A Boston-based doctor throws rocks at North Dakota

by philg

Atul Gawande, who works in Boston (though maybe soon in D.C., since he was tapped by President Biden to join Dr. Jill Biden, MD on a COVID-19 advisory board), writes in New Yorker: “Inside the Worst-Hit County in the Worst-Hit State in the Worst-Hit Country”.

How much can we trust a scientist like Dr. Atul Gawande, MD, MPH and the science-informed editors of the New Yorker? The article turns out to be about North Dakota, which enjoyed a year of relative freedom (mask law from November 14, 2020 to January 18, 2021) and experienced a lower death rate tagged to COVID-19 than 9 other states. Maskachusetts, in particular, where Dr. Gawande actually lives, has had a higher death rate than North Dakota despite a year of shutdown and masks.

How about the “Worst-Hit Country”? Can we rely on this Person of Science’s statement that the U.S. is the country that has been worst-hit by COVID-19? Statista says no. There are 7 countries, e.g., Belgium, Slovenia, and the UK, that have had a higher death rate.

How are Americans supposed to put their trust in #Science when there are at least two obvious falsehoods in the article’s headline? Or maybe there are three falsehoods, actually! The NYT page on North Dakota allows one to sort the counties of ND by death rate. Ward County, featured in Dr. Gawande’s article, is nowhere near the top: 274 deaths per 100,000, a little higher than the state average of 191, but nowhere near the worst in ND.

So… the headline contains three assertions from a scientist. All three assertions are false, as measured by the outcome that is most upsetting to humans (i.e., death). Said scientist will soon be telling Americans what they can and cannot do… #BecauseScience.

From my 1993 trip through North Dakota on the way to Alaska:

15 thoughts on “A Boston-based doctor throws rocks at North Dakota”

  1. The external hyperlinks on travels with samantha are largely broken. maybe link to archive.org versions of links that have 404’ed.

    The internet is a cultural black hole.

  2. Dr. Gawande should realize that he’s got a lot of troubles of his own, and stick to them.

    Nobody in ND wants Gawande’s or the New Yorker’s help. He should move on to places where he’ll be better appreciated.

  3. Phil, I don’t doubt that the New Yorker, like any news or magazine site, is incentivised to hook readers with dramatic headlines, and it is quite reasonable for you to call them out. Note, however, that the Statista page you link to, placing ND 10th in total deaths, is based on total integrated deaths since the start of the pandemic. As we all know, the rural/Western states were relatively free of Covid for many months from Feb/March 2020 and pretty much into August, while e.g. NJ, NY, MA, being International ports of entry, dense urban centers, etc, were the first to get wallopped, and have pretty much been suffering ever since, including an impressive second wave.

    The Dakotas and similar sparesly populated (mid-)Western states by inference would have to have impressive rates measured on shorter marginal time scales (days, week, month) to catch up as much as the have in the integrated rankings. ND is still top of this Statista page as of 11 Feb:

    https://www.statista.com/statistics/1109004/coronavirus-covid19-cases-rate-us-americans-by-state/

    despite that things have improved substantially very recently. I barely looked to find this statistic, and won’t dig into county or country in a similar way, other than to note e,g, Europe was hit heavily well before US. In summary, I don’t think your binary assessment of three falsehoods in the headline based on total integrated statistics is without counterpoint.

    It is objectively pretty amazing that such a rural sparsely populated state could find a way to infect so many so quickly.

    In my opinion, and FWIW, Atul Gawande is a pretty competent and productive guy. He has had, and continues to have, a positive impact on public health. He is well known, for example, for advocating for the use of check lists by surgical teams in the OR, inspired by their impact on safety in aviation, something which might resonate here . An obstruction to check list use, by the way, is that surgeons are a pretty egotistical bunch, something which Gawande would acknowledge, anyhow studies show 30% reduction in surgical mortality through their use. To me it’s extremely positive that he (and other competent professionals) have joined the Federal/administration Covid response team (handling vaccine strategy and distribution I think), despite the slightly disparaging whiff (as signalled for example by your dig at Jill Biden) of your reference.

    While it is true that the New Yorker and Atul don’t have their primary market in the flyover states, I found the article interesting and worth reading, again FWIW.

    • It isn’t surprising to me that COVID-19 took a while to get to remote corners of the U.S. The same thing happened with the 1918 flu, which arrived in Alaska villages a year after it was in big cities. See http://dhss.alaska.gov/dph/VitalStats/Documents/PDFs/AK_1918Flu_DataBrief_092018.pdf for example.

