Screening for autism: AAP issues new recommendations

February 2008

Philip A. Brunell, MD: Autism is a subject that has been in the news recently. What do pediatricians need to know about autism?

Scott Myers, MD: Autism spectrum disorders — a term that is now being used by the AAP and others now to include autistic disorder, Asperger syndrome and pervasive developmental disorder not otherwise specified (PDD-NOS) — occur in nearly one in 150 children, according to recent epidemiologic studies by the CDC.

The prevalence of these disorders, as they are currently defined and operationally diagnosed, has increased dramatically in the last two decades. One recent survey suggested that 44% of pediatricians reported taking care of 10 or more children with an autism spectrum disorder in their practice. Pediatricians play an important role in the early recognition of these disorders and in guiding families to diagnostic resources and in taking care of these children longitudinally because these are chronic conditions.

James Coplan, MD: One of the questions related to this discussion is whether atypicality in some people is the tail of the normal curve rather than being an abnormality?

If we think about IQ as the model, in the late 1960s, 10 million people were ‘cured’ of mental retardation by the stroke of a pen because the definition of mental retardation was changed from –1.5 standard deviations to –2.0 standard deviations.

It’s possible to think about a relatively bell-shaped distribution of intelligence with a mean IQ of 100 and a standard deviation of about 15 points. There are some people who have IQs of 55 to 70 who have nothing the matter with them from a medical standpoint. This is just their polygenic endowment for intelligence. When we try to sort out children who have a medical disorder from children on the tail of the normal curve, we consider several factors. Below an IQ of 55, the individual certainly has a problem biomedically.

Are there multiple minor dysmorphic features? If so, the child is more likely to have an identifiable disorder. Finally, you’ve got to look at the discrepancy between the child’s IQ and the parent’s IQ.

Constantino and Todd looked at 1,500 twin pairs in Missouri because they were interested in looking at monozygous vs. dizygous twins, and they gave the Social Responsiveness Scale (SRS) on all of them (Arch Gen Psychiatry. 2003;60:524-530). They found that about 1.4% of all boys and 0.3% of all girls in this unselected population met criteria on the SRS for PDD-NOS. So the question becomes: When do we begin to think about some of the milder forms as being a variation of normal biology rather than disease state? It’s the difference between disability, which is dictated by the constraints imposed by society, vs. the medical model of having something biologically wrong, like having an extra chromosome.

Myers: There is this issue of whether some individuals at the high functioning end of the autism spectrum have pathology or whether people who at one time would have been considered to be just atypical in some way have now been put into a category with a diagnostic label. Some of that plays a role in the change in prevalence.

The other things have to do with how hard we are looking for these disorders, what tools we are using to diagnose them and how the diagnostic criteria are applied. I think it’s not as much the change in the criteria from DSM-IIIR to DSM-IV as the way the criteria are interpreted and applied. It is important to note that these are behaviorally-defined criteria and there’s some subjectivity that goes into whether an individual meets the threshold for each one.

It’s not clear that just the magnitude of the change in prevalence implies a real increase in risk. We also can’t completely dismiss the possibility of the existence of an etiologic environmental factor. There is a subgroup of children with autism spectrum disorders who have clear etiologic diagnoses, such as fragile X syndrome, duplication on the long arm of the 15th chromosome, tuberous sclerosis and many others. Currently, we are able to identify this type of definitive etiologic diagnosis only abut 10% of the time. The other 90% or so is within that idiopathic group, there’s certainly a strong genetic influence. If you look at twin studies, monozygous twins who share the same DNA have a concordance rate of about 90% when you consider autism or at least social communication impairment on the autism spectrum, whereas dizygotic twins who share about 50% of the DNA have a concordance rate of 5% to 10%, which leads to heritability calculations of 0.9 or above, and that’s high. Heritability of ADHD is about 0.7 or 0.8 and asthma is about 0.4. So, there’s a strong genetic influence in idiopathic autism, yet the concordance rate isn’t 100%. As a result, people will certainly hypothesize things like environmental factors. There are also other reasons why it might not be 100%, such as epigenetic factors and other things that affect expression of genes. There are certainly some environmental causes of autism, such as prenatal exposure to substances like thalidomide and valproate and infectious agents.

Brunell: There are people who think that there may be a genetic predisposition to react adversely with environmental stress, such as measles-mumps-rubella vaccine or mercury in thimerosal-containing vaccines.

