Respiratory virus surveillance in Australia: past, present and future

Over the years, surveillance for respiratory viruses has developed from mainly being based on influenza case notifications (since 2001) to a more sophisticated system that allows for a more comprehensive picture of the burden, severity and impact of several respiratory viruses.

Prior to 2000, the National Influenza Surveillance System was largely based on influenza case notifications, initially in a laboratory system from 1978, and after 2001, nationally notifiable. As influenza diagnostics were infrequently used, the numbers of notifications was low, with only 2943 notifications nationally in 1998¹ (compared to 251,095 in 2023). Some indirect information was supplied from sentinel general practitioners reporting rates of influenza like illness, and absenteeism reported at a major national employer.¹

During the 2000s, and particularly after the H1N1 swine flu pandemic in 2009, a broader range of surveillance systems were developed and now provide a multidimensional assessment of influenza activity.² These systems have expanded in scope to include COVID-19 and other important viral pathogens.

At a community level, FluTracking, a community-based surveillance system that surveys participants weekly, provides indicators of community activity of influenza-like-illness through weekly surveys of a volunteer cohort since 2006. These are primarily based on symptoms, although questions are also asked about medical attendance, and COVID-19 and influenza testing. Data from FluTracking suggest that in 2019, 29% of participants reported fever and cough at least once during the season; during period of high respiratory viral activity, up to 3–4% of respondents have fever and cough during weekly reports with approximately one-third taking time off work and approximately one-quarter seeking medical attention.³

However, there are likely to be selection biases that are difficult to quantify; for example, in 2017, 84% of FluTracking respondents over 65 years reported being vaccinated (higher than most estimates of population coverage) and 62% reported a university bachelor degree or above (compared to ~30% in the population).⁴

The widespread use of sensitive and specific respiratory virus nucleic acid amplification assays has likely improved the ascertainment fraction. In the community, the proportion of FluTracking participants who reported being tested for influenza was 7.6%, with 56% of those testing positive.³ In hospitals, the proportion of influenza diagnoses in adults recorded as being virologically confirmed increased from ~20–35% to over 80% between 2006 and 2013,⁵ and is likely to be even higher since the widespread use of nucleic acid amplification assays during the COVID-19 pandemic.

At a primary care level, the Australian Sentinel Practices Research Network (ASPREN) has reported on influenza-like illness presentations across more than 200 general practices nationally since 1991.⁶ Although the general practitioners participating in ASPREN are likely to be representative of presentations to primary care,^6,7 and testing is systematic, there are likely to be biases that patients present to general practitioners. In this system, the criteria for patient testing and the multiplex assays used are standardised allowing surveillance for a broad range of respiratory pathogens.

At a hospital level, the Influenza Complications Alert Network (FluCAN), in collaboration with the Paediatric Active Enhanced Disease Surveillance (PAEDS) system, provides surveillance data on hospitalisations with confirmed influenza (since 2009), COVID-19 (since 2020) and more recently respiratory syncytial virus (RSV).⁸ The Short Period Incidence Study of Severe Acute Respiratory Infection (SPRINT-SARI) also provides a similar function for patients in the Intensive care unit (ICU).⁹ An important aspect of both systems is as an indicator of severity; a true infection/fatality ratio is not generally possible to estimate, and population level indicators such as excess mortality are difficult to estimate in real time. In addition to changes in hospitalisation and ICU admission numbers with confirmed infection, the proportion admitted to ICU, the length of hospital stay and in-hospital mortality provide information of severity and the impact on the hospital system.

Virological surveillance has been well established and co-ordinated by the Global Influenza Surveillance and Response System (GISRS) since the early 1950s for influenza, and is rapidly maturing for other respiratory pathogens.¹⁰ Many countries collect influenza strains through National Influenza Centres (including three in Australia), and these are further characterised at the seven World Health Organization (WHO) Collaborating Centres. Rapid advances in genomics have permitted timely analysis of changes in circulating strains.¹¹ However, there will be a continuing need for phenotypic assays to provide information on vaccine match and potentially antiviral resistance. Virological surveillance is particularly important for SARS-CoV-2, as the appearance of new variants appears to drive increases in COVID-19 incidence.

Additional information on a number of respiratory pathogens may be available from novel sources of surveillance, such as the quantification and characterisation of viral fragments in wastewater.¹² However, further work is required to understand the interpretation and limitations of these data. In the past, other novel sources of data such as search terms correlated with influenza activity were not found to be as reliable as initially thought.¹³

Although surveillance systems have generally focused on disease incidence (at varying levels of severity), there is a need to consider a broader and more dynamic vision for surveillance. The WHO has developed an approach to further expand surveillance termed the ‘mosaic surveillance framework’.¹⁴ It calls for multiple fit-for-purpose systems that aim to detect and assess emerging respiratory viruses, monitor the epidemiological characteristics of respiratory viruses in inter-pandemic periods, and to inform the use of human health interventions.

To evaluate immunisation programs, the key information needs include vaccine coverage and vaccine effectiveness. Although estimates of coverage in the past have relied on telephone based surveys,¹⁵ coverage can now be accurately estimated from notifications to the Australian Immunisation Registry (AIR).

Surveillance systems can now provide useful information of effectiveness of immunisation, which is expected to vary from year to year due to a complex interaction between the match of vaccine strains to circulating strains and incompletely understood immunological processes. Estimations of vaccine effectiveness against medical presentations were pioneered in primary care surveillance. In this variant of a case control study, case and non-case status is assigned once test results are known, reducing (but not eliminating) the potential for testing bias.¹⁶ FluCAN similarly is able to estimate vaccine effectiveness against hospitalisation using a more conventional case control design.¹⁷

There is enormous potential in using linked administrative databases for surveillance. These systems are still being developed and their use is subject to complex governance arrangements, but population-based vaccination data linked to other government data allows a comprehensive picture of gaps in vaccination coverage.¹⁸ To estimate vaccine effectiveness, large administrative datasets have enormous statistical power and can examine questions that cannot be addressed in sentinel surveillance systems, such as the relative effectiveness of different vaccine types,¹⁹ there are also methodological limitations in using administrative data that also need to be considered.²⁰ Issues include the accuracy of coded hospitalisation data, difficulty in defining comorbidities and ascertainment of immunisation status.

Ultimately, a broader surveillance framework might fulfil information needs to optimise public health control of seasonal respiratory infections, while being flexible to collate information on future pandemic pathogens. For newly emerged pathogens, these information needs vary with different phases of emergence.²¹ Information is required on biological transmission parameters (e.g. incubation period, basic reproductive rate, generation interval), the true incidence of infection, and social and personal behaviours (e.g. mask use, physical distancing, work from home and mobility data, social attitudes). Collectively, these would provide a comprehensive picture of current epidemiology (‘now-casting’) and facilitate at least short-term forecasts of disease incidence to inform public health policy to mitigate the impacts of seasonal and pandemic viral infections.