Here’s a hot topic constantly resurfacing in the news, especially with the Omnibus currently proceeding at the US Court of Federal Claims, to wit: Is autism caused by vaccines? I won’t pretend that I’m going to capture everything in this controversy; there are too many players in the drama. (Autism Diva is keeping track of the daily news on the hearing.) However, this does make for an excellent case study in the scientific method. We get to look at concepts like incidence & prevalence, correlation vs causality, testimonials vs evidence-based medicine, and some general concepts in epidemiology. Could we possibly have any more fun?! (tongue-in-cheek joke)
When you read about autism, something noted most everywhere is the increasing numbers of children diagnosed. Surely, people say, there has to be something causing that to happen!
The whole vaccines-causes-autism story starts back in 1998,
when there was a research study in England by Dr Andrew Wakefield’s research group, touting a possible link between the MMR and autism. Additionally, in 2003 Dr Mark Geier and his son David came up with idea that no, it was the Thimerosal (a mercury compound used as a bactericide) in the vaccines that causes autism, which theory was later highly sensationalized as evidence for a big government conspiracy in David Kirby’s book, Evidence of Harm, and has been touted on various Web sites. There is also a small but vociferous number of parents who insist that their child was not afflicted with autism until vaccinated.
INCIDENCE & PREVALENCE
Epidemiology can be really tricky. We have to distinguish between the incidence of new cases, and the prevalence or overall number of cases in the population, and we also have to look at how things are diagnosed and recorded. For example, if you examine old historical records for tuberculosis, you won’t find much because it used to be called “consumption”, “scrofula”, “white plague” or several other things. It’s not that tuberculosis didn’t exist, but that it was identified differently, and people didn’t keep the same kinds of records at different times and places.
Likewise, the definitions of autism have changed and been expanded over the years to include a much broader spectrum of identifiers and labels (including Asperger’s Syndrome, which was added to the ICD10 in 1992, and the DSM-IV in 1994). Children that used to be diagnosed with conditions such as mental retardation or childhood schizophrenia would now be diagnosed as autistic. The diagnostic rates (which are not necessarily the prevalence rates) likewise increased from the earliest estimates for narrowly-defined autism at 5/10,000 to current estimates for broadly-defined autism at 1/166. (As Lorna Wing and David Potter point out, “…it is very difficult to make comparisons among studies done by different workers, at different times, in different places, using different definitions, and different methods of case finding and examination.”)
It is true that we are seeing more children (including autistic children) needing special education services. Because of changes in legal requirements for public schools (in the UK, the Disability Discrimination Act of 1995; in the US, the section 504 of the 1973 Rehabilitation Act, and the IDEA), schools now have to accept all children, regardless of disability, and to provide services for them. Formerly many children with disabilities were kept at home, were placed in institutions, or were unidentified and simply struggled through school. (Just ask adults with AD/HD how much fun school was.) School staff, parents and doctors are also more familiar with various disabilities that can affect learning (it’s hard to pick up a parenting magazine without running into articles on this or that), and consequently children are more likely to be assessed.
So now there are more children getting diagnoses, and also more students finally getting services because those services now exist. Sometimes the diagnosis a child receives depends partly on which sorts of services will likely be more useful for them, especially if a child has several conditions. It’s also interesting to note that over time, in some districts there appears to be some diagnostic substitution, where increase in autistic students roughly compares to the decrease in students diagnosed with other learning disabilities. It’s also one of those facts of bureaucracy that people don’t often keep records unless they have to, so a lack of historical records doesn’t necessarily indicate a lack of prevalence — just a lack of records. (This is what we mean by “absence of evidence doesn’t mean evidence of absence”.
The Wakefield study was alarming, and made a big news-splash, which in turn helped paved the way for the hysteria created by the Geiers’ paper. One of the key features to the scientific method is repeated tests of hypotheses, so further studies looking a the Thimerosal connection were done in Sweden, Denmark, England, Canada and the United States. None of those supported the mercury-autism hypothesis. In fact, since 1993, vaccines in Sweden and Denmark have not contained Thimerosal, but those countries continue to have new cases of autism being diagnosed. Likewise, to help reduce overall exposure to mercury from various sources, all pediatric vaccines have been free of Thimerosal since 2001, but this has not decreased the incidence of new autism diagnoses in the US either. A very recent US study also shows no connection between autism and RhIg (which also used to contain Thimerosal) and is given to mothers with Rh- blood type during their third trimester of pregnancy. Over in the US, in 2000, the Institute of Medicine (IOM) at the National Academy of Sciences, by request from the Centers for Disease Control and Prevention (CDC) and the National Institutes of Health (NIH), and the American Academy of Pediatrics (AAP), both examined the evidence for the MMR as a cause of autism, and dismissed the claims. They also noted that the timing of vaccinations, including the MMR, did not have any effect on the age of diagnosis. In Japan, the MMR was dropped in 1993 in favor of individual measles, mumps and rubella vaccines, and there has not been a decrease in the rate of autism following that change.
