key: cord-0779115-0lajhqn5
authors: Misra-Hebert, Anita D; Jehi, Lara; Ji, Xinge; Nowacki, Amy S.; Gordon, Steven; Terpeluk, Paul; Chung, Mina K.; Mehra, Reena; Dell, Katherine M.; Pennell, Nathan; Hamilton, Aaron; Milinovich, Alex; Kattan, Michael W.; Young, James B.
title: Impact of the COVID-19 pandemic on healthcare workers risk of infection and outcomes in a large, integrated health system.
date: 2020-08-19
journal: Res Sq
DOI: 10.21203/rs.3.rs-61235/v1
sha: 5906b659353764be80bfb8036590f91d1862f668
doc_id: 779115
cord_uid: 0lajhqn5

Background: Understanding the impact of the COVID-19 pandemic on healthcare workers (HCW) is crucial. Objective: Utilizing a health system COVID-19 research registry, we assessed HCW risk for COVID-19 infection, hospitalization and intensive care unit (ICU) admission. Design: Retrospective cohort study with overlap propensity score weighting. Participants: Individuals tested for SARS-CoV-2 infection in a large academic healthcare system (N=72,909) from March 8-June 9 2020 stratified by HCW and patient-facing status. Main Measures: SARS-CoV-2 test result, hospitalization, and ICU admission for COVID-19 infection. Key Results: Of 72,909 individuals tested, 9.0% (551) of 6,145 HCW tested positive for SARS-CoV-2 compared to 6.5% (4353) of 66,764 non-HCW. The HCW were younger than non-HCW (median age 39.7 vs. 57.5, p<0.001) with more females (proportion of males 21.5 vs. 44.9%, p<0.001), higher reporting of COVID-19 exposure (72 vs. 17 %, p<0.001) and fewer comorbidities. However, the overlap propensity score weighted proportions were 8.9 vs. 7.7 for HCW vs. non-HCW having a positive test with weighted odds ratio (OR) 1.17, 95% confidence interval (CI) 0.99-1.38. Among those testing positive, weighted proportions for hospitalization were 7.4 vs.15.9 for HCW vs. non-HCW with OR of 0.42 (CI 0.26-0.66) and for ICU admission: 2.2 vs.4.5 for HCW vs. non-HCW with OR of 0.48 (CI 0.20 -1.04). Those HCW identified as patient-facing compared to not had increased odds of a positive SARS-CoV-2 test (OR 1.60, CI 1.08-2.39, proportions 8.6 vs. 5.5), but no statistically significant increase in hospitalization (OR 0.88, CI 0.20-3.66, proportions 10.2 vs. 11.4) and ICU admission (OR 0.34, CI 0.01-3.97, proportions 1.8 vs. 5.2). Conclusions: In a large healthcare system, HCW had similar odds for testing SARS-CoV-2 positive, but lower odds of hospitalization compared to non-HCW. Patient-facing HCW had higher odds of a positive test. These results are key to understanding HCW risk mitigation during the COVID-19 pandemic.

Understanding the risks associated with the COVID-19 pandemic 1 on healthcare workers (HCW), including the risk of acquisition at work vs other settings, is crucial. Prediction of risk can inform how to protect HCWs such as recommendations on use of personal protective equipment (PPE) at work or in the community. The presence of speci c symptoms in HCW (China, US) 2,3 and symptoms predicting SARS-CoV-2 test positivity in HCW (Netherlands) 4 have been reported as well as characteristics associated with HCW deaths (China). 5 Based upon data from the 2018 National Health Interview Survey, it was estimated that 26.6% of patient facing HCW were at increased risk for poor outcomes from COVID-19 infection because of their comorbidities or age. 6 Reported experiences in China 7 , Italy 8 and Solano County, CA without initial use of PPE 9 showed higher percentages of HCW testing positive for COVID-19. In contrast, a screening study of HCW in England showed no signi cant difference in positive results between clinical and nonclinical staff with implementation of isolation and PPE protocols perhaps suggesting predominant community rather than nosocomial transmission patterns. 10 The extent of risk modi cation with PPE remains unclear. [7] [8] [9] 11 A recent prospective study in the United Kingdom and US suggested a ve-fold increased risk for HCW caring for patients with COVID-19 compared to HCW not caring for patients with COVID-19, even with the use of PPE 12 while another study of HCW in a large healthcare system showed a decrease in positive tests for SARS-CoV-2 associated with a universal masking recommendation. 13 This heterogeneous landscape makes it di cult for the HCW community to determine actual risk of acquiring COVID-19 in healthcare vs. community settings and the effectiveness of various risk-mitigating strategies.

