The UK Statistics Authority has rebuked the government over its lack of transparency around projected Covid-19 deaths and hospital admissions, saying it could cast doubt over official figures.
A range of estimates were used to make the case for a second English lockdown in a press conference on 31 October. However, the UKSA said “the data and assumptions for this model had not been shared transparently”, potentially undermining confidence in official figures.
On Friday, Downing Street acknowledged there had been an “error” in the way the data was presented but said the underlying analysis was unchanged.
A No 10 spokesman said: “We want to make the data as clear and transparent as we can. There was no error in the underlying analysis. The consensus is that without action we will breach the first wave [peak] in terms of hospital admissions.
“We have acknowledged the mistake which we have corrected and we will continue to try and provide data in the clearest possible way.”
The upper limit of these projections has now been revised down. How much does this change our understanding of the second wave?
What was misrepresented in the data and how much have the projections changed?
At the press conference, the government showed a chart that projected daily coronavirus deaths could surpass the first wave peak and hit 1,500 deaths a day by December. The figures were based on data available up to 28 October and produced by academic modelling teams.
The upper limit of this projection has since been revised down to 1,000 deaths a day, which would be in line with the first wave peak. However, the central projection remains the same in the new version, which predicts that there could be about 800 deaths a day by December.
The upper limit for projected hospital admissions has also been revised down, from about 9,000 to 6,000 a day. Again, the central projection has not changed, and remains at more than 4,000 admissions a day by December.
How does this change our understanding of deaths and hospital admissions?
The latest modelling shows the second wave peak could be less severe than initially predicted, but the models are updated regularly and this could change again in coming weeks.
Dr Duncan Robertson, a policy and strategy analytics academic at Loughborough University, said: “The central projection remains the same: all that’s changed is what are called the ‘prediction intervals’ [the upper and lower limits of the range where the prediction might fall]. Hospitals are still going to be overrun if we don’t do something now, and adjusting the prediction intervals on the graph is not going to change that.
“The government needs to be proactive in its decision making rather than reacting when it’s too late. Scientific advice shows a projection of what could come and it’s important that policymakers put mitigation measures in place so that they are ahead of the epidemic rather than being controlled by it.”
What else has been wrong with data during the pandemic?
The inaccuracy is the latest in a long line of problems with government Covid-19 data, which has cast a shadow over how the pandemic has been managed and communicated.
Many of the issues have revolved around testing and how the number of tests being carried out have been counted. In May and June the agency reprimanded the government on its testing statistics twice. Rather than show the number of tests that had been carried out, the government included mailed-out tests in an effort to hit its target of 100,000 tests a day by the end of May.
In a letter at the time to the health secretary, Matt Hancock, the head of the UKSA, Sir David Norgrove, said: “The aim seems to be to show the largest possible number of tests, even at the expense of understanding.”
The number of tests available has also been in doubt after 1.3m inaccurate tests were quietly pulled from the government’s tally and it changed how it counted available tests in August.
There have also been concerns about the accuracy of death figures. In July, Hancock announced an urgent government review into how deaths were counted by Public Health England after it was revealed that people were included in Covid death counts even if they had recovered from the virus months earlier. However, analysis showed the likely overcount of deaths at the time was minimal.