Anders Hviid, DrMedSci; Jørgen Vinsløv Hansen, PhD; Morten Frisch, DrMedSci; Mads Melbye, DrMedSci
Financial Support: By the Novo Nordisk Foundation and the Danish Ministry of Health.
Disclosures: Dr. Hviid reports grants from Novo Nordisk Foundation during the conduct of the study. Dr. Hansen reports grants from Novo Nordisk Foundation during the conduct of the study. Authors not named here have disclosed no conflicts of interest. Disclosures can also be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M18-2101.
Editors' Disclosures: Christine Laine, MD, MPH, Editor in Chief, reports that her spouse has stock options/holdings with Targeted Diagnostics and Therapeutics. Darren B. Taichman, MD, PhD, Executive Editor, reports that he has no financial relationships or interests to disclose. Cynthia D. Mulrow, MD, MSc, Senior Deputy Editor, reports that she has no relationships or interests to disclose. Jaya K. Rao, MD, MHS, Deputy Editor, reports that she has stock holdings/options in Eli Lilly and Pfizer. Catharine B. Stack, PhD, MS, Deputy Editor for Statistics, reports that she has stock holdings in Pfizer, Johnson & Johnson, and Colgate-Palmolive. Christina C. Wee, MD, MPH, Deputy Editor, reports employment with Beth Israel Deaconess Medical Center. Sankey V. Williams, MD, Deputy Editor, reports that he has no financial relationships or interests to disclose. Yu-Xiao Yang, MD, MSCE, Deputy Editor, reports that he has no financial relationships or interest to disclose.
Reproducible Research Statement:Study protocol, statistical code, and data set: Not available.
Corresponding Author: Anders Hviid, DrMedSci, Department of Epidemiology Research, Statens Serum Institut, Artillerivej 5, 2300 Copenhagen S, Denmark; e-mail, email@example.com.
Current Author Addresses: Drs. Hviid, Hansen, Frisch, and Melbye: Department of Epidemiology Research, Statens Serum Institut, Artillerivej 5, 2300 Copenhagen S, Denmark.
Author Contributions: Conception and design: A. Hviid, M. Melbye, J. Vinsløv Hansen.
Analysis and interpretation of the data: A. Hviid, M. Melbye, J. Vinsløv Hansen.
Drafting of the article: A. Hviid.
Critical revision for important intellectual content: M. Frisch, M. Melbye, J. Vinsløv Hansen.
Final approval of the article: M. Frisch, A. Hviid, M. Melbye, J. Vinsløv Hansen.
Provision of study materials or patients: M. Frisch.
Statistical expertise: J. Vinsløv Hansen.
Obtaining of funding: A. Hviid.
Administrative, technical, or logistic support: M. Frisch, M. Melbye.
The hypothesized link between the measles, mumps, rubella (MMR) vaccine and autism continues to cause concern and challenge vaccine uptake.
To evaluate whether the MMR vaccine increases the risk for autism in children, subgroups of children, or time periods after vaccination.
Nationwide cohort study.
657 461 children born in Denmark from 1999 through 31 December 2010, with follow-up from 1 year of age and through 31 August 2013.
Danish population registries were used to link information on MMR vaccination, autism diagnoses, other childhood vaccines, sibling history of autism, and autism risk factors to children in the cohort. Survival analysis of the time to autism diagnosis with Cox proportional hazards regression was used to estimate hazard ratios of autism according to MMR vaccination status, with adjustment for age, birth year, sex, other childhood vaccines, sibling history of autism, and autism risk factors (based on a disease risk score).
During 5 025 754 person-years of follow-up, 6517 children were diagnosed with autism (incidence rate, 129.7 per 100 000 person-years). Comparing MMR-vaccinated with MMR-unvaccinated children yielded a fully adjusted autism hazard ratio of 0.93 (95% CI, 0.85 to 1.02). Similarly, no increased risk for autism after MMR vaccination was consistently observed in subgroups of children defined according to sibling history of autism, autism risk factors (based on a disease risk score) or other childhood vaccinations, or during specified time periods after vaccination.
No individual medical charts were reviewed.
The study strongly supports that MMR vaccination does not increase the risk for autism, does not trigger autism in susceptible children, and is not associated with clustering of autism cases after vaccination. It adds to previous studies through significant additional statistical power and by addressing hypotheses of susceptible subgroups and clustering of cases.
Novo Nordisk Foundation and Danish Ministry of Health.
Study flow diagram.
Table. Study Population Characteristics*
Cumulative incidences of autism (unadjusted and with 95% CI bands) in 657 461 children born in Denmark between 1 January 1999 and 31 December 2010, by vaccination status and age.
Association between measles, mumps, rubella vaccination and autism in subgroups of 657 461 children born in Denmark between 1 January 1999 and 31 December 2010.
