It is a safe bet that there would be a lot of questions arising from the latest study, which shows no link between thimerosal exposure and autism. I thought there would be some interviews in the press covering most of the obvious questions, so I decided to ask some questions of my own of the study’s lead author, Cristofer Price of Abt associates.
I was very interested in the more complete discussion in their Techical Reports and data. I was also interested in how these results might apply to the idea that there are “too many” vaccines given “too soon”. Mostly I was interested in why this study took so long to get published give the CDC’s statements after the Thompson study of 2007—statements which indicated that this follow-on study should be available within about a year or so.
Below is the exchange:
First: you cite two Abt reports from 2009 on the subject:
Price C, Robertson A, Goodson B. Thimerosal
and Autism. Technical report. Vol I. Bethesda,
MD: Abt Associates Inc; 2009
I can’t find them on your site at this time. Are they there or will they be made available when the embargo is lifted?
[Response: The tech reports will be up on the CDC and Abt web sites on Monday. ]
Will the data be made available as was done with the Thomson(2007) study? If so, how would one access it?
[Response: Yes, the process for obtaining the data will be very much like the process that was in place for the Thompson(2007) study. Instructions for how to access the data and a data use agreement, etc. will be up on the CDC web site on Monday. The terms specified in the data use agreement are similar to those from the prior study. ]
As to the paper, I see that the results are the same for autism with and without regression. Are there any other issues of severity which were checked (e.g. level of intellectual disability, seizures) which were also monitored?
[Response: We did do a sub-analysis where AD cases with low cognitive functioning were excluded (see technical report on Monday for full details and results) Analysis of the subgroup of AD cases where children with low cognitive functioning were excluded was motivated by the following concern. Because children who are non-responsive during the assessment process are more difficult to assess, it can sometimes be difficult to determine whether children with severe developmental delay actually have autistic disorder. If the imprecision of the assessment process for such children resulted in inclusion of children without AD in the AD group, then we would expect that the estimate of the relationship of exposure to AD risk could be attenuated. Therefore, an outcome category for AD with low cognitive functioning excluded was created and its relationship to exposure was estimated. The results for this subgroup were very similar to those for the overall analysis.]
There are children (both case and control) who have 0 mercury exposure from vaccines in all categories. Are there children in all these categories who are unvaccinated?
[Response: I don’t have the answer to this handy. I know that there were a few kids in the sample that had zero vaccine receipts, but I don’t think they were in all of the categories because there were few of them. Most of the kids with 0 mercury exposure received at least some vaccines, but they were thimerosal free.]
To some extent, mercury exposure from vaccines could be used as a proxy variable for vaccine exposures. I.e. the amount of mercury would be somewhat proportional to the number of vaccines received. Are there any trends in just number of vaccines and autism? I.e. anything that would address the “too many, too soon” slogan? I do see that you discuss this somewhat on page 661
[Response: In the technical report (Volume II, Chapter 16) I show data on the the cumulative numbers of vaccines recieved as children aged. It shows that the cases and controls got the same numbers of vaccines. That chapter was not designed specifically to address your question about “too many too soon”, but it does show cases did not get more, sooner than controls.]
After Thomson(2007) came out, I recall that the CDC webpage suggested that your present study would be out in about a year. Why has this study taken so long to reach the public?
[Response: I’m not sure why the CDC web page had the overly optimistic suggestion that it would be out in about a year. To understand the timeline, I will need to explain some things about the phases of analysis, then the process of drafting the paper and getting it published. This is going to be a bit long winded, but part of it I am cutting and pasting from the technical report:
The study protocol was developed by a design group led by Abt Associates, Inc. working in close consultation with Principal Investigators from the
Centers for Disease Control and Prevention (CDC), Principal Investigators, Data Managers, and Study Managers from the each of the three HMOs, and with the study’s External Expert Consultants. Prior to recruitment and data collection, a detailed analysis plan was written for the study that specified the research questions, study design, eligibility criteria, sampling plan and target sample sizes, the form of the statistical models that would be used, the specific hypotheses to be tested, decision rules for categorizing outcome classifications, the coding of exposure variables,
the list of covariates to be used as statistical control variables, the coding of each of those variables, and decision rules for the retention or omission of each covariate in the final analysis models.
By agreement among the members of the design group, data analysis for the study was to be completed in two phases. In the first analysis phase, analysts at Abt Associates were to carry out as closely as possible the analyses specified in the plan and to do only the analyses specified in the plan. At the end of this phase, all members of the design group were invited to a meeting in Washington, DC where the first round, preliminary results were presented to the group. Prior to that meeting, the results of analyses linking exposures to outcomes had not been shared with anyone outside of Abt Associates. The second phase of analysis began with the meeting in Washington, DC. At that meeting, the design group considered the results and generated new hypotheses and questions that were to be pursued in the second phase. Over the ensuing months design team members provided written comments on the results of the preliminary analyses and made suggestions for additional analyses. The current report includes results from both phases.
The meeting in DC described in the paragraph above took place in May of 2008. We gave all of the members of the design group a couple of months to give feedback and suggestions on the analyses that they wanted in Phase II. There was a lot of back and forth there. The technical report includes results from both phases. We were well into 2009 before we (at Abt) had made it all the way through those second phase analyses. Then, drafts of the manuscript had contributions from a large number of authors (which takes a lot of time) and we sent drafts to our External Expert Consultants, made changes, replied to queries etc, then a draft had to go through CDC review which takes time, then we the publication process (getting a manuscript published in a peer-reviewed journal) takes a surprisingly long time. So, here we are in 2010.]