Cumulative Cases by Specimen Date
In the Public Health England dashboard, there is a table of confirmed cases by specimen date. Journalist Robert Peston (ITV) sounded on Twitter:
As you can see it makes no sense at all. There is weird repetition and none of the columns reconcile.
This article aims to make sense of what the data table shows.
Thinking about cumulative cases
There are many graphs in Public Health England’s COVID-19 data dashboard. Among them, there is the total number of lab-confirmed SARS-CoV-2 cases in England. This table is by specimen date.
We need to think about what each line is showing. This is cumulative lab-confirmed cases by the specimen date.
The specimen date is the day that the tested person gave their sample. This is important for understanding the epidemic. Labs may send their analyses across in batches. Measuring cases by reporting date means we would be, in part, measuring when labs send batches.
The line for 5th July shows: all lab-confirmed cases in England with a specimen date of 5th July or earlier. There are 47 new cases, so that is the ‘change’. That figure is added to those already reported, giving the new total.
The fourth column is the sum of the second and third columns.
The cumulative total may be easier to see when you look at daily increments.
What is the total number of new confirmed cases against each specimen date?
- 05 July 2020: 47
- 06 July 2020: 100 = 47 + 53
- 07 July 2020: 207 = 47 + 53 + 107
- 08 July 2020: 444 = 47 + 53 + 107 + 237
- 09 July 2020: 776 = 47 + 53 + 107 + 237 + 332
- 10 July 2020: 806 = 47 + 53 +107 + 237 + 332 + 30
Each line shows all lab-confirmed cases in England with a specimen taken on that date or earlier.
Before 11th July, this table also included the reporting date. This may have been a source of confusion.
By definition, the reporting date has no new cases. That means the reporting day will equal the latest date with new specimens.
For example, the 11th July line would show:
- Previously reported: 248,089
- Change: 806
- Total confirmed cases: 248,985
This is also true for any date in the future.
Many thanks to PHE
Prof Oliver Johnson (Bristol) highlighted an error on the new dashboard.
The graph had intended to show COVID-19 deaths per 100,000 people — by age and sex. There appeared to be a calculation error.
Within two days, Public Health England resolved the problem. Many thanks to their analysts.