Fully adjusted autism hazard ratios comparing children who received measles, mumps, rubella vaccine with those who did not. The hazard ratios are depicted on a logarithmic scale, with bars representing 95% CIs. P values are from a test of homogeneity of effect. DTaP-IPV/Hib = diphtheria, tetanus, acellular pertussis, inactivated poliovirus, and Haemophilus influenzae type b vaccine; NA = not applicable.
The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.
March 6, 2019
MMR and autism study is fundamentally flawed
Vaccines are not created equal
The purpose of publishing this study is to use the resultsworldwide. The assumption is that the MMR vaccinecovered in the study is representative of vaccines usedworldwide. But that assumption is false. There are nospecifications that control the thousands of antigens inthe vaccine from the chicken embryo cell culture used togrow the viruses or excipients such as gelatin(non-target antigens (NTA)).
We have seen repeatedlythat adverse events can becaused by NTA.
Gelatinor egg protein containing vaccines caused thedevelopment of allergy to those antigens. (1) Thelatest example is Pandemrix vaccine induced narcolepsy.Pandemrix manufactured in Europe contained largeramounts of H1N1 nucleoproteins (an NTA) and resulted inway more cases of narcolepsy, compared to Arepanrixmanufactured in Canada (2).
Eggprotein amounts in vaccines vary by orders of magnitude,among vendors, batches and over time. (1)
Bovine casein content varied 2-fold in just eight vaccine samples.
Engerix vaccine contains 500% of the yeast proteins as the Recombivax vaccine.
Due to such variation among vaccines, the study results are rendered inapplicable and the purpose of the study is defeated.
There have been many developments in autism research in thelast two decades that the authors have ignored.Antibodies directed against folate receptor alpha,GAD65, glutamate receptors and other antigens, in the child or the mother, can cause autism. Multiple vaccines can induce these antibodies in the child or the mother (3).
Cow’smilk protein containing vaccines (DTaP) cause the vastmajority (75%) of autism cases (4). A subset of autismcan be caused by GAD65 antibodies that are induced byGAD65 proteins in the MMR vaccines. GAD65 can be of chick (cell culture) or animal (gelatin) origin.Maternal antibodies can affect the fetal brain (3). Epidemiological studies cannot account for these variables.
The US IOM committee in their 2011 report (5) wrote:
“The committee concluded the evidence convincingly supports 14 spe-cific vaccine–adverse event relationships. In all but one of these relation-ships,the conclusion was based on strong mechanistic evidence with the epidemiologic evidence rated as either limited confidence or insufficient.”
So in an overwhelming 93% of cases, mechanistic studies provided convincing evidence and epidemiological studies failed.
This study is therefore flawed, misleads and must be retracted.
1.Arumugham V. Evidence that Food Proteins inVaccines Cause the Development of Food Allergiesand Its Implications for Vaccine Policy. J DevDrugs. OMICS International; 2015 Oct;04(04):1–3.
2.Godlee F. A tale of two vaccines. BMJ. 2018 Oct4;363:k4152.
3.Arumugham V. Vaccines and Biologics injury tablebased on mechanistic evidence – Mar 2019[Internet]. 2019. Available from: https://doi.org/10.5281/zenodo.2582634
4.Arumugham V, Trushin M V. Autism pathogenesis:Piecing it all together, from end to beginning …. J Pharm Sci Res. 2018;10(11):2787–9.
5. Clayton EW,Rusch E, Ford A, Stratton K. Adverse Effects ofVaccines:: Evidence and Causality. NationalAcademies Press; 2012.
Professor Keele University
March 7, 2019
Our recent research demonstrating extremely high levels of aluminium in brain tissue in autism combined with the unique observation of the association of aluminium with immune cells and microglia in brain tissue highlights a possible role for aluminium in autism. Did the authors include prior exposure to aluminium as a factor determining possible susceptible groups? For example, vaccination including an aluminium adjuvant would constitute exposure to aluminium and especially in infants. So when different groups have been compared did you take into account their previous exposure to aluminium, for example in vaccinations that include an aluminium adjuvant. One reason to ask this would be the very real possibility that the MMR vaccine, which does not include an aluminium adjuvant, might catalyse the uptake into brain tissue of aluminium from the periphery in immune cells such as lymphocytes et c.. I would be interested to hear the authors' comments on this, thank you.
Jeffrey A. Hirschfield, MD CPI
Score Physician Alliance
Study conclusions unsupported.
This retrospective Danish database registry and a lack of formal chart review produces poor quality data. The lack of a comprehensive multiaxial assessment approach using CPT10 and a failure to perform chart reviews are a prerequisite to claims made by the authors. I am concerned about the authors' potential conflicts of interest. I would request that an independent panel review this study for its inclusion into a scientific database as this does not meet the gold standard necessary for inclusion.
