The need for adequate quality assurance/quality control measures for selenium larval deformity assessments:...

Assessing the frequency and severity of larval fish deformities is a subjective exercise that is subject to considerable parameter uncertainty unless appropriate quality assurance/quality control (QA/QC) measures are incorporated. This issue has received limited attention in the literature. Only one study was identified that contained adequate data to evaluate the reproducibility of larval deformity data. Parameter uncertainty was substantially larger than expected. There was poor reproducibility between observers for nearly all types and magnitudes of deformities, and there were particularly large differences in how mild deformities were assessed. The reproducibility of the edema endpoint was the poorest of the 4 types of deformity evaluated. Specific recommendations for improving the QA/QC aspects of larval deformity assessments include blind and nonsequential labeling; explicit effort on the development and application of an a priori framework; internal QC checks to quantify the influence of sample preservatives, observer drift, or multiple observers; and an external QC check of a minimum of 10% of all larval fish. Future selenium reproductive studies should include an explicit uncertainty analysis and disclose raw deformity data to facilitate recalculation of tissue residue guidelines as the science in this area advances.