Kristina A. Bryant, MD: Studies in the United States, England, Finland and Denmark all failed to show an association between receipt of MMR and autism. The Institute of Medicine could not find evidence to support a causal association between MMR and autism, or thimerosal-containing vaccines and autism.

Coplan: There are mountains of epidemiologic data that are negative and there have been several well-done studies within the past couple of years. There is the one out of Montreal where Fombonne, et al looked at MMR and thimerosal and it’s very clear that the prevalence of autistic spectrum disorder is highest in the most recent cohort of children who never saw thimerosal at all.

A metaanalysis by Ng looked at all data on mercury and noted no epidemiologic data to support the assertion that exposure to thimerosal or mercury is statistically linked with increased prevalence of autistic spectrum disorder. Yet any given parent can say, ‘Well maybe my child is the exception.’ It becomes hard to prove to any one parent that their child isn’t the exception to something.

Myers: The epidemiologic data don’t show an increased risk related to higher thimerosal exposure.

The recent study in The New England Journal of Medicine wasn’t looking at autism per se, but at more than 1,000 7- to 10-year-old children, their neurological functioning and its relationship with prenatal mercury exposure and postnatal exposure through age 7 months via both vaccines and immunoglobulin. Out of 42 factors researchers looked at, there were several associations, but in both directions — positive and negative — and nothing impressive. This study did not support a causal association between early exposure to mercury and deficits in neuropsychological outcomes. Probably the one thing coming out of that, because it’s similar to two previous studies, is an association between tics and mercury exposure, which needs to be investigated further.

The other thing is what we know about the neuropathology from autopsy studies of brains. Decreased cerebellar Purkinje cell number, curtailment of maturation of the forebrain limbic system and alterations in cortical minicolumns happen before the third trimester and are the most consistent findings in the brains of individuals with autism. These changes happen long before exposure to vaccines.

I’ll leave it to the infectious disease people to comment more on the issue of whether we have done harm by reacting quickly and early to the issue of thimerosal and calling for removal. In other words, was that action premature and did it add to the level of concern to a degree that in retrospect will have done more harm than good?

Brunell: I don’t think there’s any question that it has added to parents’ anxiety, and some people in retrospect feel that was not a good thing to do. The vaccine was safe. To just remove a factor that had nothing to do with it is not going to make it more safe.

Bryant: Understandably, parents and many pediatricians were confused by the decision to remove thimerosal from vaccines.

In terms of potentially doing more harm than good, we can talk about what happened with hepatitis B vaccine. In 1999, there was a recommendation to temporarily suspend administration of the birth dose of hepatitis B vaccine until thimerosal-free preparations became available. Although thimerosal-free hepatitis B vaccines became available within a few months, many hospitals did not reinstitute programs to deliver the birth dose. Consequently, not all babies born to hepatitis B-infected mothers were vaccinated promptly, and one baby in Michigan developed fulminant hepatitis B infection and died. In 2004, coverage for the birth dose of hepatitis B vaccine was still only 46%.

Influenza vaccine is another example. Some influenza vaccines for children still contain thimerosal, and parents may refuse to protect their kids against flu because they are worried about vaccine safety.

Brunell: There’s another issue that I’d like to raise and that is the legislation and the funding to take care of handicapped people. In 1990 there was an act and subsequent funding that encouraged people to make the diagnosis of autism because those children would be qualified for special educational services. Many children’s diagnoses were changed from mental retardation to other classifications.

Myers: That’s the issue of diagnostic substitution and its effect on the administrative prevalence. The information is mixed on that, but a well-done study by Shattuck did show a similar decline in the rate of mental retardation as the rate of autism increased. Most authorities would agree that there is diagnostic substitution that has gone on and that it played a role in the change in prevalence, but it is probably not the entire answer.

Bryant: We need to do a better job of screening. The latest AAP report noted that 44% of pediatricians said they cared for 10 patients with autism spectrum disorder, but a minority of pediatricians who answered one survey said they were actively screening. That’s worrisome.

Myers: The latest AAP recommendations include several things about screening and surveillance. One is this concept of general developmental surveillance at all visits, with further investigation, including the use of a specific screening measure, when concerns are raised. Also, for the first time, the AAP recommends formally using an autism screening measure at 18 and 24 months. There are a variety of screening measures available. The AAP has just released a resource toolkit that provides information on screening measures and includes some of the measures that are available in the public domain, which can be kept on the desktop and printed for use in the clinic.

Bryant: How long does the screening of an 18-month-old child take?

Myers: Something like the M-CHAT (Modified Checklist for Autism In Toddlers) takes about five minutes to complete.

