Substitute the word science for perfection.
Trifles make perfection, but perfection is no trifle.
-Michelangelo
What’s wrong with this study and Sallie Bernard’s comments published in Pediatrics
Background: Smith and Charles Woods utilized data previously collected for a 2007 Centers for Disease Control and Prevention (CDC) study investigating neuropsychological outcomes from early thimerosal exposure. SafeMinds President Sallie Bernard served as a consumer representative on the original 2007 study.
1. At the time of the first study Ms. Bernard submitted comments regarding limitations of the data including only a 30% participation rate when the standard for scientific research is at least 70%.
2. Children were tested at 7-10 years of age, and there was no attempt to control for prior interventions which may have ameliorated neuropsychological deficits.
3. Investigators did not combine prenatal and postnatal thimerosal exposures in assessing outcomes, when cumulative exposures would have been an important analysis for ruling in or out a thimerosal effect.
4. Children born weighing 5lbs 8 ounces or less, almost 9% of the population, were excluded from the study. These infants may have been more vulnerable to mercury exposure, and their exclusion limits the applicability of the findings.
5. In addition to these inherent weaknesses of the data set, the current study by Smith and Woods did not include vaccine exposure beyond 3 DTP, 3 Hib, 2 hepatitis B and 2 polio vaccines. The recommended schedule for this time period allowed for a third dose of hepatitis B and polio vaccines as early as 6 months. Some children in the study received other vaccines in the first year of life, including those for influenza, hepatitis A, MMR, pneumococcal, tuberculosis, tetanus toxoid, varicella, and meningococcal. Whether a child did or did not receive these additional doses was not factored into the analysis.
6. Smith and Woods controlled for cumulative ethylmercury exposure, without providing justification for doing so. Cumulative ethylmercury is positively associated with receipt of 3 of the 4 vaccines in the analysis (hepatitis B, DPT and Hib) and thus with overall vaccine receipt. Including this variable would have the effect of reducing variance (P value) observed for the exposure variable of interest, vaccine receipt.
7. A number of additional variables in the model such as familial and SES factors were associated with vaccine receipt as well as the outcome measures. The authors do not state whether diagnostic tests were conducted for multicollinearity and whether their model was impacted by collinearity, for these and other variables.
8. Vaccine timeliness was ill-defined. A vaccine was considered on-time if it was given within 30 days of the recommended window. In practice, this meant, for example, the birth dose of Hepatitis B vaccine, a shot of particular concern to many parents and doctors, would have been counted as “on time” if it were given any time from day of birth up to age 2 months (60 days). Many parents who “delay” vaccination are requesting a similar extension in timing like the one allowed under this study definition.
Ms Bernard’s submitted comments regarding the study to Pediatrics. These were accepted by the journal and posted online.
Comments to Pediatrics posted by others, including Dr. Larry Rosen, pointed out conflicts of interest by the study authors as well as other deficiencies in the research.
Dueling experts (and elsewhere whistleblowers). The situation isn't as one sided as you'd like to represent.