Visualising Drug Combination Efficacy

HCQ got a pretty bad rap, but originally that was mostly because of deeply flawed studies ("airbags deployed in hospitals don't prevent car accident deaths" level bad), and scientifically-illiterate political-agenda-pushing media ( yes, BBC News, I'm looking straight at you ) parroting the meaningless findings. however, this paper from Henry Ford Health System was brought to my attention today by Chris Martenson of Peak Prosperity in his video "Covid-19: Vindication! HCQ+ & Ivermectin Work!". The wall of numbers was very poorly presented, and even Chris mangled them a bit, so I decided to just visualise them, as pictures speak louder than numbers/words.

White is life, red is death. Pink is 95% confidence, the black line within it is the recorded ratio. Oh equal areas are equal proportions. Beware things like Simpson's Paradox if you even think of trying to combine groups (i.e. don't combine percentages). Here - I'll do that for you (and remember - trust, but verify!):

HCQ no +
Azno 108/409=26.4%162/1202=13.5% 270/1611=16.8%
yes 33/147=22.4%157/783=20.1% 190/930=20.4%
no + yes 141/556=25.4% 319/1985=16.1% 460/2541=18.1%

The graph was generated by perl script, as I wouldn't know how to generate such graphs in libreoffice (or whatever they're calling that bloatware monstrosity nowadays) or any other such suite.


I'm not saying these drugs do have any efficacy, this was merely an exercise in visualisation, and as an example I merely chose the first example that looked like it had data that wasn't garbage or otherwise useless - plus that data wasn't already being presented in a intuitively clear way. Since the original flurry of deeply flawed anti- reports, there's been an even bigger flow of even more flawed pro- reports. Bio/medical science publication is not in a good state. Thank goodness for places like Retraction Watch.

Page last knowingly updated: 2022-02-11.

Another hastily constructed page by Phil Carmody
Home / CoViD / visualisation.html