After pointing out in her Washington Post op-ed that media outlets have incentives to exaggerate dangers and underreport successes, economist Emily Oster provides what may be the best advice I’ve seen on making strategic decisions about SARS-CoV-2:
In the absence of complete information on risks, our overreactions can have serious consequences. One example is the Three Mile Island nuclear event, which has not been conclusively linked to any long-term negative health outcomes but did terrify Americans about nuclear power. People simply didn’t have enough baseline information about the number of nuclear plants operating safely on a given day to realize that the probability of a nuclear disaster was vanishingly small. The result was that nuclear power — a plentiful, carbon-free energy source — never reached its potential in the United States, leading to needless overreliance on dangerous fossil fuels.
We risk making similar mistakes with the coronavirus. Keeping children out of school harms their development. Shuttering businesses destroys livelihoods. These downsides may be offset by the benefits of limiting covid-19. But we cannot rationally assess the trade-offs when we have only partial information.
What we really need to know is not the anecdotes that news reports provide, but the full picture. What share of schools have cases? Moreover, what differentiates places with cases from those without? Is it differences in prevention measures? Demographic and economic characteristics? The prevalence of community-spread events?
To answer these questions, we need systematic data collection and reporting — the sort that lets us evaluate risks in all kinds of situations, from driving cars to flying on planes to, yes, ocean swimming. It should be possible to do this. As schools open, districts will have counts of at least detected covid-19 cases, as well as information on the overall enrolled population. This data could be combined in public databases with user-friendly dashboards and maps. Since this type of data collection has not been spearheaded by central authorities, I’ve partnered with a set of national educational organizations and a data team to try to put it together.
I would fault President Trump for not providing that kind of leadership but not only President Trump. Few governors have emphasized “systematic data collection and reporting” sufficiently—Indiana appears to be an exception. Our own governor seems to be collecting the data and ignoring it quite assiduously.
Free enterprise to the rescue.
Offering “data analytics” and “data story telling” helping us humans understand the numbers and avoid the pitfalls of perspective.
https://cambersystems.com/
Looking at Chicago, we see a more than 5x increase in people who visited Lake Calumet and Kilbourn Park, one of the highest increases in the nation of visits to a tract.
https://cambersystems.com/spikes-in-movement-on-memorial-day/
And they have a blog.
It should be possible, but someone has to pay for it. Do we have enough of the right kinds of people to do it? We have needed surveillance testing all along, but there has been lots of resistance against testing in general. All of this is stuff that the CDC is set up to handle, but they have performed poorly during this outbreak. Much worse than they have behaved in the past.
Steve
Opinions differ as to why that might be. Maybe it’s Trump. What if it’s generational change?
‘Opinions differ as to why that might be. Maybe it’s Trump. What if it’s generational change?’
If it’s a generational change, it happened a couple of generations ago. Most of the leadership way predates OMB’s time in office.
Creating incentives seems to nudge acts and actions into a given direction. For instance, because Medicare payments, for hospitalization and the use of ventilators, were higher with a diagnosis of COVID there have been a myriad of stories where said Dx was based on symptoms rather than a positive test for the virus. Worst yet, people dying from stroke, heart attack, diabetic coma, accidents, but also testing positive for the virus, often resulted in the virus slapped on death certificates as the official cause of death.
Consequently, putting together an honest, reliable systemic data collection and reporting heavily relies on there being no bias or hidden rewards lurking. This hopefully would discourage those in charge of subtly manipulating raw numbers, even holding data back to appear in clumps exhibiting sudden “spikes,†that are then open and ripe for misinterpretation.
The CDC bureaucracy has long been one operating under red tape constraints, creating approval delays generating possible in-house influences for policy/treatment recommendations. Furthermore, their COVID fatality numbers have been questioned by “experts†such as Dr. Birx, and their initial sloppy testing kits slowed down a more immediate response to this virus, as did their conflicted, alternating guidance on masking, virus contagion etc. Anti viral assessments by the CDC and NIH officials have also been suspect, where remdesivir was pushed (with members of the NIH being personally invested in this drug), while HCQ has continuously been discredited, despite a 60+ year track record of usage and a multitude of studies (not “gold standard†randomized ones) of impressive successes.
Basically, unbiased oversight is needed to create a more accurate data base, along with screening for any unjustified politicization of the methods employed in renderIng the final numbers.
We can analyze our fannies off, and perhaps we should and will over time, but the overall thrust is already known. Its not a mystery. Once again we fall into the trap of false precision. Is the risk to delayed physical medical exams 1.2345 x the risk of corona, or 1.2478? Is the risk in NYC 1.117 x the risk in Peoria, or 1.256?
I read today that some genius at McKinsey has determined that a year lost in school results in a loss of 4% of lifetime income. AYFKM? That’s just a guy or gal with a computer, a vivid imagination, and too much time on their hands. All you really know is that its not good. Statistical precision will not be found. Just take a look at Sweden and make a decision. Its like Monty Python’s Ministry of Silly Walks.
” for hospitalization and the use of ventilators, were higher with a diagnosis of COVID there have been a myriad of stories where said Dx was based on symptoms rather than a positive test for the virus.”
Unsubstantiated stories. Again, where I work and at all of the ICUs where I have contact this did not happen. Hospitals were losing money because of the extra costs of caring for Covid so they increased the pay. That turns into a conspiracy. However, we really do have excess deaths so this was not just taking pts who had normal heart attacks and calling them Covid deaths.
“a multitude of studies (not “gold standard†randomized ones) of impressive successes.”
I have not seen even one that showed impressive success. Could you please link to some of those. I was keeping count of studies read for HCQ but stopped at about 60 so I am sure I have missed some.
“The CDC bureaucracy has long been one operating under red tape constraints, creating approval delays generating possible in-house influences for policy/treatment recommendations.”
If you dont know what you are talking about then dont. The CDC actually has a pretty good record of responding well to infectious disease issues in the past. For those of us in the tree it is just shocking about how poorly the CDC has functioned compared with outbreaks in the past. There is an issue of scale here so I can grant them that, but this is still aberrant.
Steve
Have to note the silence here. Every time a conservative claims that a re a lot of good studies with impressive results about HCQ I ask for a citation or a link to said studies. Never happens. Never. (And I will read them. Part of being a professional geek.)
Steve
I’m going to be a contrarian and say data isn’t the problem (and hasn’t been a problem for months).
Lets take another field, stock market investing. Its been subject to decades of data analysis, with lots of neatly organized data on every stock, sector, and mutual fund. Yet it never fails that a significant portion of investors end up buying high, selling low, and losing money despite the stock market’s upward bias in the past century.
The reasons are well known, (a) disagreement on the interpretation of the data; (b) different risk tolerances — greedy when prices are high, fearful when prices are cliff diving (c) the difficulty in making decisions solely on the data; and ignoring one’s personal cognitive biases.
All three of those are at play in responding to COVID-19. Lack of neatly organized data isn’t the problem — its the grey matter between our ears (or more accurately, in decision-maker ears).
As an example — we are 6 months into the epidemic — and beaches are still being closed despite the lack of documented outbreaks from it.
People wearing facemasks on bike trails in the woods are a common sight; so is no facemasks in crowded indoor environments. You can put “One Way” or “Maintain Social Distancing” signs up but that doesn’t mean people will pay any attention to them.