OBJECTIVE: The goal of this study was to investigate the feasibility and outcome of a systematic autism screening process for all toddlers (aged 14–30 months) in a large, community-based pediatric practice.
METHODS: All toddlers who presented to the clinic during the 6-month screening period were eligible. We used 2 screening questionnaires and allowed physicians to refer directly to capture as many children as possible. Receptionists and medical assistants distributed and collected screening questionnaires; research staff did all scoring and follow-up, either by telephone or in person when indicated.
RESULTS: We obtained a high rate of screening (80% of eligible children). Of the 796 children screened, 3 had already been diagnosed with an autism spectrum disorder (ASD); an additional 10 children who showed signs of early ASD that warranted further evaluation or intervention were identified. Formal screening measures identified more children with ASD than did clinical judgment or caregiver concerns; however, no single method (ie, questionnaire, caregiver concerns, provider concerns) identified all children with signs of early ASD. We had excellent participation from racially and ethnically diverse families, including Spanish-speaking families. Thirty-two percent of the children who were screened did not present for a well-child visit during the study period and were screened at a sick visit, follow-up visit, or injection appointment.
CONCLUSIONS: A partnership between pediatricians and autism specialists resulted in effective, systematic autism screening. Future studies should examine how to create effective systems of care.
WHAT'S KNOWN ON THIS SUBJECT:
Autism screening at 18- and 24-month well-child visits can optimize early identification and outcome. Although promising screening questionnaires exist, no studies have shown how they can be systematically applied to all children who present to a health care setting.
WHAT THIS STUDY ADDS:
Developing a supportive partnership between pediatricians and autism experts resulted in a high rate of screening for all children, including those from minority groups and the uninsured. Formal screening was much more effective than relying on clinical judgment alone.
Routine screening for autism spectrum disorders (ASDs) in toddlers has been recommended by the American Academy of Pediatrics.1 The Modified Checklist for Autism in Toddlers (M-CHAT) and the Infant Toddler Checklist (ITC) are 2 promising screening instruments.2,3 However, no studies to date show how they could be implemented in a consistent manner to all children who present to a primary care setting. The M-CHAT is a 23-item parent questionnaire developed for children aged 15 to 30 months as an autism-specific screener for use in the general population. Studies have relied on the voluntary efforts of a large number of pediatricians, with an average of 20 to 32 screens obtained per provider over the course of 1 year or more.4,–,6 The ITC is a 24-item broad-based screener of language and communication that has shown promise as a screener for autism.7 Studies of the ITC are based on a large and growing pool of children voluntarily screened from a variety of public places, child care providers, and health care providers over the course of several years.7,8
Three studies have attempted population screening. The landmark study of the Checklist for Autism in Toddlers (CHAT) reached 16 235 children at 18 months of age, representing 39.8% of the target population.9,10 The Quantitative CHAT, a recent modification of the CHAT, was mailed to 2360 caregivers, with a 33% response rate (n = 779).11 The screening study with the highest participation rate involved a 2-step screening procedure with the Early Screening of Autistic Traits Questionnaire,12 with a 4-item version successfully administered to 31 724 of 31 776 (99.8%) children aged 14 to 15 months, followed by a 14-item version at follow-up.13 However, researchers concluded that the overall yield of children identified (n = 18) was low, and that parents were somewhat resistant, perhaps because of the very young age of the children.
Our Each Child Study implemented a systematic ASD screening process in a large, busy pediatric practice in Salt Lake City, UT. We screened children aged 14 to 30 months using both the M-CHAT and ITC and allowed providers to make direct referrals. We focused on 4 basic questions to (1) determine how many children with early signs of ASD could be identified in a community setting through a systematic screening process, (2) examine the extent to which formal screening could identify children before their behavior was of significant concern to either their caregivers or provider, (3) examine the extent to which families from diverse backgrounds would participate in screening, and (4) examine the extent to which screening outside the well-child visit was feasible or necessary. To study these issues, we worked closely with the pediatric practice site to develop a systematic method for distribution and collection of screening materials, and then conducted all scoring and follow-up ourselves. This process removed the potential for bias related to the individual provider's level of autism expertise or patient population.
All aspects of the study were approved by the institutional review board of the University of Utah.
The study was conducted in a large community pediatric practice that employed 9 providers, served an average of 33 000 patients per year, and routinely scheduled 10-minute sick visits and 20-minute well-child visits.
A total of 990 toddlers in the study age range were seen in the clinic during the 6-month screening period (verified against the clinic's daily schedules). All children had a 2006 date of birth, and were between the ages of 14 and 24 months at the start of the screening period and between 21 and 32 months of age at the end. Screening questionnaires were obtained on 796 (80%) of these children.
