eleven times the maximum cost that NICE will approve.
Of course, people will immediately object to this model, and for valid reasons. How do we know how many will die, how do we know the average quality of life of those who die, how do we know the average reduction in life expectancy?
In fact, we do know two things with reasonable accuracy. First, we can be pretty certain about the average age of death, and we can also be fairly clear on the average quality of life of those who have died.
What is less certain is how many will die, and the average life expectancy of those who have died. At this point we need to look at the ‘variables’ in the model in a little more detail. This is UK only.
Number who may die
The 500,000 figure for possible deaths, that I used in the calculation above, is the absolute upper range of the numbers that have been proposed, and it comes from modelling that was developed by the Imperial College in London. Their modelling has been since used around the world to guide Government responses. 1
On the other hand, the UK Government has used an estimated 250,000, for the upper limit of deaths – if nothing is done to prevent spread. Other figures have been much lower, but I am going to use 500,000 as the maximum, and 250,000 as the ‘most likely number’ in this model.
My minimum figure will be 20,000, as this has recently been suggested by the same Imperial research group. It seems low.
Average age of death
In Italy – which has had the greatest number of deaths – the average age at death is 78.5. This is comparable with age of death in other countries. I am going to use this as a non-variable 2.
Average reduction in life expectancy
This is more complicated. Using Italy, again, the average life expectancy is 82.5 years (both men and women). However, if people die aged 78.5, this does not mean you have reduced life expectancy by 3 years.
The average life expectancy in Italy, at birth, is 82.5 years. However, once you reach 78.5, you can expect another eight or nine years of additional life. [You will have avoided car crashes, early cancer, suicide and suchlike which reduce the ‘average’ life expectancy of the entire population].
On the other hand, those who are dying of COVID have multiple medical conditions. On average they have three serious underlying problems such as: diabetes, COPD, heart disease, previous stroke, active cancer and suchlike.
Which means that these 78.5-year olds do not have a life expectancy of eight or nine years. It will be far less. How much less? This is virtually impossible to calculate. I am going to estimate a half – or 4.5 years (an average).
Which means that, in this model, my lower figure of years of life lost will be three years. My upper figure is nine years and my ‘most likely’ figure 4.5 years.
Average quality of life of those who die
Again, this is difficult to establish. However, studies have been done to work out the ’reported’ quality of life in those with multimorbidity. Perhaps the most accurate figure I could find with that elderly people with three underlying serious health problems have a quality of life of 0.8.3
Using different figures in the model
Having put figures to the likely range of the variables, we can look at the cost per QALY in various scenarios. I am only going to look at three. ‘Best case’ ‘Most likely’ and ‘Least benefit.’
I am going to start by inputting the figures that would provide the greatest possible gain in QALYs. This is 500,000 deaths prevented, and an average gain in life expectancy of nine years [This assumes all 500,000 lives will be ‘saved’ with the actions taken]. Quality of life is kept constant at 0.8.
The calculation is:
500,000 x 9 x 0.8 = 3,600,000 QALYs
Which gives a cost per QALY of £97,200 [£3,5Bn ÷ 3.6m]
We can then run the ‘most likely’ scenario, which is 250,000 deaths prevented, with an average gain in life expectancy 4.5 years.
250,000 x 4.5 x 0.8 = 900.000 QALYs
Which gives a cost per QALY of £388,888 [£3.5Bn ÷ 900K]
Finally, we can tun the ‘least benefit’ scenario, which is 20,000 deaths prevented, with an average gain in life expectancy of 3 years.
20,000 x 3 x 0.8 = 48,000 QALYs
Which give a cost per QALY of £7,291,666 [£3.5Bn ÷ 48K]
As you can see, none of these models achieves a cost per QALY that would be approved by NICE.
