To ensure that scientific research of immediate concern is shared as rapidly as possible, we fast-track accepted articles from the IJTLD and PHA and publish them as edited preprints prior to publication in a journal.
All content in PHA is Open Access and free to read. PHA covers all areas of operational research including: infection control, nutrition, TB, HIV, vaccines, smoking, COVID-19, microbial resistance or disease outbreaks. Find out more about PHA or sign-up to receive the Table of Contents
M. Paleker, Y. A Tembo, M.-A. Davies, H. Mahomed, D. Pienaar, S. A Madhi, K. McCarthy
Article submitted 6 November 2020. Final version accepted 20 March 2021.
Asymptomatic COVID-19 may contribute to the pandemic trajectory. The positivity rates of COVID-19 seroprevalence studies are higher than the proportions with a history of clinical symptoms of COVID-19, suggesting a high proportion of asymptomatic infections. In Spain, COVID-19 seroprevalence was 5%, with one third of participants being asymptomatic. The proportion of asymptomatic COVID-19 in adults can reach 56%, but data from sub-Saharan Africa are limited. Globally, as in South Africa, case definitions and testing strategies to identify COVID-19 cases has changed with evolving evidence, leading to the inclusion of testing of asymptomatic contacts when resources and capacity permits.
In the present study, we aimed to describe the proportion of asymptomatic COVID-19 identified through outbreak investigations among essential workplace clusters in South Africa from 8 March until 18 May 2020, which includes the time of the highly restrictive lockdown.
Following SARS-CoV-2 importation into South Africa, a national state of disaster was declared, with a nationwide lockdown restricting movement and limiting business operations to essential services only. These included the healthcare, law enforcement, and correctional services, food and beverage producers, and essential grocery retail.
Outbreak investigations were carried out nationwide following identification of confirmed COVID-19 cases in essential services through passive surveillance of persons with COVID-19 symptoms. Contacts were defined as anyone in the workplace within 1 metre of a confirmed case for longer than 15 minutes, or those who shared a workspace (including break rooms or work-related transport) with a confirmed case. RT-PCR tests were conducted on nasopharyngeal swabs of all workplace contacts regardless of symptoms, and the presence of symptoms was recorded.
Identification of subjects included in the analysis
This was a retrospective analysis including ‘all SARS-CoV-2 PCR-positive cases identified during investigation of outbreaks at essential services that occurred during lockdown’. A subset of cases were followed up approximately 2 weeks after the first test. The presence of symptoms were documented and repeat nasopharyngeal swabs were taken. De-identified data of cases (age, sex, essential service type, initial test date, follow-up date, COVID-19 symptoms at initial and follow-up tests) were extracted from outbreak investigations across the country.
Data analysis was performed using R v3.6.2 (R Computing, Vienna, Austria). We described the proportion of asymptomatic COVID-19 cases initially and at follow-up, and used logistic regression to assess the association between having symptoms at the time of first diagnosis with age, sex and industry sector.
Informed consent was waived for this secondary analysis of routine de-identified data collected for the COVID-19 outbreak response. Ethics approval was obtained from the Human Research Ethics Committee at Stellenbosch University, Cape Town (N20/05/019_COVID-19) and the University of Cape Town, Cape Town, South Africa (460/2020).
Of 887 confirmed COVID-19 cases (75% female) identified at 13 essential workplaces from 8 March to 18 May 2020, the median age was 37 years (interquartile range [IQR] 29–48; Table 1). The proportion of asymptomatic COVID-19 cases at time of first testing was 43.7%. Men were less likely to be asymptomatic than women (adjusted odds ratio [aOR] 0.69, 95% confidence interval [CI] 0.47–1.00), and those in correctional services (aOR 47.27, 95% CI 16.80–197.98) and the pharmaceutical industry (aOR 2.03; 95% CI 1.21–3.40) were more likely to be asymptomatic than those from other sectors (Table 2). A total of 112 cases were followed up for a median of 16 days (IQR 10–20). Of the 52 who were asymptomatic, 17.3% subsequently developed symptoms; hence, 36.6% were asymptomatic throughout follow-up.
Among SARS-CoV-2-positive cases identified during outbreak investigations shortly after SARS-CoV-2 importation into South Africa, we identified a high proportion of asymptomatic COVID-19 cases in essential workplace clusters. Of the cases followed up, 36.6% remained asymptomatic. Our findings have implications for the implementation of interventions designed to control the epidemic, case identification and testing, and the force of infection generally and that prevalent in South Africa at the time. There is biological, epidemiological and modelling evidence for transmission of SARS-CoV-2 from asymptomatic people. Whereas vaccines are the most sustainable method for controlling the epidemic, the reality is that in sub-Saharan Africa, vaccines are unlikely to be widely available in the immediate future. The high proportion of asymptomatic COVID-19 cases in our study indicates that adherence to interventions such as case isolation and quarantine of close contacts, should continue to be pursued as an effective public health intervention to control SARS-CoV-2 transmission.
In resource-limited countries, careful consideration must be made regarding testing strategies. In our study, testing of asymptomatic contacts allowed identification of a substantial proportion of persons who did not meet symptom-based testing criteria. Therefore, in addition to testing symptomatic persons for SARS-CoV-2, consideration should be given to testing of asymptomatic close contacts to limit transmission. The high asymptomatic proportion of cases has epidemiological and public health implications. First, estimates of the number of infections and attack rates are likely an underrepresentation, especially if case detection relies on passive surveillance and testing of symptomatic persons. Second, high numbers of asymptomatic infections will increase the rate at which susceptible persons acquire infection (the force of infection). In our context, the force of infection during this time may have contributed to the failure of lockdown to successfully contain COVID-19. Our data are helpful for modelling studies to support pandemic response activities, which take into account the reproductive number and the proportion of infections that are asymptomatic and/or that can be transmitted while asymptomatic. In our setting, we observed variations in the proportion of asymptomatic COVID-19 within the essential service sector. Although those in correctional services had a much greater odds of being asymptomatic (aOR 47.27, 95% CI 16.80–197.98) than other sectors, the importance of this is unclear.
A source of variation and a limitation of our study lies in the subjective nature of symptoms, recall bias or contextual incentives to over or underreport symptoms, such as fear of being ostracised. Additionally, we do not to compare cycle thresholds of RT-PCR results, as these tests were done across different laboratories using different test methodologies.
The high proportion of asymptomatic cases of COVID-19 indicates the need for strict quarantine of all contacts, with or without SARS-CoV-2 testing of asymptomatic cases, in order to prevent and contain outbreaks.