In the comments to my interview with DarkSyde, another parent and I discussed “alternative” medicine and some of the autism literature. Today over on Respectful Insolence, Orac has an excellent post on how using data from the VAERS database isn’t the best way to accurately gauge adverse effects.
A bit of background on VAERS: this is the Vaccine Adverse Event Reporting System. If you or your child has a bad reaction to a vaccine–a serious illness after receiving it, an unexpected side effect, etc.–they’re the folks you contact to report it. These adverse effects are then analyzed to see if they’re being reported more frequently than would be expected due to chance alone. (For example, it’s simply a fact of life that some people will become ill, or even die, within a few days of receiving a vaccine–especially if it’s a vaccine in wide use. VAERS and other data can be used to investigate whether these reactions are likely due to the shot itself, or are simply a chance event). However, VAERS is subject to a lot of biases in design. First, it’s designed to allow self-reports, and it will accept pretty much anything the reporter submits as an “adverse event.” Orac links to a story where a woman reported to VAERS that a vaccine had turned her into the Incredible Hulk–and it could have stayed in the database if she’d not given them permission to remove it. This is good in a way–it’s better to be broad and overly-inclusive than to dismiss reported side effects. But any type of self-reporting brings a bias to the study. For example, those who are more wary of vaccines may be more likely to report side effects than those who have more confidence in them. And, as Orac points out, potential lawsuits appear to be driving up the number of side effects reported to VAERS–a reason to be skeptical of evidence used from that database to support a connection between vaccination and thimerosal.