Scientific Method for Health Risk Analysis: The Example of Fine Particulate Matter Air Pollution and COVID-19 Mortality Risk

Applied science is largely about how to use observations to learn, express, and verify predictive generalizations—causal laws stating that if certain antecedent conditions hold, then certain consequences will follow. Non-deterministic or incompletely known causal laws may only determine conditional probabilities or occurrence rates for consequences from known conditions (Spirtes 2010). For example, different exposure concentrations of air pollution might cause different mortality incidence rates or age-specific hazard rates for people with different values of causally relevant covariates. A defining characteristic of sound science is that causal laws and their predictions are formulated and expressed unambiguously, using clear operational definitions, so that they can be independently tested and verified by others and empirically confirmed, refuted, or refined as needed using new data as it becomes available. Comparing unambiguous predictions to observations (using statistics if the predictions are probabilistic) determines the extent to which they are empirically supported. The authority of valid scientific conclusions rests on their testability, potential falsifiability, and empirically demonstrated predictive validity when tested. Using new data to constantly question, test, verify, and if necessary correct and refine previous predictive generalizations, and wider theories and networks of assumptions into which they may fit, is a hallmark of sound science. Its practical benefit in risk analysis is better understanding of what truly protects people, and what does not—for example, the unexpected discovery that administering retinol and beta carotene to subjects at risk of lung cancer increased risk instead of decreasing it (Omenn et al. 1996; Goodman et al. 2004).