Understanding the role of large gatherings on the NHS
Published 28 May 2020
England has been in lockdown for nearly nine weeks. The rules, which are now being relaxed, have placed pressure on both the economy and individuals.
The five tests that have been set out for the removal of lockdown rules focus on the potential impact of Covid-19 cases on the NHS. While much is still unknown about how and when Covid-19 is spread, our analysis shows that large events held in March may have contributed disproportionately to additional cases and mortality.
Our analysis estimates that the Cheltenham Festival and Liverpool’s Champions League football match against Atletico Madrid led to an additional 37 and 41 deaths at local hospitals between 25 to 35 days later, compared with similar hospital trusts which were used as a control.
Many more people will have been affected. Based on an average infection mortality rate of 0.6%, the 78 additional deaths mean that around 12,900 people will have been infected. Many of these will have required hospital care, including in capacity-constrained critical care units.
While the NHS did have the capacity for excess demand from Covid-19 in March and April, this was only possible due to the creation of additional capacity and demand for normal NHS services plummeting. While some of the additional capacity will remain, regular demand will return – potentially with some of the demand that was lost in March and April. Our work with Sherwood Forest Hospitals NHS FT showed that in some scenarios returning demand could exceed available capacity.
>> Contact us if you are interested in our work with the NHS on Covid-19 or more generally
In the context of relaxing some of the lockdown rules, it is essential to consider the return of some of the ‘lost’ demand and regular NHS demand. These demand streams alone will place significant pressure on the NHS, which will need to operate with stricter infection control measures. In this context, large scale gatherings should be prohibited until there is a better understanding of how the spread of the disease can be mitigated in these environments.
Covid-19 is spread through respiratory droplets from individuals that are infected with the disease. At its peak, the likelihood of encountering someone infected was as high as 1 in 4 in parts of London.
Large events or mass gatherings attended by 100s of 1,000s of people create a much higher risk of encountering someone that is infected. While stadiums may be open and seating not densely packed, attendees are likely to find themselves in busy queues with other people. These queues might for travelling to and from the event, using restrooms, or for refreshments. Even if only one or two people are (highly) infectious, the possibility of encountering these individuals is increased at events.
In March 2020, several large events were cancelled ahead of the lockdown due to safety concerns. Some events did go ahead, such as the Cheltenham Races and Liverpool’s Champions League football match against Atletico Madrid.
It is likely that people attending these events had Covid-19 (symptomatic or non-symptomatic) and infected others – possibly leading to higher mortality “but for” the event.
To estimate additional mortality, we have identified the closest hospital to the events being analysed. For Cheltenham Races, this is Gloucestershire Hospitals NHS Foundation Trust, and for the Champions League match, it is Liverpool University Hospitals NHS Foundation Trust.
Both hospitals show increased mortality over the days and weeks after the event, although this same pattern was seen at most NHS hospitals in England due to the general spreading of the Covid-19 disease.
To estimate the additional mortality that may be due to the event, the expected level “but for the event” must be assessed. To do this , control hospitals are selected from the same region. For Cheltenham Races, this is University Hospitals Bristol NHS Foundation Trust, and for the Champions League match, it is Stockport NHS Foundation Trust. As these hospitals are not of the same size, some difference is to be expected – a larger hospital will tend to have more cases and higher mortality.
To account for this, the control is normalised by adjusting daily mortality by the ratio of beds between the hospitals. This does not control for all factors that may affect mortality, such as the age of the demographics. Equally, death outside of the local hospital, which may be expected as people travel to these events (including from Spain), is not included.
The chart below shows the number of additional daily deaths at the hospitals local to the events compared to the control. These start at roughly equal but then increase to between three and four additional deaths per day from day 25 to 35 – the period corresponding with increased mortality from the progression of the disease.
In total, between and including day 25 and 35 there were:
Cheltenham Races: 37 additional deaths
Champions League: 41 additional deaths
Manchester Derby: 27 additional deaths
Based on an infection fatality rate of 0.6%, these deaths correspond to additional 17,346 infections. Many of these will not have needed hospital care, although up to 500 will have required admission into a hospital bed – potentially in critical care.
This analysis is based on aggregated data, so cannot be used to establish causality.
6. Data sources
The data used and outputted from this analysis are available on request.
For further enquiries, please contact George Batchelor on email@example.com or 07980 804 956.