Modeling Nursing Home Harms From COVID-19 Staff Furlough Policies

Key Points Question What is the tradeoff between COVID-19–related harms and non–COVID-19–related harms when allowing nursing home staff with mild COVID-19 to work while masked? Findings In this modeling study that used an agent-based model of a 100-bed nursing home, understaffing (without SARS-CoV-2 infections) was associated with missed tasks, resident hospitalizations, and deaths, costing an annual $1 071 950, and furloughing staff testing positive for SARS-CoV-2 infection was associated with additional missed tasks, non–COVID-19 hospitalizations, and deaths, costing an additional $247 090 from the perspective of the Centers for Medicare & Medicaid Services (CMS). However, allowing 75% of nursing home staff who were mildly ill to work averted most of these non–COVID-19 harms, saving $85 470 from the CMS perspective without worsening staff or resident COVID-19 hospitalizations. Meaning The findings of this study suggest that allowing nursing home staff who were mildly ill with COVID-19 to work while masked was associated with less harm from alleviated missed tasks, outweighing increasing harm from COVID-19 transmission.


Introduction
Currently, nursing home staff who test positive for SARS-CoV-2 are furloughed for up to 10 days to reduce spread.However, given understaffing strains, furloughs may have detrimental consequences on resident care and burnout for remaining staff.2][3][4][5][6][7][8] From 2017 to 2018, 91% of US nursing homes were short staffed for registered nurses (RNs) more than 40% of the time. 3During the COVID-19 pandemic, 9 staff shortages negatively impacted care activities (eg, feeding, hydration, mobility support, bathing, and medication administration), resulting in adverse outcomes (eg, bedsores, dehydration, and weight loss). 9-11Thus, it is important to understand the potential tradeoffs in furloughing staff to protect residents and other staff from COVID-19 vs not meeting residents' needs due to staff absences.Therefore, we adapted a computational agent-based model of a nursing home to estimate the clinical and economic impacts of different COVID-19 staff furlough policies.
Each resident had a set of characteristics (Figure 1) designating their daily care needs (eg, mobile support, feeding assistance, turning or mobility assistance, and receiving medications).The model advanced in discrete, 1-day time steps for 1 year.Each day, staff and residents interacted (eFigure 1 in Supplement 1) until leaving the nursing home (length-of-stay elapsed, hospitalized, died, or left job).A resident's interactions with staff or other residents were greater within their housing area, designated social groups and connections, and assigned staff.Staff interactions depended on their job type and assignments.

JAMA Network Open | Infectious Diseases
Modeling Nursing Home Harms From COVID-19 Staff Furlough Policies

Staff Roles and Activities
Resident-facing staff were assigned to a hallway or housing area, with staff interacting more with residents in an assigned area for continuity of care.We represented clinical care staff (CNAs, licensed practical nurses, and RNs) performing different tasks for residents.We modeled day and evening shifts when most tasks are performed.
We represented 4 major categories of tasks: feeding and hydrating; administering essential medications; turning, exercising, and moving residents; and hygiene-related tasks (eg, toileting assistance, bathing and showering, dressing, brushing teeth, and wound and device care).These tasks were performed daily based on residents' needs (Figure 1).Staff covered tasks of furloughed staff if they had time (eg, a CNA fed residents assigned to a furloughed CNA).If staff did not have time to complete tasks (ie, total hours worked was less than hours needed), they dropped tasks for randomly selected residents.Staff dropped tasks until their task list could be done within their hours worked.Staff were less likely to drop critical tasks (eTable 6 in Supplement 1) but had a chance of dropping each task, given residents' variable needs and requests (eg, incontinence episodes causing CNAs to prioritize cleaning before feeding).

SARS-CoV-2 Transmission
eFigure 1 in Supplement 1 shows previously described 12,13 mutually exclusive SARS-CoV-2 states that each resident and staff could be in and how they moved through them.Each day, residents and staff interacted with each other, and a person who was infected could transmit SARS-CoV-2 to a person who was susceptible (eMethods in Supplement 1).After recovering, residents and staff had natural immunity against infection (waned over time) (eMethods in Supplement 1) and hospitalization (long lasting 40 ).