Both the ITC and M-CHAT were administered and published cutoffs were used. Thus, children screened positive on the ITC if they were below the 10th percentile on the Social, Symbolic, or Total scores. Children screened positive on the M-CHAT if they obtained a total score of 3 (of 23) or failed 2 of the 6 critical items. The questionnaires were initially presented in a single order (M-CHAT first, ITC second), but after the first few weeks of the study they were alternated to avoid possible order effects.
The ITC includes a yes/no question about whether the caregiver is concerned about their child's development. We added a similar yes/no question about whether the provider was concerned, and a place to indicate if a referral had already been made.
Step 1: Completion of Screening Questionnaires
Research and clinical staff worked together to develop procedures for distribution and collection of screening information. Caregivers were offered a screening packet by office staff at check-in. All appointment types were included (sick visit, follow-up visit, injection appointment, and well-child visit). A cover letter explained the screening to caregivers, stated that participation was voluntary, provided our contact information, and asked for permission to contact them for follow-up if indicated (with a release of information). Screening packets were available in both English and Spanish.
A research assistant shadowed the office staff for the first few days to evaluate the distribution and collection plan. Once this was successfully in place, we reviewed the daily clinic schedule to determine the number of eligible children and provided regular feedback about the rate of completed screens for each pediatrician. All screening questionnaires were scored by our research staff.
Step 2: Initial Telephone Follow-up
Families were contacted by telephone if the child scored in the at-risk range on either screener, or if their physician had made a direct referral. In the case of a positive M-CHAT, the published follow-up interview4,5 was conducted by graduate students with experience in early autism and child development under the supervision of the lead author. Interviews generally took 10 to 30 minutes. In the case of a positive ITC, relevant items were repeated and discussed over the telephone. If the provider had made a direct referral, both measures were readministered over the telephone. Results were discussed immediately with caregivers, followed by a letter to them and the pediatrician. Caregivers of children whose positive screen was confirmed were invited for an in-person evaluation at no cost. The parents of children who screened negative after telephone follow-up were encouraged to continue monitoring their child's development, including completing a second screening packet if one was offered by their pediatrician. All caregivers were reminded that screening results can be inaccurate.
Step 3: In-Person Evaluations
Children were evaluated with the Autism Diagnostic Observation Schedule: Toddler Version,14 Mullen Scales of Early Learning,15 and Vineland Adaptive Behavior Scales-Second Edition.16 However, because of the nature of the study and the limited research on diagnostic measures for this age group, emphasis was put on whether early signs of ASD were significant enough to warrant a complete clinical evaluation and/or referral for intervention, rather than whether the child met full Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision criteria.17 Clinical judgment of the first author, who is research trained on the Autism Diagnostic Observation Schedule18 and the Autism Diagnostic Interview–Revised19 and specializes in autism diagnostic assessment, was used with input from the clinical team to determine if the child showed significant signs of autism. Characteristics suggestive of early ASD included impaired ability to direct the attention of others; impaired responding to others' attempts to engage them or direct their attention; impaired nonverbal communication; a lack of integration between verbal, facial, and gestural communication; lack of a response to name; impairments in imitation; and repetitive use of objects. Caregivers were given oral and written feedback about their child's results, with appropriate recommendations and referrals. Caregivers were also contacted a few weeks later to see if they had additional questions about their child's screening results.
Figure 1 shows the flow of patients from initial screening through in-person evaluations. Screening packets were obtained on 796 children (80%). Of those who completed all steps of screening, 13 children (1.6%) had significant signs of early autism. Three of the 13 had already received an ASD diagnosis before the study began. Of the 10 newly identified children, caregivers expressed general developmental concern in 4 cases, and providers expressed concern in 5 cases; in only 2 of these cases, however, were both the caregiver and the provider concerned.
Three of the 10 newly identified children had screened negative on the initial questionnaires. In 2 cases, the caregivers approached a provider with questions prompted by the experience of completing the questionnaires. In the third case, the child screened negative on both questionnaires at age 18 months (and did not qualify for telephone follow-up), but then screened positive on an M-CHAT administered by the physician immediately after the official screening period ended; thus, we chose to provide follow-up. No loss of language or other obvious regression was noted, but the mother indicated that at 24 months she now noticed he was not as verbal or responsive as other children his age.
Demographic variables collected on the 796 participants indicated excellent representation of the ethnic and racial diversity of Utah, and close approximation of the diversity of the nation with the exception of blacks (Table 1). Seventy-five (9%) of the screening packets were completed in Spanish.
In-person evaluations were administered to 19 white, non-Hispanic (WNH) families, 1 black family, 8 Hispanic families (3 evaluations conducted in Spanish), and 2 families from the Pacific Islands. Of the 13 children with signs of ASD, 8 were WNH, 1 was black, and 4 were Hispanic. The 3 who had previously been diagnosed with ASD were all from WNH families. All but 1 of the 30 children seen in person displayed some type of developmental delay and received a referral for intervention.
