Safety and Health at Work Vol. 13 Supplement 2022
Reconciling epidemiological and toxicological data: Some general principles and the example of firefighters

Safety and Health at Work Vol. 13 Supplement 2022
Reconciling epidemiological and toxicological data: Some general principles and the example of firefighters

Reconciling epidemiological, toxicological data, general principles, example of firefighters

The fundamental relationship between toxicology and epidemiology is that increasing exposure results in an increasing response varying with the degree of individual susceptibility that at some point becomes either detectable and counted as an outcome or an increased density of cause that increases the frequency of a stochastic (random) response. Toxicologists count the magnitude of the response, epidemiologists count the frequency, and clinicians observe the new onset of outcomes that appear with exposure. A toxic agent may be the driver of the outcome or a contributing factor adding to or modifying risk. Firefighters demonstrate all of these effects as they occur together but outcomes are largely concealed or offset by lower personal risk factors. Epidemiology has therefore often been under-interpreted as a guide in terms of relative risk, a problem compounded by many methodological
problems (chiefly low power, illogical aggregation of disease rubrics, dilution of risk estimates, and confounding. The data cannot be assumed to tell a simple story: interpretation requires understanding, not meta-analysis of phenomenology, which has been less helpful in etiological studies of firefighters than in other applications. What the investigator is usually seeking is an indicator of risk, which is not the same thing as the frequency of past experience. Epidemiology provides a summary of experience but it is a trailing indicator, because that experience happened earlier, in a different time and place. Looking backward, assessing causation in the individual case, one asks: “Given that something bad happened, what is the probability that it was causally related to the attribute in question?” but epidemiological methods apply to populations, not individuals. Causation analysis may benefit from Bayesian methodology to individualize risk estimates. Looking forward for prediction, in order to design more effective prevention, one asks:
“Given the attribute, what is the probability that something bad will happen?” That requires a leading indicator, which more reliably emerges from an understanding of the mechanism driving the response. Looking forward, toxicology and biological markers (indicators), together with exposure science (the exposome) may have greater predictive potential than extrapolating from past experience imperfectly understood. The synthesis of epidemiology and toxicology needs to be taken further into analysis.

Tee L. Guidotti

Elsevier Korea LLC

January 2022

Sri Wahyuni

PDF

English

Text

Safety and Health at Work Vol. 13 Supplement 2022