Resident and Staff COVID-19-Related Health Outcomes
Residents and staff who were symptomatic started with mild infection with a probability of progressing to severe disease requiring hospitalization 12,13 and had a probability of being treated with antiviral medications.Each hospitalized individual had a length of stay and probability of COVID-19related mortality at the end of their stay.Residents who survived with a length of stay of 10 or fewer days returned to the nursing home (ie, their bed was held); those with longer lengths of stays left the model.

SARS-CoV-2 Testing, Isolation, and Furloughs
Staff were tested when disclosing symptoms or when overt symptoms were noted, as previously described 12,13 (eFigure 2 in Supplement 1), assuming that 50% of those infectious were tested. 25,26aff awaiting test results wore N95 respirators and continued working.Staff who tested positive were furloughed for 7 or more days from test positivity or until they tested negative according to national guidance. 41When staff who were mildly ill were allowed to work, they wore N95 respirators for their infectious period duration.
Residents who were symptomatic were isolated while awaiting results.Residents who tested positive were isolated for 10 days and their roommates quarantined for 10 days.Resident isolation and quarantine required an N95 respirator, eye protection, gown, and glove use by staff.

Costs and Economic Outcomes
Each person accrued relevant costs and health effects as they moved through the model.Residents

Statistical Analysis
Our first set of scenarios assumed all staff who tested positive for SARS-CoV-

Allowing Staff With Mild COVID-19 Illness to Work While Masked
Allowing increasing proportions of staff who were mildly ill to work while wearing N95 respirators resulted in fewer furlough days, more completed tasks, more COVID-19-related and fewer non-COVID-19-related (missed task-related) health outcomes, and lower costs (Figure 2 and Simulated nursing home staff furlough policies allowed different proportions of staff who were mildly ill to work while wearing N95 respirators.Note difference in axis scales across panels. Figure 3).Non-COVID-19-related outcomes and costs consistently outweighed COVID-19 outcomes when staff who were mildly ill worked while masked (Figure 3 and Table 2).
Specifically, allowing 25% to 75% of staff who were mildly ill to work (staff who were moderately or severely ill were furloughed) alleviated understaffing.Compared with a 100% furlough, allowing 25% of staff who were mildly ill to work was associated with 6.1 (95% CI, 0.6-11.6)additional resident COVID-19 cases and no additional COVID-19-related hospitalizations or deaths.

Increasing Virus Transmissibility and Community Infection Risk
When evaluating whether increased SARS-CoV-2 transmissibility and community infection risk may change the estimated impact of allowing staff who were mildly ill to work, we found that increasing the transmission probability to 0.1 increased staff cases by 123.7 annually (22.5% community onset) and annual furlough days to 665 when 0% of staff who were mildly ill worked and to 167 annual furlough days when 75% of staff worked mildly ill (eFigure 3 in Supplement 1).Even with more staff and resident COVID-19 cases, allowing 75% of staff who were mildly ill to work while masked resulted in 984 additional completed tasks with significantly fewer non-COVID-19-related outcomes and no association with resident additional COVID-19 outcomes, reducing total costs (Table 2).Overall, alleviating non-COVID-19-related harms with fewer staffing shortages when 75% of staff who were mildly ill worked resulted in more cost savings despite increased COVID-19 cases when increasing viral transmissibility (Table 2).
Increasing the community infection risk 1.3 times the baseline values increased staff COVID-19 cases by 9 and resident cases by 3 annually.However, allowing staff who were mildly ill to work while masked was associated with few additional resident cases and COVID-19-related hospitalizations while also being associated with substantially fewer non-COVID-19-related harms from missed tasks, thus saving costs compared with a 100% staff furlough (Table 2).

Increasing the Severity of COVID-19
Increasing COVID-19 illness severity (increasing the probability of hospitalization by 3 times) also did not change the direction of tradeoffs, with reduced furloughs, missed tasks, and non-COVID-19related harms from allowing staff who were mildly ill to work still outweighing COVID-19 outcomes.