Disability Adjusted Life Years
I fully recognise that looking at human life in from this purely economic perspective can seem harsh, almost inhumane. Can we really stand back and watch an elderly person ‘drown’ as their lungs fill up with fluid ‘Sorry, we are not spending money on more ventilators, because it is not cost-effective.’ Or suchlike.
However, there is also a health downside associated with our current approach. Many people are also going to suffer and die, because of the actions we are currently taking. On the BBC, a man with cancer was being interviewed. Due to the shutdown, his operation is being put back by several months – at least. Others with cancer will not be getting treatment. The level of worry and anxiety will be massive.
Hip replacements are also being postponed and other, hugely beneficial interventions are not being done. Those with heart disease and diabetes will not be treated. Elderly people, with no support, may simply die of starvation in their own homes. Jobs will be lost, companies are going bust, suicides will go up. Psychosocial stress will be immense.
In my role, working in Out of Hours, we are being asked to watch out for abuse in the home. Because we know that children will now be more at risk, trapped in their houses. Also, partners will suffer greater physical abuse, stuck in the home, unable to get out. Not much fun.
Which means that we are certainly not looking at a zero-sum game here, where every case of COVID prevented, or treated, is one less death. There is a health cost.
There is also the impact of economic damage, which can be immense. I studied what happened in Russia, following the breakup of the Soviet Union, and the economic and social chaos that ensued. There was a massive spike in premature deaths.
In men, life expectancy fell by almost seven years, over a two to three-year period. A seven-year loss of life expectancy in seventy million men, is forty-nine million QALYs worth. It is certainly a far greater health disaster than COVID can possibly create.4
In Lithuania, the impact of the break-up of the Soviet Union was also dramatic, and damaging. Below is a graph, looking purely at deaths from cardiovascular disease. As you can see, starting in 1989 (when the Berlin wall fell) there was an enormous spike, representing hundreds of thousands of premature deaths. These same spikes, in death and disease, were seen across most countries in the former Soviet Union. 5
These, the downsides, can be calculated, using the figure that is the opposite of the QALY, which is the DALY. The Disability Adjusted Life Year. Or, to put it another way, how much harm are you causing with your interventions? I am not doing this calculation here, because it would have about ten thousand variables and would take far too long.
Despite this, the message here is that severe damage to an economy does not simply affect bank balances, it can be deadly. If we look at the result of social deprivation in the UK, the effect is (potentially) immense
This was highlighted in a review by Michael Marmot, who studied two areas of Glasgow. Lenzie, which was rich, whilst the other area, Calton, was poor (socially deprived). The findings were stark:
…we can see this in Glasgow. When we published the report of the WHO Commission on Social Determinants of Health (CSDH) in 2008, I drew attention to stark inequalities in mortality between local areas of Glasgow: life expectancy of 54 for men in Calton, compared with 82 in Lenzie.’ 6
A twenty-eight-year difference in life expectancy between people living approximately five miles apart. The difference? Money.
This, I hope, puts into some perspective the discussion on cost per QALY. I framed it, to start with, as a discussion about money, but it is not really about money. Health does not exist in some bubble, sitting apart from the rest of society. Health and wealth are closely interrelated.
Which means that I fear that we are taking actions that could, in the longer term, if we are not very careful, result in significantly more deaths than we are trying to prevent.
Even if we restrict the analysis purely to the cost per QALY and narrow the ‘health’ analysis purely to COVID, and deaths from COVID, it remains difficult to justify spending £350 billion pounds to control a single disease.
I know that many people will violently disagree with this analysis and will think I am some cold-hearted fiend. ‘People are dying, we must do absolutely everything we can. No matter how much it costs.’ ‘What would you say if it was your mother…’ and suchlike.
Well, I have spoking to my mother, who is 92. Her view is that she has lived long enough. She thinks the Government actions are a ridiculous over-reaction. She is going out shopping and chatting to friends… she will take no advice on the matter.
So, what would I do if it was my mother that is dying? I will say that she made her choice, and who am I to argue with it.