Interpreting the literature

Abstract

Epidemiological studies do not, by themselves, answer all the important questions about the etiology (cause) of occupational disease that are required to make a judgment for compensation. They inform conclusions about “general causation”, which refers to evidence for an elevation of risk and the connection between disease and possible causes in general. Workers’ compensation and other compensation systems require individualized evaluation of each case, called “specific causation”, taking into account the individual circumstances, risk factors, and health risks for the claimant. Epidemiology can also inform specific causation by indicating whether risks are disproportionately elevated in certain job assignments, after a certain number of years of exposure, or against a background of other, non-occupational risks (such as cigarette smoking). This requires interpretation. Epidemiology normally requires a scientific standard of certainty (95 %, at least for any one observation). However, workers’ compensation adjudication, like civil litigation, requires “the weight of evidence” (>50 % certainty). Thus, although an individual study in epidemiology is conducted to scientific standards, the body of knowledge must, as a matter of law, be interpreted by the preponderance of evidence. It is not acceptable to holdout for scientific certainty in adjudicating a claim, because it is almost never achievable in an individual case and serves only to put an insurmountable obstacle in front of the claimant. Epidemiological studies of firefighters, as with other occupations, yield a relatively small number of cases for important but uncommon outcomes such as cancer, especially those that have a random component and are also incident in the general population. This means that statistical power is low for these outcomes and this, together with sources of bias, usually results in underestimating the risk. As a result, not every study should be expected to show an elevation that is truly present in an outcome such as cancer’ so-called “positive” studies are probably more informative and have more probative value than “negative” studies, which have some probability of having simply missed the association. Evidence suggesting a roughly doubled risk compared to the general population or a reference group, after close attention to subgroup analysis, checks for confounding, and exposure-response, are better evidence for a causal association than consistency, which cannot be expected in small studies of uncommon outcomes designed for etiological investigation. However, rigid application of doubling of risk is overly strict as a criterion for adjudication and will result in denying benefits in many meritorious cases.