key: cord-0819987-5lauop7l authors: Lofgren, Eric; Lum, Kristian; Horowitz, Aaron; Madubuowu, Brooke; Fefferman, Nina title: The Epidemiological Implications of Incarceration Dynamics in Jails for Community, Corrections Officer, and Incarcerated Population Risks from COVID-19 date: 2020-04-14 journal: nan DOI: 10.1101/2020.04.08.20058842 sha: 28f2f3169e77c805aedcf96c7bc5b4e69ee6d353 doc_id: 819987 cord_uid: 5lauop7l COVID-19 challenges the daily function of nearly every institution of society. It is the duty of any society to be responsive to such challenges by relying on the best tools and logic available to analyze the costs and benefits of any mitigative action. We here provide a mathematical model to explore the epidemiological consequences of allowing standard intake and unaltered within-jail operational dynamics to be maintained during the ongoing COVID-19 pandemic, and contrast this with proposed interventions to reduce the burden of negative health outcomes. In this way, we provide estimates of the infection risks, and likely loss of life, that arise from current incarceration practices. We provide estimates for in-custody deaths and show how the within-jail dynamics lead to spill-over risks, not only affecting the incarcerated people, but increasing the exposure, infection, and death rates for both corrections officers with whom they interact within the jail system, and the broader community beyond the justice system. We show that, given a typical jail-community dynamic, operating in a business as usual way will result in significant and rapid loss of life. Large scale reductions in arrest and speeding of releases are likely to save the lives of incarcerated people, staff and the community at large. Introduction 1 people to practice CDC recommendations such as social distancing [12] , the 23 incarcerated population has a higher expected rate of existing health conditions than 24 the community from which they come [13] [14] [15] , jails are dependent completely on a 25 workforce that moves in and out of the jail and the community including vendors, 26 lawyers, corrections officers, medical staff, etc., and there is strong evidence that 27 incarceration itself has profound adverse effects on the health of incarcerated 28 people [16] [17] [18] . These descriptors make jails highly likely not only to place detained 29 people at increased risk of infection and resulting severe outcomes, but also to function 30 as a driver for increased infectivity, adversely impacting attempts to contain and 31 mitigate disease spread in the broader communities in which jails are located. To study 32 the dynamics of this system and provide quantitative metrics for risk to incarcerated 33 populations and the populations with which incarcerated people necessarily interact, we 34 construct and tailor a epidemiological model of COVID-19 transmission, and then use 35 that model to consider how some possible reforms to the system (i.e. reduction in arrest 36 intake, increased rates of returning incarcerated people to their homes, and 37 improvement of conditions within the jails) will alter these baseline risks. 38 Model/Methods 39 Transmission Model 40 We begin by tailoring a standard SEIR model to the specific dynamics of COVID-19. 41 We first split our total population into four categories of risk: Children under 18 42 (denoted with the subscript K), Low-risk adults (denoted with the subscript L), 43 High-risk adults (denoted with the subscript H), and Elderly adults (denoted with the 44 subscript E). We also designate a separate population category for jail staff, O (note: consider death from any non-COVID-19 cause; this is done to highlight the 58 COVID-19-specific dynamics. Additionally, as a simplifying assumption due to their low 59 rates of both infections and complications, we do not model hospitalizations or deaths 60 in children. Similarly, once hospitalized, patients are assumed not to spread COVID-19 61 further, as additionally modeling the impact of healthcare-associated COVID-19 cases is 62 well beyond the scope of this model. Lastly, we split our population into segments 63 depending on the subsection of the community or jail system in which they are 64 currently functioning: the community at large, C, the processing system for the jail, P , 65 the court system T , and the jail system, J. A schematic for this model can be seen in 1, and the differential equations 67 comprising the model are in SI Appendix 1. The model was implemented in R 3.6.3 68 using the deSolve package, with the visualization of results primarily using ggplot2. 69 Statistical analysis of one parameter (see below) was done using the flexsurv package. 