key: cord-0961391-l5cq5xhe
authors: Gomez-Vazquez, J. P.; Garcia, y.; Schmidt, A. J.; Martinez-Lopez, B.; Nuno, M.
title: Testing and Vaccination to Reduce the Impact of COVID-19 in Nursing Homes: An Agent-Based Approach
date: 2021-03-26
journal: nan
DOI: 10.1101/2021.03.22.21254125
sha: 2b775872d9ba8b4b2c18fd67fa2f6a872d71b62a
doc_id: 961391
cord_uid: l5cq5xhe

Background: Efforts to protect residents in nursing homes involve non-pharmaceutical interventions, testing, and vaccine. We sought to quantify the effect of testing and vaccine strategies on the attack rate, length of the epidemic, and hospitalization. Methods: We developed an agent-based model to simulate the dynamics of SARS-CoV-2 transmission in a nursing home with resident and staff agents. Interactions between 172 residents and 170 staff were assumed based on data from a nursing home in Los Angeles, CA. We simulated scenarios assuming different levels of non-pharmaceutical interventions, testing frequencies, and vaccine efficacy to block transmission. Results: Under the hypothetical scenario of widespread SARS-CoV-2 in the community, 3-day testing frequency minimized the attack rate and the time to eradicate an outbreak. Prioritization of vaccine among staff or staff and residents minimized the cumulative number of infections and hospitalization, particularly in the scenario of high probability of an introduction. Reducing the probability of a virus introduction reduced the demand on testing and vaccine to reduce infections and hospitalizations. Conclusions: Improving frequency of testing from 7-days to 3-days minimized the number of infections and hospitalizations, despite widespread community transmission. Vaccine prioritization of staff provides the best protection strategy, despite high risk of a virus introduction.

COVID-19 has highlighted many inadequacies in the American healthcare system. Elderly and frail 31 residents of long-term care facilities (LTCFs) have experienced a disproportionate burden of infection 32 and death. Approximately 5% of all US cases have occurred in LTCFs, yet deaths related to in these facilities account for 34% of all US deaths as of February 12, 2021, according to the New York 34 Times [1] . successful thus far, there is a growing concern that insufficient levels of vaccine coverage will be reached. 52 As of the end of January 2021, median first dose rates among LTCF residents is 77.8%, but only a 53 median of 37.5% of staff have received at least their first dose [9] . It is unclear at this time whether the 54 lower vaccination rates among staff is a result of prioritization of residents, lack of recording alternative 55 sources of vaccination, or staff choice; however, a survey of nursing home staff conducted in the state 56 of Indiana (November 2020) found that 45% of respondents were willing to receive a COVID-19 vaccine 57 immediately once available, and an additional 24% would consider it in the future [10] . While visitors are 58 disallowed and residents only interact directly with a small number of other people, staff are the primary 59 vector for viral introduction [11, 12] ; therefore, low rates of vaccine uptake among staff should be of great 60 concern from the perspective of preventing an outbreak. Additionally, there is limited evidence about the 61 ability of vaccines to reduce asymptomatic transmission. Preliminary data from the UK suggests a 49.3% 62 reduction in infections from an asymptomatic source [13] . Recent evidence of the circulation of more 63 transmissible SARS-CoV-2 variants also raises concerns about the course of this pandemic, particularly 64 3 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) 

Model structure 80 We developed a stochastic agent-based model to simulate the spread of SARS-CoV-2 in an LTCF, based 81 on the floor plan and occupancy of a nursing home in Los Angeles County, California with 172 residents 82 and 170 staff [ Figure 1 ]. The simplified floor map shows the location of bedrooms with a capacity 83 of 3 residents, 5 quarantine rooms reserved for residents with frequent outside traffic and/or capacity 84 to quarantine exposed residents, recreation areas which are currently off limits to resident and staff 85 interactions, and rooms for staff. Table 1 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Mortality rate was set at 11.8% [21] for hospitalized agents. The average recovery time for asymptomatic 102 agents or those who never required hospitalization is 15 days [17], during which they remain infectious. 103 We assumed that recovery from a primary infection provided adequate immunity for the remainder of 104 the simulation.