Jake Crosby, MPH
March 8, 2019
Conflict of Interest:
Received funding from Autism Media Channel and Children's Medical Safety Research Institute
Study supports MMR-autism link in sibling subset
As with their 2002 NEJM study, the coauthors' conclusion is contradicted by their results. The Hazard Ratio that stands out in Figure 3 is MMR risk in children with ASD siblings. The HR from MMR was 2.64 among children with siblings who have autism. Although the magnitude is far greater than any other reported HR in this figure, it is not significant because of the small sample of this group. It would therefore be preferable for the authors to instead report the risk for autism from having a sibling with autism but no MMR vaccination and then the risk for autism from having both an autistic sibling and MMR vaccination. The increased risk in both groups would both be strong and significant, but the magnitude would be substantially higher in MMR+ASD sibling compared to no MMR+ASD sibling. The study's results are supportive of an MMR-autism link in the sibling subset with sibling status being an important effect modifier. Yet the conclusion states the opposite. Not surprisingly, the corresponding author of this study did not respond to this reader's concern in email.After the coauthors' 2002 study was published, a letter to NEJM by an epidemiologist at McGill University suggested they had artificially eliminated the association between MMR and autism by adjusting for age.(1) He suggested that age was an effect modifier and that the authors compare age-stratified rates between subjects 24-29 months post-MMR vaccination to non-MMR vaccinated. His letter was never published or responded to, and his advice was never heeded. We can now add ASD sibling status as yet another effect modifier that the authors dismissed to dismiss the very real MMR vaccine-autism link.1. Stott C, Blaxill M, Wakefield AJ. MMR and Autism in Perspective: the Denmark Story, J Am Phys Surg. 2004, 9,3, 89-91.http://www.jpands.org/vol9no3/stott.pdf
W John Martin
Institute of Progressive Medicine
Potential Misinterpretation of Epidemiological Studies
Epidemiological studies are inherently flawed if multiple factors contribute to the disease in question. In pursuing the possible adverse role of vaccines, it is necessary that the comparisons groups be equally matched for all influencing factors. In particular, the studies do not exclude possible unique reasons why some of the children have remained unvaccinated. If one or more of these reasons leads to an increase in autism, they could easily mask any autism inducing influence of the vaccine. The occurrence of an illness following vaccination warrants a detailed analysis of the affected individuals. One topic that the authors have failed to consider is the existence of vaccine-derived, stealth adapted viruses. These viruses are not effectively recognized by the cellular immune system because of deletions or mutations in the genes coding for the relatively few major virus components, which are normally targeted by the cellular immune system. Minor virus components may become immunogenic if the reactivity of the immune system is boosted with adjuvants or live vaccine virus.DNA sequence data have unequivocally established an African green monkey cytomegalovirus origin of the stealth adapted viruses isolated from a patient with the chronic fatigue syndrome and from another patient who died following a bipolar psychosis (1, 2). This topic should be covered in any discussion of vaccine safety. W. John Martin, MD, PhD. 1. Martin et al. (1995) African green monkey origin of the atypical cytopathic 'stealth virus' isolated from a patient with chronic fatigue syndrome Clinical Diagnostic Virology 4(1):93-103.2. Martin et al. (1996) Simian cytomegalovirus-related stealth virus isolated from the cerebrospinal fluid of a patient with bipolar psychosis and acute encephalopathy. Pathobiology 64(2): 64-6.
Dr Manimalar Selvi Naicker
University of Malaya
Hypothesis Needs Clarification
Vaccine hesitancy/rejection is a growing problem.I think the vexing part (for both doctors and parents) of the debate is this:1) Are we asking if MMR can cause Autism at all?In which case our evidence will almost necessarily be restricted to the occasional case reports and Vaccine Adverse Events registry reports of WELL -INVESTIGATED patients.OR2) Are we asking if MMR can cause Autism with a certain large enough probability that it can be picked up on large epidemiological studies (epidemiology being a "blunt tool")?This paper follows a long line of papers using Epidemiological research methods to prove/disprove the MMR-Autism link (risk/association etc).My concern with regard to continued "Epidemiological" research is this:There are a few things we know from previous studies (the stats I am quoting are approximates).1) Non-MMR-Autism in the population is about eg 2% (varying from 1.5 to 2.5)2) MRR-Autism (if it exists at all) is probably rarer than 1/1000000 ( this is because MMR induced encephalitis is 1/1000000 so MMR-Autism must be even rarer. In fact, in spite of so many studies it has never been picked up at all!). But lets fix it as 1/1000000=0.0001% for the sake of this argument.Even if we do an RCT (gold standard) with 2 million children: Given MMR (1M) vs Not given MMR(1M)No-MMR group= 2% children get autismMMR group = 2%+0.0001%=2.0001% get autism (that is 1 one child out of the million got MMR-Autism) and (20,000 out of a million got Non-MMR-Autism). BUT we don't know who got which type.Will (2.0001- 2.0) be "statistically significant"? Likely not.Will (2.0001-2.0) be "clinically significant"- I think not.Why?1) because we know that autism varies from 1.5% to 2.5% within a normal population and 2.0001% is within that normal range. 2) There is bound to be Random Error in the distribution of baseline Non-MMR autism as the cases get distributed to each group, so both groups won't get EXACTLY 2%3) MMR group with 2.0001 % and Non-MMR group with 2% are almost perfectly balanced.So, quite naturally we cannot conclude that there is any association between MMR and Autism even though 1 child did have MMR-linked Autism in our exampleIn short, rare causal events can be masked by natural variation in the incidence of the diseaseFor us to be able to pick up MMR-Autism on an epidemiological type study it has to be pretty much well and consistently OUTSIDE the 2.5 range.Or at least we must consistently see a larger difference between the two groups.Dr Manimalar Selvi Naicker(MBBS, M.Path (Histopathology), M.MedStats)
Robert W. Stoddard II
Carnegie Mellon University
Could I try Causal Learning to this Data?