70 As this study used only publicly available data and does not involve human subjects, 71 IRB approval was not required. i.e. we do not model people moving from jail to prison. According to [19] , the yearly 98 number of admissions to prison is about 600,000 while the yearly number of admissions 99 to jail is around 10.6 million. So, assuming that all prison admissions first had one jail 100 admissions, around 95% of all jail admissions do not go on to prison; they are released 101 back into the community as in our model. Thus, we expect that the omission of prison 102 from our model does not substantially impact the overall findings. An online database of public employees salaries in Allegheny County shows a 104 population of 384 people whose job title is corrections officer, whose job location is the 105 jail, and who are listed as active. Although this is certainly an underestimate of the 106 total number of the jail's staff, which includes other types of employees, we think this is 107 a useful approximation to the total number of staff. We use this figure as the number of 108 staff members moving between community and jail. Staff transition between community 109 and jail at a rate that assumes 8 hour shift lengths in the jail per day with the 110 remaining 16 hours per day spent in the community. Estimation Population Mixing and Contact Rates 112 We estimate parameters that describe the relative rate of transmission between each of 113 the community categories: Children, Low-risk adults, High-risk adults, and Elderly 114 adults from other studies. We use β C * qr to denote the relative rate of transmission to 115 category q from category r. We estimate this as m qr is the number of times a person in category q is in contact with a person in 117 category r. We use the category-category contact rates given in [20] . c q is the 118 proportion of COVID-19 cases that occurred for people in category q and p q is the 119 proportion of the total population in category q. The ratio of these two terms, c q p −1 q , is 120 meant to capture the relative proclivity for people in group q to be infected by the virus. 121 We use case counts and population information for South Korea as reported by 122 Statista. 45 t i is the number of daily contacts for individuals in category q, as reported 123 in [21] . The β * s are normalized such that the Child-Child transmission is equal to one. 124 The resulting values of β C * is shown in Table 1 Within the jail and processing, we assume that mixing patterns are not 126 category-dependent. We set β J * = β J * qr = β C * LL c j and β P * = β P * qr = β C * LL c p , where β C * LL is 127 the low risk adult to low risk adult base transmission rate in the community and c j and 128 c p are factors that denotes how times more contacts per day a person in jail or 129 processing, respectively, has than a person in the community. We set these values to be 130 c j = 3 and c p = 6, corresponding to an assumption of three and six times more contact 131 in jail and processing, respectively, than take place in the community. values may be found in Table 2 . In one case,γ, or the asymptomatic period −1 , the original source reported that their 151 estimate was likely an underestimation due to censoring. However, given that the 152 authors provided the data within their manuscript [23] , the data was re-estimated to 153 account for censoring using a parametric survival model assuming an exponential from one compartment to another within a compartmental model). The fit for this 156 exponential model may be found in SI Appendix 2. We represented the effects of several policy interventions or failures as changes to various 159 parameters in this model. We consider four categories of scenarios that could vary the 160 rate of spread: in addition to modeling shelter-in-place (reduced mixing) conditions in 161 the community, we modeled scenarios related to reductions in arrest rates, increases in 162 release rates, and changes to within-jail conditions. These scenarios are detailed below 163 in Table 3 . Most scenarios are additive; that is, all arrest reduction interventions 164 assume a baseline scenario of shelter-in-place in the community. The scenarios involving 165 faster release of individuals in jail all assume both shelter-in-place in the community, 166 and were each run under each of the "Arrest Reduction" scenarios to determine the 167 cumulative effects of arrest reduction, increased release rates, and community 168 shelter-in-place conditions. The mixing reduction scenario in the jail assumes 169 shelter-in-place in the community as well as a 25% reduction in arrests (equivalent to 170 the "Bail Eligible" Arrest Reduction scenario), as it is unlikely that jails will be able to 171 effectively reduce contact rates without reducing their average daily population. Finally, 172 in the reduced detection scenario, we assume shelter-in-place, but vary the likelihood 173 that serious cases of COVID-19 is caught and treated in a timely manner. Below, vulnerable populations are defined as individuals over the age of 65 or at 175 increased risk of complications form COVID-19 due to other co-morbidities. We 176 estimate that 40% of the jail population is vulnerable by this definition, according to 177 information from the Bureau of Justice Statistics [29] . We estimate that around 25% of 178 those arrested are bail eligible, based on information from Allegheny county that cash 179 bail was used in 28% of cases between February and June of 2019 [30] . . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint . Table 3 . Scenarios and parameter adjustments for a number of policy-based interventions to curtail COVID-19 in jail and the community. , 2020 8/25 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint . hospitalizations and COVID-19 related fatalities (Fig 3) . Shelter-in-place orders had no 203 discernible impact on the health outcomes of the staff of the jail. , 2020 9/25 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint . In comparison, a strategy deferring the same number of people with no regard to 223 their underlying risk (≈36.5%), resulted in a 6.8% decrease in overall infections within 224 the incarcerated population, but a 61.7% increase in deaths among incarcerated persons 225 compared to the scenario specifically targeting those at greatest risk of averse outcomes 226 for deferred arrest (Fig 4) . The deferral strategy targeting individuals for high risk . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint . Cumulative infections, hospitalizations and deaths in the community (first row), among persons in jail (second row) and among jail staff (third row) for several incarceration deferment scenarios. More aggressive scenarios, such as a 90% decrease in all incarcerations or discontinuing incarceration for low level offenses result in large improvements in the epidemic within the jail, with a smaller impact among staff and the community. Deferring incarceration for people particularly at risk for adverse outcomes results in a markedly pronounced decrease in deaths among persons in jail. Pairing increased arrest deferral with a more rapid release of persons who were 233 already incarcerated enhanced the impact of those interventions, reducing infections, 234 hospitalizations and deaths among the incarcerated -as well as the staff of the jails -235 as the rate of release increased (Fig 5) . The one exception to this was the outcomes in 236 the wider community, which, when paired with less aggressive arrest deferral scenarios, 237 experienced a slight increase in the number of infections, hospitalizations and deaths at 238 lower levels of accelerated release schedules. This phenomenon peaks at a release rate 1.5 239 times faster than the default release rate, with a 0.3% increase in community infections. 240 At rates faster than this, the effect is considerably reduced, with a release rate twice as 241 fast as the default release rate resulting in a 0.02% increase in community infections. , 2020 12/25 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04.08.20058842 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint . from the ability to physically distance persons in jail while in common areas) from baseline (dark blue) to identical to the community's shelter-in-place order (green). This shifts the epidemic curve in the community slightly, and results in both a shifted and decreased curve among persons in jail as well as staff. An increase in the detection of severe COVID-19 cases among incarcerated persons 255 from 95% to 100% (equivalent to the same detection of the need for medical treatment 256 available in the community) unsurprisingly increased the number of hospitalizations, as 257 5 out of every 100 incarcerated persons needing hospitalization were no longer missed, 258 either for lack of access to care, insufficient diagnostic capacity, or other reasons. 259 Similarly, owing to the vast reduction in the case fatality rate between hospitalized 260 (CFR = 5% for low risk and 33.3% for high risk) severe cases and unhospitalized severe 261 cases (CFR = 100% for both groups), the number of deaths dropped by 72.2% when the 262 detection of severe cases rose to the same level as the community. Between these 263 scenarios, the number of infections rose slightly with better detection, increasing by 264 0.6% (Fig 7) . This is likely due to the slightly longer time an untreated severe case 265 spends in the incarcerated population before they are removed due to death vs. when a 266 treated case is transferred for hospitalization. This effect will only be present if the level 267 , 2020 14/25 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. These results clearly follow from the features of the jail system themselves in 284 challenging ways. While only 1% of the population entering into the jail system are 285 elderly [31] , incarceration in jail itself degrades the health of incarcerated people [16] [17] [18] , 286 leaving them more vulnerable to infection and severe outcomes from infection [32] . As 287 individual robustness to disease decreases, the epidemiological result is the increased 288 vulnerability of the whole jail population. Beyond the direct implications for the health of incarcerated people, jail populations 290 have high rates of re-entry into the general community and they depend on people who 291 regularly mix with the outside community. Jail populations are largely composed of 292 individuals who have not