Bryant: I think that as we talk about expanding screening that’s important because the typical well visit lasts a median of 16 minutes, and you have a lot of ground to cover.

Myers: The recent clinical reports from the AAP recommend following the 2006 AAP recommendations on developmental surveillance and screening, with the addition of some specific guidance about screening for autism spectrum disorders. General developmental screening is primary and often that will lead you, if there’s a problem, to asking questions about autism or including an autism screening measure.

Brunell: Whenever you talk about screening, you wonder about specificity and sensitivity. It sounds as though the specificity is not as important because you may pick up other problems in the screening that are also important. But the sensitivity is important because you don’t want to be missing these kids. Some say the earlier you get a child with autism and start remediating, the better the prognosis is going to be. Do you agree?

Coplan: We might say with some proof that if we identify a child with autistic spectrum disorder based on their developmental picture and then we determine that the child has fragile X, then we say to the parents, ‘You’re a carrier for fragile X. This is what your recurrence risk is. It’s your choice to decide what to do from here.’ That’s important too and that has nothing to do with proving through a double blind, randomized controlled trial that intervening earlier leads to a better outcome.

Myers: There is also some evidence that earlier intervention is associated with improved outcomes. Sensitivity is important, but specificity should not be trivialized. High specificity leads to higher positive predictive value (PPV). Primary care providers may lose interest in screening if the PPV is low (ie, if most of the children who fail the screening test do not ultimately have the disorder), and this certainly can be the case in low prevalence disorders.

Brunell: Another issue is the inequality of care. We may identify a kid who needs to be referred and they may be living in areas where adequate treatment is not available.

Myers: The pediatrician who suspects an autism spectrum disorder, or other developmental disorder for that matter, should not wait for the diagnosis to be made by the subspecialist but should refer to the early intervention services based on suspicion or failed screening. Too often there’s the idea that we have to know the diagnosis specifically first, and if there are many months of waiting for the subspecialist, that’s time wasted.

Brunell: Looking at time as a parameter, it seems that some of the people who have claimed there has been an increase have reached this conclusion by looking at kids of different age groups, and in 8-year-olds they find the criteria that they used in 2-year-olds don’t pick up as many cases. Therefore, they assume that when these 8-year-olds were growing up, the incidence was lower.

Coplan: First of all, one hopes that therapy makes a difference and, if therapy makes a difference, then the prevalence in any given birth cohort better be going down as that group of children get older.

More compelling, Leo Kanner’s original paper from 1943 (Nerv Child. 1943;2:217-250) was in fact a five-year follow-up study of children he had been following since 1938. These children received no therapy of any kind because there wasn’t any, but Kanner described tremendous improvement in that cohort of children.

The data are clear and there have been more than a dozen studies with more than 1,000 children that next to the degree of atypicality itself, the child’s non-verbal IQ is the single most important determinant of outcome. Children who have a normal non-verbal IQ as a group do much better than kids who have coexisting mental retardation. So you would predict, based on the data, that if you start with a cohort of 5-year-olds and the prevalence is whatever it is, the kids who have normal intelligence as time goes on are going to look less and less autistic and some of them, about 10% to 15%, go right off the spectrum and they no longer have a diagnosable disorder when they get to be adults.

There are people who have a preexisting belief that it must be mercury or that it must be some other thing causing the autism. There are also people who have a preexisting belief that some particular nontraditional therapy cures autism and they will pick and choose certain information to try to bolster their case. We need to be cognizant of that.

Brunell: It’s hard to argue beliefs.

Myers: This gets into a little bit of one of the issues pediatricians ask me about the most – the issue of some of the complementary and alternative medicine interventions. The AAP report on the management of autism talks a bit about this issue and reiterates previous recommendations to keep open lines of communication and support families, but it also emphasizes the importance of critically evaluating the scientific evidence of efficacy of various treatments. The clinical report helps pediatricians to understand the various interventions and whether they are empirically supported in the literature. The report then encourages physicians to help families to learn how to evaluate scientific evidence and recognize pseudoscience and to guide them toward empirically supported interventions.

Brunell: Are there any complementary therapies that you view as potentially dangerous?

Bryant: Chelation therapy. One child has died.

Myers: That was a case of chelation with disodium EDTA, resulting in fatal hypocalcemia. The treating physician now faces manslaughter charges in Pennsylvania. Some argue that the mistake was the chelating agent, not the idea of chelation itself. But why chelate at all when the evidence suggests that there’s not a causal association between heavy metals like mercury and autism? Chelation outside of an Institutional Review Board-approved, double blind randomized clinical trial with safety monitoring is irresponsible at best.