I am practiced in both causal search and estimation and have a personal desire to use that technology to qualify these results and convince my grandchildren to get vaccinated.
The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China.
March 11, 2019
Conflict of Interest:
Measles, Mumps, Rubella Vaccination and Autism: insufficient evidence for susceptible children
I read with great interest the article by Hviid et al . This nationwide cohort study showed that measles, mumps, rubella (MMR) vaccination does not increase the risk for autism. However, we should keep in mind that although with large sample size (n=657 461), the nature of the study design (i.e., observational design) make the conclusion not definite. The controversy cannot be put to an end after this study. Leaving aside the study design, obvious misinterpretation was made when the authors made the conclusion.Since there was a hypothesized link between the MMR vaccine and autism, most of us are much more concerned to know if MMR vaccination could increase the risk for autism in susceptible children. The study showed that MMR vaccination was associated with a hazard ratio of 2.69 (95% confidence interval, 0.58-12.43) in subgroup of siblings with autism. The hazard ratios and 95% confidence intervals were 0.98 (0.84-1.13), 0.89 (0.78-1.01), and 0.89 (0.45-1.77) in subgroups of no siblings with autism, no siblings, and father’s ID missing, respectively. The authors concluded that the study strongly supports that MMR vaccination does not trigger autism in susceptible children because the hazard ratio of 2.69 was accompanied by a wide confidence interval with no statistical significance. One reader recommend the authors to report the risk for autism from having a sibling with autism but no MMR vaccination and the risk for autism from having both an autistic sibling and MMR vaccination. This reader suggested that the study's results are supportive of an MMR-autism link in the subgroup of siblings with autism. I think that it was neither able to conclude there was an MMR-autism link in the sibling subgroup, nor to conclude that there was no MMR-autism link in the sibling subgroup. In the study, there were only 37 children in group of siblings with autism. No matter what result is and if there was a statistically significant result, such low sample size means there was insufficient evidence to make a conclusion between MMR vaccination and autism risk in susceptible children (i.e., children with siblings with autism).References1. Hviid A, Hansen JV, Frisch M, Melbye M. Measles, Mumps, Rubella Vaccination and Autism: A Nationwide Cohort Study. Ann Intern Med. 2019 Mar 5. doi: 10.7326/M18-2101. [Epub ahead of print]
Anders Hviid, Jørgen Vinsløv Hansen, Morten Frisch, Mads Melbye
Statens Serum Institut
March 9, 2019
To the Editor:We appreciate the massive, global attention our paper has received and have carefully evaluated all comments. Below, we address the major claims made by some of our critics.Mr. Arumugham lists a number of specific biological mechanisms hypothesized to be involved in the etiology of autism, and he rightfully claims that epidemiological studies are not capable of addressing such mechanisms. However, Mr. Arumugham ignores the central message from our present and prior well-conducted epidemiological investigations, namely that regardless of which biological mechanisms lead to autism, there is no difference in autism risk between MMR-vaccinated and MMR-unvaccinated individuals. In this light, it seems illogical and futile to continue the quest to understand the biology of autism by focusing on theoretical adverse effects of a vaccine that seems, rather compellingly, to be safe both overall and, specifically, in relation to autism risk.Professor Exley wants to know if we took prior exposure to aluminium, for example in vaccinations that include an aluminium adjuvant, into account. Indeed, we did. According to Figure 3 in the paper, MMR vaccination was not associated with autism risk, whether children had been vaccinated with aluminium-adjuvanted vaccines (DTaP-IPV/Hib) in infancy or not. In response to Mr. Crosby’s comment it should be noted that sibling status at 1 year of age was not an effect modifier (Figure 3, p>0.20), and the confidence interval for the “Siblings with autism” effect estimate overlaps HR=1.00. In Figure 2 of the Supplementary Material, we transparently provide the unadjusted cumulative incidences according to age and vaccination status, and stratified according to child’s sex and sibling history of autism; these supplementary analyses do not support an association in children with autistic siblings. Sibling status at 1 year of age is a conservative measure of “Siblings with autism” with only 5 unvaccinated and 32 vaccinated cases (Figure 3). If we consider sibling history as a time-varying variable, that is, each child’s “sibling status” is updated during the study period, e.g. if a sibling is diagnosed with autism, we gain statistical power. In this analysis, MMR vaccination was also not associated with autism risk in children with autistic siblings (HR, 1.15, 95% CI, 0.71-1.87) (see Results).The further claim by Mr. Crosby that we “artificially eliminated” an association between MMR vaccination and autism by adjusting for age in our 2002 study is false and reveals an incomplete understanding of survival analysis. Our 2002 study and the current study share many similarities in study design, and thus, the concept of confounding by age is clearly illustrated in Figure 1 of the Supplementary Material, where the distribution of time and autism cases according to vaccination status and age is visualized; here we show that unvaccinated children are younger, and since younger children have lower rates of autism the crude rate ratio is confounded by age. Consequently, adjustment for age is required. In Figure 2, we present cumulative incidences according to age; there is no suggestion of any age interval where vaccinated children have a higher rate of autism than unvaccinated children, and a test of homogeneity yielded a P-value of 0.138 (see Results).In conclusion, we maintain that we have analyzed the current study as well as our 2002 study using appropriate statistical methods, finding no support for an association between MMR vaccination and autism risk, a lack of an association that also applied in children with autistic siblings. Sincerely,Anders Hviid, Jørgen Vinsløv Hansen, Morten Frisch and Mads Melbye