6 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Though there is large variability on the impact of COVID-19 in these facilities, tied to historic variability 128 in testing capacity and PPE availability and adherence, the most immediate risk of a COVID-19 outbreak 129 in a nursing home is the level of community transmission of SARS-CoV-2. Since we assumed that visitors 130 are disallowed completely, residents' risk for primary exposure is contact through staff who acquired an 131 infection from the wider community. A critical factor that our model aimed to study was to assess the 132 impact of the probability of viral introduction from the community on the predicted size of internal 133 outbreaks. Each scenario we investigated was simulated across three different probabilities of a staff 134 member introducing the infection: low (5% per day), medium (10% per day), and high (15% per day).

These are expressed as 'introduction probability', which is set to 0.1 for the baseline scenario 1. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted March 26, 2021. ; https://doi.org/10.1101/2021.03.22.21254125 doi: medRxiv preprint every 7 days (baseline), 5 days, and 3 days. Testing of residents in all scenarios assume that one resident 141 per room is tested weekly, systematically cycling through the each resident every three weeks. Reduction 142 in transmission probability from PPE use and vaccination were applied by modifying the shedding and 143 infection probability parameters (Table S1) 

where the odds ratio ω (OR ω ) represents the global baseline transmission probability of all agents, the 154 odds ratio π (OR π ) represents the transmission reduction from the presence or absence of PPE, and the The implementation of frequent testing, particularly every 3-days reduced the attack rate by half and 195 allowed containment of the outbreak within 9 days, despite high probability of virus introduction.

Estimates and 95% confidence intervals illustrated in Figure 3 are provided in Supplemental Table 2 .

When vaccine was prioritized among staff, residents, or both, the attack did not seem to differ except 198 when the introduction probability of was high, in which case the simulated median attack rate was 0.02 199 when staff were prioritized compared to 0.03 if residents were prioritized or no prioritization was present.

Assuming a low probability of introduction, no prioritization provided the best opportunity to control 201 an outbreak, leading to a median of 9 days (95% CI: 7-26) until eradication. We evaluate a vaccine's 202 ability to block transmission for scenarios of vaccine efficacy, staff and residents had the same efficacy, 203 and residents had reduced efficacy compared to staff. We found that the probability of virus introduction 204 was the most significant factor in determining the attack rate and days to the eradication of an outbreak.

The attack rate doubled to 0.02 with high transmission probability and the time to the eradication of 206 an outbreak was optimal only for low transmission. In all scenarios of low or moderate probability of 207 9 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Increasing the probability of an introduction increased the total number of infections from 7 to 72, and 224 5 additional individuals were hospitalized. Analyses for these outcomes revealed significant decreases 225 attributed to testing and vaccination across different frequency of testing and vaccine efficacy. Prevention 226 of hospitalizations was more effectively accomplished through vaccination and was independent on 227 age-specific vaccine efficacy assumptions. 228 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Results from our model were most evident when we assumed a larger probability of viral introduction.

In such cases, increased frequency of universal testing and isolation of positive cases (quarantine or paid 244 leave) lead to larger reductions in attack rate than any other scenario. Prioritizing the vaccination of staff 245 over residents lead to a moderate decrease in attack rate when viral introduction probability was high.

12 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) staff are the most important vectors through which introduction from the community occurs [11, 12, 28] .

Our results support using strategic prioritization of staff for universal testing and vaccination as an 249 important method for reducing the likelihood of an outbreak, especially in situations where community 250 transmission is high.

There are several important challenges that these facilities will continue to face. LTCF administrators 252 reported that staffing remains one of the primary barriers to maintaining high infection control standards 253 [29] . Additionally, facilities that had a high degree of connectedness via shared staff showed higher case 254 rates in general [30] . Expanded paid leave programs may also reduce the need for staff to seek additional 255 employment to make ends meet, generally lowering their personal risk and the risk of introduction events.