been convicted of a crime, and therefore will be released 293 quickly back into the general community rather than to further incarceration within the 294 carceral system. Jails with disease prevalence higher than the general populations they 295 , 2020 15/25 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04.08.20058842 doi: medRxiv preprint serve will therefore act as sources of infection, re-seeding infection into communities that 296 may be striving to contain or mitigate ongoing outbreaks, or even reintroducing 297 infection into otherwise disease-free populations. It is important to note that this would 298 happen even if no one were released given the volume of people coming in and out of 299 jails in staff and vendor roles, so should therefore not be construed as an implication 300 that releases should be suspended or impeded. 301 We are not helpless to effect change. Some obvious potential courses of action 302 suggest themselves immediately. New arrests of people of unknown disease status may 303 be regularly brought into jails, increasing the likely severity of outbreaks both by the 304 plausible continuous introduction of new sources of infection and by the maintenance of 305 higher rates of contact among susceptible incarcerated people due to the density and 306 structure of jail housing arrangements. If jurisdictions across the country reduce their 307 intake by significant percentages, our models demonstrate that we will meaningfully 308 directly reduce the disease incidence in the incarcerated population (as seen in Fig 4) . 309 Moreover, these same strategies also clearly produced a reduction in the source of risk 310 to incarcerated people's families and the broader community (Fig 4) . These strategies 311 could be enacted in a number of ways, such as (but not limited to) replacing 312 misdemeanor arrests with citations, avoiding recommendations for jail time or 313 prohibitive terms for bail conditions, or refusing to detain anyone for nonpayment of 314 fines or fees during the course of the outbreak. Having considered these potential strategies for categorical reduction in intake into 316 jails, we also considered the case in which the categorical consideration for reduction in 317 intake stemmed instead from the health of the arrested person. In this case, we get the 318 expected reduction in the within-jail outbreak that would have been associated with a 319 general reduction of the same percent intake (≈36.9%, allowing for a small number of 320 incarcerations in these groups), but we fail to achieve any significant reduction in 321 disease burden in the broader community by taking this action. It is therefore more 322 effective to reduce the intake rate across the entire population than to attempt to single 323 out particular categories of individuals due to their likely susceptibility to severe 324 morbidity or mortality from infection. The larger the reduction in overall intake, the 325 greater the reduction in disease achieved for all populations (incarcerated people, the 326 broader community, and jail staff, in decreasing proportion of effect). These broader 327 interventions are also likely to be relatively straightforward to implement 328 administratively, without knowledge of an individual's underlying comorbidities, if any. 329 In addition to reducing rates of intake into the jail system, another obvious, concrete 330 step we might take to reduce disease risks for everyone is to increase the rate of release 331 from jails. This should clearly be coupled with a decreased rate of intake rather than 332 enacted in isolation, since increasing release rates while maintaining the same rate of 333 intake would increase infection risks for incarcerated people, the staff who work at the 334 jails and court systems, and the broader community. This may even still occur when 335 expedited release is coupled with decreased rates of intake if the rate of release is To be maximally effective, each of these interventions should anticipate, rather than 340 react to, widespread infection incidence in jail populations. Critically, the factors that cause these outbreak dynamics and drive the resulting As with all models, the conclusions of this study depend on an accurate 363 representation of the flow of individuals between the jail system and the wider 364 community, either due to arrests or due to their employment as jail staff, as well as the 365 values of the parameters used to determine how swiftly this flow occurs. The inherent 366 nature of emerging epidemics makes both of these things uncertain -the clinical and 367 biological aspects of the pathogen might not be fully understood, and the data needed 368 to parameterize these models is often sparse and incomplete. This problem is especially 369 acute in models of this sort, which seek to present a "what-if" scenario to stave off a 370 public health crisis, rather than analyze how that crisis unfolded after the fact. Nevertheless, while the exact projected magnitudes may be sensitive to these unknowns, 372 in truth, the