There was insufficient information to determine whether statistical models were appropriate. The authors did not mention evaluating the assumption of normality or transforming metals exposure variables, which were used continuously in the models. Box plots for other trace elements (e.g., Se) indicated highly non-normal distributions. It was also unclear whether models for arsenic simultaneously included other placental metals, whether metals were highly correlated, and how many covariates were included in the models (of 13 metals and 9 other candidate variables). Mean ± SD placental arsenic was several-fold higher in cases (2.08±1.96 µg/L) than controls (0.37±0.51 µg/L, p<0.001), but associations were not significant after multivariate adjustment (p=0.137). Statistical significance could also have been affected by sample size. The authors reported that 64 participants were required in order to detect an effect size of 0.765 at alpha error of 0.05 and statistical power of 0.85 based on a similar study. Their sample size (n=90) exceeded the estimated requirement. However, the authors did not describe the study used as a basis for these calculations, and why it was considered an appropriate guide.

Baser, 2020, 7024522

Study evaluation