Tom Jefferson & Carl Heneghan

What does the Covid data really tell us?

What does the Covid data really tell us?
(Photo: Getty)
Text settings

Another week has passed with more restrictions piled on – but as lockdown measures become ever more restrictive, the demand for evidence grows. Sir Keir Starmer, for instance, has asked to see evidence for new lockdown measures. In mid-August, Andy Burnham called on the government not to put Oldham into lockdown as Sir Richard Leese, the lead for health in Greater Manchester, pointed out that there is ‘no evidence’ that additional lockdown measures would improve the chances of halting the virus.

Tomorrow, we’re told, there will be more restrictions still. But on what grounds?

The main evidence presented to us by the government is the new daily total for new Covid infections. But how severe are the cases? Severity can be assessed quantitatively starting from the clearest and unquestionable outcome – death. But even there, the evidence is not clear. Is a ‘Covid death’ someone killed by the virus, or someone who died from other reasons who also had the virus?

Our understanding of ‘Covid deaths’

Evidence from Italy casts new light on this. ISTAT, Italy’s national statistics institute, and Istituto Superiore di Sanità (ISS), an Italian health authority, published a report on the impact of Covid-19 deaths recorded up to the 16 July. They found 28 per cent of death certificates they studied have Covid-19 as the only cause of death. The remainder, over two thirds, are deaths with Covid: i.e. people with preexisting conditions such as hypertension, diabetes, heart disease and cancer. Often with more than one condition in the same person. So the figure for ‘Covid deaths’ in this instance will be almost four times as big as the number who died from Covid.

An ONS report of deaths involving Covid-19 in England and Wales in June paints a similar picture: ‘Of the deaths involving Covid-19 that occurred in England and Wales in March to June 2020, there was at least one pre-existing condition in 91 per cent of cases.’ Unlike Italy, dementia and Alzheimer’s disease was the most common pre-existing condition found among Covid deaths.

A recent update by the ISS shows that 59 per cent of Italy’s 36,000 deaths were amongst the over-80s. In August, the age of those who died with Covid shifted significantly towards the very elderly. In other words, cases are getting younger but deaths over the summer and early autumn are getting older. In traditional pandemic theory, deaths among the young should increase and several studies have demonstrated this age shift in deaths to younger populations. The major prevention strategies should therefore focus on older people — and not younger people — to minimize the loss to expected years of life.

Since the 1 August just over a thousand Covid deaths have been registered in England and Wales, most amongst the over 80s. (And one in five in the over 90s.) So Covid appears to have been — in the great majority of cases — a cofactor which tipped the frail and elderly. The current fixation on the daily diet of infection numbers is not helping anyone. Basing what happens next on comparisons with figures from the day before is leading to more anxiety, and ever-increasing restrictions. We’re told that hospital admissions are near capacity and can only get worse with hospitals in the north of England running out of Covid beds within a week. But, again we are left wondering: who is being admitted? And what for? The poor quality of data makes it very hard to gather a clear understanding.

The rise in hospital-acquired Covid cases

Let’s look at people catching the virus in hospital. Using NHS England data, we estimated the probable number of those catching Covid-19 in hospital – known as Healthcare Associated Infections (HCAIs). We found for the most recent date of reporting, on 6 October, over 18 per cent of hospitalisations were being diagnosed after over seven days in hospital. Such patients are highly likely to have caught the virus in hospital, given that they were not diagnosed with it upon entry.

Hospital-acquired Covid infection appears to be particularly notable in the North West, where 17 per cent in the last week and 24 per cent on 6 October of the newly admitted patients who were subsequently tested, were positive with Covid-19. So they were probably HCAIs.

This is an obvious danger if you mix Covid patients together with non-Covid patients. Yet, little has been done to invest in ‘fever hospitals’ that could isolate Covid patients and help prevent the in-hospital spread of infection – keeping our hospitals open for the routine care that is so badly needed.

To talk about Covid ‘cases’ and draw international comparisons is also misleading as ‘cases’ has many different meanings in Europe. In fact, there are differences in the ways cases are counted between England and Wales. The interpretation of case numbers is further complicated by the absence of data, broken down by severity and likelihood of being infectious, which are the important categories for public health. If we were able to focus on those who are symptomatic in the community (as opposed to in care homes or hospitals) we may get a better awareness of what is really happening. The current debate suggests that bars and restaurants are somehow the number one source of Covid: this risks diverting attention from hospitals and care homes where much more can be done to minimise infection.

At present it is very difficult to understand the situation given the variety of sources and partial datasets. Particularly if the focus is on daily statistics with little information to interpret their meaning.

Seven months on, and policymakers are still lost in a fog of data. There is an urgent need to provide a coherent, consistent picture that creates an evidence base that allows some understanding of the impact of Covid: how it spreads, and how to stop it. This current confusion inhibits our understanding of the virus, makes us more likely to enact harmful measures – and less likely to take action that could genuinely minimise deaths.