Selecting Appropriate Controls

Publication Details

Dale Sandler stated that the plan is to enroll 7,000 people who were unexposed to serve as a control group for the GuLF study. The investigators are “hoping to enroll non-exposed individuals, about 4,000 from the local community, 2,000 from further away, and then we have this special category of federal workers. We will include in our cohort federal workers, who may not come from the Gulf States,” Dr. Sandler said. She acknowledged that “this is tricky to do.” David Tollerud expressed concern that regarding the number of controls, “there’s just not enough to do what you need to do.”

Maureen Lichtveld agreed that exposure reconstruction “will make or break” the study, as a true baseline is no longer available; the study participants were exposed to multiple contaminants and had multiple exposure routes and exposure opportunities; and the exposure duration to each contaminant could have been brief, intermittent, or continual. This muddies the waters when distinguishing exposure between workers and controls and choosing an appropriate control group. She suggested that federal workers outside the Gulf coast are the least appropriate controls and that Gulf coast fishermen or workers outside affected areas would be the most relevant control populations. Several other participants also had concerns about the appropriateness of the proposed control groups.

Stephen Cole suggested structuring eligibility criteria to maximize the nonexposure comparability of participants. He also stressed the need to collect information on possible confounders, including those variables that are a common cause of dropout and cautioned against having a healthy worker selection effect. Roberta Ness added that it is likely that those who were trained but not selected to work on the cleanup are systematically different from those who were, and it may be difficult to figure out how they differ because such information was not necessarily recorded. She suggested using workers’ friends as controls or using other BP or federal workers as an out-of-region control group. Roxane Cohen Silver of the University of California, Irvine, noted that one should not assume that the impact, especially the mental health impact, on those who were indirectly exposed, such as family members of the cleanup workers, was minimal.