Secondary attack rate

Devastating New Data From PHE Shows Vaccine Effectiveness Down to 17% and No Reduction in Infectiousness – But Mortality Cut by 77%

Public Health England yesterday released their latest technical briefing on the variants of concern, number 20, and it has some illuminating data.

It provides us with the infection, hospitalisation and death rates for the Delta variant, broken down by vaccination status and age. This allows us to do a calculation of the real-world vaccine effectiveness in the over-50s during the Delta surge, albeit a rough one without any adjustments.

The latest report has data up to August 2nd. If we substract from these values the data in Briefing 17 (up to June 21st) then we get the data covering the period June 22nd to August 2nd, which broadly corresponds to the bulk of the Delta surge in the U.K. The vaccine rollout to the over-50s was basically complete by this point, having stabilised according to PHE at around 88% double vaccinated and 10% unvaccinated (the other two per cent remaining single vaccinated, perhaps due to a bad reaction to the first dose).

In this period the PHE data tells us there were 2,464 Delta cases in the unvaccinated over-50s and 17,926 in the fully vaccinated over-50s. From this we can estimate the vaccine effectiveness against infection in the over-50s during the Delta surge as 17% (1-[(17,926/88%)/(2,464/10%)]). This confirms using additional data the estimate I made last week.

We can make a similarly rough calculation of the vaccine effectiveness against death. Between June 22nd and August 2nd, PHE reports that among the over-50s there were 339 deaths with the Delta variant in the double vaccinated and 167 in the unvaccinated. Using the same proportions vaccinated and unvaccinated as above, this gives a vaccine effectiveness against death in the over-50s during the Delta surge of 77% (1-[(339/88%)/(167/10%)]). Interestingly, this is very similar to the latest estimate of vaccine effectiveness against serious disease from Israel, which is around 80%. This is a decent level of protection and helps explain why the Delta surge had proportionally fewer hospitalisations and deaths, but it is well below the levels suggested by earlier studies and quoted by PHE, which are north of 95%.

Why is the Government Claiming “One in Three” Test and Trace Contacts Become Infected When its Own Data Shows That to Be False?

In a desperate effort to encourage people to self-isolate when pinged by the NHS Covid app or contacted by Test and Trace – even cancelling their wedding day if necessary – the Prime Minister’s official spokesman said yesterday that: “One in three people contacted either by Test and Trace or the app go on to develop coronavirus.”

There’s just one problem with this latest nugget of fear-based nudgery: it’s not true.

Test and Trace get in touch with people’s contacts and ask them to self-isolate. The proportion of those contacts who become infected is known as the secondary attack rate (SAR). Public Health England publishes the data on this SAR from Test and Trace data in its Technical Briefings, so we know what it is. The most recent estimate for the SAR of the Delta variant (in June) is that 10.3% of an infected person’s household contacts become infected (the SAR for non-household contacts is considerably lower).

How, then, can it be true that one in three – 33% – of people contacted by Test and Trace or the app go on to develop coronavirus? That’s claiming the SAR of SARS-CoV-2 is around 33%, but the Government’s own published data says it’s more like 10%.

Can the Government back up its claims, and explain why it is stating that the SAR of Covid is more than three times the figure published in its own data?

What SAGE Gets Wrong: The Evidence that Almost Everyone is Exposed During a Surge and Most Are Immune

During a Covid surge, what proportion of the population is exposed to an infective dose of the virus, which they either fight off with no or minimal symptoms or are infected by? This is one of the most important questions scientists need to answer.

It’s closely related to the question of whether lockdowns work. If lockdowns work then, as per SAGE and Imperial orthodoxy, the restrictions successfully prevent the virus from reaching most people, who remain unexposed and susceptible – and hence in need of vaccination to protect them when the protective restrictions are lifted. If lockdowns don’t work, however, then they don’t prevent the virus spreading, and thus the majority of people will be exposed to it as it spreads around unimpeded by ineffectual restrictions.

Another related question is: What proportion of exposed people are infected? Using ONS data we can estimate that around 10-15% of the country tested positive for SARS-CoV-2 over the autumn and winter. How many were exposed to the virus to produce this number of infections? Was it, say, 10-20%, with half to all of them catching the virus? Or was it more like 80-90%, with around 10% being infected? It’s a question that makes all the difference in our understanding of the virus and how to respond to it.

If almost all are exposed during a surge, and relatively few of them are infected, then a number of things follow. First, most people have enough immunity to fight off the virus when exposed to it, and only a small minority become infected. Second, the surge ends when enough of that small minority who are particularly susceptible to this virus or variant acquire immunity through infection, i.e., when herd immunity is reached. Third, there won’t be another surge or wave until there is a new virus or variant which evades enough of the existing population immunity to require herd immunity to be topped up via a further spread of infections.

If, on the other hand, very few are exposed during a surge, and most of them are infected, none of these things is true. It means: Most people have little immunity and are highly susceptible. A surge which infects 10-20% of the population has exposed not much more than that. The surge does not end because of herd immunity but because of restrictions. And there will be another surge as soon as restrictions are eased or behaviour changes and the unexposed begin to be exposed again. SAGE orthodoxy, in other words.

