STRENGTHS AND LIMITATIONS OF THIS STUDY

This longitudinal self-testing-based sentinel surveillance programme for respiratory diseases recruited participants via frontline workplaces that were chosen as sentinels for early detection of waves of disease.

The use of central swab drop boxes located at workplaces for weekly self-collected swabs improved ease of participation and retention in the programme.

Weekly testing of recruits regardless of symptoms allows detection of asymptomatic or pre-symptomatic COVID-19.

Participant recruitment was hampered by resourcing limitations and COVID-19 restrictions requiring case isolation that were in place at the time of the programme, affecting representativeness of the study population.

Testing for influenza and respiratory syncytial virus only in symptomatic participants may have reduced detection of these infections and the increasing use of rapid antigen test to self-diagnose COVID-19 during the programme delayed notification of these results, impacting timely detection of changes in incidence.

Introduction

Queensland, a state in Australia, was largely free of COVID-19 until late 2021. The Australian international borders were closed after an initial small peak of cases in March and April 2020, and the only cases in Queensland were linked to international or interstate travel or quarantine system failures.1–3 PCR testing was the primary method of identifying COVID-19-positive cases with testing readily available from private pathology providers and state-run COVID-19 respiratory clinics. People were able to get tested without a fee, with testing encouraged by the media, websites and where possible by direct SMS (short message service) messaging for people with even tenuous links to confirmed cases. Contact tracing for every case identified was comprehensive, and all close contacts were required to have evidence of a negative COVID-19 PCR result before leaving mandated quarantine. As a result, notifications of COVID-19 to Queensland’s Notifiable Conditions System (NoCS) through 2020 and up to late 2021 were close to an accurate representation of the COVID-19 infections present in Queensland.

However, in late 2021 and early 2022, several aspects of the COVID-19 response changed in Queensland as there were very high rates of two-dose COVID-19 vaccination (above 80%).4 Despite ongoing transmission in neighbouring states, on 13 December 2021, the state borders were opened allowing people to enter Queensland from interstate.4 This coincided with a transition of the dominant COVID-19 genotype from Delta to the more highly transmissible Omicron variant. With an unexpected surge in demand, lines at PCR testing locations rapidly grew and the turnaround time for receiving test results lengthened to up to a week. This is likely to have deterred many people from seeking testing. While rapid antigen tests (RATs) were approved for use in early January 2022, they were expensive and difficult to obtain. As a result, we suspected that the number of COVID-19 notifications in Queensland during this time was an underestimate.

In January 2022, during the first significant wave in Queensland, we developed a surveillance programme to estimate the prevalence of COVID-19 in the Gold Coast region of Queensland, by door-knocking randomly selected residents and asking them to do supervised self-collected PCR swabs.5 This programme exhibited some promise as a surveillance system but was expensive, resource-intensive and time-consuming.

Notifications of influenza had also reduced dramatically after Australian international borders were closed in early 2020, with some contribution from the non-pharmacological measures put in place to reduce transmission of COVID-19.6–8 After international arrivals were again permitted from early 2022, influenza notifications were expected to increase during the usual Queensland ‘influenza season’ in the southern hemisphere winter months.

In Queensland, respiratory syncytial virus (RSV) typically has a spike in cases during late autumn, before the increase in influenza cases.9–11 As RSV only became notifiable in Queensland in July 2021 (Public Health Act 2005 Public Health Amendment Regulation (No. 2) 2021), previous data were not complete. The winter of 2022 was to be the first time COVID-19, influenza and RSV would be circulating in the population together.

Sentinel surveillance is used to actively monitor disease trends in a population. Most sentinel surveillance for respiratory disease is syndromic reporting of acute respiratory illness. This could be individual instances of illness12 or longitudinal reporting by a cohort,13 or reporting of specific confirmed pathogens via opportunistic testing of people presenting to healthcare, usually with acute respiratory illnesses.14–17 Some recent COVID-19 surveillance programmes have included people seeking testing due to higher risk of infection, for example, being a close contact.18

A COVID-19 sentinel surveillance programme at Duke University in the USA used longitudinal testing of symptomatic and asymptomatic students and staff across the university.19 If someone tested positive for COVID-19, they were isolated and their close contacts quarantined. The authors cite this programme as the reason why the incidence of COVID-19 at the university was lower than in the surrounding area.19 A similar programme enrolling healthcare workers across the UK investigated reinfection rates by testing participants fortnightly but incorporated the results of non-programme tests via data linkage with the national notification system.20 21 A San Francisco area longitudinal surveillance programme for COVID-19 (TrackCOVID) recruited people via mail or telephone, and asked people to go to a clinic for testing (nasopharyngeal swab for PCR and venous blood for antibody testing) once a month,22 and a similar longitudinal study in Spain (ENE-COVID) recruited participants by telephone for point of care and antibody tests 2–4 weeks apart.23

