Establishing Guidelines for the Critical Appraisal of Animal Toxicology Studies on the Developmental Toxicity of PCBs
On this page:
A key aspect of hazard identification for environmental pollutants is understanding the strengths and limitations of the available animal toxicology studies. The U.S. Environmental Protection Agency’s (EPA) Integrated Risk Information System (IRIS) Program uses a study evaluation framework aimed at discerning risk of bias (i.e., factors that may distort the magnitude or direction of effect) and insensitivity (i.e., factors that limit the ability of a study to detect a true effect). There is a recognized need to develop outcome-specific criteria for this framework since study evaluation considerations can vary widely among specific health outcomes. In this work, we proposed criteria for evaluating developmental toxicology studies that report on offspring survival, growth, and structural alterations using the IRIS framework. We then applied these criteria to evaluate studies on polychlorinated biphenyls (PCBs). The IRIS framework evaluates animal studies based on the following domains: Reporting Quality; Animal Allocation; Observational Bias/Blinding; Selective Reporting and Attrition; Chemical Administration and Characterization; Exposure Timing, Frequency, and Duration; Endpoint Sensitivity and Specificity; and Results Presentation. Ratings are applied to each domain (Good, Adequate, Deficient or Critically Deficient). These domains are considered together to arrive at an overall judgment of confidence in the study (High, Medium, Low or Uninformative). We identified several areas for which criteria needed to be developed or defined in greater detail to facilitate the evaluation of developmental toxicity studies. For instance, in the Exposure Timing, Frequency, and Duration domain, reviewers must consider whether the exposure period encompasses the critical/sensitive window for an outcome, if known. For Endpoint Sensitivity and Specificity domain, reviewers must consider whether a study used well-established methods for evaluation of developmental toxicity, or whether methods were poorly defined (e.g., evaluation of “abnormalities” or “malformations” with no definitions provided). In the Results Presentation domain, an important consideration for many developmental studies is whether authors controlled for intra-litter correlation, e.g., using the litter as the statistical unit. These proposed criteria were used to highlight factors that may affect the reliability of studies on the developmental toxicity of PCBs and will ultimately be useful to inform the weight of evidence for hazard identification.