from Hacker News

Ohio Prison Death Updates [pdf]

by burgreblast on 5/13/20, 1:55 PM with 56 comments

• by burgreblast on 5/13/20, 1:55 PM

  17 days ago we heard that 3,330 inmates tested positive, 96% without symptoms.

  https://news.ycombinator.com/item?id=22980932

  There was much speculation, but many people agreed that in 2 weeks we would have super interesting data.

  It's been 17 days. We have an update from ohio.gov that tested individuals climbed to 7536, 4439 are positive (59%), total 49 deaths (.01)

  Not an epidemiologist. Does this data fit the Diamond Princess model? Or more broadly, which model fits this data best?

  Is there other data to show how many became symptomatic? How do we interpret this update, more than 2 weeks after initial reports?

• by sdenton4 on 5/13/20, 10:04 PM

  Just as an extra wrench in the works, an unknown number of prisoners have been released or transferred, probably after being deemed higher risk: https://www.google.com/amp/s/thehill.com/regulation/court-ba...

  So death rates may still be reduced by interventions. Take care, arm chair statisticians.

• by romaaeterna on 5/13/20, 9:40 PM

  30 of the deaths are at PCI, which is basically a nursing home for older prisoners.

  EDIT: “Marion houses a high number of older individuals, many who have pre-existing health conditions. Pickaway houses our long-term-care center similar to a nursing home, and Franklin is our state prison medical center.”

  Marion is 25% of the deaths. Pickaway 59%. And Franklin 10%.

• by dr_faustus on 5/13/20, 9:32 PM

  Considering they probably tested everyone so there are no unrecorded cases, a 1% death rate is pretty much in line with studies finding the CFR to be in the 0.5-1.5% neighborhood [1,2]. I would guess they are younger on average, but the prison population probably has some risk factors in terms of nutrition, insufficient Vitamin D, lack of exercise, etc.

  [1] https://www.bmj.com/content/369/bmj.m1327 [2] https://www.medrxiv.org/content/10.1101/2020.04.14.20062463v...

• by whycombagator on 5/13/20, 9:53 PM

  I'd also be interested in how many have detrimental side effects from the virus, after recovery. Mostly all I see from covid numbers are black & white: died/recovered.

  There was an article recently on HN that talked about the possible lasting effects of the virus[0]. I am interested to know how frequently lasting effects occur

  [0] https://news.ycombinator.com/item?id=23127167

• by subsubzero on 5/14/20, 12:31 AM

  I think the better report to look into will be Lompac(southern california)[1], they have the higher % of infected prisoners, I think its been roughly 70% of the prison population has tested positive for covid. Analyzing these closed systems for a true death rate seems to be the only way we will get decent data regarding mortality. I don't see govt. officials going door to door in cities asking for people to prick their finger to analyze for covid antibodies. A few things to watch out for is prisoner transfers due to extreme sickness, also getting a age distribution for the prison like others have said here, another thing to note is prison food and env. is bad on health.

  [1] - https://beta.trimread.com/articles/14963

• by rayiner on 5/13/20, 11:01 PM

  It’s an interesting question: how do you figure out whether these numbers are good or bad? Ohio has a prison population of about 44,000. So these numbers (actual plus probable) represent a death rate of 120 per 100,000. That’s ten times the death rate in Ohio as a whole. But it’s about half the death rate of NYC (actual plus probable). What’s the correct reference point? NYC seems like a reasonable reference, given that prison is an inherently high density living situation.

• by kregasaurusrex on 5/13/20, 11:59 PM

  While this is a small dataset of tested individuals which may be proportionally representative to larger groups, it comes at odds with releasing PII with respect to individual privacy to see potential comorbities (old age, obesity, diabetes, lack of an auto-immune response, etc). We know that people with exposure to recycled airspace and live in close proximity are more likely to transmit covid to one another. Mask wearing and social distancing are actions that will have to be obeyed by everyone as a collective in order to work-- one may never know if they're an asymptomatic super-spreader without widespread testing. Cloth masks aren't for protecting yourself as much as they are everyone around you.

  Within the data itself, it shows that quarantine and isolation are effective practices against spreading covid in a hotspot. This may be a good stop-gap measure while researchers are able to study it more, but government's responsibility at all levels of keeping people safe in returning to work has has greatly fallen short of expectations.

• by malkia on 5/14/20, 1:10 AM

  Stab in the dark - AFAIK, Testosterone level in males in Prisons is much higher than usual (I can't back up that claim right now, but have been reading posts related to it).

  Then there is this - https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7185012/

  just the two clicked together somehow, but might not be related after all..

• by lvs on 5/13/20, 11:17 PM

  This is likely just about population age distribution in prisons vs the general population.

  16% of the US population is 65+, but only 2.7% in (federal) prisons. So that must account for a decent chunk of it.

  https://www.bop.gov/about/statistics/statistics_inmate_age.j...

• by m0zg on 5/14/20, 12:14 AM

  I've read somewhere (but don't know if it's true) that prison meals are enriched with vitamins, and in particular with vitamin D. Could someone confirm?

  Also, there are fewer old prisoners than old people in the general population.

• by mellosouls on 5/13/20, 10:34 PM

  I can't see the document, but when a similar story recently popped up on here the average age of prisoners was 38.

  fwiw.

  edit: the previous story https://news.ycombinator.com/item?id=22941493

  age stats: https://www.cleveland.com/news/erry-2018/08/84f4aab48f389/oh...

  (the underlying source is not accessible to me either at the moment)