Crunching the data from England’s arrival testing programme

18 Jun

The Department of Health and Social Care (DHSC) is responsible for England’s COVID testing programme. It publishes weekly statistics and has just released its latest update. Most statistics are updated every week, but the data from the testing of international arrivals is only revealed every three weeks “to align with the Joint Biosecurity Centre's assessments of international travel restrictions”.

Yesterday was one of those special updates which includes the arrivals testing figures for the three-week period May 20 to June 9. What do they show?

Overall statistics

First a quick technical caveat. All the figures which I will share on the arrival testing programme come from crunching the data from “Table 21”, which shows the figures broken down by country. Any figures which are below three are suppressed and I have set these figures to zero. So all the total figures will be very slightly underestimated because of some missing 1’s and 2’s, but I don’t think that will change any of the conclusions.

During the three week period ending June 9, about 71,000 people were tested. To put that into context, a total of 12.8 million people were tested in England over the same period.

0.9% of the travellers tested positive, a slightly higher rate than for the overall testing programme (0.65%).

The chart below shows the figures split by the red / amber / green country classification. That can only be done where the country is known, and there is a remarkably large number of cases where the government doesn’t seem to know where people have come from, which is a bit worrying. The statisticians provide the following note:

“… the country / territory the passenger returned from is self-reported. In cases where a travel destination has not been reported, this is classified as 'Unknown'.”

There is also a category for travellers where multiple countries had been visited. Given how many of these tested positive, it looks to me that at least one of those countries was probably a red list destination. The most likely explanation for multiple countries is that these are people that travelled to the UK via a hub in a third country.

38% of the positive tests were sequenced and 49% of the sequences showed a “variant of concern”. That was almost certainly the Delta (“India”) variant, given the country mix (see later for details of this). The sequencing does look to have been concentrated on arrivals from red list countries, with 47% of the positive samples from red list and “multiple” destinations sequenced. Only 11% of positive cases from amber list countries were sequenced during the period.

*Source: GridPoint analysis of Table 21 data from NHS Test and Trace Statistics (England), published June 17, 2021*

Other than the surprising number of “unknowns”, the other thing that jumps out at you from this data is the extremely low number of travellers from green countries. The country classifications are the current ones, with Portugal rated amber. There were just 503 passengers from the eleven countries that are still on the green list during these three weeks. Portugal was on the green list for much of the period, and there were 1,439 arrivals from there. But even if Portugal was included in the green category, it would still have accounted for fewer than 3% of the overall passengers. That demonstrates the extent to which the current UK traffic light system is essentially a “two colour” system in practical terms.

Over a third of passengers came from red countries. This seems a surprisingly large number, given that the only travel that is permitted from red countries is UK citizens returning home. Almost 1,300 people a day are having to check into government managed quarantine hotels for 10 days at a cost of £1,750. They are testing positive at a rate of 1.8%, almost three times the national test positivity rate.

The test positivity for arrivals from amber countries is much lower, at 0.4%. That’s actually lower than the 0.65% national average. However, given that the general testing programme is biased towards people with a higher risk, a better comparison would with the ONS estimates of community prevalence in England. That was 0.18% during this period, meaning that arrivals from amber list countries were twice as likely to be infected as the general population in England at the same time.

Who is travelling anyway?

The following chart shows the number of arriving passengers for the top countries. India still tops the list, despite having been classified as red since April 23. Number two is Pakistan, which has been on the red list even longer.

Which are the highest risk countries?

The next chart shows the countries with the highest volumes of positive tests. I’ve also shown the test positivity rate. Many people, including me, questioned the reclassification of Portugal from green to amber. To be fair, this data does suggest that the country would look a bit out of place if it was still rated as green, with test positivity of 1.1%, which is higher than many amber rated countries on the list.

The arrival testing data is only one of the things that inform the Joint Biosecurity Centre’s recommendations on country classification. However, based on this data, I wouldn’t be surprised to see Iraq (3.7% positive) and Uganda (5.0%) added to the red list at the next update. In terms of moves the other way, perhaps there is a case for Turkey moving from red to amber. Finally, both Romania and the United States have very low test positivity rates of 0.3%, with Bulgaria also low at 0.5%. Those are still above England’s 0.18% community prevalence rate, but could they be low enough to allow a move to green status?

Variants

The final bit of insight we can extract from this data relates to the detection of variants. Several countries are on the red list because of the presence of variants of concern. Which countries look bad on this measure?

There are only five countries where three or more cases of variants of concern were detected by the UK’s arrival testing programme over this three week period. They are all on the red list - don’t expect any of these countries to come off it any time soon.

What level of risk reduction is the system delivering?

It is worth reminding everyone that every traveller had been required to get a negative test before even being allowed on the aircraft. The fact that anyone tested positive after arrival demonstrates the fact that pre-departure testing alone does not remove all risk of importing cases.

The arrival testing caught 623 cases in a three-week period, or 30 a day. Almost all of that came from red list countries.

No green list country generated three or more positive tests. The arrival testing system caught 89 positive cases from amber list countries in three weeks, or about 4 per day. No variants of concern were detected. Is that a meaningful enough reduction in risk to justify the costs and inconvenience of all that testing and home quarantine? The aviation industry would probably argue not.

But the number of cases actually being caught by the system is not the real risk reduction. The big impact is the depressing effect on travel volumes caused by the cost, inconvenience and uncertainty created by the system.

A cynic might argue that it is purposely designed to achieve just that.

COVID-19TestingUKTraffic Lights

Robert Boyle