[BostonCard]

This is a story to keep track of.

https://www.npr.org/sections/coronavirus...oronavirus

73% of inmates (1828 in total) in an Ohio prison tested positive for coronavirus, after Ohio decided to test every inmate.

"Because we are testing everyone — including those who are not showing symptoms — we are getting positive test results on individuals who otherwise would have never been tested because they were asymptomatic,"

So far, no deaths.  The breakdown of who is symptomatic and who isn't is provided.  If things stay this way, the 95% confidence interval around the IFR is 0 to 0.2%, and would be consistent with the Stanford and Los Angeles county studies (not adjusting for age and comorbidities in the prison population).  Obviously this would be updated if you started seeing some deaths.  But it will, unfortunately, be an interesting cohort of people to follow, since the testing was done early and it is a different population than the cruise ships.

BC

[BostonCard]

This is a story to keep track of.

https://www.npr.org/sections/coronavirus...oronavirus

73% of inmates (1828 in total) in an Ohio prison tested positive for coronavirus, after Ohio decided to test every inmate.

"Because we are testing everyone — including those who are not showing symptoms — we are getting positive test results on individuals who otherwise would have never been tested because they were asymptomatic,"

So far, no deaths.  The breakdown of who is symptomatic and who isn't is provided.  If things stay this way, the 95% confidence interval around the IFR is 0 to 0.2%, and would be consistent with the Stanford and Los Angeles county studies (not adjusting for age and comorbidities in the prison population).  Obviously this would be updated if you started seeing some deaths.  But it will, unfortunately, be an interesting cohort of people to follow, since the testing was done early and it is a different population than the cruise ships.

BC

[Snorlax94]

Will this mean we can keep some of the violent criminals in prison?

[Snorlax94]

Will this mean we can keep some of the violent criminals in prison?

[M T]

This is a story to keep track of.
https://www.npr.org/sections/coronavirus...oronavirus

73% of inmates (1828 in total) in an Ohio prison tested positive for coronavirus, after Ohio decided to test every inmate.

So far, no deaths.  The breakdown of who is symptomatic and who isn't is provided.  If things stay this way, the 95% confidence interval around the IFR is 0 to 0.2%, and would be consistent with the Stanford and Los Angeles county studies (not adjusting for age and comorbidities in the prison population).  Obviously this would be updated if you started seeing some deaths.  But it will, unfortunately, be an interesting cohort of people to follow, since the testing was done early and it is a different population than the cruise ships.

If the population has that percentage of positives, it is probably too early in that facility's epidemic to have many deaths (so one can NOT begin to calculate an IFR at all, in the same unwarranted fashion that was being used by everyone in January & February while active cases hadn't progressed to deaths).  I hope they closely monitor them for symptoms, and give them the appropriate medical treatment.  This is a great opportunity to see if any treatments that may prevent worsening of the problems that happen in the untested & unmonitored general public.  Whether it is Remdesivir, or monitoring of oxygen saturation levels, or whatever...

It is unfortunate that this is happening at multiple facilities (prisons & long term care facilities), but it does provide a testing ground for the ideas of what wide-spread testing can do.  Here, one probably won't be doing test & trace, but should be able to do test & treat.

I hope some forward-looking medical research organizations are readying studies to try out at future outbreaks at prisons/LTCF.
Another one (not in Ohio).

By the way, if you follow the link in the newspaper article,  you will find that prisons with cells have few active cases.  The really bad outbreaks are in "combo" facilities.

---

Back of the envelope calculations for illustrative purposes only:
(I'm treating the inmates as an isolated population, but the guards may have infected some.)

R0 here is going to be very high in that one prison, with 1828 positives & 667 negatives among the inmates.(2495 total)
I do believe that a high R0 means that there has been little time for symptoms to develop or for anyone to die.

I tried to match the numbers using a SEIR calculator, initially set with some of the COVID-19 parameters.

As I've adjusted the numbers, exposed & infectious mean what you expect.  "Incubation period" is not the time until symptoms appear, but is the time until infectious (2.5 days before symptoms).  I'm treating infectious as equivalent to "testing positive".

I kept bumping up R0, but I reached a limit of 100 in the program (GUI limit is 10, but as a parameter, it can go to 100).  So, then I adjusted the duration of infectiousness.  Infectious topped out (at 1857) if I pushed infectious duration to 24 days (GUI limit).  I believe that duration should be lower (with R0 higher), so I've set it lower at 18 days (which I still think is too high).

It's hard for me to see how R0 could be 100 or higher, unless an early case was serving food to everyone else.  I don't think adding in the guards could somehow help.

