Autism in Little Kids: The Arguments for Universal Screening

Editor's Note:
In August 2015, the US Preventive Services Task Force (USPSTF) released a draft recommendation statement^[1] regarding screening for autism spectrum disorder (ASD) in children. The task force concluded that current evidence is insufficient to assess the balance of benefits and harms of screening for ASD in asymptomatic children—those in whom no concerns have been raised by their caregivers. This recommendation stands in contrast to those from the American Academy of Pediatrics (AAP), which recommend universal screening for ASD in all children at age 18 and 24 months.^[2]

Medscape spoke with David S. Mandell, ScD, associate director of the Center for Autism Research at The Children's Hospital of Philadelphia (CHOP), and Juhi Pandey, PhD, a pediatric neuropsychologist at CHOP, about these two disparate recommendations and the implications for pediatric primary care.

Is Anyone Screening Out There?

Medscape: How widespread is universal ASD screening at 18 and 24 of age months now? What percentage of US children are screened at these ages?

Dr Mandell: There aren't a lot of good data on how much physicians screen for autism. The small amount of data available about community practice suggests that about one half of kids are screened for autism at those ages. That's up from less than one quarter about 7 or 8 years ago—so it's a promising improvement, but it's still not as widespread as we'd like.

Medscape: What is the purpose of universal screening for ASD? Is it to identify children who might have ASD and refer them for further evaluation? Or is it to "diagnose" ASD, and begin therapy?

Dr Mandell: It's rarely expected that the primary care provider is going to make the diagnosis of autism. Making a diagnosis of autism, especially in very young children, is complicated. So the goal of the screening is really to get that kid referred for more comprehensive evaluation.

Medscape: It seems that the USPSTF is suggesting that improved developmental outcomes should be used as the measure of benefit for universal screening. Is this an appropriate outcomes measure for screening?

Dr Mandell: This is where I think that the task force is hampered by the rules that it has established—a bizarre set of rules for relatively rare conditions and conditions that don't have a biological marker, such as autism. What the USPSTF is saying is that there are no studies that have followed a group of kids from screening through treatment to show that those kids have better outcomes than if the screening had not been put in place.

Now, the authors of this recommendation acknowledge that they were able to find 26 treatment studies of good or fair quality demonstrating that there is substantial cognitive gain for children who receive evidence-based treatment compared with those who don't. However, they note that those kids were clinically referred and were not identified through screening: That is, nobody was waiting to enroll them into a treatment trial when they were screened at the pediatrician's office.

I find this to be a very strange metric. They agree that we have screening tools that can identify kids with autism. They agree that the bulk of the evidence suggests that treatment is effective in improving outcomes for children with autism. They're saying that no single study has combined those two things. I think that's a really high bar, a really expensive bar, and probably an unnecessary bar.

Dr Pandey: A better outcome measurement for screening would be to assess how many of those children who were identified by the screener then go on to get the diagnostic evaluation that is warranted, in order to know the outcome of that diagnostic evaluation. A better marker to assess the value of a screening tool is to determine whether we are identifying the right kids who require further evaluation.

Ideally, the whole point of screening is to get people onto the next stage, which is the treatment that they need. To base the outcome of screening on treatment outcomes is skipping important steps.

Medscape: How strong is the evidence for early intervention in autism? Is there evidence of benefit in those with mild, moderate, or severe ASD? Conversely, what evidence do we have that delayed or no treatment is harmful?

Dr Mandell: We don't know now what predicts treatment response for young children with autism. So there may be kids who are very severely impaired who respond very well to treatment, and some who don't respond at all. What evidence we do have suggests that kids who have better communication, better language, and higher cognitive ability at baseline respond better to treatment than kids who are more severely impaired. Of course, those more moderately or mildly affected kids are the ones who are going to be missed if you don't screen.

Barriers to Screening

Medscape: One of the objections raised about another universal screening recommendation for children—the adoption of universal newborn screening for critical congenital heart disease— was access. There are concerns that infants born in some rural areas or communities did not have access to pediatric cardiologists to evaluate whether the infant who screened "positive" truly had critical congenital heart disease. In the meantime, the wait created stress, anxiety, and expense for the parents. Is access to pediatric developmental specialists also a barrier to universal ASD screening?

Dr Mandell: There are two kinds of access that are really important here. One is access to a multidisciplinary team that can do a complete evaluation and provide both a gold-standard diagnosis and also identify other conditions that may commonly co-occur with autism. Wait lists are very long for evaluations by a developmental pediatrician or a tertiary care autism clinic. That's a huge problem, and we haven't figured out how to expand access to those very specialized services.

There's another type of access that might be equally or more important than that kind of access, and that's access to early intervention services. All 50 states participate in the Individual With Disabilities Education Act (IDEA) Part C, which mandates that for any child referred to the program, the state must provide an evaluation and services if warranted within a relatively short period.

What we strongly encourage physicians to tell families, and what I strongly encourage families to do, is to pursue both lines concurrently: Get yourself on a waiting list for an autism specialist, but also start getting the early intervention services to which your child is entitled.

