Latest PHE Data Shows Vaccine Effectiveness Down to Just 15% in the Over-50s, 37% in the Under-50s. Deaths Cut by 76% in Over-50s, But Just 12% in Under-50s

Public Health England (PHE) has released a new technical briefing on the variants of concern, number 21, and this allows us to update our estimate of (unadjusted) vaccine effectiveness against the Delta variant using the data it provides on confirmed Delta cases.

We subtract the figures in briefing 21 from those in briefing 17 to give the figures for the period June 22nd to August 15th. We also use figures for proportions of the population vaccinated by age derived from the PHE Covid surveillance reports.

Starting with the over-50s, for the period June 22nd to August 15th, PHE reports 29,282 Delta infections in the double vaccinated and 3,915 in the unvaccinated. PHE figures show that in this period the proportion of the over-50s double vaccinated was stable at 88% and the proportion unvaccinated was 10%. Calculating the vaccine effectiveness against Delta infection in the over-50s (1-(29,282/88%)/(3,915/10%)) gives a figure of just 15%, down from 17% using data from the briefing two weeks ago. This is very different to the estimate in the recent Oxford University study using ONS survey data, a study which I criticised for numerous implausible findings.

With regard to deaths with Covid (within 28 days of a positive test), PHE reports 602 in the double vaccinated and 280 in the unvaccinated in the over-50s in this period. This works out (1-(602/88%)/(280/10%)) at a vaccine effectiveness against death of 76%, down slightly from 77% using data from the previous briefing. This is a 76% reduction in mortality including any reduced risk of infection, not in addition to it. It’s an encouraging figure, albeit lower than earlier studies have suggested.

For the under-50s, for the period June 22nd to August 15th, PHE reports 36,855 Delta infections in the double vaccinated and 125,394 in the unvaccinated. PHE figures show that on June 20th 61% of under-50s were unvaccinated while 18% were double vaccinated. On August 15th those figures were 52% unvaccinated and 35% double vaccinated. Taking the average of these gives 56% unvaccinated and 26% double vaccinated for the period. Using this to calculate the vaccine effectiveness against Delta infection in the under-50s (1-(36,855/26%)/(125,394/56%)) gives a figure of 37%. This is higher than in the over-50s, but still very low and much lower than earlier studies (including the trial) indicated.

For deaths, PHE reports 27 in the double vaccinated and 66 in the unvaccinated in the under-50s in this period. This works out (1-(27/26%)/(66/56%)) at a vaccine effectiveness against death of just 12%. Why this would be so much lower than in the over-50s is unclear, but it’s worth bearing in mind that these are small numbers of deaths which may make the estimate unreliable.

These figures are much lower than those commonly quoted and used in modelling, and if they are closer to the truth then they mean the official, self-congratulatory estimates of “100,000 deaths” and “24.4 million infections” prevented by the vaccines are huge overestimates.

By plotting the differences between the reported total Delta cases in the last three briefings we can also get a picture of how they are changing over time in the different age and vaccine-status cohorts. The red and yellow lines in the chart below show that new Delta infections in the unvaccinated have stayed largely flat over the past fortnight, but those in the vaccinated (for this purpose, all who are at least 21 days after their first dose) have surged, so that a majority of new infections (44,038 out of 76,117, or 58%) are now in the vaccinated. This reveals that the recent modest increase in positive cases in England is being driven primarily by infections in the vaccinated. The fainter lines show the trends in the over- and under-50s, indicating that in both age cohorts new Delta infections in the vaccinated now outnumber those in the unvaccinated. This helps to explain the declining vaccine effectiveness estimates given above.