Skip to main content
. Author manuscript; available in PMC: 2023 Sep 1.
Published in final edited form as: Neurotoxicology. 2022 Jul 19;92:49–60. doi: 10.1016/j.neuro.2022.07.003

Table 6:

Performance of competing methods to impute missing Hg data at 19 months and at 19 years among SCDS Main Cohort participants, based on a comparison of imputed values to observed Hg values. For each method, table entries show mean differences, sum of squared (SS) differences, and the ratio of the mean differences (in absolute value) and SS differences to the average Z-score method. Comparisons are based on the subset of n = 621 participants in early childhood and n = 379 in early adulthood who have complete Hg data at all ages within the time window.

Average Z-score Average Geometric average AUC Average Z-score* FCS-9 single* FCS-9 mean* FCS-4 mean*
19 months
 Mean difference −0.50 1.64 1.24 1.60 −0.38 0.25 −0.72 −0.79
 SS differences 6278 9378 8016 8717 6337 10,068 5705 5564
 Mean ratio 3.30 2.50 3.22 0.65 1.90 2.08
 SS ratio 1.48 1.26 1.38 1.59 0.90 0.88

19 years
 Mean difference −0.65 −4.31 −4.96 −4.16 −0.56 0.25 −1.01 −1.07
 SS differences 10,004 16,142 18,572 16,109 10,419 18,945 10,415 10,314
 Mean ratio 6.60 7.59 6.37 0.44 1.78 1.87
 SS ratio 1.55 1.78 1.55 1.82 1.00 0.99
*

Imputation uses Hg data at the age of interest from 50% of the sample.

Denominator for ratio calculation is average Z-score; the mean ratio is an absolute value