COVID-19 QALYs

We shall have to wait a long time for a formal cost-benefit analysis of lockdown in the UK. ‘Defeat of the virus’ remains uncertain, the bills continue to roll in, and the Government is less than anxious to provide a calculation.

What is sure is that the budget deficit widened from £63bn in fiscal 2019/20 to £355 bn in fiscal 2020/21 reflecting significantly increased spending together with a collapse in tax revenue wrought by the biggest decline in GDP (9.9%) since the early 1700s. This is financed by quantitative easing, whereby the Government issues bonds which are then bought by the Bank of England, putting money into the Government’s hands. The scale dwarfs QE in the 2007/8 financial crisis. In his recent budget speech, Rishi Sunak, the UK Chancellor stated:

“Taking into account the significant support announced at the Spending Review 20 … our total COVID support package, this year and next, is £352 billion.

“Once you include the measures announced at Spring Budget last year, including the step-change in capital investment, total fiscal support … amounts to £407 billion.

How many lives have been or will be ‘saved’ by this £407bn is contentious, but let’s start with Professor Ferguson’s original estimate of around 500,000 deaths with ‘Do nothing‘ and subtract the 125,000 COVID-19 deaths to date. That suggests 375,000 lives saved, at £1.085 million apiece (£407bn/375,000).

These figures are open to considerable criticisms outlined below, and I do not actually believe them.  Nonetheless, they have some official stamp and so provide a starting point. Moreover, they give lockdown the benefit of the doubt – I do not hear assertions that it has saved more than 375000 UK lives, nor that the Government deficit will rapidly reverse.

£1.085 million per life saved can be tested against NICE’s standard criterion that a new treatment is potentially fundable by the NHS if it costs less than £30,000 per ‘Quality Adjusted Life Year’ (QALY). Therefore, for lockdown measures to meet NICE’s criterion, they need to buy around 36 QALYs per ‘saved’ patient.

 Life expectancy in the United Kingdom is c. 79 years for men and 83 for women, with the last few years not counting as full QALYs owing to impaired health. So, to gain 36 QALYs, it’d be necessary to be preventing deaths among people in their mid-40s. This isn’t happening. Few people in their 40s die of COVID. Rather, the median age of a COVID-19 death, preventable or not, is 83 years.

UK life expectancy at 83 is around eight years and will be less in terms of QALYs. What is more, care home residents make up a considerable fraction (c. 33%) of deaths and, on entry to a nursing home, one has a median life expectancy of only 462 days. In Scotland, around 30% of COVID deaths were among patients with recent (non-COVID) hospitalisation.  Since routine hospital admissions were restricted, this implies that many who succumbed to nosocomially-acquired COVID-19 had serious health conditions.

Once we factor these points into account and describe the sort of patient likely to die of COVID, it becomes unlikely that more than 5 QALYS are being gained by those protected from fatal infection. This gives a cost per QALY (c. £200,000), or around seven times NICE’s normal maximum.

Several criticisms can be raised to these calculations. First, it will be objected that there are many approximations and simplifications, which is true. However, none is loaded to exaggerate the cost per QALY, and several are biased to underestimate it, not least that the costs are still accruing. 

Second, it will be objected that no account is paid to Long COVID-19 and the loss of QALYs that it will cause. That is also true but is balanced by omitting the costs of mental health damage engendered by lockdowns.

Third, some will assert that spending on pandemic control, including the economic stimulus, has big economic multipliers. This seems inherently dubious because it dismisses that much of the cost arises from furlough payments to those now idle and from the collapse of productive activity that normally provides a tax base. These aspects, not productive investment. It also dismisses vast spending on importing PPE at inflated prices and on a £22bn ‘Test, Track and Trace’ system that failed to prevent the UK’s second and third lockdowns, all of which are in the included public spending. Stimulus may have accelerated the development of, for example, innovative mRNA vaccine technology that will find wide future uses, and that is to be welcomed. However, arguing that this makes the whole stimulus positive is like asserting that WWI spending was positive because it accelerated aviation development.

Fourth, and with more justification, it will be asserted that the calculation (like all QALY sums) are callous and that ‘Something had to be done’. I agree that one cannot decently shrug at the prospect of 500,000 deaths or the hazard of ICUs being overwhelmed. Moreover, even Sweden, with a more measured response, has accrued a deficit of Kr402bn in 2020 (£34 bn, with a six-fold lower population than the UK). Their cost per QALY must be considerable too.  The response here is not that nothing should have been done but that the wrong things were done. Long-made, scientifically-based pandemic plans were abandoned, and the alternative of Focused Protection was summarily dismissed. 

This, in turn, links to a valid criticism: that there is no proper evidence that 375,000 deaths have been prevented by lockdown. The number comes from the simple subtraction of attributed deaths from a disputed prediction. The actual number is likely far lower. The death trajectory per 100000 population in Sweden, which did not lockdown, has been no worse than in countries, including the UK, that did. Florida, with a larger elderly population and no lockdown since September, has not had a higher death rate than California, with far greater restrictions. The trajectories in North and South Dakota have been similar, despite fewer restrictions in the latter. Numerous countries with strict lockdowns, from the UK to Peru, have recorded heavy death tolls. The UK’s death toll to date is 2.5-times the 48000, to which it was suggested, by Prof Ferguson, that multiple and repeated non-pharmaceutical interventions should reduce the 500000 ‘do nothing’ estimate.

These exemplars and a broader lack of correlation between lockdown stringency and COVID-19 death rates call lockdown’s efficacy into question, except maybe as an early (and costly) response to a small, localised outbreak, as in Melbourne.

Moreover, there is little apparent reason why generalised community lockdown should be efficacious in preventing deaths if severe and fatal infection is substantially care-home and hospital-associated.

If these criticisms are accepted, it follows that the estimated cost of £200,000 per QALY is far below the actual figure. And, worse, much of the vast spending and economic dislocation has done very little to prevent the most dangerous infections.