Nick H Ogden, Aamir Fazil, Julien Arino, Philippe Berthiaume, David N Fisman, Amy L Greer, Antoinette Ludwig, Victoria Ng, Ashleigh R Tuite, Patricia Turgeon, Lisa A Waddell, Jianhong Wu
4 min
This paper outlines the predictive modeling efforts conducted by the Public Health Agency of Canada (PHAC) during the early stages of the COVID-19 pandemic. The researchers aimed to evaluate the impact of various non-pharmaceutical interventions (NPIs)—such as physical distancing, case detection, and contact tracing—on the transmission of SARS-CoV-2. By utilizing both agent-based and deterministic compartmental models, the study sought to provide evidence-based guidance for public health decision-making in an environment characterized by high uncertainty and a lack of vaccines or specific treatments.
The study employed two primary modeling frameworks to simulate the epidemic in Canada. The deterministic compartmental model, based on SEIR (Susceptible-Exposed-Infectious-Recovered) dynamics, provided a broad view of population-level transmission. The agent-based model offered a more granular approach, simulating interactions within specific community settings like schools, workplaces, and homes. Both models were calibrated using evolving data on virus transmissibility, incubation periods, and the proportion of asymptomatic cases, allowing researchers to compare the effectiveness of different intervention intensities.
The modeling results highlight a clear distinction between "delaying" the epidemic and "controlling" it. Without any interventions, the models predicted an attack rate exceeding 70% of the Canadian population. Implementing partially effective NPIs successfully flattened the epidemic curve and reduced the peak, but failed to stop the spread, often leading to a rebound if measures were lifted prematurely. In contrast, high-intensity interventions that pushed the effective reproduction number below one were identified as the only strategy capable of significantly reducing the total attack rate to between 1% and 25%. The authors conclude that while disruptive measures like lockdowns are effective, their removal must be paired with robust surveillance and contact tracing to prevent new transmission chains.
This research underscores the critical role of mathematical modeling in managing emerging infectious diseases. By quantifying the potential outcomes of different policy choices, these models helped public health officials understand the trade-offs between economic disruption and epidemic control. The findings emphasize that in the absence of pharmaceutical solutions, sustained and high-intensity public health efforts are essential to prevent healthcare systems from being overwhelmed.
BACKGROUND: Severe acute respiratory syndrome virus 2 (SARS-CoV-2), likely a bat-origin coronavirus, spilled over from wildlife to humans in China in late 2019, manifesting as a respiratory disease. Coronavirus disease 2019 (COVID-19) spread initially within China and then globally, resulting in a pandemic. OBJECTIVE: This article describes predictive modelling of COVID-19 in general, and efforts within the Public Health Agency of Canada to model the effects of non-pharmaceutical interventions (NPIs) on transmission of SARS-CoV-2 in the Canadian population to support public health decisions. METHODS: The broad objectives of two modelling approaches, 1) an agent-based model and 2) a deterministic compartmental model, are described and a synopsis of studies is illustrated using a model developed in Analytica 5.3 software. RESULTS: Without intervention, more than 70% of the Canadian population may become infected. Non-pharmaceutical interventions, applied with an intensity insufficient to cause the epidemic to die out, reduce the attack rate to 50% or less, and the epidemic is longer with a lower peak. If NPIs are lifted early, the epidemic may rebound, resulting in high percentages (more than 70%) of the population affected. If NPIs are applied with intensity high enough to cause the epidemic to die out, the attack rate can be reduced to between 1% and 25% of the population. CONCLUSION: Applying NPIs with intensity high enough to cause the epidemic to die out would seem to be the preferred choice. Lifting disruptive NPIs such as shut-downs must be accompanied by enhancements to other NPIs to prevent new introductions and to identify and control any new transmission chains.
Alex: That's the most significant limitation. They had to use surrogate contact data from the European POLYMOD studies because Canada didn't have equivalent granular contact surveys at the time. They also lacked local demographic resolution for hospital capacity, so ICU demand estimates carried quite wide uncertainty bounds.
Sam: Which is an understandable trade-off given the timeline — this was built within months of the virus emerging. But it does constrain how literally you can read the absolute numbers. [[RP_SECTION:policy-implications-of-modeling|Policy Implications of Modeling]]
Alex: Precisely. The right way to read this work is as a framework for comparing intervention scenarios under high uncertainty, not as a predictive oracle. The value is in the relative ordering of outcomes — unmitigated versus suppression versus delay-and-reduce — not in the point estimates.
Sam: That distinction matters a lot for how policymakers should have been using it. You're not reading off a forecast; you're stress-testing assumptions about what happens if you move too fast.
Alex: And the authors were explicit about that. The models provide the comparative structure for the decision, but the decision itself involves tradeoffs the models don't adjudicate — economic costs, social costs, political feasibility.
Sam: Looking at where the field went after this — real-time mobility data, genomic surveillance, municipality-level adjustments — it's clear how much this kind of work identified the gaps. The blunt, national-level mandate was a first-generation tool.
Alex: Right. The modeling infrastructure that came later allowed for much more surgical, dynamic responses. But in May 2020, this was the best available framework, and it made the central point clearly: if you don't break the transmission chain, you're only ever delaying the rebound. Flattening the curve buys time, but time has to be spent on something — vaccination, treatment capacity, test-and-trace — or the epidemic simply resumes where it left off.
Sam: That's the finding that should have been front and center in every policy briefing. The curve doesn't stay flat on its own.
Alex: It doesn't. And that's the durable lesson from this paper — not any specific parameter estimate, but the structural logic that connects intervention strength, transmission dynamics, and the inevitability of a second wave if the underlying immunological gap isn't closed. It's a clean demonstration of how mathematical modeling earns its place at the policy table, not by predicting the future, but by making the consequences of different choices legible before you have to live with them.
Sam: Thanks for listening to ResearchPod.