UK Edition
Hopes that life may soon be back to normal were dashed by Boris Johnson this week when he indicated that restrictions would remain in place to prevent a deadly third coronavirus wave.
Predictably, the announcement relied on unduly pessimistic modelling, which suggested a full release from lockdown in June could trigger a new wave of hospital admissions every bit as bad as the January peak and result in up to 59,900 deaths.
It seems absurd that Britain should find itself facing a similar situation to the second wave after an extremely successful vaccination programme.
Look more closely at the modelling from Imperial College, Warwick University and the London School of Hygiene and Tropical Medicine (LSHTM) and it soon becomes clear how such dire forecasts have emerged. Much of the data is needlessly negative and often out-of-date – but here are the reasons why things are not quite so bleak.
Immunity
One striking problem with the Imperial model is it has underestimated the number of people protected from coronavirus.
The paper, dating from March 30, says: "Assuming optimistic vaccine efficacy, even if 2.7 million vaccine doses/week are given up to August 1 (2.0 million thereafter), only 44.6 per cent of the population will be protected against severe disease (due to vaccination or recovery from infection) by 21 June 2021 when NPIs (non-pharmaceutical interventions) are due to be lifted."
It's not clear where this figure is from, but data published by the Office for National Statistics (ONS) on March 30 showed that 54.7 per cent of people in England had antibodies to Covid by March 14. Wales and Northern Ireland are also around 50 per cent, and Scotland 42 per cent.
With the vaccination rollout still going strong, a significant proportion of the population will be protected by the end of June.
The modellers say they have not factored in waning immunity, so their low figure cannot be accounted for by imagining that some people will lose their protection in coming months.
There is also growing evidence that other immune responses are keeping Covid at bay, and even without antibodies there may still be some protection against the virus.
The models of viral spread in the months to June are also based on immunity levels of 34 per cent – again far below what we currently have. We know that even by mid-March it was far higher.
Risk among the vaccinated
There is an extraordinary paragraph buried in the summary of modelling submitted to Sage by the Scientific Pandemic Influenza Group on Modelling (SPI-M) which discusses who will die in a third wave.
It says: "The resurgence in both hospitalisations and deaths is dominated by those that have received two doses of the vaccine, comprising around 60 per cent and 70 per cent of the wave respectively. This can be attributed to the high levels of uptake in the most at-risk age groups."
Yes, you did read that correctly. Third wave deaths will predominantly be driven by people who have been vaccinated.
The reasoning is that around 10 per cent of vaccinated over-50s will not be protected by the vaccine, based on around 90 per cent efficacy – equating to about 2.9 million people.
Imperial has predicted that full release could bring up to 40,000 deaths, while Warwick suggests 59,900. Yet these numbers seem to be extraordinarily high based on what we know of the virus.
In the first and second waves, around 147,000 of the over-50s and vulnerable died from Covid, roughly one in 200. But under the new scenario it would rise to around one in 70.
Clearly the death rate in the first and second waves was kept down by restrictions, but are we really to believe that a mass vaccination programme will more than double the risk of risk of dying for the unprotected over-50s?
Imperial has also ignored the fact that vaccines substantially reduce transmission, AstraZeneca by 67 per cent and Pfizer by 75 per cent. And it's worth bearing in mind that a substantial proportion of these unvaccinated people will already have immunity from a prior infection, or at least some natural immunity from other coronaviruses.
Likewise, AstraZeneca trials show that the jab may offer 100 per cent protection against hospitilisation and death, so suggesting 10 per cent will still be at risk could be far too high.
Effectiveness
The modellers have updated some of their effectiveness data since they were heavily criticised for being too pessimistic in their assessment, in the models used to draw up the roadmap, of how well the vaccine would work in real life.
Yet many of the assumptions are still wide of the mark when compared to real world data.
LSHTM is the most negative on how well the AstraZeneca jab will perform, estimating that it will reduce infections by just 31 per cent after two doses, while Imperial suggests 63 per cent and Warwick 65 per cent.
Yet trial and real world data suggests that it may have around 76 per cent efficacy at preventing a symptomatic infection.
Data from Public Health Scotland in February showed that the AstraZeneca jab reduces the risk of Covid-related hospitalisation by 94 per cent after the first dose, yet Imperial estimated it to be 70 per cent, LSHTM modelled it as just 72.5 per cent and Warwick estimated 80 per cent.
The models also underestimate real-world effectiveness of the Pfizer jab which appears to have between 94 and 97 per cent efficacy against symptomatic disease. LSHTM and Imperial put it at just 85 per cent after two doses.
Seasonality
One big omission from both the LSHTM and Imperial modelling is the effect of more clement weather in the spring and summer.
Respiratory viruses do not tend to surge in the summer months, and Warwick's models found that including seasonal effects in modelling would reduce the total number hospitalised in the third wave by 43 per cent.
Imperial's model also suggested that seasonality would not affect the results significantly, although expecting it to "reduce and broaden the peak of hospitalisations".
The R-rate
In February, LSHTM estimated that schools returning would cause the 'R' number to rise to between 1.1 and 1.5.
Its latest modelling accepts that the current 'R' is around 0.85, but says it is increasing and estimate it would be 1.8 if there was no population immunity.
The model suggests it will rise to 2.2 by May 17, the equivalent of previous Tier Two restrictions in which shops and pubs were open but people could only meet inside in "bubbles" and could not travel widely outside their local areas.
Yet when similar restrictions were in place in Britain last autumn, the 'R' number never rose beyond 1.6.
Data released on Tuesday shows that far from rising, infections are continuing to fall – down 35 per cent in a week despite the release of restrictions and mass testing in schools. Deaths are down by 45 per cent and hospital admissions by 23 per cent.
The lengthy time period between each roadmap release date was deliberately chosen so that real-world data could be used to judge moving to the next step. We are clearly moving in the right direction.
Continuing to base policy on models which are already out of date by the time they are presented cannot be a sensible way to make decisions about the nation’s freedom. It's time to move on – cautiously, but resolutely – with the release of restrictions based on data not dates, and certainly not models.