Staff Adherence to Masking While Working Mildly Ill and Percentage of Staff Being Tested
When accounting for the possibility that staff working with mild COVID-19 did not adhere to masking, the findings were robust when staff wore surgical masks instead of N95 respirators (Table 2).When staff who were mildly ill did not wear masks or N95 respirators, there were significantly more staff and resident COVID-19 cases but not hospitalizations, as well as additional CMS costs ($15 833).These costs were still outweighed by the cost savings from completed tasks that were associated with the prevention of non-COVID-19-related harms ($77 109).Similarly, the results were robust to changes in the percentage of staff who were infectious getting tested and to increased vaccination coverage, nursing home transmission, and importation of COVID-19 from the community (Table 2).

Discussion
Pervasive staffing shortages currently limit the provision of basic care needs in US nursing homes. 42,43 The intent of regulations to furlough all staff with COVID-19 earlier in the pandemic was to protect staff and residents when COVID-19 frequency, severity, and sequelae were much greater.
Since then, vaccination and prior illness have substantially reduced severe outcomes.This in turn may have tipped the balance between the benefits of mandatory furlough and unintended consequences on resident care, as well as the balance between staff furlough and burnout of remaining staff. 42,44,45r simulation experiments may be conservative.We assumed staff were unlikely to disclose illness and would work with hidden symptoms or presume mild symptoms were from noninfectious ailments.Thus, in our experiments, only half of staff infected with SARS-CoV-2 reported or had sufficiently overt symptoms to trigger testing.The rest of the staff worked while contagious, thus undermining the current mandated furlough policy.Reasons for underreporting and working ill include a need for workers to protect limited paid sick days.Substituting a policy that allows staff who are mildly ill to work could renew and compel efforts to train staff on masking; frequent hand disinfection; and distancing, when able, from coworkers and residents.[48] All of this is not to say that nursing home employees should be encouraged to work while sick with a contagious pathogen.Rather, our work underscored the reality that the development of appropriate nursing home staffing policies involves tradeoffs (eg, chronic understaffing means that reducing available staff time can have consequences).Models can help with assessing tradeoffs such as whether worsening understaffing by mandated furloughs may lead to more or less harm than allowing staff to work while contagious.We found that effective vaccines (representing current annual uptake), acculturation to masks (access, expectations, and the ability to wear them), and SARS-CoV-2 strain adaptations may now allow a clear answer of worsening harm from COVID-19 furloughs.This finding was unchanged when increasing virus transmissibility or severity, with non-COVID-19-related costs outweighing COVID-19-related harms and costs from all perspectives.This is because even a greater increase of COVID-19-related hospitalizations in a population that was relatively immune did not approach the severity experienced when SARS-CoV-2 entered a fully This study constitutes an important example of how infection-control decision-making should account for broader and downstream impacts throughout a system and how systems approaches can elucidate these impacts. 49The ultimate solution would be to substantially increase nursing home staff.This would not only address the issue of missed tasks even absent infectious disease threats, but would allow greater leeway for furlough strategies to control pathogen spread.New federal quality standards require 3 nursing hours per resident day (0.55 hours of RN care, 2.45 hours of nurse aide care). 50Another study has estimated that over 4 nursing hours are needed for quality care. 51wever, most nursing homes are severely understaffed with wage levels unable to compete with some fast food opportunities. 42,43The pandemic further exacerbated pervasive staffing shortages, which have persisted. 8,42,52,53In 2021, 94% of nursing homes were understaffed; in 2024, 72% had fewer staff than before the pandemic.If nursing homes become sufficiently staffed to handle crosscoverage for furloughing staff while preventing burnout, 42,54 then the tipping point may change.