To examine whether screening outside the well-child visit was feasible or necessary, screening packets had been offered at all visits during the 6-month period. Thirty-two percent of children screened did not present for a well-child visit during the screening period, but were screened at a sick visit, follow-up visit, or injection appointment. A χ2 analysis was conducted to determine if children from underrepresented groups were more or less likely to present for a well-child visit. A 3 × 2 χ2 analysis of race/ethnicity (WNH, Hispanic, other) and well-child examination (yes/no), was nonsignificant (P = .33). However, χ2 analysis of type of insurance (private, public, not insured) and well-child examination (yes/no) was significant (P = .01). To examine this further, 2 orthogonal tests were conducted. A 2 × 2 χ2 analysis of type of insurance (private or public) and well-child examination (yes/no) was not significant (P = .67). However, a χ2 analysis with insurance in general (insured/not insured) was highly significant (P = .003). Results indicated that 78% of uninsured children never presented for a well-child visit during the screening period, compared with 52% of insured children. A Fisher's exact test was conducted because of the low number of uninsured families, and the results remained highly significant (P = .003).
Because we presented the measures in a single order for the first few weeks, data were more complete for the M-CHAT (n = 781) than for the ITC (n = 728). A direct comparison between the M-CHAT and ITC should exclude children who were missing data and should consider the time required for telephone and in-person follow-up, which is beyond the scope of this article. Both measures yielded a high number of false-positives that were reduced with a telephone interview, and neither measure alone identified all children. If screening results depended solely on the 6 critical items of the M-CHAT, only 3 of the 10 newly identified children with early ASD would have been identified.
To our knowledge, this study is the first attempt at systematic ASD screening within a community-based pediatric practice in the United States. We examined 4 basic questions and found that (1) 1.6% (n = 13) of children screened showed early signs of ASD, 10 of whom were newly identified through screening questionnaire scores or as a result of completing the questionnaires; (2) the majority of newly identified children were not yet of concern to their caregivers or providers; (3) families from racially and ethnically diverse backgrounds participated equally; and (4) 32% of all children, and 78% of uninsured children, would have been missed if screening had been restricted to well-child visits. This represents an important step toward determining a method for achieving systematic screening and estimating the base rate of early signs of autism in a community sample, and suggests that visits for reasons outside of health supervision represent valuable opportunities to screen for autism, especially for the uninsured.
Our rate of participation (80%) was significantly higher than any study of similar-aged children screened with full questionnaires,9,–,11 and was achieved by developing a partnership with the practice whereby pediatrician resources focused on systematic collection and autism experts focused on scoring and follow-up. Our method required little time from pediatric staff or providers, making it feasible to access autism screening for all patients regardless of the other issues that needed to be addressed during the appointment. Because we scored and conducted follow-up of positive screens ourselves, we could ensure a consistent level of autism expertise at each stage of interpretation. This partnership increased the chance that all children would have equal access to high-quality screening.
The act of completing a formal screening questionnaire caused some parents to reconsider their child's developmental progress, and in at least 2 cases prompted parents to initiate a conversation about the child's development with their provider that resulted in an appropriate referral.
A limitation of this study was that we were unable to evaluate children who screened negative. Thus, we cannot examine the true sensitivity and specificity of individual questionnaires or the screening process as a whole. Clearly, many more children were identified through their participation in the screening process (n = 10) than had been identified through existing caregiver- or provider-initiated referrals (n = 3). A follow-up study to reassess the children evaluated in person is underway, at which point prediction values may be informative.
It was unfortunate that not all families who screened positive chose to participate in follow-up, which is a concern for many health and developmental issues and is a complex problem for separate study. In our study design, follow-up required additional time from parents and was conducted outside of the medical home, which may have been barriers for some families. However, our results suggest that if implemented today, a significant number of families would participate.
To our knowledge, this is the first study to observe Spanish-speaking families from community pediatric screening through an in-person evaluation. Our high rate of participation among racially and ethnically diverse patients suggests that our method was acceptable to families from diverse backgrounds. This is important in light of recent research indicating that children from ethnic minority groups are less likely to have a documented ASD diagnosis than white children.20 Other research has also found an autism diagnostic bias against children from ethnic minority groups when primary care providers relied on clinical judgment but more equivalent rates of autism diagnoses when structured diagnostic rating systems were available.21 Systematic screening methods could reduce these disparities.
