An Update on the Human Experiment in Marion Prison
I’ve had a lot of good things to say about Ohio Governor Mike DeWine’s response to the COVID crisis. In days ahead, we’re likely to see more cynical decisions.
For example, OH has asked employers to inform the state when their employees don’t return to work out of fear of the virus. If people don’t have a good medical reason to stay home, they may lose their unemployment payments.
As Ohio reopens some businesses, employees who don’t return to work could lose unemployment benefits – even if they have a health condition that makes them more susceptible to the novel coronavirus or problems getting child care.
Ohio Department of Job and Family Services is asking employers to report workers who quit or refuse work – potentially cutting those former employees off from unemployment benefits. Those not eligible for unemployment in Ohio would also lose access to the additional $600 a week approved in the federal stimulus package.
Before someone loses benefits, both the employer and employee would have a chance to weigh in with the Ohio Department of Job and Family Services. Any decision can be appealed to the Unemployment Compensation Review Commission.
“It’s a case-by-case basis, but if you’re just saying ‘I’m afraid of the virus,’ that would not be sufficient,” Ohio Department of Job and Family Services Director Kimberly Hall told reporters Monday. “The analysis would need to be that your work environment, the conditions there, are such that you’re at risk from a health and safety standard and you have good cause in that regard.”
It was in anticipation of the return to work that I got so uncomfortable about what I called an uncontrolled human experiment in herd immunity at Marion prison. Marion was one of the first places where an entire population that could be tracked over time was tested; that process started on April 11, about 25 days ago.
By April 20, 76% of the population had tested positive. By the time all the tests came back, up to 86% of prison inmates had tested positive.
By the time you learn almost an entire prison is positive, you’re running a controlled experiment in what the real fatality rates of the virus are. According to yesterday’s data, just over two weeks later, those stats show:
- Of 175 prison staffers who “reported” positive, 1 has died, .5%.
- Of those 175, 116 are considered “recovered,” 66%.
- Around 86% of around 2,500 inmates tested positive.
- Of around 2,176 inmates who tested positive, 11 have died, .5%.
- Of those 2,176, 636 are considered “recovered,” 29%.
The fatality in Pickaway is, thus far, notably higher, with a 1.6% fatality rate among prisoners and 1% among prison staff.
The full life cycle of the virus is at least a month, so these numbers are not final, even ignoring the chronic effects seen in some patients.
What Ohio is doing (and other prison systems have replicated) is started to learn what the true fatality rate of this disease is. For better and worse, that data will explain what the risks are to exposing entire populations to the disease.
Once you have that data, it becomes possible to make cost-benefit decisions weighing the cost of human life against the lost production of a factory or a store, decisions that might be used to push towards herd immunity for everyone. And I have every expectation that such data will be used to make those cost-benefit calculations, weighing the cost of human life in terms of dollars.
I wonder what the real emergency will turn out to be when DeWine deems it is finally time to tap into that rainy day fund. Meteor strike?
Maybe the expected second wave?
It will be interesting, in this experiment, to see when herd immunity stops the spread of the virus. From what I understand, at 86%, they should already be there.
Applying those stats to overall population, we would see roughly 1.5 million deaths. Trump will claim anything short of that is a win. TPM linked to a video montage a day or two ago in which his predictions inexorably rose higher and higher.
I rapidly spreading epidemics, the level of infection overshoots the level of herd immunity because you still have many infecteds when you hit the proportion of infected that would lead to each infection producing fewer than 1 new infection. So lets say herd immunity occurs at 60% infected, but you still have some number, N (200), who are infected, the number of new infections they will produce is less than N, say 150, but you still produce new infections. This overshoot can be reduced by mitigation factors that reduce the increase in the first place.
Right, and in this case, they realized they had a cluster and tested everyone, and realized they had a huge number of asymptomatic people. That 2-week window before you’d be a statistic makes it easy in this case.
It’s too bad the data doesn’t include any mention of symptoms; it would be really nice to know how many of those early asymptomatics were really just pre-symptomatic. The proportion of true asymptomatics in a population would be a really nice number to know.