The Cleveland Clinic Health System (CCHS) is a large, integrated health system with 55,574 eligible employees in Ohio & Florida. The CCHS initiated multiple COVID-19 related public health initiatives to mitigate the spread of the disease and its impact on the HCW community. In parallel, we maintained a rigorous, comprehensive, and prospective registry capturing disease risk and progression in all individuals tested for COVID-19 in our health system. In this study, we aimed to assess whether HCW are at higher risk for COVID-19 infection, COVID-19 related hospitalization, and intensive care unit (ICU) admission compared to non-HCW using advanced statistical methodology to account for various confounders.

Cohort de nition COVID-19 Cleveland clinic enterprise registry: All patients, regardless of age, who were tested for COVID-19 at all CCHS locations in Ohio and Florida were included in this research registry. For this study, all individuals who were tested for COVID-19 in the CCHS between March 8, 2020 and June 9, 2020 were studied. This registry provides better representation of the overall population than testing restricted to one geographic health system site. Registry variables were chosen to re ect available literature on COVID-19 disease characterization, progression, and proposed treatments, including medications initially thought to have potential for bene t after drug-repurposing network analysis. 14 Capture of detailed research data was facilitated by the creation of standardized clinical templates implemented across the healthcare system as patients were seeking care for COVID-19-related concerns. Data were extracted via previously validated automated feeds from electronic health records 15 (EPIC; EPIC Systems Corporation) and

manually by a study team trained on uniform sources for the study variables. Study data were collected and managed using REDCap electronic data capture tools hosted at the Cleveland Clinic. 16, 17 The COVID-19 Research Registry team includes a "Reviewer" group and a "Quality Assurance" group. The reviewers were responsible for manually abstracting and entering a subset of variables that cannot be automatically extracted from the electronic health record (EHR). Reviewers were also asked to verify high-priority variables that have been automatically pulled into the database from EPIC. The Cleveland Clinic Institutional Review Board approved this study and waived the requirements for written informed consent. 21, 22 weighting was performed to address potential confounding in comparing HCW to non-HCW given their baseline differences. The overlap propensity score weighting method was chosen given its bene ts of preservation of numbers of individuals in each group and of achieving higher levels of precision in the resulting estimates. This methodology is preferred when the propensity score distributions among the groups are dissimilar and when the propensity scores are clustered near the extremes (i.e. close to zero or one). A propensity score for being a HCW was estimated from a multivariable logistic regression model. For the outcome of being test positive for COVID-19, the propensity score logistic regression model included covariates that were found to be associated with a positive COVID-19 test outcome in our previous work. 23 For the outcomes of hospital and intensive care unit (ICU) admission of COVID-19 test positive patients, the propensity score covariates are those that were found associated with COVID-19 hospitalization outcome in our previous work including age, race, ethnicity, gender, smoking history, body mass index, median income, population per housing unit, presenting symptoms (including fever, fatigue, shortness of breath, diarrhea, vomiting), comorbidities (including asthma, hypertension, diabetes, immunosuppressive disease), medications (including immunosuppressive treatment, non-steroidal anti-in ammatory drugs [NSAIDs]), and laboratory values (including pre-testing platelets, aspartate aminotransferase, blood urea nitrogen, chloride, and potassium).

The overlap propensity score weighting method was then applied where each patient's statistical weight is the probability of that patient being assigned to the opposite group. 21 Overlap propensity score weighted logistic regression models were used to investigate associations between HCW status and the probability of testing positive for SARS-CoV-2, hospital admission for COVID-19 and ICU admission for COVID-19 illness. The results are thus reported as weighted proportions, odds ratios and 95% con dence intervals. All statistical analyses were performed using R 3.5 and SAS version 9.4 (SAS Institute). P values were 2-sided, with a signi cance threshold of .05.