This paper describes a novel longitudinal testing-based sentinel surveillance programme developed to estimate the incidence of notifiable respiratory diseases (COVID-19, influenza and RSV) during a time of reduced testing, in parts of Queensland over the southern winter 2022. The objective of the programme was to detect a winter wave of COVID-19, influenza or RSV before tertiary hospitals were impacted. The results of the surveillance programme are presented to show the potential utility of this type of programme in detecting respiratory disease in an environment of reduced testing rates.

MethodsDesign

This longitudinal testing-based sentinel surveillance programme based on the protocol by Ginige et al 24 was designed to detect winter respiratory viruses notifiable under the Public Health Act 2005 (Queensland).25 All participants were tested for SARS-CoV-2 and symptomatic participants were also tested for influenza A and B, and RSV. When participants tested positive for SARS-CoV-2, they were excluded from the rest of the programme but could re-enrol after 90 days according to the reinfection period in place at the time.26 Participants who tested positive for influenza or RSV remained in the programme.

Recruitment

Participants were recruited from staff of selected workplaces in three Queensland Hospital and Health Service (HHS) regions:

Gold Coast HHS, in the south of the state on the border with New South Wales (Australian Bureau of Statistics (ABS) 2020 estimated resident population (ERP): 650 996), recruited participants between 4 April 2022 and 17 July 2022 .

Metro South HHS, covering half the state capital of Brisbane (ABS 2020 ERP: 1 205 022), recruited participants between 4 April 2022 and 3 October 2022.

Central Queensland HHS, on the central coast of the state (ABS 2020 ERP: 220 865), recruited participants between 4 April 2022 and 26 June 2022.

A variety of workplaces expected to be at higher risk of exposure to SARS-CoV-2 (for example, healthcare or frontline emergency services) were contacted in each region and asked if they would be interested in allowing their employees to participate. These workplaces were selected because frontline workers (and similar) are expected to act as early warning sentinels due to higher risk of infection. Once this high-level consent was obtained, a link to an electronic registration form hosted on the Gold Coast HHS Citizen Space platform was sent to employees via broadcast emails or flyers. The registration form included a consent check box, demographic details (name, date of birth, sex, Aboriginal and/or Torres Strait Islander status), contact information for provision of communications relating to the programme (address, mobile number, email address), workplace details for logistics of sample collection, and pathogen-related information, including details of previous COVID-19, influenza or RSV positive tests and vaccination details for COVID-19 (ever vaccinated) and influenza (2022 seasonal vaccine).

Participants had to be 18 years or older and able to complete the English-language online survey. Potential participants were excluded if they had tested positive for COVID-19 in the previous 90 days (the Australian ‘reinfection period’ at the time of the programme26).

Household members were not excluded from participating, and the Metro South site actively encouraged their inclusion in the study.

On enrolment, participants were informed they could withdraw at any time and were provided with a link to a withdrawal form on Citizen Space. Upon withdrawal, they were given the option to allow the use or removal of their previous data for analysis. Participants also contacted the programme convenors directly via email to withdraw or suspend their involvement.

Participant retention was encouraged by allowing participants to skip weeks if needed (when absent from the workplace or forgetting to submit a swab) and to return to the programme when able.

A desired sample size of 380 participants for each region was calculated to enable detection of a change of 1% from a baseline positivity rate of 1% based on a 95% CI.

Due to difficulty recruiting participants, rolling recruitment continued throughout the programme.

Participant process

After registration, participants were provided with a pack delivered to their workplace containing:

Information sheets detailing the programme.

Pre-labelled nasal swabs for self-collection of specimens for SARS-CoV-2 testing with a sheet detailing the process for self-collection using the nasal swab.

Prefilled ‘asymptomatic’ pathology forms for submission with the swabs.

Prefilled ‘symptomatic’ pathology forms for presenting at COVID-19 respiratory clinics if unable to submit a self-collected swab to their workplace due to presence of symptoms consistent with COVID-19.

Once a week, asymptomatic participants were asked to swab themselves using the swab provided, place the swab and prefilled pathology form in the provided bag and deposit it in a secure box at their workplace. If participants developed symptoms anytime during the week (even if an ‘asymptomatic’ swab was previously provided via the standard process), participants were asked to attend a state-run COVID-19 fever or respiratory clinic in their area and present the prefilled ‘symptomatic’ pathology form. This pathology form included a request for the addition of a respiratory panel including SARS-CoV-2, influenza A and B, and RSV and allowed identification of their results as part of the programme. In some areas, symptomatic participants were also able to drop off their self-collected swabs at their usual collection point if done safely (without risk of infection to others). Participants often also notified us of COVID-19 RAT positive results or results of testing at private pathology laboratories via dedicated programme email addresses. Only positive PCR or RAT results notified in this way were included in analysis.