greatest utility of models such as these in in determining best courses of 373 action and likely magnitudes of the effects that can be gained from those actions, rather 374 than exact predictions of precise numbers of individuals [33] . Due to the logical nature 375 of the processes studied, so long as errors in the parameters used are consistent across 376 scenarios, they will not impact the understanding that results from our projections 377 about which courses of action achieve the best outcomes, even if those errors would alter 378 our understanding of the precise amount of effect achieved by each intervention. Conditions within jails must be immediately improved to decrease the probabilities of 381 disease transmission and support better health for incarcerated people to protect not 382 only themselves, but also jail staff and the community at large. Decreasing population 383 density both directly decreases disease exposure, interrupting transmission dynamics, 384 and also facilitates many other interventions. It is a natural result of reduced intake. 385 We can achieve/enable many desired benefits with just that one, simple action, but to 386 achieve maximal benefits to society, preventing the greatest burden from disease both 387 within the jails and without, broad actions that include alterations to both intake and 388 release and also many of the within-jail strategies for improving the individual means to 389 enact personal hygiene, protection through social distancing, and access to medical care 390 are all needed. . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint SI Appendix 1. According to the logic presented in the Model/Methods section 1 define the following system of equations to capture the epidemiological dynamics of our system: Within the Broader Community: , 2020 20/25 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04.08.20058842 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04.08.20058842 doi: medRxiv preprint , 2020 25/25 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04.08.20058842 doi: medRxiv preprint See Which States and Cities Have Told Residents to Stay at Home The Impact of School Closure for COVID-19 on the US Healthcare Workforce and the Net Mortality Effects. Available at SSRN 3555259 The Incarceration Nation: Interpreting the United States Imprisonment Rate Measuring the Exposure of Parents and Children to Incarceration Examining the relationship between US incarceration rates and population health at the county level. SSM-Population Health What percentage of Americans have ever had a family member incarcerated?: Evidence from the family history of incarceration survey (FamHIS) Discrete Event Simulation of Jail Operations in Pursuit of Organizational Culture Change Jail Inmates at Midyear 2010 -Statistical Tables The Company Store: A Deeper Look at Prison Commissaries Infection control in jails and prisons An Epidemic of Pneumococcal Disease in an Overcrowded, Inadequately Ventilated Jail Health care needs and service use among male prison inmates in the United States: A multi-level behavioral model of prison health service utilization Health issues of incarcerated women in the United States Juvenile incarceration and health The need for power, stress, immune function, and illness among male prisoners Vitamin D deficiency and inadequacy in a correctional population Do people who experience incarceration age more quickly? Exploratory analyses using retrospective cohort data on mortality from Ontario Mass incarceration: The whole pie. Prison Policy Initiative The impact of stratified immunity on the transmission dynamics of influenza Contact network structure explains the changing epidemiology of pertussis Impact of non-pharmaceutical interventions (NPIs) to reduce COVID19 mortality and healthcare demand Clinical characteristics of 24 asymptomatic infections with COVID-19 screened among close contacts in Nanjing, China COVID-19 Epidemic Risk Assessment for Georgia. Available on GitHub The incubation period of coronavirus disease 2019 (COVID-19) from publicly reported confirmed cases: estimation and application. Annals of internal medicine Estimates of the severity of COVID-19 disease Transmission interval estimates suggest pre-symptomatic spread of COVID-19. medRxiv Severe Outcomes Among Patients with Coronavirus Disease 2019 (COVID-19)-United States Medical problems of state and federal prisoners and jail inmates Punishing Poverty: Cash Bail In Allegheny County Allegheny County Jail Population Management: Interactive Dashboards The exponential survival model fit to Hu et al., 2020 estimated a mean asymptomatic shedding period of 11.79 days (95% CI: 7.86, 18.61), which is indeed a higher estimate than that found in the original manuscript, which did not account for censoring. As 5.1 of those days are already accounted for in the original estimate for γ − 1, this yielded an estimate forγ −1 of 6.7 days. Compared to a Kaplan-Meier fit of the available data, the exponential model fit well, and adequately models underlying survival function (Fig 8) .