by Nic Lewis, Watts Up With That:
- A new variant, B.1.1.7, of the SARS-CoV-2 virus has recently spread rapidly in England
- The public health agency’s best estimate of B.1.1.7’s weekly growth rate advantage is 1.51x
- They mis-convert this in a reproduction number ratio of 1.47; converting appropriately gives a ratio of 1.25
- Confident claims by the UK government scientific advisers that the higher growth of B.1.1.7 is due to increased transmissibility are misplaced; it could be partly of wholly due to other factors
- 1.1.7 has not shown a greater growth rate advantage than two previous variants did, both of which are now thought to have no greater transmissibility than previously existing variants
- There is little evidence that B.1.1.7 is more virulent, or likely to be resistant to existing vaccines
TRUTH LIVES on at https://sgtreport.tv/
The apparent rapid growth in England of a new variant of the SARS-CoV-2 virus that causes COVID-19 has led to dire warnings by those advising the UK government. Their advice suggested only that the new variant was more transmissible (more infective), not that it was more virulent (causes more serious illness). Nevertheless, it resulted in swift (many would say panicky) actions first by the UK government and then by governments of many other countries. The UK government imposed further restrictions on people’s freedom to mix and to move, while other countries banned travellers from the UK. Many millions of people in the UK had to cancel their plans for the Christmas holiday at very short notice, in addition to having their freedoms further curtailed thereafter. In this article I examine to what extent the advice that led to these damaging government actions was justified.
The new strain, B.1.1.7, and its spread in the UK
The UK government agency Public Health England (PHE) termed the new variant VUI-202012/01, and now VUC-202012/01, but I shall refer to it by the scientific name given to its lineage, B.1.1.7 (Rambaut et al.), or just as “the new variant”. The lineage involves 8 amino acid changes (6 mutations and 2 deletions) in the gene for the important spike protein, along with 9 amino acid changes in genes for other proteins. The lineage has sometimes been referred to just by the name of the best known mutation it possesses, N501Y, but doing so is to be avoided as there are other variants that also have this mutation.
Rambaut et al. have this to say about the new lineage:
The B.1.1.7 lineage carries a larger than usual number of virus genetic changes. The accrual of 14 lineage-specific amino acid replacements prior to its detection is, to date, unprecedented in the global virus genomic data for the COVID-19 pandemic.
They also identify three of the mutations in particular (including N501Y) as being suspected of having potential biological effects.
B.1.1.7 was first detected in SARS-CoV-2 sequenced from a sample collected in south-east England on 20 September 2020, since when the cluster of cases has grown rapidly and spread to other locations. The UK sequences many more SARS-CoV-2 genomes than any other nation, and more than the rest of Europe combined, so the fact that B.1.17 was first detected in the UK does not necessarily imply that it originated there. The lineage has also been detected in several other countries and may well now be widespread.
Growth of the B.1.1.7 lineage in the UK can be tracked in sequencing data uploaded to GISAID. I used the COVID-CG processing facility to select each day’s sequences with all eight B1.1.7 spike gene mutations. As the daily data were noisy and few sequences were dated after 12 December 2020, I took 7-day moving averages, centred up to 9 December. Figure 1 shows the resulting proportion of all UK sequences represented by the B.1.1.7 lineage since its first emergence. It should be noted that the proportion of non-B.1.1.7 sequences represented by the areas in which B.1.1.7 first grew to prominence may have increased over time, resulting in the growth shown overstating how fast it grew in individual areas or in the UK as a whole.
Figure 1. The proportion of all SARS-CoV-2 genomes sequenced in the UK made up by the B.1.1.7 lineage
The higher growth rate of B.1.1.7
A PHE report published on 21 December 2020 presents epidemiological evidence about the growth rate of B.1.1.7 relative to non-B.1.1.7 lineages. By using a proxy marker for B.1.1.7 they were able to utilise data from a significant proportion of the UK ‘pillar 2’ testing programme. Doing so provided a much larger dataset than that of sequenced SARS-CoV-2 genomes, and enabled stratification of weekly data for each of 42 NHS “STP” areas. Figure 2 reproduces Figure 1 of the PHE document, which shows the multiplicative advantage in weekly growth rates of B.1.1.7 cases (the ratio of B.1.1.7 to non-B.1.1.7 week t+1 cases divided by week t cases). The x-axis is for the B.1.1.7 proxy, S gene test negative. The week stated is the base week, so the yellow points reflect ratios of week 49 (week ending 5 December) cases to week 48 cases.
Figure 2. Empirical data analysis of the multiplicative advantage in weekly growth rates. Each point represents the ratio of weekly growth rates between B.1.17 [VOC] and non-B.1.1.7 for an NHS England STP area and week, based on the pillar 2 data shown in Figure S1 of the PHE report. Colours and shapes differentiate calendar weeks. Numbers above 1 show a multiplicative advantage. The blue line represents the mean value for a particular frequency, and the grey lines the 95% envelope. Scatter at low frequencies largely reflects statistical noise due to low counts.
When the new variant represents a small proportion of total cases (under ~ 25%, say), the proxy used is less satisfactory, and there is also a lot of scatter. Nevertheless, the variant’s proxy-based mean multiplicative advantage (ratio) in weekly growth is remarkably independent of its relative prevalence. That supports PHE’s methodology, although the week 48 data suggests that the multiplicative advantage might decline once the variant makes up more than ~25% of the total cases. PHE compute from this data a mean multiplicative advantage in weekly growth of 1.51 for B.1.1.7. By assuming a fixed generation interval of 6.5 days, they convert this into a reproduction number (Rt) multiplicative advantage of 1.47 for B.1.1.7 relative to other variants,.
PHE also estimated the effect of B.1.1.7 on Rt using genomic (sequencing) data for the same areas and weeks. They estimated an additive effect on Rt of 0.57, or 0.74 when the effect was allowed to vary between areas. PHE also estimated the effect on Rt using the PCR test S gene proxy data, adjusted for specificity (which is poor when the S gene negative proportion is low). Their estimates of the additive effect on Rt using that data were 0.52, or 0.60 when the effect was allowed to vary between areas. Using a Bayesian regression model their estimate of the effect was 0.56. However, since any biological difference in infectivity would be expected to cause a multiplicative effect on Rt, and Rt was varying during the analysis period, an estimated additive effect on Rt is less useful and also liable to be less accurate than a multiplicative estimate. In addition all these estimates involve more complicated statistical models, further assumptions and estimates of other variables. I therefore prefer their estimated multiplicative advantage of 1.51 (for weekly growth, prior to conversion to Rt scale), which is directly derived from underlying data. This is equivalent to a logarithmic daily growth rate advantage of 0.059.
Other evidence regarding the faster growth of B.1.1.7
A meeting of the NERVTAG committee – which advises the government on the threat posed by new and emerging respiratory viruses – on the new variant took place on the 18 December 2020. The minutes refer to an estimate from genomic data of a growth rate 71% higher than other variants; none of the documents that was considered by the committee contained such an estimate. It appears from the minutes of a subsequent meeting on 21 December 2020 that this was one of several undocumented estimates from NERVTAG member Professor Neil Ferguson of Imperial College. An alternative regression estimate that he apparently presented indicated that lineage B.1.1.7 had a Rt 0.39 higher than non-variant lineages from early November to early December. This is presumably an additive effect estimate, and is noticeably lower than PHE’s estimates using much the same method. Two other estimates stated in the minutes to be from Professor Ferguson both appear to actually be slightly misstated versions of the PHE estimate of a multiplicative Rt advantage of 1.47 for B.1.1.7.