How Canada’s decentralised covid-19 response affected public health data and decision making

How did fragmentation and decentralisation contribute to covid-19 outcomes?

Canada faces systemic challenges in evidence informed public health decision making, partly because of its complex, decentralised governance of healthcare and public health (appendix).33 Within Canada’s disaggregated federation, constitutional responsibility for public health is shared between a national, federal government and 10 provincial governments. The three territories (Yukon, Northwest, and Nunavut) derive their responsibility for aspects of health from their founding federal legislation. Within each province and territory, healthcare and public health systems operate side by side, and here we focus on public health, which, in 2019 accounted for only 5.2% of Canada’s total health spending.34

SARS brought international attention to shortfall in the Canadian public health system. After the 2003 outbreak, the country’s handling of SARS was described as an “international embarrassment.”35 The chair of the National Advisory Committee on SARS and Public Health, David Naylor, described “squabbling among jurisdictions, dysfunctional relationships among public health officials from the three levels of government (federal, provincial/territorial, and municipal), an inability to collect and share epidemiological data, and ineffective leadership” – which, taken together, held hostage the health of Canadians.136

PHAC was created in 2004 in response to concerns about Canada’s ability to respond effectively to public health threats, with a commitment to reduce health inequalities, while recognising that public health was a shared responsibility across government levels.37 PHAC is responsible for developing national clinical and public health guidelines, technical advice, border quarantine services, and specialised services for early testing of emerging pathogens.38 However, it lacks the powers to direct provincial and territorial health agencies or other bodies with similar mandates to implement its recommendations.

In Ontario, Public Health Ontario (PHO), an arm’s length scientific and technical organisation created after SARS,39 led data and surveillance reporting and laboratory testing for covid-19, starting in early January 2020. Like other provincial public health agencies, the PHO does not create policy but its work informs provincial and local policies and guidelines—for example, through covid-19 surveillance, scientific and technical information, laboratory testing, and evidence synthesis of best practices.

Although each province or territory has a chief medical or public health officer, they can delegate public health decisions, responsibilities, and implementation to regional or municipal health authorities, with or without sufficient resources.40 This means that decision makers at different jurisdictional levels lack coordination in a pandemic. For example, each province or territory devised its own interventions and timelines for protective measures such as school closures, border controls and closures, prohibition of gatherings, and masking requirements, leading to substantial variation in policy and practice across the country, widely varying hospital admission rates, and public confusion (fig 1).17294142

Some health professionals joined official advisory committees across the country (appendix). Others engaged in social media, sometimes becoming trusted voices that challenged public health leaders. The conflicting messaging undermined public trust over time.43 Organisations representing health professionals, including the Canadian Medical Association, Canadian Nurses Association, and Canadian Pharmacist Association, called for improvements in the shortages of drugs and personal protective equipment, workplace conditions, harassment, and burnout that were affecting health workforces, long term care staffing and conditions, and for vaccination access and coverage.444546 Federal and provincial health professional associations both amplified the distribution of public health guidance and released their own to support decision making.47 Most notably, provincial associations enabled the rapid pivot to virtual care. However, overall the advocacy by health professions’ leadership remained fragmented and hampered by inadequate data production, access, and use across provincial, territorial, and regional jurisdictions.

In the absence of a coordinated pandemic planning authority, the supporting evidence and rationale for different rules in different places were often unclear. PHAC provided national guidance48 on infection prevention and other control measures as well as national vaccine guidelines through the National Advisory Committee on Immunisations (NACI). NACI made evidence based recommendations 49 on covid-19 vaccine eligibility but had no responsibility for implementation. Each province or territory then created its own vaccine eligibility plans, allocation, and mandates. For example, vaccine mandates in Ontario applied only to those working in long term care homes while in New Brunswick, Newfoundland, Nova Scotia, and Yukon vaccines were required for those working in healthcare, the public sector, long term care, and schools.50 Similarly, provincial public health authorities differed in their guidance on masks, school closures, curfews, and lockdowns.2951 Outbreak investigation, management, and risk mitigation happened at local and municipal levels. Adding to the complexity, in some provinces, entities such as school boards and acute care hospitals implemented mandates to augment provincial public health guidance.

While some variation may be justified given localised risk factors, insufficient pooling of data and analyses during the pandemic made it hard to understand these variations or share their rationale with the public. PHAC and some provincial public health organisations used their own researchers to conduct knowledge syntheses and primary research whereas other provinces created scientific advisory committees with external membership.4252 Indeed, in Ontario, medical academics persuaded the government to establish an expert, public facing scientific advisory committee that was able to challenge public health leadership.

Deficiencies in data and data infrastructure for decision making

Public health data derive from multiple sources, including electronic health records, administrative data on healthcare services utilisation, and public health surveillance data, including testing data and genomics data, such as viral sequencing. All government levels collect, access, and use these data to some degree, but data were in short supply to support public health decision making during covid-19. This situation will persist without major reform. The national auditor reported that when the covid-19 pandemic began, PHAC “did not have regulations or finalized agreements with the provinces and territories to clearly outline what public health surveillance information to share and how to share it,”32 including case level information on vaccine safety with Health Canada, vaccine developers, and the World Health Organization,32 delaying information sharing on vaccine safety.

In addition, public health decisions across jurisdictions in Canada were hampered by outdated health information systems and lack of integration and interoperability between data sources after eight years and C$130m (£76m; €90m; $98m) of investment failed to establish a pan-Canadian technical solution for public health data.53 The numerous pieces of federal, provincial, and territorial legislation covering health information or privacy create a “legal interoperability” challenge, which was amplified by “human interoperability” challenges related to risk averse interpretations of those laws by data custodians. A clearer vision for a public health data system in Canada will require coordination and interoperability of systems under the control of disparate and geographically distanced organisations to support analyses and public health decision making.5455 Data may not need to be pooled or moved across jurisdictions, but access to those data could be enabled through federated data structures.

As the pandemic progressed, public confusion arose from jurisdictional inconsistencies in advice and case reporting.56 Most jurisdictions routinely reported the raw number of covid-19 cases. More detailed information about the demographics or location of cases was often more guarded because of privacy concerns, meaning that public health agencies did not respond to community interest about local risk and tailoring of implementation strategies to mitigate covid-19 risks was slow or non-existent. Lack of local data contributed to a lack of understanding of local transmission dynamics and contributed to loss of public trust over time.31 Similarly, contact tracing information was not centralised within provinces or territories. Collection of data on ethnicity, occupation, and other factors was slow to evolve, and in some places never happened, even though this was clearly related to virus spread.6

Data sharing lagged even for non-personal data such as viral genomes, which is perplexing given funder commitments to health and genomic data sharing.5758 For example, Canada was one of the slowest industrialised countries to submit SARS-CoV-2 genome sequences to Global Initiative on Sharing All Influenza Data (GISAID), which supports public health decision making through global surveillance of SARS-CoV-2 variants.59 The rationale for limited sharing included that data might be misinterpreted, partly because of differences in data related practices. For example, provinces had different policies about testing eligibility, test types—PCR or rapid antigen tests—and how test results were reported. Variation made it difficult to compare case rates, which was partly addressed by reporting the proportion of tests that gave positive results (“test positivity”). Lack of testing had broader implications for attribution of deaths to covid, further complicated by definitions of death “with” versus death “from” covid-19 across time and jurisdiction. While these concerns could be overcome with good data documentation, or data and testing standards, inconsistent and changing definitions during the pandemic created public confusion and distrust.