We then used locally weighted regression smoother (LOESS) to summarize the trend of COVID-19 test positivity through the study period for HCW and non-HCW as related to the public health measures instituted at the state level in Ohio and those speci c to the CCHS.

Overall COVID-19 cohort characteristics and outcomes: There were 551 HCW and 4,353 non-HCW who tested positive for COVID-19 (Appendix Table 2 ). Of those who tested positive for COVID-19, a lower proportion of HCW were hospitalized compared to non-HCW (38 or 6.9% HCW vs. 1205 or 27.7% non-HCW) or were admitted to the intensive care unit (10 or 1.8% HCW vs. 470 or 10.8% non-HCW). In the group who tested positive for COVID-19, there was a greater proportion of HCW of Asian and White race compared to non-HCW (2.9 vs. 0.8% and 61.0 vs 56.4%, respectively), a similar proportion of HCW with a positive COVID-19 test had presenting symptoms of cough, fatigue, diarrhea, loss of appetite, and vomiting and a lower proportion had fever or shortness of breath. Lower proportions of HCW testing positive had COPD/emphysema, diabetes, coronary artery disease, heart failure, cancer, or immunosuppressive disease and were previously prescribed carvedilol, angiotensin converting enzyme inhibitors, angiotensin receptor blockers or melatonin compared to non-HCW. The neighborhood population characteristics of population density or population per housing unit did not differ for those HCW who tested positive and median income was slightly higher compared to non-HCW.

Overlap propensity weighting: Using the variables in the prediction model for COVID-19 test positivity, 23 overlap propensity score weighting ( Table 2 ) resulted in propensity score weighted proportions of 7.7 vs. 8.9 for non-HCW vs. HCW having a positive test and produced an overlap propensity score weighted odds ratio of 1.17 with a 95% con dence interval (CI) of 0.99-1.38 for a HCW having a positive test compared to a non-HCW (Figure 1a ). Then using the variables which predicted hospitalization for COVID-19 infection, overlap propensity score weighting was applied ( (Figure 1a ).

We then compared characteristics of HCW identi ed as having positions that required direct contact with patients ("patient facing") and those that did not. There were 5,159 HCW with patient-facing positions and 986 HCW in non-patient facing roles (Appendix Table 3 

The summary of the trend of SARS-CoV-2 positive test results in the study period is shown in Figure 2 . The overall proportion of positive COVID-19 test results decreased during the study period and the trend for HCW and followed that of non-HCW.

Our analysis of HCW compared to non-HCW who were tested for SARS-CoV-2 in one health system with 2 geographic locations (Ohio, Florida), and which controlled for signi cant differences in baseline characteristics between the HCW and non-HCW groups, showed that the odds of having a positive COVID-19 test were not signi cantly different for HCW compared to non-HCW, and HCW had lower odds of subsequent hospitalization, and without statistically signi cant differences in ICU admission compared to non-HCW once they tested positive. The HCW classi ed as having patient-facing positions had higher and signi cant odds of a positive COVID-19 test with insigni cant differences detected compared to nonpatient facing HCW in outcomes of hospitalization or ICU admission. We found a similar proportion of HCW with a positive COVID-19 test had presenting symptoms of cough, fatigue, diarrhea, loss of appetite, and vomiting while a lower proportion had fever or shortness of breath. We note that we were not able to capture the symptoms of loss of taste and/or smell and that these symptoms may be common especially with mild cases of COVID-19. 24, 25 The overall proportion of COVID-19 positive tests in HCW was low and decreased during the study period corresponding with implementation of risk-mitigation measures in our health system such as the recommendations for universal masking and physical distancing but also followed the trend for non-HCW. Several of the previous studies of HCW risk for infection during the COVID-19 pandemic were limited by their sample sizes, 7-9 lack of generalizability for healthcare systems that have adequate access to PPE, 7-9 methodology relying on self-report, 12 limited ability to adjust for known risk factors of disease susceptibility and progression [7] [8] [9] [10] 12 and lacking data to investigate the relative effects of dual exposure of HCW to COVID-19 in the community versus the workplace. [7] [8] [9] [10] 12 The fact that HCW identi ed as patient-facing had a signi cantly higher odds for SARS-CoV-2 test positivity suggests an increased risk of COVID-19 infection with work exposure. However, it is important to note in our study that that over 70% of the HCW group reported an exposure to COVID-19 with 28% reporting exposure to a family member with COVID-19. In our study, we were not able to con rm if the patient-facing HCW were working in patient-facing areas the 14-day period before the test was ordered when exposure could have occurred, or whether the exposure occurred with or without PPE -both in the workplace or in the community, or the relative contribution of initially prioritizing testing availability to HCW with reported exposures. While the risk to HCW attributed to community spread may not be captured in our available data, the reported exposure risk including the higher proportion of HCW vs. non-HCW reporting exposure to a family member with COVID-19 suggests a degree of community acquisition of infection. A potential contributing factor to community acquisition is that HCWs, particularly patient-facing HCW, are less able to follow stay-at-home guidelines or work remotely from home. Indeed, while PPE use is associated with decrease risk of infection from coronavirus, 26 a recent report estimated less than 5% risk to HCW inadvertently exposed to patients not known to be SARS-CoV-2-positive at the time of initial exposure with exposure likely occurring without appropriate PPE 27 suggesting that the work exposure risk may actually be low. However, universal pandemic precautions have been recommended for optimal risk mitigation for HCW. 28 

