The Davis et al Paper ( Arch Ped Adolesc Med 2001; 155: 354-359)

 Submitted by Walter O. Spitzer, M.D., M.P.H., F.R.C.P.C., Emeritus Professor of Epidemiology, McGill University, Montreal

  This scope of this opinion is limited to the Davis paper, to the extent that it is relevant to the objectives of the 06/19/02 congressional hearing at Washington D.C. (as summarized in Chairman Burton’s letter of invitation).

 *The Davis paper cannot evaluate the relationship between Measles-Mumps-Rubella vaccine and/or Measles-Containing Vaccine (MCV) and an increased risk of Inflammatory Bowel Disease (IBD). The paper is flawed.

 *The fatal flaw that negates the ability of the study to reach conclusions about the explored association is that it is grossly underpowered.

 *The power for the key results (Table 2) is 12%. It was not reported in the paper. We (methodologists at the Royal Victoria Hospital clinical epidemiology unit and I) calculated power from the data in the paper. In non-jargon English, a power of 12% means that one has a chance of 88% of declaring no increase in risk if indeed there was a two-fold increase in the risk.  In Table 3 of the Davis paper all power calculations for each cell demarcated by vaccination ages are similarly very low. Providing confidence intervals (CIs) is correct. But that is insufficient for the intelligent layman, the clinician, the patient and the policy-maker to detect how much power is lacking and to understand that such low power does not allow conclusions or decisions.

 *It would appear that the Davis group reached their decision about necessary sample size based on advance assumptions of exposure of 70% (from their projection overheads) compared to 30% unexposed. In fact, the published paper reports exposure of 94% among the controls, compared to only 6% unexposed.  It is this small 6% unexposed group that reduces the power.  This should have alerted the investigators, early during the field work, that the proposed sample size would be inadequate, resulting in the serious underpowering.


*Given controls with exposure of 95 %, the sample sizes that would have been needed to detect a two-fold risk increase:

-         For a power of 80% (conventional): cases 730, controls 2215;

-         For a power of 90%: cases 930, controls 2820.


*As an epidemiologist focused on direct or indirect potential associations of vaccines, particularly MMR, with autism, and, by extension, IBD in autistic persons, I find the choice of outcomes strange. Restricting the outcomes studied to Crohn’s Disease and ulcerative colitis reduces the relevance of Davis’ work to current hypotheses about vaccine-IBD associations emerging from laboratory and clinical research. 

 *A hallmark of valid science is that it is verifiable and replicable. When a research project is done with governmental funds for equipment, material, space, etc. by investigators also supported with public funds, ethics and social responsibility dictate that resources be made available to replicate the project. In epidemiology, the size of typical undertakings is large. Usually, databases created are also large. The time for total replication would often be prohibitively long. Thus, such databases should be available to competent legitimate investigators with valid concerns also on grounds of ethics and social responsibility. “Fishing expeditions” by verifiers should be prevented with a priori research plans. If necessary, the plans should be reviewed by entities other than the database owners or funders. Controversial or inconclusive findings by honest investigators can be expected to benefit from verification and/or replication. Issues dealing with confidentiality of patients, doctors and other providers have been worked out for many years. Reasonable time should elapse from original creation of data and release of initial conclusions to verification. Neutral monitoring committees are generally important in such situations.

 *All of the foregoing in the previous paragraph also applies to privately funded research with few exceptions such as patents and legitimate commercial considerations.

  *I believe that the Vaccine Safety Datalink Project data should be made available to academic, governmental and competent private research groups for all the reasons given above. In this particular case, the VSDP, a public, publicly funded resource should be open with reasonable safeguards unless matters such as national security or an epidemic justified closing access. Temples of secrecy have no place in science. Secrecy always suggests the question, “What do they have to hide?”

 *I reiterate that the case control study from the Vaccine Safety Datalink project cannot determine whether measles-containing vaccines do or do not increase the risk for inflammatory bowel disease.

  I do not have any conflicts of interest. I have no nuclear or extended family members with any relevant disease, notably an autistic syndrome.. I have not had any grants for research or for any other purpose from US or other national sources since December 2000. As required by age, I retired this year I advise families of autistic persons in the UK pro bono. I do not know Dr R.L Davis personally nor any co-author of the paper commented upon. I do not further any view on association or causation with respect to MMR as a potential determinant of autistic syndromes. I am a worried agnostic, I do not know.  I strongly believe we need to rule in or rule out a relationship. As an epidemiologist and public health doctor, my bias, if any, is that I would like to see MMR exonerated on safety, given its efficacy.  Long and short versions of my CV have been submitted separately.