March 16, 2019
Conflict of Interest:
Received funding from Autism Media Channel and Children's Medical Safety Research Institute
Authors' response does not answer questions about either study
Questions about the authors’ current and 2002 study remain unanswered. The statistical insignificance of autism risk from MMR vaccination in the sibling group was already acknowledged as being due to small sample size. An anomaly of such elevated magnitude in Figure 3, however, requires further investigation.
For example, what is a child’s risk of ASD associated with having an ASD sibling with no MMR vaccination? Now what is the risk associated with the combination of having an ASD sibling and an MMR vaccination? There is clear evidence of an interaction. Yet nothing in the study or supplementary material answers these questions, including the supplementary figure referred to by the authors where the sibling sample is reduced to even smaller groups.
Analyzing sibling status as a “time-varying variable” is also inappropriate when sibling status is supposed to be treated as a marker for innate susceptibility. That susceptibility would be present from birth; it wouldn’t begin with the diagnosis of the sibling.
Also unaddressed are the points made about the 2002 study in the unpublished letter to NEJM by McGill University epidemiologist Dr. Samy Suissa. In his letter, Dr. Suissa demonstrates that the age variable is an effect modifier that was inappropriately treated as confounder in the study. Here is the portion of his letter explaining what the study did wrong:
“Indeed, the rates of autistic disorder by age at vaccination, although not the age at follow-up, are 18.9, 14.8, 24.6, 26.9 and 12.0 per100,000 per year respectively for ages <15, 15-19, 20-24, 25-35 and >35 months. These rates are all above the overall rate of 11.0 for the reference group of no vaccination, over all ages. It is then somewhat implausible for the adjusted rate ratio to fall below 1, unless the risk profile by age in the unvaccinated is vastly different than in the vaccinated (effect-modification). In this case, the adjustment for age could have been artificial. It would be useful then to present rates on subjects 24-29 months since vaccination and on the unvaccinated (crude rate ratio 2.5) stratified by age. Otherwise, one could be tempted to conclude that the figure is in fact suggestive of an association between MMR vaccination and the risk of autism.”(1)
None of the figures from the current study provide the measurements Dr. Suissa suggested. Moreover, those figures are not even for autistic disorder but rather for the total category of ASDs in general.
1. Stott C, Blaxill M, Wakefield AJ. MMR and Autism in Perspective: the Denmark Story, J Am Phys Surg. 2004, 9,3, 89-91.
March 20, 2019
To the Editor:Mr. Crosby repeats the claim that the age-adjustment in our original 2002 paper has “artificially eliminated” an association between MMR and autism. He references an unpublished letter by Dr. Suissa, claiming that the rates in Table 2 of our 2002 paper and our adjusted RRs are incompatible and that this might reflect effect modification. This is incorrect. From Table 2 in our 2002 paper, we can calculate the crude relative risk (RR) of autistic disorder as 1.45 (95% CI, 1.08-1.95). Adjusting for age yields a RR of 0.91 (0.68-1.23). As explained previously, the discrepancy between the crude and age-adjusted RRs is due to the follow-up in the unvaccinated group comprising person-time from 1 year of age until age at MMR vaccination from all the MMR vaccinated children. Now, if we start follow-up at 2 years of age instead of 1 year of age we get a crude RR of 0.90 (0.66-1.22) and an age-adjusted RR of 0.88 (0.65-1.19). Clearly, starting follow-up at an age where the unvaccinated and vaccinated groups have similar age distributions yields almost identical results as the main result in the paper; we maintain that adjustment for age is both appropriate and necessary with follow-up from 1 year of age. Similarly, for our current study, we repeat our previous response: In Figure 2, we present cumulative incidences according to age; there is absolutely no suggestion of any age interval where vaccinated children have a higher rate of autism than unvaccinated children, and a test of homogeneity yielded a P-value of 0.138 (see Results). Mr. Crosby claims that there is clear evidence of an interaction between sibling status and MMR vaccination. We tested this and found no such evidence P>0.20 (see Figure 3). Mr. Crosby also claims that analyzing sibling status, as a time-varying variable is inappropriate. This, too, is not correct. By continually updating sibling status in the event of a sibling receiving an autism diagnosis, all information this variable can supply as a proxy for genetic susceptibility is utilized. Our claim that the two approaches measure the same attribute with differing degrees of statistical precision is supported by a comparison of autism HRs: status at study start HR 7.32 (5.29-10.12) and time-varying status HR 7.64 (6.77-8.62) (Table 3, Supplementary Material). In conclusion, we maintain that we have analyzed the current study and our 2002 study using appropriate statistical methods, finding no support for an association between MMR vaccination and autism risk, a lack of an association that also applied in children with autistic siblings.Sincerely,Anders Hviid, Jørgen Vinsløv Hansen, Morten Frisch, Mads Melbye.