I think that was a really interesting finding of a recent study^[3] of the Modified Checklist for Autism in Toddlers (M-CHAT). About 54% of kids who screened positive on this tool met criteria for ASD. But almost all of the screen-positive children—about 98%— had some kind of developmental delay that warranted intervention.

Medscape: This study would seem to provide support for the value of screening then.

Dr Mandell: I agree. I think the task force is constrained by looking very narrowly at whether screening for autism results in better outcomes for children with autism.

Dr Pandey: Right. The USPSTF is not wrong in saying that we need to think more about autism screening, because ideally you want a measure that is 100% specific, which would be the best-case scenario. The truth of the matter is that ASD is a behavioral diagnosis, with parents who are sometimes informed and sometimes not, and have different life experiences, which leads them to answer questions in unique ways. All of those barriers to developing the perfect screener are all true.

What is nice about autism screeners is that you do pick up children with ASD earlier, which is a good thing, but you also pick up children who have language, motor, or social delays.

So in terms of positive predictive value, perhaps these tools do not pick up only children with ASD. But they do indeed pick up children who require, and benefit from, early intervention. I don't see how that is a bad result. I have never met a parent who said they wished they hadn't done that.

Medscape: The USPSTF did not definitively say "Do not screen," but rather noted that there was "insufficient evidence for screening." Would you anticipate that this might lead to denial of payment for diagnostic services by third-party payers?

Dr Mandell: I think it depends a lot on how autism screening is integrated with developmental screening more broadly, because certainly, developmental screening is highly recommended. This recommendation is clearly a step backward in trying to identify autism in very young children. It's probably too soon to know whether it's going to affect physician or payer behavior.

Screening Without a Screening Tool?

Medscape: If pediatric clinicians do not screen every child at age 18 and 24 months, what key things should they be looking for to decide which children should be screened? Or would it be more appropriate, in children with high-risk behaviors, to skip the screening and just go straight to more targeted testing?

Dr Pandey: We have known for a long time that parental concerns are highly predictive of developmental challenges in kids. I think physicians should pay a lot of attention to those parental concerns.

An interesting study conducted by Katherine Zuckerman and colleagues^[4] looked at differences between child age at first parental concern and age at first parental discussion of concerns with a healthcare provider among children with ASD vs those with other developmental concerns. What they found was that parents who have children with autism have concerns much earlier than other parents, but the age at diagnosis was about the same.

What that means is that physicians either aren't listening to those parents, or they are minimizing their concerns. This is common. Children develop differently, and professionals can find many reasons why development may be delayed. He's a boy; he's a second child—all of those things that people come up with to rationalize the behavior or developmental delay that they're observing. One of the things that screening does is provide a concrete measure and take some of that bias away.

This is particularly important when thinking of traditionally underserved minority groups. Disadvantaged families, those for whom English is a second language, and families whose ethnicity varies from that of the provider can present a whole set of other challenges. Those families may not be as familiar with developmental milestones, or they may expect the doctor to know any concerns and therefore do not bring them up themselves. They may have communication problems, or the physician may attribute some of the delays that he or she is observing to English not being spoken in the home. For those families, standardized screening and response to that screening is even more important. I fear that a recommendation to not screen is going to disproportionately affect disadvantaged families.

Which leads to a second concern: Even among kids who screen positive, as many as one quarter or one half of them don't get referred to early intervention for one reason or another. Watchful waiting is common.

Medscape: Dr Pandey, I have a two-part question for you as a neuropsychologist: What should pediatric clinicians be looking for, and what is the next step?

Dr Pandey: I will tell you that I, like other clinicians, use my gut a lot of time, perhaps to estimate a child's developmental functioning or IQ. I might make assumptions about what a kid can and cannot do and base my testing battery on those assumptions. But I will tell you that without standardized, structured testing, I am often wrong—and that is true of most clinicians. We either under- or overestimate on the basis of our own biases and clinical assumptions. Without grounding in a structured kind of assessment, be that testing or screening, we often are not correct and not specific enough.

Targeted screening at specific ages is not based on a concern. Screening on a schedule takes away the bias that comes with clinical judgment.

Using a screener to identify a concern—not only regarding autism, mind you—you are more likely to pick up children who might be missed because of either your bias or the parents' bias.

Screening is hard to do for a number of reasons, especially in underserved communities, where the load is so high and the professionals are so few. At CHOP, we have been looking at ways to improve services, including screening, in underserved populations.

For clinicians who are not using a screening tool, surveillance is extremely important. The 2007 American Academy of Pediatrics guidelines^[2] encourage surveillance and screening together. And the research does show that surveillance and screening together is the best package. Surveillance is that ongoing history and is focused on language and motor skills, such as first words, babbling, sitting, trunk control, pulling up to stand, first steps. Those observations, very much geared toward language and motor skills, should continue.