In our analysis of one healthcare system which implemented signi cant risk mitigation strategies to prevent the spread of COVID-19 infection, and which controlled for signi cant baseline differences in HCW compared to non-HCW, the odds for SARS-CoV-2 infection were similar for HCW and non-HCW and HCW had lower odds for COVID-19 related hospitalization .The patient facing HCW had higher odds of SARS-CoV-2 infection. 

Centers for Disease Control and Prevention

Clinical characteristics of 80 hospitalized frontline medical workers infected with COVID-19 in Wuhan

Characteristics of Health Care Personnel with COVID-19 -United States

Strong associations and moderate predictive value of early symptoms for SARS-CoV-2 test positivity among healthcare workers, the Netherlands

Characteristics of deaths amongst health workers in China during the outbreak of COVID-19 infection

Health Insurance Status and Risk Factors for Poor Outcomes With COVID-19 Among U.S. Health Care Workers: A Cross-sectional Study

Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China: Summary of a Report of 72 314 Cases From the Chinese Center for Disease Control and Prevention

Covid-19: The daunting experience of health workers in

Transmission of COVID-19 to Health Care Personnel During Exposures to a Hospitalized Patient

First experience of COVID-19 screening of health-care workers in England

Masks for Prevention of Respiratory Virus Infections, Including SARS-CoV-2, in Health Care and Community Settings

Risk of COVID-19 among frontline healthcare workers and the general community: a prospective cohort study

Association Between Universal Masking in a Health Care System and SARS-CoV-2 Positivity Among Health Care Workers

Network-based drug repurposing for novel coronavirus 2019-nCoV/SARS-CoV-2 | Cell Discovery. Accessed

Extracting and utilizing electronic health data from Epic for research

Research electronic data capture (REDCap)--a metadata-driven methodology and work ow process for providing translational research informatics support

Department of Health COVID-19 Outbreak

The State of Florida Issues COVID-19 Updates | Florida Department of Health

Addressing Extreme Propensity Scores via the Overlap Weights

Understanding Observational Treatment Comparisons in the Setting of Coronavirus Disease 2019 (COVID-19)

Individualizing risk prediction for positive COVID-19 testing: results from 11,672 patients

Loss of Taste and Smell as Distinguishing Symptoms of COVID-19

Evolution of Altered Sense of Smell or Taste in Patients With Mildly Symptomatic COVID-19. JAMA Otolaryngol--Head Neck Surg

Epidemiology of and Risk Factors for Coronavirus Infection in Health Care Workers

COVID-19 infections among HCWs exposed to a patient with a delayed diagnosis of COVID-19

Universal pandemic precautions-An idea ripe for the times

The authors report no con ict of interest related to this work.Dr. Misra-Hebert receives funding from the Agency for Healthcare Research and Quality grant # K08HS024128 and reports grants from NHLBI, grants from Novo Nordisk, Inc, grants from Merck Inc.,