Study population and setting
Community members and frontline healthcare workers who were prospectively recruited via social media and direct invitations through existing cohort studies, were followed longitudinally and reported on whether they received a positive SARS-CoV-2 test. Data was reported through the COVID Symptom Study smartphone application and were collected between March 24 and April 23, 2020 from the United Kingdom and the United States. Data on demographic information, comorbidities, symptoms, and SARS-CoV-2 testing were captured through the app by self-report. For healthcare workers, defined as those with direct patient contact, additional information was collected about patient interactions and access to personal protective equipment. Cox proportional hazards were used to model adjusted hazard ratios of a positive SARS-CoV-2 test. Secondary analyses were conducted to account for likelihood of receiving a test. Subgroup analyses among healthcare workers also examined factors related to PPE availability.
Summary of Main Findings
A total of 2,810,103 individuals were enrolled– 670,298 individuals were excluded who had follow-up time less than 24 hours, and 4615 who previously tested positive at enrollment. Among the remaining 99,795 frontline healthcare workers and 2,035,395 community members, a total of 34,435,272 person-days were included; 5545 incident reports of a positive SARS-CoV-2 test were reported. Incidence was 48.2 per 100 person-years among healthcare workers and 4.01 per 100 person-years among the general community. Healthcare workers had an adjusted hazard ratio of 11.6 (95% CI 10.9-12.3) for reporting a positive test compared with the general community population. Using methods to adjust for the likelihood of receiving a test, healthcare workers remained at significantly greater risk of a positive test (aHR=3.40, 95% CI 3.37-3.43). In the subgroup analyses looking among healthcare workers, those with adequate personal protective equipment (PPE) had lower risk than those reporting reuse of, or inadequate PPE.
The use of the novel mobile application to collect data allowed for the prospective collection of data on over two million community individuals and close to 100,000 healthcare workers. The study utilized two different approaches to account for higher likelihood of testing among healthcare workers, and the significantly higher risk among healthcare workers were robust to these adjustments.
Recruitment for this study was not random, and may not capture certain underrepresented groups, including older adults. There is also the potential for emigrative selection bias in this study. A large proportion of those originally enrolled did not have any follow-up time greater than 24 hours. Participants were followed until they reported a positive test or the time of last data entry. This assumes that those who stopped providing data were represented by those who continued to provide data; incidence may be overestimated if those less likely to test positive were less likely to engage in follow-up.
This study directly compares frontline healthcare workers and community individuals to understand differences in risk and the role of personal protective equipment.
This review was posted on: 25 June 2020