Only half of total cases by ONS numbers are diagnosed
The ONS carries out weekly swab sampling of a random sample of the population and uses this to estimate the total number of SARS-CoV-2 cases in the community as a snapshot in time. Assuming each infection leads to a two week window of positivity, the number of new infections each week can be calculated by halving the total. Although this may seem crude, the total estimated number of cases from June 2020 until September 2021 by this methodology is 13 million for England. The ONS’ own estimate for aggregate new cases per day works out at 13 million for the same time period.
Despite extensive testing — at great expense — the UK Government testing programme has diagnosed 6 million cases in that same time period. If you believe they are both right, that would beg the question: what is the point of a mass population testing programme that leaves half of cases undiagnosed?
Some might argue that a proportion of these ONS ‘cases’ are asymptomatic. However, ZoeApp estimates of symptomatic cases in the community are similarly high. In fact, combining ZoeApp data with ONS would suggest that at least 80% of the ONS ‘cases’ are symptomatic.
Take the 6 weeks, from 16 August until 27 September. ONS estimated there were 2.1 million new ‘cases’ in England and ZoeApp estimated a similar number for the UK, which proportionally works out at 1.7 million symptomatic ‘cases’ in England.
| Source of data | Asymptomatic | Symptomatic | Total | Notes |
| PHE (Pillar 2 community testing), new positive tests in week 33-38 | 310,000* | 679,000* | 989,000 | No modelling. Results from people who were tested. |
| ONS population estimate (Sum of estimates each week 33-38 divided by 2) | 2,118,000 | Modelled data based on random sample tested. | ||
| ZoeApp (Estimates at end of each week 33-38 divided by 2) (Adjusted by 84% to give estimate for England not UK) | 1,746,000 | Modelled data based on sample surveyed about symptoms and tested. |
Table 1: Estimates for number of COVID cases by different methods and by symptomatic status. Calculations for symptomatic and asymptomatic pillar 2 PCR positives are described here.
Perhaps testing has improved, with 47% of ONS ‘cases’ diagnosed in pillar 2 in the most recent 6 weeks. However, there is something odd here. Only 39% of the symptomatic ‘cases’ predicted by ZoeApp were diagnosed on Government testing. However, over 80% of the possible asymptomatic ‘cases’ predicted by ONS and ZoeApp were diagnosed.
How have we created a testing system, based on symptoms, that is 2 times better at diagnosing asymptomatic ‘cases’ than symptomatic ‘cases’?
It makes no sense.