March 12, 2019
Authors’ inability to address vaccine content variation effect and lack of control for multiple mechanisms is very troubling
In their response the authors were unable to address my point about the effect of wide variation in the unregulated non-target antigen content of vaccines. They should therefore accept that this study has no predictive value and misleads the scientific community. The study conclusions may not even apply to Denmark (much less the rest of the world), as the current vaccines in use there may not have the same characteristics as the ones administered during the study period.
Regarding mechanisms, the authors write: “regardless of which biological mechanisms lead to autism, there is no difference in autism risk between MMR-vaccinated and MMR-unvaccinated individuals.”
Not true. What if more mothers of children in the MMR-unvaccinated group were making maternal autism related antibodies? What if the MMR-unvaccinated group received more cow’s milk protein containing vaccines? That will mask MMR induced autism. (1)
As previously noted, 75% of autism cases are caused by cow’s milk protein containing vaccines such as DTaP, Prevnar 13, Hib. So MMR’s contribution can be easily masked if all other contributors are not carefully controlled.
As I wrote previously, per the US IOM report, epidemiological studies are already weak and rarely provide convincing evidence. On top of that when you fail to control for multiple biological mechanisms, it leads to type II errors, with devastating consequences for millions.
The authors write: “it seems illogical and futile to continue the quest to understand the biology of autism by focusing on theoretical adverse effects of a vaccine”
Filtering the lines of scientific inquiry of autism mechanisms using such flawed and misleading epidemiological studies makes no sense. It is the reason why after billions and decades wasted (2,3), we do not have “official” root cause for autism. This approach is wrong, unscientific and extremely dangerous.
The correct approach to this problem is the far more reliable mechanistic approach involving bioinformatics analysis and autoimmune serology as suggested by Wraith et al. (4)
Further, as others have noted, declaring conflicts of interests and funding does not erase their undesirable impact on the study’s conclusions (5).
1. Arumugham V. Epidemiological studies that ignore mechanism of disease causation are flawed and mechanistic evidence demonstrates that vaccines cause autism [Internet]. 2017. Available from: https://doi.org/10.5281/zenodo.1041905
2. Glasziou P, Chalmers I. Research waste is still a scandal-an essay by Paul Glasziou and Iain Chalmers. BMJ. England; 2018 Nov;363:k4645.
3. Gyles C. Skeptical of medical science reports? Can Vet J = La Rev Vet Can. Canadian Veterinary Medical Association; 2015 Oct;56(10):1011–2.
4. Wraith DC, Goldman M, Lambert P-H. Vaccination and autoimmune disease: what is the evidence? Lancet (London, England). England; 2003 Nov;362(9396):1659–66.
5. Opinion | Transparency Hasn’t Stopped Drug Companies From Corrupting Medical Research - The New York Times [Internet]. [cited 2019 Jan 22]. Available from: https://www.nytimes.com/ 2018/09/14/opinion/jose-baselga-research-disclosure-bias.html
The issues remain
The authors’ use of relative risks for two years of age at the beginning of follow-up in their 2002 study is a flawed comparison. MMR vaccination is not routinely administered until after the first year of life. Beginning follow-up at two years of age would disproportionately cut MMR-unvaccinated person-time compared to MMR-vaccinated person-time. The result would yield a greater incidence rate in the MMR-unvaccinated group and a lower relative risk for autism in the MMR-vaccinated compared to the MMR-unvaccinated. If beginning follow-up at age two explains the higher risk for autistic disorder in the MMR-vaccinated, then why does the incidence of autistic disorder peak in children who received the MMR vaccine between 25-35 months of age?
Still unaddressed are the disparities in autistic disorder incidence for children by age at MMR vaccination. Also unaddressed is why the incidence for every one of these age groups exceeds the incidence in the MMR-unvaccinated. The supplementary figure from the current study does not explain these disparities in incidence of autistic disorder. The figure does not even display a graph of the incidence of autistic disorder. Age-stratified incidence rates for the MMR-unvaccinated group and the MMR-vaccinated group around the time since vaccination where incidence for autistic disorder peaked would help explain these incidence disparities. That is why Dr. Suissa suggested these stratifications.
