Advancing evidence to enable optimal communicable disease control

Background

In principle, evidence-based decision-making relies on the use of data, empirical evidence and analytical approaches rather than intuition or past experiences alone to inform real-world policy decisions.¹ The volume of papers published during and since the emergency phase of the COVID-19 pandemic demonstrates how epidemiological data were used and their influence on the public health response, but these papers also expose data gaps and a limited capability in evidence synthesis that impeded real-time evidence-based decision-making.²

From the beginning of the pandemic, data inconsistencies contributed to slow or poor pandemic policy decisions. For example, China’s very narrow case definition for novel coronavirus infections and deaths masked the severity of the threat.³ This affected early risk assessments made by the World Health Organization (WHO) and other Member States, thus leading to delays in public health response decisions that undermined their effectiveness.⁴

Case counts, using agreed definitions, and descriptive studies of outbreaks provide an initial indication of the size of the problem and transmission dynamics. These data are critical for supporting early decision-making. However, in the pandemic they only provided a limited snapshot of the problem, and more in-depth analytic epidemiological studies were needed to fully characterise risk within and across populations. The paucity of epidemiological studies during the emergency response phase limited our understanding of differential risks and impacts within populations, the effectiveness of interventions and the changing epidemiology of SARS-CoV-2.

A key aspect of analytic epidemiology is the use of epidemiological data to examine risks in specific population groups. One way to enable this during an epidemic or pandemic response is to ensure that data linkage platforms are ready and enabled for simple case count data to be linked to demographic, socioeconomic and other health data to quickly determine where a pathogen is circulating in the community and who is most at risk.^5,6 Linked data are also critical for determining if the balance of risks at a given time and place supports interventions being introduced, scaled up or de-escalated.

Well-designed analytical studies such as case-control studies, cohort studies and emulated target trials can provide higher-resolution identification of specific at-risk subpopulation groups in relation to infection rates, disease severity, intervention impacts and treatment outcomes. Importantly, they also have the potential to assess prevention and health protection measures in real time. This would provide the granularity of evidence required to meet the needs of policymakers and public health communication.⁷

Real-world evidence challenges for decision-making

Case data for infections are nearly always biased, influenced by methods of detection and testing accessibility that vary within and between populations and over time, especially in emergency settings.⁸ Without a consistent approach to setting case definitions, testing and reporting protocols, the data and any trends derived from these will be influenced by case ascertainment. Other population-level time-varying factors can also influence the data and outcomes, even where public health policy is universal.

Following the detection of an epidemic, it is important to rapidly determine the at-risk populations that need to be prioritised as the response is scaled up. A comparative study of programmatic tools and policy measures used during the COVID-19 pandemic in 15 countries across WHO regions highlighted the vulnerabilities of many health systems in responding effectively to rapidly rising caseloads in settings with existing health inequities.⁹

Early experiences from China and Italy during the pandemic demonstrated an increased risk of hospitalisation and death in the elderly.¹⁰ However, the significant delays in more granular assessments of disease risk and distribution across populations resulted in an inequitable response. Failure to fully understand the relationships between infection risk and occupation, social mixing patterns and access to culturally appropriate health services, for example, left some communities disproportionately affected. Singapore experienced their highest infection rates in foreign worker dormitories, comprising 88% of 17,758 confirmed cases by May 2020.¹¹

In Australia, the overseas-born population was not recognised as a particularly at-risk population until late in the vaccine rollout when culturally and linguistically diverse communities in low socioeconomic status local government areas were prioritised in vaccination outreach programs. This was too late to avert higher COVID-19 death rates in the overseas-born population which was 2.5 times higher than the rest of the population in Australia by 2022.¹² Disproportionately high hospitalisation and death rates should not be used as a trigger to reprioritise public health investment. Real-time epidemiological and behavioural science studies carried out with potentially vulnerable communities during the early stages of a pandemic can provide a more comprehensive understanding of risk and the potential need for targeted mitigating measures. Operational research platforms developed during pandemic preparedness phases can enable data generation.

