Investigation of A SARS-CoV-2 Delta (B.1.617.2) Variant Outbreak Among Residents of a Skilled Nursing Facility and Vaccine Effectiveness Analysis — Maricopa County, Arizona, June–July 2021

Abstract Background Short-term rehabilitation units present unique infection control challenges due to high turnover and medically complex residents. In June 2021, Maricopa County Department of Public Health (MCDPH) was notified of a SARS-CoV-2 Delta outbreak in a skilled nursing facility short-term rehabilitation unit. We describe the outbreak and assess vaccine effectiveness (VE). Methods Facility electronic medical records were reviewed for residents who spent >1 night on the affected unit between June 10–July 23, 2021, to collect demographics, SARS-CoV-2 test results, underlying medical conditions, vaccination status, and clinical outcomes. COVID-19 VE estimates using Cox proportional hazards models were calculated. Results Forty (37%) of 109 short-stay rehabilitation unit residents who met inclusion criteria tested positive for SARS-CoV-2. SARS-CoV-2 positive case-patients were mostly male (58%) and white (78%) with a median age of 65 (range: 27-92) years; 11 (27%) were immunocompromised. Of residents, 39% (10 cases; 32 non-cases) received 2-doses and 9% (4 cases, 6 non-cases) received 1-dose of mRNA vaccine. Among non-immunocompromised residents, adjusted 2-dose primary-series mRNA VE against symptomatic infection was 80% (95% CI: 15, 95). More cases were hospitalized (33%) or died (38%) than non-cases (10% hospitalized; 16% died). Conclusions In this large SARS-CoV-2 Delta outbreak in a high-turnover short term rehabilitation unit, a low vaccination rate and medically complex resident population were noted alongside severe outcomes. VE of 2-dose primary-series mRNA vaccine against symptomatic infection was the highest in non-immunocompromised residents. Health departments can use vaccine coverage data to prioritize facilities for assistance in preventing outbreaks.

M a n u s c r i p t 3

Background
Skilled nursing facility (SNF) residents are at increased risk of contracting SARS-CoV-2 compared to non-congregate living persons and were prioritized for early vaccination [1]. Outbreaks have been previously described in long-term care facilities [2,3] but data on the impact of the Delta(B.1.617. 2) variant in these settings are only now emerging. Within SNFs, short-term rehabilitation unit residents are unique due to varying disease acuity; outbreaks in this population have not been well characterized. Additionally, real-world vaccine effectiveness (VE) estimates against the Delta variant among rehabilitation unit residents are limited and can help inform state and local health department response.
In June 2021, Maricopa County Department of Public Health (MCDPH) was notified of a SARS-CoV-2 outbreak in a SNF short-term rehabilitation unit. Weekly SARS-CoV-2 antigen testing was initiated for staff and residents following notification to MCDPH [4]. Our primary objective is to describe the extent of a Delta variant outbreak in a short-term rehabilitation unit in addition to assessing estimated VE against significant clinical endpoints.

Methods
MCDPH was notified June 18, 2021, of a positive SARS-CoV-2 result from a SNF resident upon admission for respiratory symptoms to a local acute care on June 17, 2021. Facility-wide testing of residents was initiated the same day using the Abbott BinaxNOW antigen test; staff had been completing weekly testing in accordance with moderate community transmission levels [4,5].
Weekly antigen testing was provided to all staff and residents, and all positive antigen tests were confirmed by reverse-transcriptase polymerase chain reaction (RT-PCR). Available specimens were sequenced at their respective laboratories.
Additionally, trained data abstractors reviewed facility electronic medical records (EMR) for residents who spent >1 night in the affected unit during June 10-July 23, 2021. Abstracted data A c c e p t e d M a n u s c r i p t 4 included demographic information, admission and discharge dates, underlying conditions associated with severe COVID-19 outcomes [6], SARS-CoV-2 test results, vaccination history, COVID-19 symptoms, and hospitalization and death outcomes. Testing, vaccination, and death certificate information were supplemented with data received from the Arizona Department of Health Services.
The facility admitted SARS-CoV-2 positive individuals during their isolation periods; these persons were excluded from the outbreak analysis. A case patient was defined as a resident with a SARS-CoV-2 positive RT-PCR test while at the facility. Additionally, residents who were transferred to another healthcare facility and tested positive within 24 hours or died <24 hours after discharge and tested positive on autopsy were considered cases. Resident and staff vaccination status was categorized as 1) unvaccinated (no documented COVID-19 vaccination); 2) vaccinated (completion of a 2-dose primary series of the COMIRNATY (Pfizer-BioNTech) or SPIKEVAX (Moderna) mRNA vaccines or one dose of the Johnson & Johnson (Janssen) vaccine; all doses received ≥14 days before the risk period or positive SARS-CoV-2 test) and 3) partially vaccinated (at least one dose of an mRNA vaccine; ≥14 days before the risk period or positive SARS-CoV-2 test, or two doses, with the 2 nd dose received <14 days before the risk period or positive SARS-CoV-2 test. An epi curve was created using specimen collection dates comparing cases and non-cases by demographic and clinical characteristics. We calculated attack rates by vaccination status; however, given limitations in interpreting simple attack rates by vaccination status with varying resident lengths of stay, we used more rigorous methods to calculate VE. VE estimates were calculated using a time-to-event analysis. Person-time began on the facility admission date or June 10, 2021 (one week prior to index case), whichever occurred later, and ended on the date of first positive test, discharge, death, or July 23, 2021, whichever occurred earliest. Days between facility readmissions were excluded for residents readmitted during the investigation period. Residents were included in the VE analysis if they received ≥1 SARS-CoV-2 test at the facility and were not positive for SARS-CoV-2 within 90 days prior to admission.
A c c e p t e d M a n u s c r i p t 5 Cox proportional hazards models were used to calculate primary series VE and 95% confidence intervals (CI) as (1-hazard ratio) X 100%. Our modeling strategy was informed by literature describing COVID-19 VE and disease severity [6,7]; possible confounders were systematically added to develop adjusted models (Supplementary Table 1). Our primary analysis was to assess mRNA VE in nonimmunocompromised persons by excluding severely immunocompromised residents (e.g., having current cancer, solid organ transplant, HIV, or immunosuppressive condition/therapy) [8]. Secondary VE analyses were conducted to include those vaccinated with Janssen and to other clinical endpoints including infection, symptomatic infection, hospitalization, and death, and to exclude those with a documented history of SARS-COV-2 infection. All analyses were conducted using SAS statistical software (version 9.4; SAS Institute). This activity was reviewed by CDC and was conducted consistent with applicable federal law and CDC policy [9].

