Being a scientist working with Trump under public scrutiny has to be like walking along a arête, a knife like ridge between two mountain peaks. Falling over one side is loosing one’s position (and influence) by clearly contradicting Trump, even when he is wrong. Falling over the other side is saying something that is questionable scientifically, thereby loosing ones credibility, in order to pacify Trump. It seems as if one of the key scientists (Dr. Birx) may have fallen off the ridge on the side of trying to justify Trump’s optimism. If so, we should question everything she says. Credibility is hard to gain, and easy to loose.
Whom am I to say?
This is a hard thing for me to write. I am not an epidemiologist. In any per review in an epidemiological journal, I would not qualify as a reviewer. So I am out of my discipline , I admit it. The scientist in me (Physicist) says, I am not qualified to judge, so rather than risk my own credibility I should remain silent. Credibility is hard to acquire, and easy to loose, and is the currency of the successful scientist. But in this pandemic the risk of staying silent perhaps outweighs the risk of losing ones credibility. I also need to walk an arête.
On the other hand, as a computational physicist I find I can read the technical literature on the mathematics of epidemiological modeling and understand it. In fact, my work in modeling physical systems (from graduate student thru retirement) involved far more involved mathematics than is present in at least the basic epidemiological models based on differential equations. Some of the models developed are impressive team efforts. The simpler, well known, epidemiological models can give plenty of insight.
My weakness with the modeling is in correlating policy to the mathematical constants in the equations. This is made more difficult because the analysis of COVID-19 is still in process. Perhaps it is even more difficult to know societies response to the various public policies.
Misleading statements on Attack Rate
Dr. Birx suggested that the models are inaccurate and overstate the threat of Covid-19. If you like you can see a video of her presentation on 3/26/2020 here
I have a few questions about her presentation:
“In no country to date have we seen an attack rate over 1 in 1,000” (1:04 in the above video).
First, what is the definition of attack rate? If I go by the CDC’s online self study course, which also agrees with other authors.
“Incidence proportion is the proportion of an initially disease-free population that develops disease, becomes injured, or dies during a specified (usually limited) period of time. Synonyms include attack rate, risk, probability of getting disease, and cumulative incidence. Incidence proportion is a proportion because the persons in the numerator, those who develop disease, are all included in the denominator (the entire population).” {I added the Bold}
Now the population needs to be specified, and she did, a country. But the period also needs to be specified. She added “to date.” When you are in the middle of the pandemic the attack rate “to date” doesn’t tell you much! What you would like to know is how large the attack rate will become! However, the statement was almost certainly not true when she said it.
Italy, on 3/25/2020 had 74,386 cases with a population of 60.5 million or an attack rate of 1.2 per thousand. This data is from the John Hopkins tacking website on 3/25/2020 at 1:12 pm. The same number is in the World Health Organization Special Report number 66.
Worse than being wrong, this is highly misleading. Few countries have yet been through the entire pandemic sequence. So the number for Italy has already risen. Right now (3/28/2020 2:30 pm) Italy has 92,472 cases for “to date” attack rate of 1.53 per thousand and rising.
Restricting the geography to country allows for a larger denominator and lower rates. Why does this matter? New York city (population: 8.7 million) the evening before her talk,3/25/2020 had 20,011 cases according to the John Hopkins site. This is 2.3 per thousand and climbing! Now that has climbed to 29,158 so the “to date” attack rate is 3.34 per thousand and climbing.
In addition, in the US we are not testing everyone with symptoms. So the figures for New York City are lower that the actual number of infections. It takes time for symptoms to emerge. So the current testing measures what has happened a few days ago (or longer).
Ironically at 4:13 she acknowledges that some of us will look up the numbers (yup) and that we will only find “small countries” that will top 1 in 1,000. She didn’t do her homework, even though between 1:41 and 1:57 she talks about Italy.
At best her statement wasn’t up to date (she should have checked). At worst it was deliberately misleading.
Can 20% of the Population become infected?
She suggested that models that predict that 20% of the US could be infected were over estimating the problem. (2:11) She didn’t actually say that, but it was surely implied. Just before that was a general downplaying of the accuracy of the models, and assertions that they were not predicting the correct behavior.