Good-quality public health data that are reliable, consistent and sufficiently granular to inform nuanced decision-making and policies are central to a speedy and effective pandemic response. Importantly, high-quality data also improve the robustness of modelling estimates and build confidence among decision-makers who are crafting policy based on modelled estimates.¹³

Although efforts have been made to systematically coordinate mathematical modelling consortiums across the world (e.g. the Doherty-led modelling consortium;¹⁴ the Lancet Commission on Strengthening the Use of Epidemiological Modelling of Emerging and Pandemic Infectious Diseases;¹⁵ the Modelling and Simulation Hub, Africa (MASHA);¹⁶ the National Disease Modelling Consortium (NDMC) in India¹⁷), limited efforts were observed post-pandemic to improve and coordinate surveillance data and epidemiological data globally (Australia-Aotearoa Consortium for Epidemic Forecasting & Analytics¹⁸). Despite lessons from the pandemic, efforts for ongoing surveillance to provide real-time data remain largely invisible and under-resourced.

Pandemic Agreement

The role of data and evidence came into focus at the global level through the drafting of revisions to the International Health Regulations (IHR).¹⁹ Pandemic monitoring, prevention and response require global coordination. The IHR provides an overarching legal framework that defines member states’ rights and obligations in managing public health events and emergencies. The 2024 draft of the IHR includes the requirement to establish and maintain core capacities for surveillance and response, with amendments approved by the World Health Assembly in June 2024.²⁰

Alongside revision to the IHR, the world’s first Pandemic Agreement for countries to work together cooperatively to protect against future pandemics was finalised in April 2025 and approved for adoption at the May 2025 78th meeting of the World Health Assembly.²¹

The original 2023 draft Agreement specifically mentioned evaluation in relation to preparedness monitoring and functional reviews; ‘The Parties shall, building on existing tools, develop and implement an inclusive, transparent, effective and efficient pandemic prevention, preparedness and response monitoring and evaluation system’ (Article 8, paragraph 3).²² In Article 10, paragraph 2, the draft included ‘Each Party shall initiate or strengthen, as appropriate, the conduct of disease burden studies relevant to pathogens with pandemic potential, with a view to ensuring the sustainability of investments in facilities for the production of vaccines and therapeutics that could support pandemic response’. However, the updated text for Article 10 in the May 2024 version only focused on the global production of pandemic-related health products.²³

The Agreement relies on countries having monitoring systems to undertake epidemiological investigations, including disease burden and vaccine effectiveness assessments. This capability is not universal, and therefore global coordination of the procurement and synthesis of evidence across settings is required and must be a priority function of coordinating institutions such as the WHO.

An example of how data collection and analysis capability limited the use of evidence in the COVID-19 pandemic was the inability of most countries to deliver the recommended practice of completing ‘first 100 days’ studies (see, for example, Cambodia 100 day report).²⁴ Few countries were able to report critical details of the first cohort of cases because of the lack of resources and data intelligence system infrastructure. Australia did establish some important early cohort studies. For example, the first 100 COVID-19 cases across jurisdictions were recruited and followed for 14 days in 2020 to help understand the severity and household transmission dynamics in the early stages of the pandemic.²⁵

Only a handful of low- and middle-income countries conducted vaccine effectiveness studies for COVID-19.²⁶ For routinely administered vaccines, data on vaccine effectiveness are generated by high-income countries, highlighting the gaps that also exist in routine data systems. Impact studies are more often conducted as formal research rather than routine public health evaluations and are therefore more challenging to mobilise in pandemic settings.