Results
The outbreak was limited to one unit at the SNF. For 68 available beds in this unit of the SNF, 161 residents spent ≥1 night during June 10-July 23, 2021. EMR abstractions were conducted for 161 residents; 109 residents met inclusion criteria for additional analysis (supplementary Figure 1). Fiftytwo residents were excluded from the epi curve and additional analyses because they were admitted to the facility with COVID-19, spent less than 24 hours in the facility, or had not received a SARS-CoV-2 test during their stay.  Among 109 residents included in the VE analysis, most residents were male (n=60; 55%) and of white race (n=76; 70%). Age range was 27-92 years for case patients (median=65) and 32-97 years for non-cases (median=69). Among all residents meeting inclusion criteria, the median number of underlying conditions associated with risk of severe COVID-19 illness was 4 (range 1-8) ( Table 1).
Among 74 non-immunocompromised residents (22 case-patients, 52 non-cases) series, adjusted 2dose primary-series mRNA COVID-19 VE against SARS-CoV-2 symptomatic infection was 80% albeit A c c e p t e d M a n u s c r i p t 7 with a wide confidence interval (CI; 95% CI: 15, 95%) ( Figure 1). Primary-series VE (13 case-patients, 69 non-cases) with any vaccine against hospitalization was 48% (95%CI: -118, 87%) and increased to 69% (95%CI: -20, 92%) against death (Figure 1). Two-dose primary mRNA vaccine series VE (27 casepatients, 59 non-cases) against SARS-CoV-2 infection was 51% (95%CI: -27, 81%) and increased to 80% (95%CI: -10, 96%) against death, following a similar pattern of increasing VE against increasingly severe endpoints such as hospitalization or death. Additionally, two-dose primary mRNA vaccine series VE estimates were higher than VE with any vaccine. There was no substantial difference when excluding those with a previous SARS-COV-2 infection (data not shown). Additional VE estimates are presented in supplementary material including variables used in model adjustment. This large SARS-CoV-2 Delta variant outbreak within the SNF short-term rehabilitation unit occurred as the Delta variant was emerging in Arizona [10]. While COVID-19 SNF outbreaks and the increased risk for severe disease and outcomes among residents has been well-documented [2,3,11], fewer data are available on Delta variant outbreaks in these settings, particularly in the unique setting of a short-stay rehabilitation unit. The cohort described here had a wide age range (27-97 years) and included many residents with complex underlying conditions and other risk factors associated with poor outcomes, including high proportions of substance use disorders and/or experiencing homelessness [12]. Widespread presence of the highly contagious Delta variant in the United States Real-world VE estimates can guide health departments to prioritize distribution of resources and support outbreak response to mitigate COVID-19 spread. Although this outbreak was large, the sample size was too small to be able to draw substantive conclusions about VE. However, increasing trends in the VE point estimates with outcome severity is consistent with other studies [13]. The use of a real-world COVID-19 outbreak to calculate vaccine effectiveness estimates has a strong purpose; it characterizes vaccine performance in a unique setting and helps translate what might otherwise be considered technical epidemiologic methods into real-world context that is more easily appreciated by nursing home staff, residents, and others. This analysis is subject to several limitations. First, EMR abstraction did not adequately capture timing or initiation of a do-not-resuscitate order or hospice care, which limited our understanding of death outcomes. Second, due to changes in commercial laboratory processes and specimen retention times, only ten of 40 samples were sequenced (all Delta). However, it is less likely that another variant contributed to this outbreak given the prevalence of Delta variant circulating during the outbreak period and bioinformatic analyses showing multiple introductions into the facility [10].