The UK Leicester COVID-19 ‘exceedance’ May–July 2020: An analysis of hospitalised cases

We noted with interest the findings of Wang et al. 2020,¹ who used viral phylogenetic analysis to identify the origin of the SARS-CoV-2 infection in their case report of a Taiwanese traveller, near the beginning of the COVID-19 pandemic (January-February 2020). This application of viral phylogenetics can be much more conclusive at the start of an outbreak or epidemic when the numbers of infected cases circulating in the community are very low.

Six months later, on the other side of the world in the UK at the end of the first wave of the COVID-19 pandemic, most cities were emerging from lockdown restrictions. However, a few cities, including Leicester, were showing persistently high numbers of community COVID-19 cases, despite several months of national lockdown measures.²

In June 2020, Leicester became the city with the highest number of ongoing community COVID-19 cases, with 420 cases per 100,000 during 1-24 June 2020, compared to other nearby cities during the same period, e.g. Nottingham with 44 per 100,000 and Coventry with 19 per 100,000, as detected under so-called ‘Pillar 2’ testing. ‘Pillar 2’ testing was performed at community (i.e. not hospital-based) ‘lighthouse’ laboratories on samples received from community testing centres.³ Between March and June 2020 ‘Pillar 2’ testing in the area increased 27 fold from 4911 to 130497 (Open Leicester data: https://data.leicester.gov.uk).

A contemporary report by Public Health England (PHE) documented the ‘exceedance’ of these Leicester COVID-19 cases in the community,² culminating in the imposition of a local lockdown on 30^th June 2020 at a time when hospital admissions for COVID-19 were falling. Unfortunately, this Pillar 2 test data is not routinely available to hospital teams. In contrast, ‘Pillar 1’ testing is performed on hospitalised cases of COVID-19 in hospital-based diagnostic laboratories, who then have immediate access to the results. So it is possible to see an apparent disconnect between the COVID-19 situations perceived in community (Pillar 2) versus hospital (Pillar 1) settings.

Ironically, this same PHE report also showed that the numbers of hospitalised COVID-19 cases in Leicester were actually decreasing leading up to the June 2020 community ‘exceedance’. At the same time only a marginal increase in the number of COVID-19 related calls to NHS 111 online or primary care were observed in June 2020, according to analysis of the local LLR SAGE group (comprised of leads from Public Health, Social Care and health care providers). So it was a surprise to many working in healthcare in Leicester when an extended local lockdown was imposed in Leicester on 30 June 2020. Thus, Leicester became the first city in the UK to undergo a prolonged local lockdown, which has lasted until into the third national lockdown (January-April 2021).⁴

Here we analyse SARS-CoV-2 viral sequences obtained from the Pillar 1 testing of samples from hospitalised COVID-19 patients, during the May-July 2020 COVID-19 ‘exceedance’ period in Leicester, UK.

Patients whose respiratory swab samples tested SARS-CoV-2 positive by RT-PCR (reverse transcription polymerase chain reaction) were sent to a designated COG-UK laboratory in Nottingham for full-genome sequencing.⁵ The resulting viral sequences were tagged with the following anonymised metadata: sample collection date (yyy-mm-dd)_patient age (y-year)_sex (Mal/Fem)_ethnicity (White-Wh, Asian-As, Black-Bl)_outer postcode_ward code (from A to AA).

This allowed us to assess whether these hospitalised patient viruses were epidemiologically clustered in any way, e.g. by age, sex, ethnicity, geographically in the community, or within the hospital by ward (i.e. indicative of possible nosocomial infection and transmission).

Maximum likelihood phylogenetic tree construction was performed using the MAFFT online alignment tool (https://mafft.cbrc.jp/alignment/server/ ), with further manual editing using BioEdit (v.7.2.5), then using the double precision version of FastTree (v.2.1.11: http://www.microbesonline.org/fasttree/) to resolve very short branch lengths, as these SARS-CoV-2 sequences were likely to be very similar. The final tree, with Shimodaira-Hasegawa-like branch support values (as implemented by FastTree), were displayed and edited within Figtree v.1.4.4 (https://github.com/rambaut/figtree/releases/tag/v1.4.4).

In total, 176 COVID-19 patient samples had sufficient viral load for successful full-genome sequencing (aligned and trimmed to 29,491 bp length). The results are presented as phylogenetic trees (Figs. 1 , 2 ), based on full genome SARS-CoV-2 sequences tagged with the relevant metadata.

Fig. 1.

Showing outlying COVID-19 ‘Pillar 1’ Leicester patient viral sequences (more details can be seen by zooming in). The tree includes a total of 177 full-length SARS-CoV-2 sequences (29,491 bp in length), including 176 Leicester patient sequences and the Wuhan reference sequence (NC_045512). See main text for further details of phylogenetic tree construction.

Fig. 2.

Showing cluster patterns after removing outlying sequences (more details can be seen by zooming in). The tree includes a total of 147 full-length SARS-CoV-2 sequences (29,491 bp in length), including 146 Leicester patient sequences and the Wuhan reference sequence (NC_045512). See main text for further details of phylogenetic tree construction.

Fig. 1 shows the full tree with the 177 Leicester Pillar 1 SARS-CoV-2 sequences including the original Wuhan virus reference sequence (NC_045512), highlighting the outlying sequences (representing the most diverse viral sequences). Although there is some clustering of similar sequences (highlighted in coloured fonts), there is no distinct pattern of epidemiological linkage between the sequences discernible from any of the tagged metadata.

Similarly, for Fig. 2 (146 Leicester Pillar 1 sequences with the Wuhan reference sequence, where the more outlying sequences seen in Fig. 1 have been removed for clarity), although there are multiple clusters of similar sequences (highlighted in coloured fonts), again, there is no obvious epidemiological linkage from any of the tagged metadata.

Further data about these individual patients’ movements and community contacts would be needed to confirm such epidemiological clustering, which was not available on this occasion. However, this method of tagging relevant epidemiological metadata to the viral sequences will be a useful approach for future analyses.

Despite the surge in detection of community COVID-19 cases in Leicester, the accompanying decrease in the numbers of COVID-19 cases being hospitalised during May-July 2020,² indicated that most cases were not clinically ill enough to require hospitalisation, consistent with the shift to rising infection rates in younger subgroups of the population.

Note that this ‘exceedance period’ occurred before the emergence of the more transmissible UK Kent (B.1.1.7) variant in September 2020.^{5 , 6} This suggested that the ongoing high community transmission of SARS-CoV-2 during this Leicester ‘exceedance’ was most likely driven by human behaviour without any additional, enhanced intrinsic viral transmissibility.

Several related reports have since suggested that populations in which multi-generational households are common, whose family members work in sectors that cannot benefit from the UK Government ‘furlough’ scheme, may experience higher levels of SARS-CoV-2 transmission.⁷ This is despite national lockdown restrictions, where such household members, understandably, need to continue to work in the community to maintain a basic income, when self-isolation should be in place.