Infrastructure for evidence generation

As highlighted within the WHO health emergency preparedness, prevention, response and resilience framework, health systems are at the core of what is required, and this includes data systems.²⁷ Investment in the capacity to evaluate effectiveness and safety, or unintended consequences, in the real world is particularly important in pandemic settings where there is rapid deployment of measures, especially for vaccines. Formal vaccine trials quickly become impossible as ‘control’ or unvaccinated populations rapidly diminish. This is where the power of good-quality epidemiological, real-world data comes in – to evaluate interventions in real time in relevant local populations, and to provide essential intelligence to respond to false claims about vaccine effectiveness and safety.²⁸

The effort put into data collection, synthesis and reporting by the United Kingdom was relied on globally as a trustworthy real-world evidence source.²⁹ A real-world data platform established in Scotland in the first half of 2020 is another example of enabling data infrastructure. In this case, the platform links individual-level data on vaccination, testing, viral sequencing, primary care, hospital admissions and mortality, and it was used to assess local vaccine effectiveness.^30,31

There are other exemplars from around the world where real-world evidence was successfully used to support decision-making that optimised vaccine dosing regimens, assessed safety and provided guidance on priority populations.³² These case studies were showcased in a meeting in September 2024 on ‘The Role of Real-World Evidence for Regulatory and Public Health Decision Making for Accelerated Vaccine Deployment’ as specific scenarios where real-world effectiveness is both viable and beneficial, especially in emergency contexts. The meeting identified the essential conditions needed to achieve this, including epidemiological and regulatory aspects, robust scientific support, regulatory collaboration and a paradigm shift towards public–private partnerships in vaccine development and production.³²

The inter-pandemic phase offers an opportunity to identify and strengthen data sources and establish systems to support the highest-quality real-time epidemiological and modelling studies. As part of pandemic readiness, a broader view of evidence infrastructure is required, including pre-drafted data governance and sharing protocols adaptable to changing risk settings, privacy agreements and expedited ethical clearance processes at the ready.³³

If pre-planned data linkage and sharing can be stood up quickly, then prospective empirical studies can be conducted early and can inform public health responses. Only then can we deliver pandemic responses that minimise or address differential impacts and effectiveness.

The Pandemic Agreement acknowledges the importance of real-time analysis through recognition of the need for research and development during a pandemic – but the Agreeement focuses solely on genomics and clinical trial readiness and not on applied public health research. Failure to acknowledge the value of real-world evidence generation risks effective pandemic responses and trust in policymaking.

Real-time and real-world effectiveness and evaluation studies of public health interventions can only be fully tested in the actual pandemic setting. Early COVID-19 vaccine trials were conducted in countries with high rates of community exposure, yet the direct relevance to countries that prevented community-wide transmission was uncertain given the very low rates of natural immunity in these populations. Where evidence from other populations is relied on, local public health systems should be in place to enable testing of these interventions locally and as soon as possible to ensure they work as intended, or to adjust the response to local conditions as soon as possible.

Ahead of the next pandemic, there is a clear need to build capacity in a coordinated approach to applied public health research that encapsulates epidemiology, social and behavioural sciences, risk assessment and communication, and clinical practice. Epidemic and pandemic responses require anticipatory action and application of the precautionary principle to act before evidence is available. However, at the same time, there is also a mandate for real-time data collection and analysis to support the evaluation and refinement of measures in place. Local, national and global cooperation is critical for this when resources are patchy and stretched. A coordinated approach can generate a few high-quality studies that are generalisable rather than many weaker studies, so as to minimise resource wastage.

One of the key lessons from the COVID-19 response globally was the uncertainty about when to step down from the use of public health orders.³³ It has been argued that better use of key parameters, agreed a priori, could provide objective guidance on when to end measures such as lockdowns.³⁴ Knowing the key points for disengagement is critical and will be clearer with careful monitoring and real-world real-time data analysis. Communication of good-quality real-world evidence is also essential for building and maintaining public confidence and trust in public health decisions during uncertain times.³³

The central role data should play in pandemics has been highlighted in reviews of the pandemic response.³⁵ Although the International Pandemic Agreement adopted at the 78th World Health Assembly (May 2025)²¹ does not detail disease surveillance capability requirements, it does nonetheless rely on public health responses being underpinned by reliable surveillance data. Enabling and evaluating nuanced policy and targeted disease control measures in pandemic settings requires the right data infrastructure to be in place ahead of the crisis, and the corresponding synthesis and reporting capability.