Communications

Participants were sent an SMS and sometimes an email reminder every week on the day before their scheduled swab day. Participants were also provided with a generic Queensland Health email address for their region for any queries or concerns, to obtain further information and/or withdraw from the programme.

All participants who tested positive for SARS-CoV-2 received notification of their result via SMS from the laboratory as part of the standard nationwide COVID-19 response. In addition, these participants were sent a follow-up SMS to inform them they could no longer participate in the programme. Although participants could re-enrol after 90 days, few participants met this criterion and there was no formal process to notify people of their eligibility. Participants who tested positive for influenza or RSV were sent an SMS to inform them of their result (and that they were not excluded from the programme) as at the time of the programme, there was no standard pathology provider process for informing a person of positive influenza and RSV results directly.

A summary of interim programme results was emailed to participants once a week.

Laboratory testing

Swabs collected from workplace secure drop-off sites were delivered directly to Pathology Queensland for testing. Swabs collected from fever clinics entered the laboratory process as usual.

All swabs were tested at Pathology Queensland for the presence of SARS-CoV-2 by PCR using the GeneXpert (XpertXpress SARS-CoV-2 assay), Hologic Panther (Aptima SARS-CoV-2 assay or Panther Fusion SARS-CoV-2 assay), Alinity m (SARS-CoV-2 AMP kit), cobas 6800 (cobas SARS-CoV-2 assay) or BGI (real-time fluorescent reverse transcription-PCR kit for detecting SARS-CoV-2 (two-gene) assay), depending on the regional laboratory.

Swabs with pathology forms marked as ‘symptomatic’ were also tested for influenza A and B, and RSV using GeneXpert (Xpert Xpress SARS-CoV-2/influenza/RSV assay) or Hologic Panther (Panther Fusion influenza A/B/RSV assay).

Programme test results were extracted for analysis from the laboratory database using the identification code on the pathology form.

Data analysis

All data cleaning and analyses were performed using Microsoft Excel and Stata/BE V.17.0 (StataCorp, College Station, Texas, USA). Participant demographics were analysed based on information provided in the Citizen Space enrolment survey. Participants who withdrew their consent for data inclusion were excluded from the final analysis. Missing data were coded as unknown.

Laboratory results were matched with participant demographics recorded at enrolment using a unique identifier. Results provided by participants (RAT or private pathology PCR) were manually entered into the results for the corresponding week.

The age distribution of the underlying population was based on 2020 estimated resident populations (ABS) for the three regions.

Households were identified by comparing the address recorded in the enrolment form and removing those where the address was clearly a workplace address entered in error.

Results were aggregated by week, and positivity was calculated using the number of tests positive for SARS-CoV-2 out of those reported that week. Response rate per week was assessed by comparing the number of swabs returned each week with the number of people enrolled at that time who had not reached the programme endpoint (ie, SARS-CoV-2 positive test or withdrawn).

PCR and RAT positive test results were exported from Queensland’s NoCS and aggregated by week of specimen collection based on the weeks used in the programme. Total numbers were presented rather than rates as the underlying population did not change during the short period of the programme.

Statistical analysis

To assess representativeness of our programme, positivity rates were compared with the notifications of COVID-19 (PCR and RAT) in the community using Spearman’s correlation (significance assessed at p<0.05). Spearman’s correlation was selected due to outliers in the data.

Patient and public involvement

Participants in the programme were provided with details of the programme before deciding to be involved in the study. Recruitment was voluntary. Participants were not involved in the design and conduct of the study, except for some authors who were also participants in the study.

ResultsDemographics

Across the three sites, 578 people enrolled in the programme:

42.9% of participants (248 people) recruited through the Gold Coast site (population is 31.3% of total study area ERP).

52.1% (301 people) recruited through the Metro South site (population is 58.0% of total study area ERP).

5% (29 people) recruited through the Central Queensland site (population is 10.6% of total study area ERP).

Participants were mostly female (77.7%) with a median age of 45 years (table 1), corresponding to the median age of the general community over 18 years population of the three areas combined. However, the age distribution of participants differed from that of the population of the community (figure 1). There were more programme participants in the 30–39, 40–49 and 50–59 years age groups than the general community the sample was recruited from, with other age groups under-represented. Most participants (79.6%) were employees of the state health department (table 1).