The evidence, however, is strongly supportive of the first position – ubiquitous exposure – not the second, limited exposure.

Indian Variant Infectiousness Falls Again, PHE Data Shows – Just 10.7% of Contacts Become Infected. So Why Isn’t COVID-19 More Infectious?

The latest Public Health England (PHE) Technical Briefing on the variants of concern is out, and it shows that the infectiousness of the Delta (Indian) variant has dropped again, so that it is now at the same level as the Alpha (British) variant was at the end of March.

Data from PHE

The Delta variant secondary attack rate (SAR – the proportion of contacts an infected person infects) is down to 10.7% this week, from 11.4% a week ago (see graph above). That means that around 90% of the close contacts of people infected with the Delta variant are not infected by them. Once again, this shows how claims that the Delta variant is “60% more transmissible” or similar fail to make clear that transmissibility changes and often declines over time, and that even if the Delta variant is currently more infectious than the Alpha variant it is still less infectious than the Alpha variant was when it first appeared on the scene.

You might be wondering, though, how a disease that only infects 10-15% of close contacts, even at the height of an epidemic wave, can be called highly infectious. And indeed, how such a disease can cause an explosive epidemic wave at all, and why those waves crash down as quickly as they arise despite not exhausting the pool of susceptible people.

If you do wonder that, you would not be alone. A famed epidemiologist called Robert Edgar Hope-Simpson spent his life observing the idiosyncrasies of seasonal influenza and summarised some of his thinking in a 1981 article entitled “The role of season in the epidemiology of influenza“. More recently, in 2008, a group of scientists took up Hope-Simpson’s investigations (which were inconclusive) and summarised seven conundrums of influenza that he identified, all of which apply equally to COVID-19:

Claims the Indian Variant is “Hyper-Transmissible” are Nonsense – And Here’s the Graph that Proves It

Yesterday I wrote about the latest Public Health England (PHE) report claiming that the Delta (Indian) variant is much more infectious than the Alpha (British) variant. I noted that the main measure of transmissibility – the secondary attack rate (i.e., the proportion of contacts that an infected person infects) – has varied over time.

It occurred to me that it would be useful to plot these attack rate values to show them graphically. So I went through the 15 technical briefings released by PHE so far and extracted the secondary attack data for the three variants (Wuhan, Alpha, Delta) and plotted them in the graph above.

There are a few points worth making from this about the infectiousness of the Covid variants.

First, between 85% and 92% of the contacts of all those infected with any of the COVID-19 variants do not get infected. This is an indication of the high level of immunity (i.e., low level of susceptibility) in the population and the low absolute infectiousness of the virus.

Second, new variants appear to start off with – or quickly acquire – a high attack rate compared to established variants, a rate which then declines. The decline for the Alpha variant occurred prior to any significant vaccine coverage meaning it cannot all be attributed to the vaccines. Conversely, despite the high vaccine coverage in April and May the Delta attack rate spiked. Nonetheless, it has already sharply declined. (The reason for the sudden drop in the Alpha attack rate from 10% to 8% at the end of April is unclear.)

Third, the Alpha variant was once more transmissible than the Delta variant at its recent peak, but is now much less transmissible.

While more data from more variants would help to confirm the patterns here, the data so far suggest that new variants will often be more infectious than established variants to begin with, but this will not last. Further, the degree of infectiousness appears to arise more from factors such as the epidemic phase or the season than inherent properties of the variant, save insofar as the new variant is slightly better at evading our immune defences for a time. And I mean slightly. Only 3% more contacts of those infected with the Delta variant are being infected compared to the Alpha variant, a difference that is dropping week on week.

The upshot is we should stop being frightened by claims that the latest new variant is “50% more infectious”, which are a distraction, and focus on levels of serious disease and how best to cope with them.

PHE Briefing Claims Indian Variant is 64% More Infectious – But Dig Down and the Finding Falls Apart

Public Health England (PHE) released its latest technical briefing (number 15) yesterday on “variants of concern” which claimed the Delta (Indian) variant is 64% more infectious than the Alpha (British) variant. But look closer and you find this headline finding is not all it seems.

In the underlying study the researchers admit they did not control for the crucial factor of household size – bigger households will tend to have more secondary infections because there are more people in the household being exposed. The authors acknowledge that many of the Delta variant households may have been larger than Alpha variant households (say, because of different proportions of different ethnic groups), and also that many of the matched controls may have lived alone – they have no way of knowing.

We did not have information on household size, which is likely to have an effect on the estimates of transmissibility. For example, some controls (sporadic cases) will have lived alone and have no chance of onward transmission within their residence and therefore becoming a household cluster. However, we were unable to identify and exclude these cases in the analysis. Further studies of household transmission that includes denominators of all individuals in the household and their vaccination status are needed to provide improved estimates of household transmission and allow for the calculation of household secondary attack rates.

This factor by itself undermines the entire 64% claim and means it should be ignored.

Separate to this, the technical briefing provides some raw data on secondary attack rates (the proportion of contacts infected people infect) that give us an important insight into the real transmissibility of the variants.