Coplan: What do you think about hyperbaric oxygen?

Myers: There is no published information that I’m aware of to suggest any efficacy in autism spectrum disorders.

Coplan: But in terms of the potential risk for recurrent hyperoxy?

Myers: Risks include seizures and barotrauma, among others. Those have to be talked about. These types of unproven treatments with weak theoretical underpinnings, besides the medical risks, do have the potential to compete with more well-proven interventions in terms of financial resources, time, and effort, and that’s also a risk. The bottom line is, the more you can speak the lingo and know what the families have read, the more you are able to make a convincing argument about the pseudoscientific theories, lack of empiric evidence of efficacy and potential harm of unsubstantiated treatments.

Bryant: It’s interesting because those skills that you’re talking about and the tools that you’re giving the families are not really specific to working with families with autism. You have the same situation when you talk to families who have concerns about vaccines. Families are offended when they come in with their litany of concerns that they’ve read on the internet and you can’t discuss these concerns knowledgeably.

Brunell: It’s helpful to be able to present the things that are going to be useful. So what’s going to be useful?

Myers: Educational interventions, including applied behavior analysis-based treatments, and habilitative therapies such as speech and language therapy. There is some evidence supporting things like social skills instruction.

For challenging behaviors that aren’t part of the core symptoms of autism but occur at increased rates in these children, things like tantrums with aggression or self-injury or irritability, sleep problems, and those sorts of things, there are behavioral interventions, again based on principles of learning theory and behavior analysis. There can be a role for medicines, psychopharmacology, for certain target symptoms. There is no medicine for autism, but you can target certain interfering symptoms and potentially allow the child to more effectively participate in and benefit from the educational interventions.

Coplan: There’s an old Kentucky recipe for rabbit stew, and this is what I talk to my patients about. It begins, ‘First, catch a rabbit.’ So I tell parents, ‘the first thing we need to do is to get Johnny to sit down at the table, give us eye contact and when we do X, whatever it is, Johnny needs to reciprocate.’ So you begin there, with applied behavior analysis-discrete trial training and then you go on to what’s called ‘natural environment training’ and/or floor time, where you see what the child is interested in, you let the child get engrossed in it, then you gently and playfully sabotage it. The child needs to look at you and attend to you and say something to you for the child to continue getting access to what the child wanted in the first place.

So you start with the in-your-face operant conditioning and then you move on to more complex things.

Language comes along, awareness comes along, and eventually you get to “Social Stories” (www.thegraycenter.org). Then you get on to things like self-awareness, which is discussed in books like Asperger’s: What Does It Mean to Me?, and Asperger’s syndrome. It helps the child to take ownership.

All along, there’s this progression of therapies that runs parallel to the natural history of ASD itself.

Myers: The goal of behavioral interventions is to increase the likelihood of behaviors that you want to see such as communication, eye contact, functional or symbolic play and to decrease the likelihood of behaviors you don’t want to see, such as tantrums or self-injury, through application of principles of learning and operant conditioning.

Brunell: And who is going to do this? How much participation by the pediatrician?

Myers: The resources vary in different communities. Pediatricians can play an important role in directing people to resources. When parents say, ‘I’ve got these options, what do you prefer, the DIR or Greenspan floortime model or more of an applied behavior analysis-based model, or something else? What do I do?’ It’s not that you have to say, ‘well, you do this one,’ but to be able to give a short summary of what those methods have to offer and what we know so far about the evidence of efficacy is helpful. There are few good comparative studies. So, you can direct them toward the more scientifically supported intervention strategies, but it’s more than just the degree of empiric support. It’s what fits for that family and child.

For more information:

• Dr. Coplan’s book for parents of children with newly-diagnosed autistic spectrum disorder (working title: Off the beaten path – A parents’ guide to Autistic Spectrum Disorder), is scheduled to be published later this year by Bantam/Dell.
• Johnson CP, Myers SM, the Council on Children with Disabilities. Identification and evaluation of children with autism spectrum disorders. Pediatrics. 2007;120:1183-1215.
• Shattuck PT. The contribution of diagnostic substitution to the growing administrative prevalence of autism in US Special Education data. Pediatrics.2006;117:1028-1037.
• Thompson WW, Price C, Goodson B, et al. Early thimerosal exposure and neuropsychological outcomes at 7 to 10 years. N Engl J Med. 2007;357:1281-1292.