CORRELATION VS CAUSALITY
Question: does washing your car make it rain? It’s a funny joke because it’s absurd. But although one event might happen after another event, but that doesn’t mean the first caused the second. But it’s easy to understand how parents could believe that vaccinations caused their child to be autistic, as vaccinations are major and memorable series of events in a child’s life, and occur during the same time period as when many children are diagnosed with autism. However, correlation does not equal causality, post hoc ergo propter hoc. Vaccinations do not cause autism any more than does toilet-training, another major and memorable series of events in a toddler’s life. Current research indicates that various autism spectrum disorders are likely polygenic, which means caused by combinations of several genes. Various presentations of autism tend to run in families, and it affects more boys than girls.
Some parents who still struggle to come to terms with their child’s disability will seek a source of blame for causing the problem, and a cure. But what about those children who are purported to have been cured by, for example, chelation treatments? Chelation is used to remove toxic levels of heavy metals from the body. But an examination of the symptoms of mercury poisoning and those of autism show they have little in common. Although chelation therefore is not approved by the FDA as a treatment for autism, in 2005, a 5-year old autistic boy died of a heart attack resulting from chelation treatment. The chelator used drew too much calcium from his blood, causing cardiac arrest.
TESTIMONIALS VS EVIDENCE-BASED MEDICINE
There are a wide variety of unproven therapies out there that will supposedly cure any number of things, and those Web sites will have lots of testimonials. But of course, the plural of testimonial is not data. Human beings are prone to a number of (unconscious) cognitive biases, such as selective perception, investment bias (also post-purchase rationalization — we value something more when we’ve spent a lot of money on it), and/or the placebo effect. Sometimes the improvements seen in children being given treatment are simply maturation. Children with developmental disabilities by definition don’t develop “on schedule”, but over time they will still continue to mature, improve and make gains in abilities, albeit sometimes unevenly or in spurts. People want to see their children improve, so they associate the positive changes seen with any previous treatment, or treatments, as some people will employ several at once in desperate efforts for something to cure the problem. To get around these bias problems in scientific research, we use control groups, double-blind studies, and statistics.
In addition to potentially dangerous and potentially lethal “cures”, we still have the ever-present issue of dangerous and potentially lethal diseases. Just a couple decades after large-scale immunisations were put into place, it’s easy to forget just how dangerous these highly-infectious diseases were. In decades previous, children used to die at much greater rates; diseases such as tuberculosis, diphtheria, pertussis (“whooping cough”), polio, smallpox, measles (rubeola), mumps, rubella (“German measles”) and chicken pox would run through families, neighborhoods and cities, and one child of every ten would not even live long enough for their first birthday party. For those who lived, these diseases could lead to other issues, such as meningitis, deafness, blindness, sterility, and mobility impairment. (No, chicken pox is not nearly as serious as many of the other diseases, but a flare-up of shingles is hardly a walk in the park.) After the vaccine scare, immunisation rates in the UK dropped from 92% to 78.9%, and in 2006 a child in Britain died from measles for the first time in 14 years. In some areas of the UK, only one in nine children were getting the MMR. Of all the infectious diseases that can be prevented by vaccine, measles was and still is the most deadly, and is the cause of half of the one million deaths that could be prevented.
A vaccine is only effective in 85-95% of the immunized population (and a few people with compromised immune systems cannot take vaccines), so we rely upon “herd immunity” to help protect those vulnerable people by keeping the level of pathogen down to a very low level. According to the World Health Organization, it takes a 95% vaccination rate for herd immunity to prevail and prevent significant outbreaks, such as the mumps outbreak in the American Midwest last year. There were over 1,000 people infected, the largest in over 20 years. The outbreak was likely caused by a visitor from the UK, where the same virus genotype (G) has caused over 70,000 cases of mumps between 2004 and 2006. In the 1990’s, vaccination shortages in the former Soviet Union resulted in over 200,000 cases of diphtheria, leading to the deaths of 5,000 people.
Parents should be reassured that their children are much safer by being immunized, because diseases such as measles, mumps and rubella are not a thing of the past. These diseases can be very debilitating and life-threatening. They should be reassured that there is not good evidence showing any causal link between vaccines and autism.