How Much Evidence Is Enough? The Limits to Epidemiology

By David Michaels

Later today, the California Occupational Safety and Health Standards Board will hold a public meeting to consider issuing an Emergency Temporary Standard to protect workers exposed to the chemical diacetyl. This chemical, a primary component of artificial butter flavor, has been implicated as a cause of bronchiolitis obliterans, a debilitating and sometimes fatal lung disease.

Yesterday, on behalf of the Project on Scientific Knowledge and Public Policy (SKAPP), I sent the Standards Board new evidence: an unpublished Dutch study reporting three cases of the rare lung disease among workers at a diacetyl factory.

Bronchiolitis obliterans is rare, but numerous cases have occurred among workers employed in factories where flavorings containing diacetyl are produced or used. Dozens of workers employed at popcorn plants have developed occupational lung disease, and at least one has died. Several animal studies also found significant respiratory damage following exposure to artificial butter flavoring or to pure diacetyl, even with short exposures -- in one case, only a single four-hour exposure.

OSHA has been aware of these bronchiolitis obliterans cases and animal studies for years, but has failed to issue an exposure standard to protect workers in the flavorings industry.

Regulators cannot wait for complete information before issuing rules to limit exposure to potentially toxic substances. Diacetyl poses the classic (but easily addressable) regulatory dilemma. The evidence for the toxicity of diacetyl is limited by an obvious problem: we do not have (and cannot have) controlled studies of humans exposed to diacetyl but to no other potential toxins. There are multiple chemical exposures at factories where diacetyl is used. Regulators must rely on these studies of the patterns of disease in workplaces, in addition to evidence gathered in experiments with laboratory animals.

It is hard to imagine what additional evidence could still be gathered on diacetyl. We will never find a workplace in which only diacetyl is present. The Dutch study comes close, since it deals with a diacetyl production plant rather than a plant producing multiple flavorings, but reluctant regulators could still argue for the presence of other confounding factors. The animal evidence is very strong. It is time to assume that diacetyl causes obstructive lung disease at extremely low levels and prevent all exposure â probably by banning it.

It has been almost six month since a group of unions and scientists petitioned OSHA for an emergency standard to protect workers from diacetyl. To date, the federal agency has taken no action. The OSHA website even maintains the statement that:

A cause-effect relationship between diacetyl and bronchiolitis obliterans has not been established.

California is actively working on the problem.

How much evidence is necessary before OSHA will act?

David Michaels heads the Project on Scientific Knowledge and Public Policy (SKAPP) and is Professor and Associate Chairman in the Department of Environmental and Occupational Health, the George Washington University School of Public Health and Health Services.