All the authors’ supplementary HRs show is that a child’s risk for autism with an ASD sibling is the same regardless of when that sibling is diagnosed. However, children’s siblings may not be diagnosed until after the children’s MMR vaccinations. Those children will be counted and followed up as if they have no ASD siblings until the siblings are diagnosed; this may not even be until after the children themselves are diagnosed. This is not an appropriate method to measure a susceptibility variable. Even though there were nearly 18,000 children in the current study with no MMR vaccination and only non-ASD siblings, the authors chose for their reference group a sample of just 79 participants (Table). The risk of having ASD should have been separately analyzed in the group with an ASD sibling but no MMR vaccination as well as in the group with a combination of both.
The authors’ conclusions are not supported by the results they generated using the methodologies that were employed in either study.
Anthony R. Mawson
Department of Epidemiology and Biostatistics, School of Public Health, Jackson State University
March 28, 2019
The Measles, Mumps and Rubella (MMR) Vaccine and Autism: Commentary
In a nationwide Danish cohort study, Hviid et al. estimated the risk of autism according to MMR vaccination status. They reported a nonsignificant fully-adjusted autism hazard ratio of 0.93 (95% CI, 0.85 to 1.02). They also compared autism rates in those who received the MMR plus two or more doses of DTaP-IPV/Hib, and those who received two or more doses of the latter but without the MMR. They found no difference in autism rates and concluded that MMR vaccination does not increase the risk for autism. However, comparing autism rates among children receiving and not receiving the MMR would not rule out a possible contributory role for the MMR vaccine in causing autism. Many factors, including the MMR and other vaccines, could be combining with as yet unidentified host and environmental factors to cause autism. The effect of these analyses was therefore to compare autism rates in two highly vaccinated groups, thereby obliterating the potential impact of vaccines in general. Comparisons should instead be between fully vaccinated and completely unvaccinated groups of children, or between MMR-only vaccinated and completely unvaccinated children in terms of autism. So the question whether MMR-vaccinated children are more likely to be diagnosed with autism than non-MMR-vaccinated may not be the appropriate question to start with.Evidence now suggests that autism is not an exclusively neuropsychiatric disorder but has many co-morbid features such as allergies, middle ear infections, and numerous systemic complications.1,2 The author and colleagues found that autism and most of these other features were significantly more common in vaccinated than in unvaccinated children, and in a vaccination dose-response relationship;3 moreover, preterm birth coupled with vaccination nearly doubled the odds of a neurodevelopmental disorder.4 These observations suggest that the trigger for autism and related chronic conditions may be related to the cumulative physiological impact of multiple vaccinations administered synchronously or near-synchronously, against a background of prematurity, rather than one or more individual vaccines administered separately. Discovering a biomarker associated with this mechanism would enable potentially susceptible children to be identified in advance and exempted from vaccination.While measles is generally a mild and short-lived disease, autism is chronic and seriously disabling. Considering that rates of autism among boys are currently 3.63% compared to 0.0005% (1 in 2,000) only 50 years ago5 – the importance of continuing to research the MMR/vaccine-autism question cannot be over-emphasized, not just as a public health problem but as a national emergency. Conflict of Interest None.References1. Xu G, Snetselaar LG, Jing J, Liu B, Strathearn L, Bao W. Association of food allergy and other allergic conditions with autism spectrum disorder in children. JAMA Network Open 2018; 1(2):e180279. doi:10.1001/jamanetworkopen.2018.02792. Adams DJ, Susi A, Erdie-Lalena CR, Gorman G, Hisle-Gorman E, Rajnik M, Elrod M, Nylund CM. Otitis media and related complications among children with autism spectrum disorders. J Autism Dev Disord 2016; 46(5):1636-42. doi: 10.1007/s10803-015-2689-x.3. Mawson AR, Ray BD, Bhuiyan AR, Jacob B. Pilot comparative study on the health of vaccinated and unvaccinated 6- to 12-year-old U.S. children. J Transl Sci 2017; 3: DOI: 10.15761/JTS.1000186 4. Mawson AR, Bhuiyan AR, Jacob B, Ray BD. Preterm birth, vaccination and neurodevelopmental disorders: a cross-sectional study of 6- to 12-year-old vaccinated and unvaccinated children. J Transl Sci 2017; 3: DOI: 10.15761/JTS.10001875. Zablotsky B, Black LI, Blumberg SJ. Estimated prevalence of children with diagnosed developmental disabilities in the United States, 2014–2016. NCHS Data Brief, no. 291. Hyattsville, MD: National Center for Health Statistics. 2017.
