Yeah, it did occur to me that maybe there are restrictions on releasing the hospital names if that somehow veers close to potentially identifying individual cases...not really my area of legal/ethics expertise...
-
-
Replying to @ADAlthousePhD @reverendofdoubt and
Though, I don't see how a list of the hospitals (without any counts of the number of cases from the individual hospitals in any given paper) would have that risk here.
0 replies 0 retweets 0 likes -
This Tweet is unavailable.
-
Replying to @reverendofdoubt @mikejohansenmd and
Right, there are separate questions about the quality of the analysis, which I'm not touching yet. At this point, some are wondering about the overall integrity of whether the data or analysis even exist, which is what I'm trying to understand.
1 reply 0 retweets 0 likes -
Replying to @ADAlthousePhD @reverendofdoubt and
There's a distinction between what they did and what you're describing here though. There's no way the analysis they reported could be done with aggregate data only. It would have required individual patient data.
0 replies 0 retweets 1 like -
This Tweet is unavailable.
-
Replying to @reverendofdoubt @ADAlthousePhD and
aside from the methods problems, I just want some explanation of how the number of deaths at the five hospitals in their Australia sample exceeds the total number of deaths in Australia at that time. There could be a simple explanation.
1 reply 0 retweets 0 likes -
Replying to @walidgellad @reverendofdoubt and
I think the "deaths by April 14" and "hospitalizations by April 14" distinction *could* explain that. If patients hospitalized before April 14 died after April 14, they could appear in the analysis as deaths, but would not appear in the country's death count on April 14.
2 replies 0 retweets 1 like -
Replying to @ADAlthousePhD @walidgellad and
I haven't been monitoring the pandemic in Australia, but any patient that was "hospitalized" on, say, April 12, could be included in their analysis. If they died on April 17, they don't show up in the country's death count on 4/14 but could be in the analysis as a death.
2 replies 0 retweets 1 like -
Replying to @ADAlthousePhD @walidgellad and
I find this somewhat unlikely. Even if you accept that every death could be captured in such a database, it would require more than 5 hospitals - 73 deaths includes every single state and a territory. At minimum that's 7 hospitals
1 reply 0 retweets 1 like
But even more, I'm fairly certain not every death occurred in a single hospital in each state (I'd have to check). We've only had ~100 deaths altogether, so even if they include every death until now from 5 hospitals it seems weird
-
-
Replying to @GidMK @ADAlthousePhD and
Also, the sort of database that's described above is fiendishly hard to do in Aus. We have very secure data protection laws - if nothing else, the data usually has to be housed here. Moreover, in most cases doing research of this kind without ethics is illegal
0 replies 0 retweets 1 like -
This Tweet is unavailable.
- Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.