Query check CCI InExtubation consistent: Difference between revisions
No edit summary |
Ttenbergen (talk | contribs) mNo edit summary |
||
| Line 16: | Line 16: | ||
{{Discuss | who = Julie | question = | {{Discuss | who = Julie | question = | ||
* this cross check would not know the difference between (on one day, intubated, extubated, and reintubated) vs (intubated and then extubated); while this hopefully not too common of a thing, would it cause false positives for Pagasa to run after? Ttenbergen 11:33, 2018 October 29 (CDT) | * this cross check would not know the difference between (on one day, intubated, extubated, and reintubated) vs (intubated and then extubated); while this hopefully not too common of a thing, would it cause false positives for Pagasa to run after? Ttenbergen 11:33, 2018 October 29 (CDT) **yes that is correct because both have a difference of zero. Using the cut off >= |2| will get true negatives and not false positives. This query considers only the counts and not the dates which will have numerous scenarios and too complicated to define. For those having a difference of -1,0,1 , there are also the possibility for incorrect sequence of dates of intubation or extubation - this is not captured in the query. If you have other suggestions, let us know.--[[User:JMojica|JMojica]] 12:19, 2018 October 29 (CDT) | ||
*yes that is correct because both have a difference of zero. Using the cut off >= |2| will get true negatives and not false positives. This query considers only the counts and not the dates which will have numerous scenarios and too complicated to define. For those having a difference of -1,0,1 , there are also the possibility for incorrect sequence of dates of intubation or extubation - this is not captured in the query. If you have other suggestions, let us know.--[[User:JMojica|JMojica]] 12:19, 2018 October 29 (CDT) | *** What I was trying to say is that I am worried this would be a false positive, which will then create work load for Pagasa and/or Collectors to confirm. Do we really want cross checks that bring up ''potential'' errors? We have talked about this before, but never really come up with a general answer. Should we take it to task meeting? Ttenbergen 23:45, 2018 October 29 (CDT) | ||
}} | |||