March 29, 2019
Additional point: authors’ inappropriate reasoning about 2002 study person-time
Incomplete person-time should be the reason against delaying follow-up, not the reason for delaying follow-up. In justifying age two as the start of follow-up in their 2002 study, the authors falsely state that all MMR-vaccinated children contribute person-time to the MMR-unvaccinated group until their MMR vaccination. However, only children still at risk of autistic disorder by the time of their MMR vaccine contribute MMR-unvaccinated person-time until they receive the vaccine. Children who receive an autistic disorder diagnosis before they receive the vaccine would be justifiably counted as an MMR-unvaccinated case. The point of using person-time is to account for unequal follow-up by comparing risk of developing a disorder in persons during exposed and unexposed periods. Although the exposed at-risk are not counted as exposed before the exposure, the unexposed at-risk had the opportunity to develop the outcome before ever becoming exposed. While everyone at risk for autistic disorder at MMR vaccination would have first contributed varying quantities of MMR-unvaccinated person-time, everyone at risk had the opportunity to develop autistic disorder before MMR vaccination. That is why delaying the start of follow-up to two years of age produces a skewed measure of association. The number of cases in both groups remain the same but are now divided against incomplete person-time in each group. Since person-time is disproportionately incomplete in the MMR-unvaccinated group, the result is biased and should not be used to justify any conclusion as the authors have now done.
Asociacion Espanola de Pediatria
May 8, 2019
Both the control group and the study have previously received a cokctail of vaccines: DTaP-IPV/Hib by 12 months of age. Obviously, if some or all of the vaccines in the cocktail caused autism in susceptible kids, by the time the MMR is adminsitered all the susceptible kids would already be autistic and the extra MMR wouldn't affect much the incidence in the study group relative to the control group. Therefore the correct conclusion would be that "Later MMR vaccination does not increase the risk in comparison to a cocktail of previously administered vaccines". By omitting this last point from the conclusion the reader is misled into believing that the MMR vaccine has been proven safe in absolute terms, that is, compared to lack of vaccination. By making a grand universal statement that "MMR vaccination does not increase the risk for autism, does not trigger autism in susceptible children, and is not associated with clustering of autism cases after vaccination" the authors unduly extapolate their observations to unvaccinated subjects. Lawmakers maybe mislead into taking wrong decisions affecting public health.
R. Cavallo, PhD
May 15, 2019
Insufficient sample sizes and possible selection bias should temper conclusions
Hviid et al. (, Apr. 16 issue) emphasize that their study contributes to the existing literature by addressing hypotheses of susceptible subgroups, such as those with a sibling history of autism. On that front, the authors report (in their central associations figure) on the restricted tiny subset of children with an autistic sibling at age 1. The result is uninformative due to small sample size. It is unclear why they don’t provide the same analysis for the group of all children that ever acquire an autistic sibling; however, they do go on to leverage the broader dataset by considering sibling status as a time-varying covariate (aHR, 1.15 [CI, 0.71 to 1.87]). While that finding is pertinent to the question of whether the susceptible group’s risk is doubled or more by receiving the vaccine, it does not provide compelling evidence either way for those curious about, say, a 50% increase in risk. The authors’ claim that the study “strongly supports that MMR… does not trigger autism in susceptible children” may, in that light, be considered incommensurate with the findings.We should also be alert to the fact that the authors were unable to measure even some of the most conspicuous potential sources of selection bias, such as family history beyond the sibling level. The authors don’t remark on the issue, but if families with increased risk — say, parents who manifest the disorder or have autistic siblings, nieces, or nephews — chose to vaccinate at a markedly lower rate, that could skew results and, incidentally, offer explanatory fodder for the findings of significantly reduced hazard amongst girls and the entire 1999-2001 birth cohort. The latter in particular would cohere with a speculative narrative of greater vaccination fears  — especially in those years immediately following publication of Wakefield et al.’s infamous Lancet article — among parents aware of a genetic pre-disposition or early autism symptoms in their infants.One can hope that a solid mechanistic understanding of the disorder’s development, earlier diagnosis , or future availability of a more informative dataset can ultimately ground the underlying conclusions here. Meanwhile, we should be wary of the ill-effects that contentious evidentiary claims on either side of this issue can bring.1. A. Hviid, J. V. Hansen, M. Frisch, and M. Melbye. Measles, mumps, rubella vaccination and autism: A nationwide cohort study. Annals of internal medicine, 170(8), 513–520 (2019).2. O. Zerbo et al. Vaccination Patterns in Children After Autism Spectrum Disorder Diagnosis and in Their Younger Siblings. JAMA pediatrics, 172(5), 469-475 (2018).3. R. W. Emerson et al. Functional neuroimaging of high-risk 6-month-old infants predicts a diagnosis of autism at 24 months of age. Science Translational Medicine, 9(393) (2017).
Hviid A, Hansen JV, Frisch M, Melbye M. Measles, Mumps, Rubella Vaccination and Autism: A Nationwide Cohort Study. Ann Intern Med. 2019;170:513–520. doi: 10.7326/M18-2101
Download citation file:
Published: Ann Intern Med. 2019;170(8):513-520.
Published at www.annals.org on 5 March 2019
Infectious Disease, Neurology, Prevention/Screening, Vaccines/Immunization.
Results provided by:
Copyright © 2019 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use