Agree,
I continue to wonder if our obsession with the CFR (case fatality rate) is an ongoing blind spot with regard to this pandemic. It’s such a tidy number, that obscures not only all the other deaths that were COVID-19 but not confirmed as such, and all the excess deaths from other causes due to the COVID-19 stress on the healthcare system, but also the far larger number of serious cases requiring hospitalization, and the so-called “mild” cases that, although they don’t require hospitalization, are massively debilitating and may carry long-term impairments even after “recovery”.
That said, I’m unconvinced that the population at Marion, although a large enough sample, is diverse enough to extrapolate meaningfully to the general population. No kids, massive vulnerabilities in certain areas, probably a much lower rate of obesity, all male. I’m not arguing that the hazard from the virus is lower for the rest of the country, just noting the limitations of this “experiment”.
I’d sure like to know the breakdown on severity, and how many of the “recovered” will require physical and/or cognitive therapy.
Good point — the Financial Times has been doing an unexpectedly good job of charting Excess Mortality which suggests CFR (apart from Marion) is inaccurate.
I would read EW’s forecast as a promise. There is ample precedent for the use of prisons and prisoners as medical guineau pigs. At times, medical ethics has been an oxymoron. Business executives have often used ruthless forms of cost-benefit analysis, excluding many hidden costs and focusing on management benefits. Enron’s and Amazon’s entire management team and Ford Motor’s Pinto gas tank team come to mind, as well as GM’s planned bankruptcy team.
Many of those executives are still around and would be happy to volunteer their time. Any number of private equity executives would be to, in exchange for free access to the resulting data sets. Public-private partnership proposals will temptingly arrive in stacks as long as Trump’s tie.
This data is fascinating, but heavy caution should be used when extrapolating from the prison population to the general population.
First, to point out the obvious: this is an adult-only population, not an all-population measure. And the age distribution among prisoners is skewed somewhat younger than among the general adult population.
Second, prisoners have, on average, much higher prevalence of relevant medical comorbidities that are likely to increase the severity of a COVID-19 infection. These include higher rates of chronic alcohol and substance use, chronic liver or kidney disease, Hepatitis B and C, plus all the “usual suspect” conditions associated with being disproportionately black/brown and poor: heart disease, diabetes, COPD, etc. Being imprisoned also means you have less access to high-quality medical care, exacerbating any conditions you have.
Even the guards are not representative of working-age population, since prison guards tend to be from lower education and SES backgrounds, and regularly interact with prison populations, with all the consequent associations ramifications for their health that that entails.
So, very interesting, but we’d need a lot more data from this particular prison to be able to adequately adjust for these factors and make an apples-to-apples comparison. We likely won’t get it.
All very good points, thanks for making them.
And while I think that’s true from a scientific standpoint, I doubt the managers at Tysons Food care about the differences (and their work populations are probably pretty similar). That is, now people can point to these numbers and says, “Just 1.5% will die, I can open my factory, burn through cases with just a few deaths, then get back to business.”
That’s my fear, anyway.
And even that presupposes that hospitals never get overwhelmed from exponentially increasing case loads.
Right, which not only cannot be presupposed, but IMO is the biggest missing piece here. I didn’t see any data presented for Marion-related hospital utilization (at link). Given general history — plus the fact that prisons have infirmaries — prisoners would be under-hospitalized (less likely brought to ER in first place, even leaving inpatient aside ) vs others.
‘Letting the experiment run’ in the general population would certainly cause this overwhelm and chaos far beyond places like Detroit, NYC, and many other less -well- known locales which were* sentinels for steep-curve effects.
That would affect the mortality rate, but also the pandemonic, cruel suffering rate.
—
*Past tense only applies to the shape of the curve: in hard-hit areas it is still corona time. With MSM horserace type coverage of the worst/highest/biggest, it’s easy to forget plateaus and all that.
