A comment on a systematic review of the safety of vaccines published in Pediatrics

In its issue of July 1, 2014, Pediatrics (the official journal of the American Academy of Pediatrics) published a paper “Safety of vaccines used for routine immunization of US children: a systematic review”, which updated the 2011 Institute of Medicine (IOM) consensus report on vaccine safety.

The following comment was also posted on Pubmed Commons.

By the distance between what it demonstrates and what it claims, this paper (as well as the preceding IOM report this one is supposed to update) illustrates that the issue of vaccines safety is still a matter of serious concern for anyone endowed with a minimum of expertise in drug safety or pharmacoepidemiology.

To start with methodological consistency, one may wonder why the McHarm instrument used by the authors to assess the quality of the reviewed studies was not published in a peer-reviewed journal (the internet link given by the authors [their ref. 7] being not accessible) whereas the only investigations they included had to be published (in contrast with Cochrane reviews which, inasmuch as possible, also take into account unpublished investigations). This inconsistency in the authors’ way of referencing sources exposes their review to a number of biases which are well known in general, but reach unparalleled levels as far as vaccines are concerned. To take just one example, the Ascherio et al.’s study, which suggested a lack of neurotoxicity with hepatitis B vaccination, was triumphantly published in The New England Journal of Medicine (2001;344:327-32) (despite its worrying biases, some of them detailed in further correspondence) with an accompanying editorial celebrating the study as a milestone, whereas, in spite of the professional fame of its authors and of its intrinsic quality, a later study on the same subject by Hernan et al., which suggested opposite conclusions, was first rejected by The NEJM, The Lancet, The BMJ (personal communication), before being published in Neurology (2004;63:838-42) with an accompanying editorial contending that nothing significant had changed regarding the safety of this immunization: as a matter of fact, Hernan was a co-writer of both studies… Another example of selective assessment, extracted from Maglione et al.’s paper itself: it remains unclear why the study by Gallagher and Goodman (which suggests a link between hepatitis B vaccine and autism) would display “high risk of bias and low quality”, whereas the “protective effect [of some vaccines] against acute lymphoblastic leukemia” does not deserve the slightest word of caution… Dozens of additional examples of publication biases could be cited: this is the personal experience of anyone working in the field of vaccines that positive results or enthusiastic reviews are far more easily published than negative investigations or critical comments. Finally and as a number of respected authors have already emphasized, a major part of vaccines studies are performed or supported by manufacturers or governmental agencies responsible for previous recommendations, a situation which clearly maximizes conflicts of interests.

Regarding now experimental designs in vaccine studies, the following list of methodological defects is concerned with safety assessments (but could be easily extended if efficacy issues were also concerned).

  • During development, use of false placebos as comparators (i.e. not devoid of pharmacological effects: adjuvants, other vaccines) is a frequent practice.
  • Compared to the supposed duration of the beneficial immunological effects of the tested vaccines, the duration of the safety studies aimed at assessing a potential for delayed immunological hazards is often ridiculously short.
  • Required on a standard regulatory basis with any new pharmacological entities, the interactions studies are weak, scarce, if not nonexistent with vaccines, whereas most of them are now administered as combinations.
  • Likewise, the dose-ranging studies are generally defective, accounting for impressive changes in the booster recommendations once the vaccine is on the market, which would be inconceivable for any other drug.

Overall and as far as safety issues are concerned, Cochrane reviewers (e.g. in their reviews of flu vaccines) frequently identify blatant weaknesses in available studies which, apparently, are beyond of the reach of the McHarm instrument… On the contrary, Maglione et al. expressed frequent reservations about the “the strength of evidence” suggesting potential safety issues, but none about the power of the reviewed studies to effectively grasp evidence of vaccine hazards… Likewise, the authors are clearly not concerned with the tremendous tendency of vaccine studies (pre- or post-marketing) to underreport adverse events: experience suggests that physicians have a worrying reluctance to accept that vaccines might have adverse effects and, besides their indisputable tendency to brush aside any such suggestions from their patients, they do not hesitate to present as reassuring that “immunizations are up to date” when confronted with an unexplained disease, without even considering that these immunizations could have triggered the disease in question… Another illustration of the same bias: when reassuring, VAERS data are unchallenged, whereas the shortcomings of the system are immediately pointed out each time they may suggest a safety problem…

Maglione et al.’s review is regrettably silent on two crucial issues of the continuous extension of immunizations against trivial diseases:

  • maybe acceptable for one vaccination against a severe disease, the autoimmune risk related to the administration of foreign material increases arithmetically when vaccinations are multiplied beyond any sound limit;
  • extended immunizations greatly alters the natural ecology of a number of infectious diseases (e.g. measles), a situation the assessment of which would be far more complex than the basic inventory of straightforward side-effects which, as mentioned above, is already severely defective from a methodological point of view: no reason to believe that vaccine promoters are better in complex assessments than in trivial ones…

As is easy to document, the never-ending extension of immunizations against anything is based upon the dramatization of anecdotic stories, sometimes tragic but fairly rare or even exceptional at a community scale. Yet, experience of drug assessment suggests that below frequencies of, at best, 1-2% of exposed patients, clinical trials fail to identify drug side-effects with a minimum of reliability (the statistical power of postmarketing surveillance being even lower by far). In a country like the USA, this detection threshold is consistent with a shadow area on iatrogenic risk of about 40,000-80,000 persons per vaccine for each vaccinated class of age: it should be obvious that risk-taking of such a size is simply disproportionate to the potential benefits of reducing the morbidity of trivial diseases (even taking into account the natural tendency of vaccine promoters to exaggerate the efficacy of immunizations…). The stubborn obfuscation of this evident arithmetical imbalance by health professionals or governmental agencies suggests that there is something rotten in the kingdom of immunization…