The Australian story

After an initial bumpy start and challenges with data coordination across the jurisdictions, Australia made some significant strides in understanding local COVID-19 transmission.³³ Although Australia did conduct a multi-jurisdiction ‘first 100’ study examining the transmission of COVID-19 among the first 100 cases that informed the advice provided to policymakers, the study was not published until 2022.²⁵ The Inquiry into the Australian Government COVID-19 Response found that, even when preliminary analyses were completed, health departments were reluctant to publish them for fear they would be over-scrutinised and taken out of context.³³

After the initial reliance on the United Kingdom data and experience, data linkage efforts were expedited. The Victorian Post-Acute COVID-19 Study (VPACS) group was formed in February 2021 to collate and share developments in the rapidly evolving field of Long COVID.³⁶ VPACS aimed to improve data linkage, promote evidence-based practice and share clinical experience to enable ongoing research.³⁶ In another example, COVID-19 vaccination data from the Australian Immunisation Register (AIR) was linked with records held in jurisdictional surveillance systems.³⁷ This was extended to other Australian Government datasets, including the establishment of Person Level Integrated Data Asset (PLIDA) data held by the Australian Bureau of Statistics, but it was too late to support policy during critical periods of the vaccine rollout.³⁸

In contrast, there were also examples where existing data and evidence reports were not forthcoming, even where specific data were collected to evaluate interventions. In 2022 as schools reopened, the Victorian government instituted mask requirements that were more strict than any other Australian jurisdiction, at the same time installing HEPA filters, providing free Rapid Antigen Tests (RATs) and requiring negative RAT results for school entry. However, data were not released to support the interventions and encourage ongoing participation, and when the data were finally made public under a freedom of information application, they revealed higher rates of infection in the primary school children who were required to wear masks.³⁹ Overall, infection rates in school-aged children were as high or higher in Victoria when compared with other states by the end of that year.⁴⁰

Evidence generation and use was a key pillar in the independent Inquiry into the Australian Government COVID-19 pandemic response. Data availability and quality are implicit in many of the 19 recommended immediate actions required of the Australian Government. However, there are also dedicated recommendations regarding the generation and use of evidence, including immediate actions to ‘Improve data collection, sharing, linkage, and analytic capability to enable an effective, targeted and proportionate response in a national health emergency’ (Action 11, page 11).³³ Coordination of evidence generation, synthesis and communication was also identified as a central function of the Australian Centre for Disease Control.³³ By the end of 2020, the immunological profile of the Australian population was quite different to its overseas counterparts, yet we still relied heavily on estimates of infection risk and vaccine effectiveness based on overseas data in generating Australian public health policy.

Overall, the ‘aggressive suppression’ policy direction served the Australian response well.^33,35 However, the response at times had an adverse impact and was limited by a lack of an evidence base in estimating risk differences across target and priority populations. This encouraged broad, far-reaching and lengthy interventions that impeded human rights and were deeply inequitable.³³

Conclusion

Pandemic responses should be evidence-based and data-driven, supported by real-world evaluations in Australia and globally. To achieve this, there is a need for coordinated approaches to build epidemiological evidence into pandemic planning. The fundamentals of data collection, analysis and sharing should be prioritised and clarified. The pandemic real-time research agenda must expand beyond genomics and clinical trials to include operational applied public health research that provides real-world data of high granularity to identify those at most risk of exposure and disease as well as those most impacted by the interventions, for better and worse. This is all the more important for low- and middle-income countries where health resources are limited and there is greater potential for disproportionate impacts within populations.