Same Problems in "New" Research on Autism and Vaccines
"New" study from DeStefano et al., rehashes flawed data - says SafeMinds
WASHINGTON, April 8, 2013 /PRNewswire-USNewswire/ -- A new book chapter on statistical methods in Recent Advances in Autism Spectrum Disorders by University of Northern Iowa researchers, DeSoto and Hitlan documents major methodological flaws in the data set said to disprove the link between Thimerosal-preserved vaccines and autism by Price, et al. More recently DeStefano et al.  used the same data set while attempting to disprove a link between the number of vaccines and autism.
"Both researchers tested whether any increased risk of autism was associated with increased exposures via vaccination. To do so, they must compare different levels of exposure. They failed to do so when they matched cases to controls on birth-year and HMO. Birth-year, by itself, defines exposure level and HMOs further ensures similarity guaranteeing cases and controls were nearly identical on the exposure," said lead author Dr. Catherine DeSoto. "This is a design flaw called overmatching, it forces cases and controls to be artificially similar and renders the results invalid."
The number, type and timing of vaccines US children receive is a function of birth-year. Formulations purchased and administered would be the same within a given HMO.
"By matching on birth year and HMO, they eliminated the variability right from the start," said Sallie Bernard, President of SafeMinds. "Here's a perfect example of the cascading impact from problem data analysis. The same statistical flaws in Price's 2010 research resurface in De Stefano's 2013 research."
"Once again, media reported exactly what CDC released last week without critically reviewing the information," said Eric Uram, Executive Director at SafeMinds.
Prior to approving Price et al. research, an appointed panel reviewed the study methodology raising design concerns. Ultimately, the Centers for Disease Control and Prevention (CDC) National Immunization Program dismissed them. In 2010, SafeMinds challenged the Price et al., article's claims by highlighting flaws related to errors in the design.  "Peer reviewers should reach agreement," concluded Bernard. "With Price, final approval processes were tipped to accept flawed methods."
SafeMinds maintains listings on peer reviewed research performed by independent scientists documenting neural risks associated with thimerosal and vaccination. Reporters are encouraged to review and then contact the doctors and scientists doing high-quality work on the issues.