July 22, 2008

Vaccination and Asthma Shown to be Unlinked

Recently in the news there has been rumblings about a link between childhood immunization and asthma, which causes many parents to be concerned. Epi Wonk has written a post detailing a meta-analysis study published in the journal Pediatrics in November 2007 which found no statistical significant association between childhood vaccination and asthma.

I fully bow to Epi Wonk's writing on the subject (he is a now retired medical school professor and senior epidemiologist at the CDC) and thought I'd pass along some of his more pertinent information on the subject. Some of this information is very important, as the authors of the Pediatrics paper put it:

“It is important that researchers clarify this issue, because… the
perception that immunization causes asthma may become a significant
determinant of parents’ attitudes toward routine vaccination of their
A few important points I took away from Epi Wonk's discussion of the study. 1) The literature was scoured for randomized controlled and relevant birth-cohort studies from 1966 to March 2006. 2) To be included, each study had to meet three criteria: Directly compared vaccinated and unvaccinated children; Validated vaccination status by medical charts; Used preset criteria to define asthma.

This front-end criteria are very important for any review or meta-analysis because it will define the significance of the results received. Epi Wonk's last statement sums this up nicely. After pouring over the data and confirming the results it is important to qualify the significance of the results based on the input criteria, here's Epi:
A recent systematic literature review of high-quality studies that directly compared vaccinated and unvaccinated children, validated vaccination status by medical charts, and used preset criteria to define asthma found no association between childhood vaccination and asthma.
This is a good time to relate how the science works to the larger public community. I have been stressing the front end criteria of this study for a reason, by using these criteria the subject of the study is narrowed significantly, meaning this is not an overriding, large-scale study that was seeking to define the complete link between vaccination and asthma. This study was simply to define correlations within birth-cohort studies.

Birth-cohorts themselves are subject to a great deal of bias (something the author of the study acknowledges) and are extremely intensive in the front-end planning stages. A mini symposium on birth cohort studies published in 2002 in the journal Pediatric Respiratory Reviews examined the experience of a large, 10-year birth-cohort study conducted in Germany (Nickel R, Niggeman B, Gruber C, Kulig M, Wahn U, Lau S. Paediatric Respiratory Reviews. Sep 2002; Vol 3; Issue 3: 169-176) . The report detailed some difficulties encountered including changing nursing staff over the years which led to different bookkeeping and even patient evaluation scores. Racial differences need to be accounted for by including as many different genotypic backgrounds. The births need to be carefully selected to be evenly distributed over a 12 month period in order to account for seasonal affects. One of the largest problems is lost subjects, those who tend to not be affected by a disease will more regularly drop out of the study, pushing a bias towards those who are affected and are more likely to continue making their appointments. Interestingly enough it was found that as the personnel at the study centers changed the loss of subjects jumped significantly.

Recall bias is also an interesting phenomenon within studies where those affected by a disease under study, or who have had family members with the disease, can more frequently recall symptoms or are more observant of their child.

Epi Wonk himself shows how these studies need to be taken in the context in which the plan is laid out. He details a study conducted in Canada which shows that the timing of vaccination affects the risk of developing childhood asthma. This should show an obvious deficiency about the study, a variable that the study was not concerned about and cannot speak about. It doesn't mean that the study was not significant or not performed correctly, in fact in means the exact opposite; that the authors correctly narrowed down the significance of the results to mirror the criteria that was established before the study took place. The authors themselves even say "the multitude of biases in studies that have used a birth-cohort design stress the need for additional adequately controlled, large scale studies."

And there you have it, being able to identify the significance of data based on the parameters in which the data is being examined.

Sphere: Related Content