Major Concerns About Study Design and Clinical Biomarker Interpretation

We are writing this letter in response to the recent article in The American Journal of Medicine by D'Andrea and Reddy. Our primary concerns are with their convenience sampling and case-control mismatching, and the authors' conclusions about the clinical results. The clinical results indicate that both groups in their study, unexposed and exposed, have clinical biomarker values for both liver function and blood profiles that essentially fall within the normal ranges for healthy adult individuals (Table).^{,  ,}  Results from clinical laboratories tend to vary only slightly for such routine and standardized tests. Statistical differences among such unmatched groups do not necessarily equate to biologically meaningful differences. Their results should include an explicit reference and comparison to well-documented clinical ranges for normal, healthy adult individuals. An alternative conclusion about their clinical results is that while these age- and sex-biased convenience samples differ from one another statistically, neither group actually differs from accepted values for normal, healthy adult individuals. This equally valid conclusion might well serve to comfort the patients under evaluation in that, as individuals, none appear to have any clinical indications of abnormal liver or hematological function. We hold that this should have been explicitly discussed, and we are concerned that the conclusion reported by the authors may well exacerbate health problems through psychosocial and behavioral means in those studied, as well as others that participated in the cleanup process. How, then, would the authors and physicians in general interpret the relevance of their statistical findings, especially when comparing a predominantly male group of younger cases to a predominantly female group of older controls, and how would they communicate these biomarker results to their patients?

Beyond biases in interpreting their biomarker data, no attempt at reconstructing exposure also is a problem. We acknowledge the difficulty and uncertainty in retrospectively reconstructing exposures, but some level of exposure assessment beyond the label of cleanup worker is necessary. The authors do not include critical elements of an exposure assessment such as individual environmental and occupational health histories. Environmental and occupational histories would shed light on relevant concomitant non-oil-spill exposures, pre-, during-, and postspill activities and exposures that need to be considered, and any use of personal protective equipment during cleanup activities that should be accounted for. In short, these authors do not appear to have followed acceptable clinical practice as established by members of the Association of Occupational and Environmental Health Clinics.

We do commend these authors for their focus on the health of oil spill cleanup workers. Although they do describe, in part, some limitations of their study, the obvious influences these have on the results of their statistical testing and ultimately, on their biased conclusions, are problematic. These problems affect the validity of their conclusions and indirectly threaten to further stress our Gulf Coast communities.