Neurodevelopmental Disorders Following Thimerosal-Containing Childhood Immunizations: A Follow-Up Analysis. Geier DA and Geier MR. International Journal of Toxicology, 2004;23:369-376.
Explanatory note: Both thimerosal—a mercury containing preservative—containing vaccines and MMR vaccine have been hypothesized as being a cause of neurodevelopmental disorders (NDDs), including autism and autism spectrum disorders. In 2004, the Institute of Medicine Vaccine Safety Review Committee thoroughly reviewed all evidence available concerning any possible associations of vaccines and autism. The Committee concluded that the evidence favored rejection of those hypotheses.
Prior articles by Geier and Geier have considered the possible relationship of vaccines and autism and were reviewed in the IOM report and elsewhere. The prior studies were found to be uninterpretable because of methodological limitations. This report is a follow-up to one of their prior studies.
Was there an increase in neurodevelopmental disorders (NDDs) related to the use of thimerosal-containing vaccines (TCV) in the United States?
The authors examined reports of NDDs including autism, mental retardation, ataxia, speech disorders, thinking abnormalities, and personality disorders submitted to the Vaccine Adverse Event Reporting System (VAERS). They planned to compare children who had received thimerosal containing vaccines between 1992 and 2000 to children who received vaccines from 1997 to 2000 that did not contain thimerosal. Their estimates of the number of children who received doses of thimerosal containing vaccines and non thimerosal containing vaccines were based on data from the Biological Surveillance Summaries (BSS) database.
The study does not allow the reviewer to understand what is being compared and how the comparisons were made.
The authors do not adequately address limitations of the VAERS database that would affect their analyses. There are issues about both the selection of the cases of adverse events as well as exposure to thimerosal. For example, while the VAERS database has proven useful for identifying possible uncommon adverse events, it contains selection biases—factors that lead to systematic errors in the true estimate of effects—which severely limit how the data can be utilized. These limitations are well described in the literature. The most obvious selection bias for this study occurs because case reporting to VAERS is voluntary; reports are more likely to be submitted for those untoward events which the observer considers more likely to have been caused by the vaccine. Reporting to VAERS is well known to be highly affected by media publicity of which there has been a great deal about this topic. In addition, diagnoses of events that are submitted to VAERS are not validated from medical records. The authors stated that “Specific vaccine-adverse events following vaccination are required to be reported to this database as mandated by law.” However, the events that they sought to examine are reported voluntarily and are not mandated by law.
The authors do not specify a time period between vaccine receipt and NDD diagnosis.
The authors determination of thimerosal exposure focused on doses of DTaP with or without thimerosal that were distributed, but does not take into consideration the transition period when formulation switchover to reduced thimerosal containing vaccines occurred; does not address the number of doses administered; ignores exposure to (or non exposure to) other thimerosal containing vaccines during that time period. For example, the VAERS database contains limited data on the prior doses of vaccines a child has received. Therefore, children reported to VAERS may have received mixed schedules of vaccines, some containing thimerosal and some not. It is unlikely that the authors were able to look at cumulative thimerosal exposure.
It is unclear as to how the authors categorized individuals into the two exposure groups. Indeed, it is unclear that they could have been able to determine the “…amount of mercury received from thimerosal-containing childhood immunizations.”
Their methods used to analyze VAERS data differed from the methodology that they cite in important ways in both this and their preceding study (see IOM report), further limiting the validity of the data on thimerosal exposure.
The authors interpret their data as demonstrating the occurrence of NDDs more frequently in children who received thimerosal-containing vaccines than children who did not.
The authors further describe their estimates of the incidence (number of new cases) and odds ratios (a calculation of the risk of the disease after exposure) for a number of reported NDDs. What the authors measure in the so-called “incidence rates” is unclear: whether these are rates per year, cumulative incidence, or other. Thus the authors report an “incidence” rate for autism of 0.77 to 1.3 per million DTaP (doses or children who received the doses?) whereas current estimates of the prevalence of autism are 1 to 2 per 1,000 children. Similarly, their figures for other NDDs differ from published rates.
Although this report suggests a link between NDDs, including autism, and thimerosal-containing vaccines, it has serious methodological limitations that make the results uninterpretable. For example, incidence rates—such as reported in this study—cannot be calculated from VAERS data.
Epidemiologic studies can be subject to potential biases that can affect the validity of the results—as in the study reported here. VAERS is a valuable hypothesis generating system. Signals observed from VAERS require testing in other types of studies.