Table 1

Demographics of participants in the winter respiratory viruses sentinel surveillance programme

Figure 1

Proportion of population 18 years and over represented by each age group, compared with 2020 estimated resident population (Australian Bureau of Statistics).

Household recruitment was actively encouraged at one site and not prevented at the other two sites. There were 25 households enrolled (including one household twice, re-enrolling after the reinfection period passed). In 14 households, none of the participants tested positive, and in 7 households, only 1 person tested positive. Of the four households where two people tested positive, three tested positive the same week and one tested positive 3 weeks after their household member. With little evidence of household transmission within participants of the programme, and as each individual would have exposures outside the household, that is, workplaces, all remaining analysis was continued on an individual basis.

Incidence of respiratory viruses

Over 26 weeks, 4290 tests were reported to the programme with 140 positive for SARS-CoV-2 (3.3%). Only 79 (56.4%) were from swabs with pathology forms provided as part of our programme. Positive RAT results were reported to us by 48 participants (34.3%) and positive PCR results from a test at an alternative private pathology provider were reported by 13 participants (9.3%). Of the 113 people who presented to a COVID-19 respiratory clinic with the ‘symptomatic’ pathology form or who reported influenza or RSV results from other pathology providers to us, 3 (2.7%) were positive for influenza A and 1 (0.9%) was positive for RSV.

The number of COVID-19 test results (including RAT and PCR from alternative providers) reported each week varied, with a maximum of 272 results in week 6 (figure 2). Similarly, the percentage of people who tested out of all those enrolled in any given week (not including those excluded, not yet enrolled, complete (ie, previously tested positive for COVID-19) or withdrawn in that week) varied from 25.9% in week 25 to 72.1% in week 8 (figure 2). The highest number of cases of COVID-19 detected was 16 in week 6, week 13 and week 15. Despite choosing to enrol, 56 participants never submitted a swab.

Figure 2

Number of participants enrolled each week, by COVID-19 test result and proportion of participants tested. The percentage of those enrolled who tested each week is shown in the black line on the secondary axis. *One site ended participation in week 12, and a second site in week 15.

There was a strong, statistically significant positive correlation between per cent positivity of participants in the programme and notifications of COVID-19 in the community for all sites combined (Spearman’s rs=0.6500, p=0.0003; figure 3).

Figure 3

Percentage of participants who tested positive for COVID-19 (lines) and NoCS notifications of COVID-19 (PCR and self-reported RATs—bars) in the population of the region. Correlation is significant (Spearman’s rs=0.6500, p=0.0003). NoCS, Notifiable Conditions System; RAT, rapid antigen test.

Discussion

This sentinel surveillance programme demonstrated the utility of a crowd-sourced testing programme in estimating disease incidence. While there were some limitations in the design and setting, learning from this programme could be used to design similar sentinel surveillance systems in the future for COVID-19 and other respiratory pathogens or even other diseases that are easily self-tested.

Negative public sentiment and remaining COVID-19 isolation rules for cases in place at the time of the programme are likely to have hampered recruitment. There were lower levels of engagement than were hoped with both workplaces and staff (potential participants) as there was a mandatory 7 days of isolation required for those who tested positive for COVID-19, even without symptoms. With difficulties in staff availability in many industries during the pandemic response, some workplaces were not willing to risk losing asymptomatic staff to isolation that would not have been required if they had not been tested. Similarly, even when workplaces agreed to participate, employees were concerned about having to isolate. After the government-mandated restrictions of the previous few years, people were fatigued with government programmes. As a result of these factors, none of the sites achieved the desired sample size.

The reluctance of workplaces and potential participants to be involved in this programme led to our programme being underpowered, and our selection of workplaces based on risk (ie, frontline) resulted in our cohort not being representative of the underlying population. Compared with the underlying population of the study areas, we had more women, people in their middle age and people who worked in healthcare. These factors may have biased the positivity rates detected in the programme. Many healthcare workers are at higher risk of becoming infected due to their proximity to ill people, but at the time of our programme, they were still required to wear masks while at work and were highly vaccinated, reducing the risk of COVID-19 spread. Representativeness of the underlying population could be improved by recruiting from a wider range of workplaces or from the general community, but this approach was not feasible with the time and resources available and the use of self-collected PCR swabs rather than RAT. We chose to focus our recruitment resources on population pockets that act as sentinels to detect waves of disease earlier (frontline workers and others in public-facing roles), strengthening the ability of this surveillance system to achieve the aim of detecting waves of disease prior to impacts on hospitalisation.

