In my thread last night on just a few of the reasons for D-I-R-T-Y data on #COVID19 #vaccination status, I failed to explain the potential impact of duplicate records being created.
Here's that simplified explanation.
1/5
Here's that simplified explanation.
1/5
https://twitter.com/JasonSalemi/status/1473125673258471424
In the fig, we have what actually happened (top), a situation in which the booster got assigned to a different record (treated as a different person in calculations), and a hopefully rare situation in which all three doses were treated as though they belonged to diff people.
2/5
2/5
Here's an example when 7% of people completing a 2-dose series had duplicate records that were created.
Those w/ a completed initial series was UNDERestimated by 6.3%.
Those w/ "only first dose" was OVERestimated by 12.6%.
The "at least one dose" exceeds the pop count!
3/5
Those w/ a completed initial series was UNDERestimated by 6.3%.
Those w/ "only first dose" was OVERestimated by 12.6%.
The "at least one dose" exceeds the pop count!
3/5
This is a major part of the reason CDC has simply chosen to cap its coverage metrics at 95%, why their "% with at least one dose" is still likely OVERestimated, and why their % fully vax and % with booster doses are UNDERestimated.
h/t @EpiMegan
4/5
h/t @EpiMegan
4/5
Please keep all this in mind when reviewing each state's vaccination status breakdown on my dashboard!
covid19florida.mystrikingly.com
5/5
covid19florida.mystrikingly.com
5/5
• • •
Missing some Tweet in this thread? You can try to
force a refresh