Limitations
Limitations of our study include that all models are simplifications of actual situations and cannot account for every scenario and outcome.We made multiple conservative assumptions.We assumed that a non-COVID-19-related outcome was associated with a specific missed task; however, outcomes may be associated with and exacerbated by multiple missed tasks.Furthermore, residents may experience multiple negative outcomes; however, we assumed residents incurred only the first.
We assumed that staff performed tasks at the same efficiency regardless of staffing; in reality, staff may initially complete tasks more quickly if short-staffed but may ultimately burn out.Additionally, we did not represent prioritizing residents with a higher risk level or residents not attended to recently.We also assumed specialty care staff would not perform basic care; however, they may if greatly understaffed.Lastly, residents only accrued costs and health effects for hospitalizations and deaths.However, missed tasks may be associated with other detrimental effects to quality of life and costs.

Conclusions
This modeling study found that mandatory furlough of nursing home staff with mild COVID-19 illness was associated with more resident harm from missed care tasks than with harm from increased COVID-19 transmission.In the current climate of extreme nursing home understaffing, allowing those with mild COVID-19 illness to work may prevent more harm from staffing shortages and missed care tasks, ultimately saving substantial costs.
with multiple negative health outcomes from missed tasks incurred only the first outcome by days to harm.The model generated health effects and costs (in 2024 values) from the Centers for Medicare & Medicaid Services (CMS), total third-party payers, and the societal perspective (eMethods in Supplement 1).

Figure 2 .
Figure 2.Estimated Impact of Nursing Home Staff Furlough Policies Compared With Current Understaffing Levels on Nursing Home Operational Outcomes

Figure 3 .
Figure 3.Estimated Impact of Nursing Home Staff Furlough Policies Compared With Current Staffing Levels on COVID-19-Related and Non-COVID-19-Related Resident Outcomes and Costs Simulated nursing home staff furlough policies allowed different proportions of staff who were mildly ill to work while wearing N95 respirators.Note the difference in axis scales across panels.The third-party payer included combined direct medical costs for residents and staff.CMS indicates Centers for Medicare & Medicaid Services.

Table 1 .
Key Model Input Parameters, Values, and Sources

Table 2 .
Annual COVID-19-Related and Non-COVID-19-Related Health Outcomes and Costs Comparing Staff Furloughs in a 100-Bed Nursing Home a cost savings accrued when allowing 75% of staff who were mildly ill to work compared with furloughing all staff who tested positive for SARS-CoV-2.281 770 in COVID-19-related costs (4.1 times higher) to $1 134 700 in non-COVID-19related costs compared with $130 100 in COVID-19-related costs (8.7 times higher).
a Data are presented as mean (95% CI).b Based on allowing 75% of staff who were mildly ill testing positive for SARS-CoV-2 to work while wearing N95 respirators compared with furloughing all staff who tested positive.c All resident outcomes were averted counts per year for a 100-bed nursing home and equivalent to a rate per 100 person-years.d Negative values are JAMA Network Open.2024;7(8):e2429613.doi:10.1001/jamanetworkopen.2024.29613(Reprinted) August 19, 2024 8/15 Downloaded from jamanetwork.comby guest on 09/01/2024 compared with $ Our simulation experiments in this modeling study resulted in an annual 22.1% deficiency in care tasks at current staffing levels.Our simulations suggest that a mandatory staff furlough for COVID-19 illness may have exacerbated current understaffing and was associated with additional non-COVID-19-related hospitalizations and potentially an additional death per year (0.7) among residents in a 100-bed nursing home associated with missed care tasks.In contrast, if staff with mild COVID-19 illness worked while masked, 73% of those missed care tasks could be completed with the tradeoff of a small number of additional COVID-19 illnesses with minimal COVID-19-related hospitalizations for either staff or residents.Thus, under current staffing conditions, furloughing staff who were mildly ill with COVID-19 was associated with non-COVID-19-related harms that outweighed COVID-19-related harms among nursing home residents.Changing current regulation to allow staff with mild COVID-19 to work with an N95 respirator also may be cost saving for both the CMS and society.For each 100-bed nursing home, the CMS could save $85 470 from harms and hospitalizations.With 15 000 US nursing homes, such a change in regulation could translate to savings well over $1 million for the CMS while alleviating substantial harm.

JAMA Network Open | Infectious Diseases
Other models can similarly assess such tradeoffs for other pathogens, as these results may not be generalizable beyond COVID-19.