      If Dr. Gawande and the New Yorker editors’ technique of picking a time window and a statistic retrospectively is legitimate, why not wait for a bus crash in a small town and then use a one-week window around the crash to write about how that town has the most dangerous roads in the United States?

  4. Thanks for the quick response, Phil. I don’t have an issue with your first paragraph, and that’s interesting. I strongly don’t agree that the bus crash example is an apt comparison, and I am sure you realize that you are taking time window distortion to a hyperbolic extreme. Since North Dakota essentially didn’t even have Covid until July or August, it is not unreasonable to measure how hard hit they are now on some other choice time span. In other words you pointing to the total integrated death count is as much a distortion as any other choice.

    I found it relevant that ND tops the charts for current infection rates. Pointing to that is nothing like “waiting for a bus crash”. The New Yorker article is not a scientific paper, and, as I said in my first comment, I have no doubt their headline crafting is designed to snare clicks. It is justified that you pointed this out, but your arguments to my mind include similar biases, albeit of opposite sign, to those you object to.

    • Jonathan: Why is the total death rate the best measure of whether a place is “worst-hit”? In my experience, humans care more about whether they live or die than they do about what comes out of the toaster-oven-with-timer (PCR machine): “you’ve been infected with a disease that is so bad that you didn’t know you had it until we gave you this piece of paper.”

    • Paul: Those are some fun graphics about “cases”, but I don’t know how meaningful they are. A death is fairly straightforward to notice and it will always be tied to COVID-19 whenever possible due to the financial incentives in our system. A “case” depends a lot on whether testing is available, whether people are required to get tests, etc. That has varied quite a bit over time and by state.

    • Paul: Indeed, interesting, and thanks for some support.

      Phil: Of course dying from Covid is way more of a bummer than getting a positive PCR test while asymptomatic. As are any number of possible long term consequences including lung function or loss of taste or smell.

      But why is the point under debate here a moving target? We were talking about time scales of statistics, now there’s a flavor in your response of “Covid is not much worse than the flu” or similar. Death rates tend to track infection rates with about a two to three week time lag. I can’t seem to find a ranking of North Dakota deaths per 100k with other states month by month, however it seems likely that rate would place North Dakota considerably higher than 10th in the rankings over the months from Sep/Oct 2020 to Jan 2021. And I think that is very relevant to that state’s populace, and to even partly justifying the New Yorker headline, your bus crash example notwithstanding.

      Further tracking cases with higher time resolution is key to finding causal links, which is what people like Gawande do. I certainly don’t want to get into a debate about shut downs or mask effectiveness, about which I have nuanced opinions with a fair dose of skepticism–we’re not all brainwashed–however I do clearly see the data has to come from somewhere!

    • Jonathan: See https://c19graphs.net/. Clunky but you can see death rates / day by choosing “States” and “Deaths”, and using the date slider while hovering your mouse over the state of interest. Would be easy to get what you would like to do if one could download the data.

  6. The print edition of the Feb 15-22 New Yorker that reached London doesn’t feature the same headline and subhead that Phil links in the online edition. Instead “Don’t tell me what to do” and “As North Dakota’s hardest-hit county battled the pandemic, a mask mandate became another battleground”. In a speed-scan of the 11-page article, I didn’t see the other claims repeated, and the hardness of Ward County’s hit was more carefully described in the text.

    In this case, the NY’s online staff seem to have gone rogue.

    • Thanks, /df! You really do have to wonder what is special about North Dakota that the New Yorker and this policy-making doctor decided to feature the state. Statistically speaking, they’ve had a similar COVID-19 experience to many other places on Planet Earth. If they’re not being singled out for their heretical decision to continue their lives in 2020, rather than shutting down, why are they being singled out? https://www.statista.com/statistics/1109011/coronavirus-covid19-death-rates-us-by-state/ suggests that it would make just as much sense to look at a variety of other states, most of which had lockdowns and masks.

  7. I won’t weigh in on how to judge “worst”. I will however mention the title of the article as a good example of intent-to-scare journalism. The title “Inside the Worst-Hit County in the Worst-Hit State in the Worst-Hit Country” *sounds* really scary. What would be *actually* be more scary would be “Inside the Worst-Hit County in the World”. The multiple “worsts” in the original title intend to scare, but are actually qualifiers limiting the field of counties vying for to be the “worst”.

    I’m a NYT subscriber, but this is not a good look IMHO.

    • NYT and the NewYorker are different publications by the way. No paywall at the NewYorker, but journalism used to be much better eons ago under Tina Brown, however liberal she was, at the times when world “liberal” still carrier part of it original meaning. If I am replying to a ML bot post trained on this blog then it is quite a good bot.

Comments are closed.