Evidence indicates that staff may be more hesitant to get the vaccine than residents [10] , and certainly facility. Maintaining vaccine coverage goals will likely require an active program that includes acquiring 263 accounting for staff who receive vaccines from a different source (i.e. a local pharmacy or a different job). 264 We have even less data about the risks presented by reopening nursing homes to visitors, prompting 265 questions about vaccine and testing requirements for visitors. An extension to this model that adds a 266 visitor agent could help answer these questions before observational data becomes available. on an outbreak once introduction has occurred, and is therefore not meant to model the processes that 277 lead to an introduction in the first place. Simulations were run for 150 days or until the facility was 278 disease-free for up to 7 days; thus, it is also not able examine the impact of multiple introductions over 279 longer periods of time or waning immunity from recovery or vaccination in its current form. 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Not all data-derived parameters were made equally. The estimated effect of PPE use on transmission 282 varied widely, thus making a reliable parameter difficult to define and the model sensitive to changes.

Testing was also oversimplified in our model, as we assumed instantaneous results and all tests were 284 equally sensitive. Additionally, we assumed that the effects of immunity, natural or from vaccination, 285 was constant over the course of an outbreak and did not wane over time. We also assumed that staff 286 agents had an equal chance of interacting with each resident agent, which is not reflective of intervention 287 strategies that silo staff into daily routines focused on a specific subset of residents, such as dedicated 288 staff for specific wards within the nursing home or for positive, isolated individuals. 289 14 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted March 26, 2021. ; https://doi.org/10.1101/2021.03.22.21254125 doi: medRxiv preprint

More than one-third of u.s. coronavirus deaths are linked to nursing homes

Long-term care providers 294 and services users in the united states

Outcomes 296 of universal covid-19 testing following detection of incident cases in 11 long-term care facilities

Mass-Testing for Coronavirus Disease 2019 in Long-Term Care Facilities for the

Covid-19 in nursing homes: Calming the perfect storm

An illustration of sars-cov-2 305 dissemination within a skilled nursing facility using heat maps

Optimizing covid-19 surveillance in long-term care facilities: a modelling study

First-dose covid-19 vaccination coverage among skilled 311 nursing facility residents and staff

Early covid-19 first-dose 314 vaccination coverage among residents and staff members of skilled nursing facilities participating in 315 the pharmacy partnership for long-term care program -united states

Willingness of long-term care 318 staff to receive a covid-19 vaccine: A single state survey

Presymptomatic transmission of severe acute respiratory syndrome coronavirus 2 among residents 321 and staff at a skilled nursing facility: Results of real-time polymerase chain reaction and serologic 322 testing

Efficient sars-cov-2 surveillance strategies to prevent deadly outbreaks 324 in vulnerable populations

Safety and efficacy of the chadox1 ncov-19 vaccine 327 (azd1222) against sars-cov-2: an interim analysis of four randomised controlled trials in brazil, south 328 africa, and the uk

Disease Control and Prevention

Temporal dynamics in viral shedding and transmissibility of covid-19

High proportion of asymptomatic sars-cov-2 infections in 9 long-term 335 care facilities

detection, viral load and infectivity over the course of an infection

Disparities in outcomes among covid-19 patients in a large health care system in 343 california: Study estimates the covid-19 infection fatality rate at the us county level

Predictors of mortality of covid-19 347 in the general population and nursing homes

Accessed on the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted

Covid-19 hospitalizations in 352 five california hospitals

Physical distancing, face masks, and eye protection 355 to prevent person-to-person transmission of sars-cov-2 and covid-19: a systematic review and 356 meta-analysis

Efficacy and safety of the mrna-1273 sars-cov-2 vaccine

Safety and efficacy of the bnt162b2 mrna covid-19 362 vaccine

Building, composing and experimenting complex spatial models with the gama 365 platform

Covid-19 nursing home data -data.cms.gov

Nothing much has changed: Covid-19 nursing home cases and 369 deaths follow fall surges

Mitigation of a coronavirus disease 2019 outbreak in a nursing 372 home through serial testing of residents and staff

Persistent challenges of covid-19 in skilled nursing facilities: The administrator 375 perspective

CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity