I analyse an anonymised data dump of 1.9 million admissions records to the emergency departments of an undisclosed NHS Trust for the period June 2017 – September 2021
1) So you stated that there are 1.9 million records in this dataset. The raw chart from part 1 suggests like 1,000 admissions per day, over about 4.5 years, only covers a small fraction of that. Where are the rest?
2) Is it possible people did a bad job coding and that maybe a chart of (A or B or C) is relevant? Maybe that would be a best-case scenario in trying to demonstrate pandemic "volume". It looks like this would end up with two peaks of significant volume. It also looks like doing so might make covid in Nov 2019 pop.
3) "Respiratory proportion" is the most generous chart for indicating pandemic volume. Why no apparent respiratory disease season in winter of 2021? Pull-forward/harvesting effect? Policies making even "regular" viral cases stay home unless severe? Something else? If policy was making the flu disappear for the ER, then that might conceal the existence of a covid pandemic.
1) 1000 admissions per day every day for 4.5 'standard' years does indeed generate 1.64 million but mean daily admissions fetches-up at 1120, with a median of 1156, and the time span equates to 4.71 standard years, hence nothing has got lost!
2) Yes, we must expect coding errors and OR statements should help - a great idea! I will add this refinement.
3) Yes indeed, some curious things emerge and this must be borne in mind when trying to grasp what has been going on through a consideration of the numbers alone. Let's see what that OR count turns up...
In part 3 you'll see me start out with a different definition for a probable ILI/COVID case. What I've done is add a new section below this that embraces your excellent suggestion.
1) So you stated that there are 1.9 million records in this dataset. The raw chart from part 1 suggests like 1,000 admissions per day, over about 4.5 years, only covers a small fraction of that. Where are the rest?
2) Is it possible people did a bad job coding and that maybe a chart of (A or B or C) is relevant? Maybe that would be a best-case scenario in trying to demonstrate pandemic "volume". It looks like this would end up with two peaks of significant volume. It also looks like doing so might make covid in Nov 2019 pop.
3) "Respiratory proportion" is the most generous chart for indicating pandemic volume. Why no apparent respiratory disease season in winter of 2021? Pull-forward/harvesting effect? Policies making even "regular" viral cases stay home unless severe? Something else? If policy was making the flu disappear for the ER, then that might conceal the existence of a covid pandemic.
1) 1000 admissions per day every day for 4.5 'standard' years does indeed generate 1.64 million but mean daily admissions fetches-up at 1120, with a median of 1156, and the time span equates to 4.71 standard years, hence nothing has got lost!
2) Yes, we must expect coding errors and OR statements should help - a great idea! I will add this refinement.
3) Yes indeed, some curious things emerge and this must be borne in mind when trying to grasp what has been going on through a consideration of the numbers alone. Let's see what that OR count turns up...
In part 3 you'll see me start out with a different definition for a probable ILI/COVID case. What I've done is add a new section below this that embraces your excellent suggestion.