Settings: R0 = 100
Population: 2490 (closest I could get)
Initial infections: 1
Incubation period: 2.72 days  (Adjusted: I don't care about symptoms showing up. I care about infectiousness.  I adjusted this to 5.2 days real incubation - 2.5 days of infectiousness before symptoms)
Duration patient is infectious  18 days
CFR: 0.0%   (Don't care  I'm not looking at deaths.)
Time from end of incubation to death: 32 days
Recovery time for severe cases: 28.6 days
Recovery time for mild cases: 11.1 days
Since I'm assuming nobody was symptomatic:
  Hospitalization rate: 0
  Time to hospitalization: 100 days

The total number of infectious cases peaks on day 12:
    0  Susceptible (not yet infected)
 228 Exposed (infected but not yet infectious)
1756 Infectious
 506 Removed (previously, but no longer, infectious.  Includes recovered)
       109 Recovered

1756 Infectious  (Actual: 1828 positives)
  734 Susceptible + Exposed + Removed (Actual: 667 negatives)

Adjusting infectious duration didn't change the peak total of Exposed + Infectious much, but it did change the date and the ratio at the peak.

If I assume the test can pick up the disease from the moment of exposure (ie, positive = exposed+infectious), then I can get the right numbers at a more reasonable R0 = 14.   SEIR calculator,

[M T]

This is a story to keep track of.
https://www.npr.org/sections/coronavirus...oronavirus

73% of inmates (1828 in total) in an Ohio prison tested positive for coronavirus, after Ohio decided to test every inmate.

So far, no deaths.  The breakdown of who is symptomatic and who isn't is provided.  If things stay this way, the 95% confidence interval around the IFR is 0 to 0.2%, and would be consistent with the Stanford and Los Angeles county studies (not adjusting for age and comorbidities in the prison population).  Obviously this would be updated if you started seeing some deaths.  But it will, unfortunately, be an interesting cohort of people to follow, since the testing was done early and it is a different population than the cruise ships.

If the population has that percentage of positives, it is probably too early in that facility's epidemic to have many deaths (so one can NOT begin to calculate an IFR at all, in the same unwarranted fashion that was being used by everyone in January & February while active cases hadn't progressed to deaths).  I hope they closely monitor them for symptoms, and give them the appropriate medical treatment.  This is a great opportunity to see if any treatments that may prevent worsening of the problems that happen in the untested & unmonitored general public.  Whether it is Remdesivir, or monitoring of oxygen saturation levels, or whatever...

It is unfortunate that this is happening at multiple facilities (prisons & long term care facilities), but it does provide a testing ground for the ideas of what wide-spread testing can do.  Here, one probably won't be doing test & trace, but should be able to do test & treat.

I hope some forward-looking medical research organizations are readying studies to try out at future outbreaks at prisons/LTCF.
Another one (not in Ohio).

By the way, if you follow the link in the newspaper article,  you will find that prisons with cells have few active cases.  The really bad outbreaks are in "combo" facilities.

---

Back of the envelope calculations for illustrative purposes only:
(I'm treating the inmates as an isolated population, but the guards may have infected some.)

R0 here is going to be very high in that one prison, with 1828 positives & 667 negatives among the inmates.(2495 total)
I do believe that a high R0 means that there has been little time for symptoms to develop or for anyone to die.

I tried to match the numbers using a SEIR calculator, initially set with some of the COVID-19 parameters.

As I've adjusted the numbers, exposed & infectious mean what you expect.  "Incubation period" is not the time until symptoms appear, but is the time until infectious (2.5 days before symptoms).  I'm treating infectious as equivalent to "testing positive".

I kept bumping up R0, but I reached a limit of 100 in the program (GUI limit is 10, but as a parameter, it can go to 100).  So, then I adjusted the duration of infectiousness.  Infectious topped out (at 1857) if I pushed infectious duration to 24 days (GUI limit).  I believe that duration should be lower (with R0 higher), so I've set it lower at 18 days (which I still think is too high).

It's hard for me to see how R0 could be 100 or higher, unless an early case was serving food to everyone else.  I don't think adding in the guards could somehow help.

