Reply to correspondence letter "Krakow's children cohort and long term follow-up of thimerosal exposure--design and statistics".

Abstract

Dear Editor J.P.K. Rooney and J. Dorea raised some important issues related to the design and statistics of our last study on thimerosal-containing vaccines and children cognitive development, which we feel should be explained. They are right that we did not discuss our previous results published inNeurotoxicol Teratol [5] in our last paper [6]. The last paper was focused only on children’s cognitive development and the results did not differ in that issue with the previous study, when follow-up was limited to 3-year-old infants. We found no relationship between neonatal thimerosalcontaining vaccine (TCV) exposure and mental developmental delays in children up to 3 years of age [5], which was confirmed in the last paper related to longer exposure to TCVs (both neonatal and up to the sixth month of life) and longer follow-up of children’s cognitive development (up to 9 years of age) [6]. On the other hand, our previous results have shown that ethylmercury is not completely harmless for the first stage of life and may be responsible for poorer outcomes of psychomotor development in children. The lower psychomotor scores associated with TCV neonatal exposure were observed only during the first 2 years of life. That negative association disappeared by 36 months [5]. We did not discuss it because in older children, we did not perform the psychomotor tests and we could not assess the potential influence of TCVon psychomotor development in later age. Differences in size of analyzed population between papers [5, 6] were associated with different attitudes as during the first 3 years of life, the same test was conducted [5], and repeated analysis was in our opinion the best way of analysis. The period analyzed in our last paper [6] was much longer and represented by different tests (and different scales) of children’s cognitive development. It was hardly possible to compare them in repeated analysis techniques. Furthermore, as we noted [6], there was substantial dropout from the basic number of children (at the first year of life) to 45 % at the age of 8. So each analysis was made for a particular age category and differed in number of children as a result of lack of some test performances. The full enrolled cohort was indeed 505 children; however, due to drop out, the lack of the possibility to check vaccination history, and missing cord blood results of heavy metals, the biggest possible number available for analysis was 318 [6]. It was smaller in a 2012 paper [5] as only children with Bayley (BSID-II) tests who performed repeated measures during all 3 years with a score result were analyzed. Furthermore, analyzing a period of 9 years in a rather small cohort, we needed to choose which cord blood metal, lead or mercury, introduces in final models for adjustments, as adjusting for both makes the group too small. Finally, we decided to present models adjusted for mercury level as more important as thimerosal issue was considered. During the analysis process, adjustments to lead level instead of mercury one were also performed; however, they were not changing the inference of thimerosal. They were not present in the manuscript, similarly as full models, due to the interest of simplicity. We do not This is a reply to the correspondence found at http://dx.doi.org/10.1007/ s00431-015-2568-7