The high incidence of depression, anxiety and other mental health issues, count too. Also a substandard diet, low in vitamins, proteins and fiber, lack of sun/fresh air, and depending on the prison, limited exercise.
Medical care in prisons is frequently limited and substandard. As with food – and regardless of rules – withholding it or better versions of it is used to motivate prisoners to do good things. Like not dissing guards or being obstreperous, and getting prisoners to work for a pittance at the prison call center or other business.
Already bad food and health care can be made worse for the non-cooperative, a trait often found in abundance among prison populations. As you mention, prison drug abuse is common, as is a certain prevalence of sexual predation, with resulting health consequences.
The many businesses which have adopted PE’s resource extraction at any cost model will use whatever data sets allow them to reopen without incurring legal liability or the added cost of employee health and safety precautions. Examples of the latter include slower line speeds; greater distancing; extra clothing, masks, and laundering; taking temperatures and other health checks, recording and reporting the foregoing, and so on. (Once you keep records, you document your failings and force yourself to do something now, or pay big later.) Eugene Scalia – the darling of anti-labor lawyers – is running the Labor Dept, so no worries there.
As the Inquirer story highlights, prison rhythms have changed considerably, not to prisoners’ benefit. In Pennsylvania, prisoners are in their cells and on lock down 23 hours a day.
https://www.inquirer.com/health/coronavirus/a/prison-pennsylvania-coronavirus-lockdown-doc-sci-phoenix-20200506.html
That was a poignant read; still, folks find humor:
All the data we see from various sources converge to the same numbers. Covid-19’s IFR is probably somewhere in the range of 0.5%-1%. It’s true that the prisoners population is not necessarily a representative sample, and it is also true that mortality rate can change between different populations (different states, different countries etc.) due to different demographics and comorbidities. But this seem to be the typical range in most western countries.
This number is high enough to take Covid seriously and justify the social distancing measure taken, but low enough to be cautiously optimistic about the future. It is 5x-10x more deadly than seasonal flu, with larger susceptible population (no prior immunity), but probably somewhat less deadly than the Spanish flu. Without any social distancing measures, we would end up with 70-80% of the population infected, and somewhere between 1.2 million to 2.6 million dead.
South Carolina prisons are advertising (on tv) in Georgia for new health workers, not sure what is happening up there.
Experiments, even uncontrolled ones, need names, to distinguish them from one another. Might I suggest that this be named “Tuskegee II”?
If the protocol fits . . .
Ooooh, exactly! I just said that before seeing this comment as I worked my way down.
at some point one could expect some class action law suits. people were sentenced to jail for specific terms. what they got was a death sentence.
I’m surprised the guards even bothered to show up for work.
So, this is basically the forced Tuskegee syphilis experiment on black men redux? Swell.
[Edit as a PS: The inestimable Peterr had already mentioned Tuskegee.]
EW—thank you for reporting on the Marion and Pickaway prison story and thank you for linking earlier to Professor Dolovich’s “COVID-19 Behind Bars Data Project.” One quick observation: Marion and Pickaway had about the same number of staff employed before the outbreak, 446 at Marion and 475 at Pickaway. The positive case rate among the staff is high 175/446 (40%) and 101/475 (21%) at Pickaway, but nowhere near as high as that of the men imprisoned. Working as a staff member at Marion or Pickaway is comparable to working at Smithfield or Tyson, but serving a sentence at Marion or Pickaway with little possibility for social distancing is like serving on the USS Roosevelt, or worse, serving on a nuclear sub patrolling under the polar ice cap.
I have family working as physicians in Marian County Ohio. They report that facilities have been heavily burdened by the infections at the MCI. When inmates need medical attention they requires two persons to transport them to different facilities. Those folks often lack training and equipment to protect themselves. Whole families of MCI workers are getting infected and not proscribed from interacting with the larger community. The general belief is that the asymptomatic are the spreaders of the virus. This only reinforces the need for testing of the larger population. If good news can come from such a thing, then event is creating an empirical/actionable finding for an approach to better deal with the problem.