Our results support the recommendation that all children be formally screened for autism, even if the child's presentation has not elicited caregiver or provider concerns. Unfortunately, screening for autism is neither quick nor easy, and our results would not support relying on well-child visits or on questionnaire-based screening in the absence of direct support for follow-up. The use of formal screeners is superior to clinical judgment alone, but the screeners currently available require a significant amount of time and autism expertise to interpret accurately. In our sample, neither the ITC nor the M-CHAT identified all cases, and both questionnaires missed some children completely. The use of the 6 critical items of the M-CHAT would have identified only 3 of the 10 new cases. Skipping a telephone follow-up and referring directly from the questionnaire to an in-person evaluation would have resulted in well over 100 possibly unnecessary evaluations, compared with 1 in our sample.
Cost of care can be measured in multiple ways, including the cost of not providing care, the burden on existing systems, and the effects of screening on families. The first step to understanding cost, however, must be to identify a clinically effective practice. Our partnership provided the support needed to successfully implement an effective autism screening process into a busy pediatric practice. Given that parents of children with ASDs report dissatisfaction with the medical home,22 and that pediatricians report a low self-perceived efficacy in caring for children with ASDs,23 identifying ways to help bridge this gap are important.
Pediatricians are uniquely positioned to facilitate autism screening but not well positioned to conduct the entire screening process on their own. For example, in our 6-month screening period, each pediatrician saw an average of 112 toddlers each (well-child visit or other), 3 to 4 of whom merited further evaluation. To conduct systematic screening themselves, the pediatrician would have to be well-versed in a field that is essentially a moving target, as researchers try to identify signs of ASD at younger and younger ages. With the current tools, they would have had to follow-up on 27 (average) positive screens to identify the 3 to 4 children who needed a referral. In contrast, by partnering with experts in autism, even busy or reluctant providers could access high-quality autism screening for all their patients, resulting in earlier referrals for evaluation and thus the possibility for earlier intervention.
Our screening partnership identified children with early signs of autism before they were of concern to parents or providers. The study had a high rate of participation (80%) in a busy community-based pediatric office and had excellent participation among families with ethnically and racially diverse backgrounds. Future research should examine this type of screening method in a wider set of pediatric practices, determine the costs associated with such a program relative to the possible benefits of earlier diagnosis and intervention, and study the impact of screening on children and families.
This study was supported by grants to Dr Miller from the Centers for Disease Control and Prevention (U01DD000068-01), the Department of Pediatrics at the University of Utah, and the Utah Department of Health (Health Resources and Services Administration Maternal and Child Health Bureau BC B04MC000321-02).
We are grateful to the providers and staff at Granger Pediatrics; the families who participated in the study; and Deborah Bilder, MD, Dale Cannon, PhD, and the faculty and staff of the Utah Autism Research Program. For their contributions to the pilot work, thanks also go to Wee Care Pediatrics; Developmental Disabilities, Incorporated; Judith Pinborough-Zimmerman, PhD; and the Utah Department of Health's Baby Watch Early Intervention program.
- Accepted January 12, 2011.
- Address correspondence to Judith S. Miller, PhD, Center for Autism Research, Children's Hospital of Philadelphia, 3535 Market St, 8th Floor, Suite 860, Philadelphia, PA 19104. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- ASD =
- autism spectrum disorder •
- M-CHAT =
- Modified Checklist for Autism in Toddlers •
- ITC =
- Infant Toddler Checklist •
- CHAT =
- Checklist for Autism in Toddlers •
- WNH =
- white, non-Hispanic
- 1.↵American Academy of Pediatrics, Council on Children With Disabilities, Section on Developmental Behavioral Pediatrics, Bright Futures Steering Committee, and Medical Home Initiatives for Children With Special Needs Project Advisory Committee. Identifying infants and young children with developmental disorders in the medical home: an algorithm for developmental surveillance and screening [published correction appears in Pediatrics. 2006;118(4):1808–1809]. Pediatrics. 2006;118(1):405–420
- Wetherby AM,
- Prizant BM
- Robins DL
- Pandey J,
- Verbalis A,
- Robins DL,
- et al
- Wetherby AM,
- Brosnan-Maddox S,
- Peace V,
- Newton L
- Baron-Cohen S,
- Cox A,
- Baird G,
- et al
- Baird G,
- Charman T,
- Cox A,
- et al
- Swinkels SH,
- Dietz C,
- van Daalen E,
- Kerkhof IH,
- van Engeland H,
- Buitelaar JK
- Dietz C,
- Swinkels S,
- van Daalen E,
- van Engeland H,
- Buitelaar JK
- Lord C,
- Rutter M,
- DiLavore P,
- Risi S
- Mullen EM
- Sparrow SS,
- Cicchetti DV,
- Balla DA
- 17.↵American Psychiatric Association. Disorders usually first diagnosed in infancy, childhood, or adolescence. In: Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision. Washington, DC: American Psychiatric Association; 2000:70–84
- Golnik A,
- Ireland M,
- Borowsky IW
- Copyright © 2011 by the American Academy of Pediatrics