Despite being underpowered and not representative of the general community, positivity rates in our programme correlated with notifications in the community and show the start of the third Omicron wave in Queensland in July 2022. However, this wave was only evident in retrospect, largely due to the use of RATs. RAT results were often reported to us only after people returned to work, sometimes weeks later, delaying their inclusion in the data. While many participants self-reported their positive RAT results to us, this was not a planned part of the programme, and it is unclear how many positive RAT results were not reported. Similarly, very few negative RAT results were reported impacting the accuracy of the denominator used in positivity calculations as people with negative RAT results who failed to report may not have submitted a swab that week.

RATs were often used by symptomatic people or close contacts of COVID-19 cases. The process for testing symptomatic people in our programme was to take our pre-populated pathology form to a nearby fever clinic. However, when our programme was operating, fever clinics had reduced hours or were beginning to close, making it more difficult for participants to be tested. In addition, people are less likely to want to leave their home to be tested when they are ill with respiratory symptoms. RATs are an attractive and easy alternative and became widely available during our programme. Incorporating an easy method for reporting of RAT results, whether positive or negative, such as an online form or portal, would have improved timeliness and completeness of reporting of RATs in this programme. Due to the lower sensitivity of RATs than PCR,27–29 symptomatic people who test negative for COVID-19 on RAT should still have been encouraged to get a multiplex PCR at a fever clinic to improve detection of COVID-19, influenza and RSV.

Despite high rates of both influenza and RSV in the community at the time of this programme, few cases of either disease were detected. Only 113 people presented to a fever clinic and were tested for influenza and RSV, as only those who presented to a fever clinic with our specific pathology form were tested for these diseases. The number of people with respiratory symptoms who had a negative COVID-19 RAT and did not proceed to get a test for influenza or RSV remains unknown.

Self-testing may have impacted the positivity rate if participants had performed their swabbing incorrectly. To mitigate this, we provided participants with detailed pictorial instructions on the swabbing process. Previous studies have shown that while self-collected nasal swabs are not as accurate as clinician-collected swabs, the convenience outweighs the reduced sensitivity.30–33

While this programme required fewer staffing resources than the previous door-knocking prevalence programme conducted in Gold Coast,24 significant staff time was required for the daily messaging, creation of personalised swab packs, dropping packs to workplaces, picking up swabs from workplaces and delivery to the laboratory, in addition to the data analysis and additional pressure on the laboratory staff. Funding was also required for the consumables used during the programme. However, implementing RAT as the primary method of testing for asymptomatic participants, despite poorer sensitivity, will reduce this burden and allow this programme to be used for much larger numbers of participants, including those in the community. Providing RAT kits instead of PCR swabs will remove the requirement for personalised packs with pathology forms and swabs, and there will be no need to provide a secure location for swab drop-off or to pick up swabs to deliver to the laboratory.

With management of these identified limitations, our programme shows promise in a setting where COVID-19 is normalised and testing of people with respiratory symptoms is less common. This longitudinal sentinel testing programme not only detects mild disease in people who might not otherwise have tested, it can also detect asymptomatic infections. With an appropriately representative selection of participants, incidence of respiratory diseases such as COVID-19 and influenza, in surveillance such as this, could be extrapolated to the whole population. This would allow guidance of public health action with education and communication strategies to reduce the incidence and impact of these diseases.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplemental information. All deidentified aggregate data are available in this paper. No additional data are available.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This study involves human participants. Gold Coast Hospital and Health Service Human Research Ethics Committee granted an exemption (EX/2022/QGC/84722) as surveillance is considered part of routine Public Health Unit work. Participants consented to participate by checking a box in the registration form.

Acknowledgments

The authors would like to acknowledge the invaluable assistance of the staff at Pathology Queensland for facilitating the testing of our swabs, in particular Emma Jones-Perrin, Joel Douglas and Matt Ford. This programme would not have been possible without the help of many staff members at each Public Health Unit (PHU). The Gold Coast PHU would particularly like to acknowledge Jennifer Doust for logistics; Christobel Mak, Abraham Duncan, Shane Wilson, Damien Smith and Jacqueline Pittaway for data management; Jayden See, Barbara Kahi and Danika Delos Santos for other essential assistance; and Ian Hughes for biostatistical advice. Metro South PHU would like to acknowledge Emma Parker and Courtney Swindells for logistics; Jasmine Royds, Jarrah Gibney and Harprit Kaur for other assistance. Central Queensland PHU would like to acknowledge Amanda Wyatt, Margaret Charles, Suzie Le Brasse, Danielle Odorico and Renarta Whitcombe for their ongoing assistance throughout the programme.