Can 20% of the United State become infected? Of course we cannot fully know the what will happen in the United States, much of that depends on how we react to the situation. We can look at the ordinary flu data to see if it is POSSIBLE for 20% to become infected. Again lets go back to the CDC website.
This is directly from the website (3/28/2020)
“A 2018 CDC study published in Clinical Infectious Diseases
looked at the percentage of the U.S. population who were sickened by flu using two different methods and compared the findings. Both methods had similar findings, which suggested that on average, about 8% of the U.S. population gets sick from flu each season, with a range of between 3% and 11%, depending on the season.
Why is the 3% to 11% estimate different from the previously cited 5% to 20% range?
The commonly cited 5% to 20% estimate was based on a study that examined both symptomatic and asymptomatic influenza illness, which means it also looked at people who may have had the flu but never knew it because they didn’t have any symptoms. The 3% to 11% range is an estimate of the proportion of people who have symptomatic flu illness.”
We vaccinate people for the flu. Looking around the web perhaps 50% of the population gets a flue shot. However there is no vaccine for the current SARS-CoV-2 virus. So the 3% to 20% range of the number above is more like 6% to 40% of the population that is not vaccinated. It should be noted that the H1N1 pandemic in 2009 is estimated to have effected 19.9% of the population.
https://academic.oup.com/cid/article/52/suppl_1/S13/498323
(see last paragraph before the summary)
So yes, 20% of the population COULD (not will, COULD) become infected. It is our job to prevent that.
Again, her statements are at best misleading, playing down the dangers that we confront.
60% to 70% in 8 to 12 weeks is …. (3:46-3:59)
Dr. Birx closes with a statement about no models show 60% to 70% of the US affected within the next 8-12 weeks. Let me start this off by saying that I do not think that the pandemic will get that bad in the US. I have faith that enough people will act appropriately to prevent this. Notice I didn’t base my comment on models. Non the less, we should parse the numbers and ask if this is possible.
Before Dr. Brix spoke John Hopkins reported 68,572 confirmed cases in the US. The population of the United States is roughly 327 million. 70% of that is 229 million. Thus the number of cases would have to grow by a factor of 4,768 (327,000,000/68,572). If (and its a very big if) we were to continue to grow exponentially this would require a little over 12 doublings. (212 = 4,096). 12 doubling in 12 weeks is one doubling a week. Is that possible? Take a look at the data yourself from the WHO, plot it up in excel and fit an exponential to it. So yes, IF the current trends continue, this is mathematically possible. Is it likely? Only if we are very foolish and twiddle our thumbs, refusing to change our behavior, while the epidemic runs through the country.
My issue with Dr. Birx’s statement is the potential that it be interpreted as saying: we don’t have to worry. But with Trump’s wanting to start the company up, and Dr. Birx as his adviser, this is the wrong message to send. At Best Dr. Birx should have qualified the statement by indicating whose models she was considering.
Be concerned, but funnel that concern into actions that help. Practice social distancing, wash your hands and do it right, know the symptoms and if you have them, self isolate. We can do this, but not without some self sacrifice and effort.
The Biggest Issue
Perhaps the biggest problem with Dr. Birx’s statements is she hasn’t shown us her data. Perhaps she would make the argument that she was only talking about the one study she mentioned. But she inferred much more.
Scientists are data based. If a scientist is unwilling to show the data, then they cannot be taken seriously. Statements about models that are hidden from public view are not to be trusted. A scientist is obligated to share their work, their data, and their reasoning so that others can examine, and critique their work. Any scientist that makes statements about the data, that they are unwilling to submit to per review, are not to be trusted. Period, end of story, drop the mic. It is per-review that puts the disinfectant of public disclosure on ones work.
Conclusion
I morn the loss of a trust worthy insider in the Trump inner circle. Perhaps this was all a show to keep her job, thinking that she can more good in her job than outside. But until proven otherwise, I will treat anything that Dr. Brix says with a bit of suspicion. That doesn’t mean that anything she say is false. It just means we need independent verification before we can believe what she says.
Sadly a scientific colleague may have fallen off the arête, and I don’t have a rope to throw to her.
