This paper is well worth a read, according to me. Given that we have all kinds of problems doing testing for COVID-19 in the US, the authors do an analysis of the ratio of people seeking medical help for flu-like symptoms, but testing negative for the flu, to people testing positive (compared to previous flu seasons). The idea being that if COVID-19 is spreading, we should be seeing increases in flu symptoms relative to flu diagnoses.
However, I’m confused about Arizona. (See figure 4 in the paper). It looks like its score on this axis has been 2 to 3 standard deviations from the mean for the decade since November.
What’s going on?
Hypotheses:
The Flu season in Arizona was so bad this year that the number of people who came in with flu symptoms skyrocketed, relative to other kinds of ailments, and even though most of those people did test positive for the flu, the wILI term swamped the (1 − proportion of tests positive for influenza), for an overall much higher than usual ILI-.
Is that mathematically plausible?
There was some other flu-like non-flu that was spreading in Arizona since November.
The flu diagnosis process spontaneously failed this season, producing orders of magnitude more false negatives, in Arizona, but not in any other state.
COVID-19 has been spreading in Arizona since November, but somehow there are only 9 serious cases.
COVID-19 has been spreading in Arizona since November, but testing is so ineffective, that there are hundreds of people with severe phenomena-like symptoms, who have not been recognized as COVID-19 cases, somehow.
Based on figure 4, this would predict that the midwest would have a worse epidemic than the pacific NW right now, and twitter reports from hospitals don’t seem to bear that out.
This paper is well worth a read, according to me. Given that we have all kinds of problems doing testing for COVID-19 in the US, the authors do an analysis of the ratio of people seeking medical help for flu-like symptoms, but testing negative for the flu, to people testing positive (compared to previous flu seasons). The idea being that if COVID-19 is spreading, we should be seeing increases in flu symptoms relative to flu diagnoses.
However, I’m confused about Arizona. (See figure 4 in the paper). It looks like its score on this axis has been 2 to 3 standard deviations from the mean for the decade since November.
What’s going on?
Hypotheses:
The Flu season in Arizona was so bad this year that the number of people who came in with flu symptoms skyrocketed, relative to other kinds of ailments, and even though most of those people did test positive for the flu, the wILI term swamped the (1 − proportion of tests positive for influenza), for an overall much higher than usual ILI-.
Is that mathematically plausible?
There was some other flu-like non-flu that was spreading in Arizona since November.
The flu diagnosis process spontaneously failed this season, producing orders of magnitude more false negatives, in Arizona, but not in any other state.
COVID-19 has been spreading in Arizona since November, but somehow there are only 9 serious cases.
COVID-19 has been spreading in Arizona since November, but testing is so ineffective, that there are hundreds of people with severe phenomena-like symptoms, who have not been recognized as COVID-19 cases, somehow.
What’s going on here?
Based on figure 4, this would predict that the midwest would have a worse epidemic than the pacific NW right now, and twitter reports from hospitals don’t seem to bear that out.