Settings: R0 = 100
Population: 2490 (closest I could get)
Initial infections: 1
Incubation period: 2.72 days  (Adjusted: I don't care about symptoms showing up. I care about infectiousness.  I adjusted this to 5.2 days real incubation - 2.5 days of infectiousness before symptoms)
Duration patient is infectious  18 days
CFR: 0.0%   (Don't care  I'm not looking at deaths.)
Time from end of incubation to death: 32 days
Recovery time for severe cases: 28.6 days
Recovery time for mild cases: 11.1 days
Since I'm assuming nobody was symptomatic:
  Hospitalization rate: 0
  Time to hospitalization: 100 days

The total number of infectious cases peaks on day 12:
    0  Susceptible (not yet infected)
 228 Exposed (infected but not yet infectious)
1756 Infectious
 506 Removed (previously, but no longer, infectious.  Includes recovered)
       109 Recovered

1756 Infectious  (Actual: 1828 positives)
  734 Susceptible + Exposed + Removed (Actual: 667 negatives)

Adjusting infectious duration didn't change the peak total of Exposed + Infectious much, but it did change the date and the ratio at the peak.

If I assume the test can pick up the disease from the moment of exposure (ie, positive = exposed+infectious), then I can get the right numbers at a more reasonable R0 = 14.   SEIR calculator,

[burger]

This is a story to keep track of.

https://www.npr.org/sections/coronavirus...oronavirus

73% of inmates (1828 in total) in an Ohio prison tested positive for coronavirus, after Ohio decided to test every inmate.

"Because we are testing everyone — including those who are not showing symptoms — we are getting positive test results on individuals who otherwise would have never been tested because they were asymptomatic,"

So far, no deaths.  The breakdown of who is symptomatic and who isn't is provided.  If things stay this way, the 95% confidence interval around the IFR is 0 to 0.2%, and would be consistent with the Stanford and Los Angeles county studies (not adjusting for age and comorbidities in the prison population).  Obviously this would be updated if you started seeing some deaths.  But it will, unfortunately, be an interesting cohort of people to follow, since the testing was done early and it is a different population than the cruise ships.

BC

I would be extremely careful about any extrapolations from this group.  Prison populations must be younger than average for the US, right?*  At the very least, older age groups are probably underrepresented.  So any CFR or IFR is going to be low compared to what you'd get in the US.

*I don't have Ohio data, but for US federal prisons, older age groups are definitely underrepresented.

Edit: found some data. 31% of all Ohio residents are 55+. As of 2010, 14.5% of Ohio prisoners were over 50. And I'll bet that the oldest age groups, that have the highest covid mortality, are the most underrepresented in prisons.

[burger]

This is a story to keep track of.

https://www.npr.org/sections/coronavirus...oronavirus

73% of inmates (1828 in total) in an Ohio prison tested positive for coronavirus, after Ohio decided to test every inmate.

"Because we are testing everyone — including those who are not showing symptoms — we are getting positive test results on individuals who otherwise would have never been tested because they were asymptomatic,"

So far, no deaths.  The breakdown of who is symptomatic and who isn't is provided.  If things stay this way, the 95% confidence interval around the IFR is 0 to 0.2%, and would be consistent with the Stanford and Los Angeles county studies (not adjusting for age and comorbidities in the prison population).  Obviously this would be updated if you started seeing some deaths.  But it will, unfortunately, be an interesting cohort of people to follow, since the testing was done early and it is a different population than the cruise ships.

BC

I would be extremely careful about any extrapolations from this group.  Prison populations must be younger than average for the US, right?*  At the very least, older age groups are probably underrepresented.  So any CFR or IFR is going to be low compared to what you'd get in the US.

*I don't have Ohio data, but for US federal prisons, older age groups are definitely underrepresented.

Edit: found some data. 31% of all Ohio residents are 55+. As of 2010, 14.5% of Ohio prisoners were over 50. And I'll bet that the oldest age groups, that have the highest covid mortality, are the most underrepresented in prisons.

[BostonCard]

Keep in mind that a prison setting is much different than a community setting, and prisons present many challenges for infection control.



Thus, you would expect the basic reproductive number for any respiratory virus to be much higher in a prison than in the community.  I tried to find literature calculating the basic reproduction number for influenza in prison but could not.  Nonetheless, this article (from which the table above is pulled) shows the challenges.

The points about the patient population being different (younger) and that the IFR cannot be calculated until the outcomes are known are both well taken and were made in the OP.  It is still a very interesting cohort to monitor.

BC

[BostonCard]

Keep in mind that a prison setting is much different than a community setting, and prisons present many challenges for infection control.



Thus, you would expect the basic reproductive number for any respiratory virus to be much higher in a prison than in the community.  I tried to find literature calculating the basic reproduction number for influenza in prison but could not.  Nonetheless, this article (from which the table above is pulled) shows the challenges.

The points about the patient population being different (younger) and that the IFR cannot be calculated until the outcomes are known are both well taken and were made in the OP.  It is still a very interesting cohort to monitor.

BC