Publications : 2018

Urban JD, Harvey S, Haws LC, Wikoff D. Assessment of study quality (risk of bias) in understanding the relationship between congenital heart defects (CHDs) and exposures to trichloroethylene (TCE). Society of Toxicology Annual Meeting. March 11–15, 2018. San Antonio, TX.


In the US EPA IRIS assessment of TCE, the Agency selected CHDs observed in a rat drinking water study as a sensitive endpoint for development of both the RfD and RfC. This decision has been controversial due to notable study limitations and the inconsistency of the findings relative to the rest of the TCE literature. Due to “concerns raised about study quality,” the TCE-CHD literature was recently re-evaluated in what was termed a “systematic evaluation” (Makris et al., 2016). While this re-evaluation has the appearance of a “systematic review” (SR), it lacks several critical elements of an SR, most notably an assessment of risk of bias (RoB) – a measure of confidence in whether the design and conduct of a study compromised the credibility of the link between exposure and outcome. Herein, we evaluated RoB for the TCE-CHD human and animal literature and integrated such using the NTP OHAT approach. To account for critical study design features of animal studies, some of the relevant RoB domains in the NTP OHAT RoB tool were refined and expanded. The single animal study reporting TCE-CHD effects was found to have high RoB for multiple domains, including selection bias and performance bias. Most of the remaining animal studies in the evidence base demonstrated low RoB for these and other domains. RoB was present for all studies for study group concealment and blinding as these elements were generally not reported by study authors. The human studies tended toward high RoB in the domains of detection bias (e.g., deficiencies in exposure characterization) and confounding bias; one human study was insufficient for RoB analysis (ecological study). Integration of RoB results with other factors (e.g., inconsistency, lack of dose-response, low magnitude) supports a moderate to high level of confidence that TCE is not associated with CHDs. Results of this evidence-based approach support that the single CHD-positive study is unreliable for use in developing toxicity factors based on: high RoB, inconsistency with animal studies of lower RoB, and a limited human dataset that does not demonstrate causality. These results demonstrate the importance of assessing study quality (recommended by NAS and a required component of TSCA assessments) and highlight the need to re-evaluate the current toxicity values for TCE using an evidence-based approach that integrates data quality.