What needs to be added is more of the social domain. Asking parents about their child's eye contact, which starts at a young age and is noted when feeding or nursing, is important. From 9 to 18 months, as infants are looking at objects and things of interest, do they look back at parents to share their enjoyment? In later infancy, do children point with their index finger to show things that are of interest to them? Do they point at pictures in a book, or point to faraway objects? Do they respond to their name beginning at 6 months of age? That doesn't have to be a purely verbal response, but the child can look up or be aware of being called by name. Eye contact is extremely important in terms of nonverbal communication.

One thing that should raise immediate concern is repetitive behavior. These are children who start to flap their hands at a really young age, or look like they are becoming very interested in spinning objects or have developed nonfunctional routines. Early repetitive behaviors are less likely at a really young age. These observations do not immediately equate to a diagnosis of autism if present, but they should raise alarm bells and warrant further investigation.

Medscape: Should that further investigation be a standardized screening tool or referral to a developmental specialist?

Dr Pandey: I would encourage pediatric providers to complete the screening, because that will help inform the evaluation that is needed.

It takes 10 minutes to do the M-CHAT, and then another 10 minutes for the follow-up interview. This is not a huge amount of time. As a clinical psychologist and a neuropsychologist, I don't change my evaluation on the basis of the screening results, but it does help to inform my clinical judgment.

Medscape: Are there specific screening tools that you would recommend? Is there a role for providing parents with a screening tool before the office visit, so they come in with that screening tool completed?

Dr Mandell: The M-CHAT is the one that has the most evidence to support it and that has been tested in the largest population sample. Combined with the physician follow-up interview, it has good positive predictive value. A big chunk of the kids you identify through that screener are going to end up having autism.

The M-CHAT is a good one because it's very easy for the parents to fill out ahead of time. Give it to them in the waiting room and have them fill it out there, and then scan quickly to see whether there are more than three positive items.

Medscape: Without clinicians using a screening instrument to guide them, is it realistic to expect that the observations made in the primary care setting are sufficient to identify children with ASD, especially those who are not at the severe end of the spectrum?

Dr Mandell: Here is my message to clinicians. The task force acknowledged that screening is successful: That is, the task force says in this report that the screening tools we have are good and identify autism. What they then go on to say, however, is that we don't have evidence that when we screen those kids and they're identified and they enter treatment, they do better than if we didn't identify them in the first place, because the treatment studies don't recruit from primary care. They recruit from kids already diagnosed with autism or at very high risk for autism. The task force notes that this is the gap in the literature.

What I would tell physicians is that the USPSTF says screening identifies autism. If you believe it's important to identify autism in these kids early, you should be using those validated screening tools.

The reality is that you have 15 minutes at most with an individual child in the middle of a busy clinical day. You're rushing through all of the other things that the AAP says you need to do at each visit: immunizations, asking about guns in the house, car seats, nutritional questions, growth charts, and all those other things you are asking parents about the child's development. You're looking at the computer screen and your electronic health record. You have very little time where you're actually examining the child, and then you are focusing on their physical well-being. You have very little time to assess social, emotional, and psychological well-being, which is primarily done via parent report.

A lot of kids are shy around strangers, and so how they act in the office is not how they act when they're at home or in the community. If they've been waiting 45 minutes in the office, they're probably going to be bouncing off the walls, too. So that's probably not an accurate assessment of their well-being. These screeners are critical because they give physicians information that they otherwise wouldn't have and don't have a way to collect during that very brief interaction with the child.

Dr Pandey: This is where the scheduled screening takes away bias. There is just not enough time to pick up some of these more subtle behavioral markers. They develop over time and appear more subtle and more severe at different stages in development, and that varies by child. The symptoms that we are talking about with autism are not as homogenous as we once thought. They are very heterogeneous, so you do need that structured screener to help you identify those cases. The office visit just isn't enough.

Take the example of a young child who is exhibiting repetitive behaviors in the way that they manipulate nesting cups—maybe they roll them instead of stack them. If you didn't have nesting cups in your office, how would you, as a practitioner, see these behaviors? And maybe a first-time parent doesn't know that this is not how you play with those nesting cups, right? So you could be losing valuable time in terms of getting that diagnostic evaluation and eventual treatment.

Medscape: Any concluding thoughts for our readers?

Dr Mandell: The USPSTF recommends a long-term study following large samples of screen-positive and screen-negative children to look at outcomes—a study that they note would be resource-intensive while providing valuable information. Given the prevalence of autism, such a study would require thousands of kids, the overwhelming majority of whom will end up not having autism and not needing any kind of intensive intervention. I think that's a ridiculous bar. Even if we were to have that study funded tomorrow, the results would not be available for 10 years.

I think that the information that we have now is strong enough to suggest both that screening identifies children with autism and that high-quality treatment is effective. That should be enough.

Editor's note: The researchers at CHOP's Center for Autism Research have developed a website as a resource families navigating diagnosis of ASD. Launched in 2014, it serves as a clearinghouse of reliable information for families throughout the lifespan of an autism diagnosis—from suspicion through adulthood—and allows parents to access information that is evidence-based, reliable, and vetted by experts at CHOP. Articles on the site are searchable by topic and life stage.