Evidence is in the Past, Risk is in the Future: Tail Events and Foresight in Public Health

Abstract Background Public health faces risks from Gaussian (thin-tailed) and Paretian (heavy-tailed) domains. Gaussian risks (e.g., local accidents) allow prediction via stable averages. Paretian risks (e.g., pandemics) feature extreme, unpredictable outcomes dwarfing prior observations. Risk assessment assuming Gaussian dynamics is perilous for heavy-tailed systems’ potentially catastrophic shifts. This distinction is vital for robust public health monitoring, foresight, and policy, especially for complex threats like infectious diseases. Methods This comparative conceptual analysis contrasts thin- and heavy-tailed risks using secondary data: scaled historical pandemic casualties (Paretian; Cirillo & Taleb, 2020) and Finnish traffic accidents (Gaussian; Statistics Finland, 2024). Graphical analysis (time-series, Zipf plots) demonstrates divergence of distributions, differential nature of extreme events, and scaling behaviour. Results Pandemic data show extreme variance and power-law scaling (heavy tails), where past extremes do not preclude larger future ones; single events dominate impact (largest observation accounts for 45% of top-quartile casualties). Traffic data show constrained variance and rapidly decaying tails (Gaussian, largest observation accounts for 6% of top-quartile casualties). This highlights the fundamentally different nature of Paretian risks relevant to foresight. Conclusions Public health monitoring and foresight must differentiate between Gaussian and Paretian domains, as increased global connectivity introduces new fragilities. Average-based prediction fails for heavy tails. Managing these requires a paradigm shift towards anticipatory awareness, when systemic harm is plausible but hard to predict. We discuss in-progress solutions to complement monitoring and foresight via participatory community surveillance networks, rapidly repurposable during crises for decision support and learning, improving preparedness for unpredictable, high-impact events. Key messages • Standard risk models assuming stable averages (Gaussian) dangerously underestimate unpredictable, heavy-tailed threats like pandemics. Public health foresight must adapt. • Recognize heavy-tailed risks; shift from prediction to anticipatory awareness using adaptable methods like participatory community surveillance for better preparedness.