Factors influencing autism spectrum disorder screening by community paediatricians

Abstract

BACKGROUND:

In most cases, autism spectrum disorders (ASD) can be reliably diagnosed at two to three years of age. However, Canadian data reveal a median age at diagnosis of approximately four years.

OBJECTIVE:

To examine general paediatricians’ practices regarding ASD screening and identify factors that influence decisions regarding the use of ASD screening tools.

METHODS:

Using a qualitative inquiry-based interpretive description approach, 12 paediatricians from four practice groups participated in four focus groups and one individual interview. These were conducted using semistructured interviews, digitally recorded and transcribed verbatim.

RESULTS:

Five main domains of themes were identified related to screening tool use: benefits; needs not addressed; elements that limit utility; elements that encourage utility; and implementation challenges. Factors influencing practice included availability of time, comfort with screening tool use, previous use and knowledge about specific tools. Systemic factors included knowledge and access to community resources, as well as the ability to provide support to the child and family.

CONCLUSION:

The results from the present study identified important factors that influence paediatric practice in ASD screening. As screening tools improve, it will be important to examine the implementation and effectiveness of screening tools and strategies for increased uptake. Future research will also need to attend to the practical needs of physicians and communities in the aim of earlier diagnosis and rapid access to interventional resources.

Autism spectrum disorders (ASD) are characterized by abnormalities in social interactions and patterns of communication, and a restricted, stereotyped, repetitive repertoire of interests and activities (International Classification of Diseases, 10th revision). The most recent estimation of ASD prevalence is one in 68 children (1). In many cases, ASD can be accurately diagnosed at two to three years of age (2,3) or even earlier (4). However, Canadian data show a median age at diagnosis of 39 to 55 months (5). Early intervention has significant benefits for children with ASD (6); thus, early diagnosis may lead to improved long-term outcomes.

A survey of paediatricians conducted by the American Academy of Paediatrics (AAP) revealed that 23% used a standardized developmental screening tool (7). Reported challenges to implementing ASD screening in paediatricians’ practices include lack of familiarity with tools, patients being referred to a clinical specialist instead or lack of time (8). There has been limited uptake, and some in the field have expressed reservations about the evidence supporting ASD screening. However, recent studies have indicated that the Modified Checklist for Autism in Toddlers (M-CHAT) identifies toddlers with ASD earlier and more consistently than surveillance by community paediatricians (9), as does the Infant Toddler Checklist (10). The AAP recommends universal ASD screening at well-child visits at 18 and 24 months (7). In contrast, a review by Al-Qabandi et al (11) concluded that there was insufficient evidence supporting routine ASD screening and recommended careful surveillance as an alternative strategy. In Canada, the updated Rourke Baby Record suggests “specific screening for ASD at 18 to 24 months of age using the M-CHAT on all children with any of the following: failed items on the social, emotional or communication skills inquiry, sibling with autism, or developmental concern by parent, caregiver, or physician.” Some provinces (eg, British Columbia) have recommended ASD screening, but targeting at-risk children (eg, those already identified with clinical concerns) rather than adopting a universal approach (12). This approach is consistent with consensus best-practice guidelines published by a private foundation with input from Canadian clinicians, researchers and families (12,13). To date, there has not been an official position statement on ASD screening from the Canadian Paediatric Society.

While studies of ASD screening implementation are limited, the challenges of implementing general developmental screening are well documented (14). In 2006, the AAP released a statement on developmental surveillance and screening followed by a nine-month pilot project. Participating paediatric primary care practices had difficulty completing screening during busier periods and tracking referrals (14). These challenges likely overlap with those in ASD screening. One community paediatric office-based ASD screening study showed that only 346 (26%) of the 1318 children who failed the screening test were referred for further assessment (15).

To date, there is a dearth of evidence examining the use of ASD screening tools among paediatricians in Canada and the factors that influence practice decisions. Improved understanding of factors influencing ASD screening could lead to better understanding of paediatric practice and guidance for enhanced paediatric screening, ASD detection and earlier access to interventions.The objective of the present study was to examine factors that influence general paediatricians’ decisions to include or omit ASD screening tools in clinical practice.

METHODS

Participants, recruitment and setting

Participants included 12 paediatricians from four paediatric practice groups in a large multicultural urban centre in Western Canada; 50% were men and the mean time since medical school graduation was 18 years (range four to 34 years). Seven participants practiced in hospital-based paediatric clinics and five practiced in community-based clinics. The paediatricians were contacted through their office by the study investigators. Focus groups were arranged based on physician availabilities and were conducted at their practice. Four focus groups and one interview were conducted. Ethics approval was obtained from the University of Alberta (Edmonton, Alberta) before study commencement.

Focus group procedure

Participants took part in a focus group facilitated by the principal investigator (AWSI) using a semistructured interview guide with targeted open-ended questions and follow-up prompts (Appendix 1). One participant was interviewed individually due to scheduling difficulty. All participants were queried based on the same set of questions. Each focus group/interview lasted 40 min to 60 min and was digitally recorded. Brief notes were taken by the interviewer.

Coding and analysis

The focus groups and interview were transcribed verbatim and read thoroughly by the author to generate an overall impression before initial coding. A qualitative interpretive description approach was chosen to present the participants’ perspectives by a process of intentionally remaining close to the data (16,17). All meaningful text units were coded using the NVivo10 software package (QRS International). Initial coding was categorized and reviewed with the coinvestigators (LZ and DN), and consensus on codes was established. Following initial discussion among the investigators, themes were re-examined and final themes were determined, again with care taken to stay close to the data and describe the participants’ experiences in their terms and perspectives.

RESULTS

Five broad domains of themes were identified: benefits; needs not addressed; elements that foster screening tool utility; elements that limit utility; and process of application in ASD screening. Each domain is presented below, with exemplar verbatim text quotes drawn directly from the data.

Benefits of ASD screening

Participants identified concern that insufficient screening resulted in children with ASD unnecessarily remaining undetected and untreated. A benefit of ASD screening, therefore, was identified as the likely identification of more children with ASD being missed on a routine visit. Participants identified a variety of clinical or environmental factors that, without screening, could heighten the risk of missed identification, as illustrated below:

The office is such an artificial setting, and if they’re a quiet kid and culturally that’s what’s expected, and that’s the type of [child with autism] the kid is, by watching alone you might not know. Sometimes visits are very short, so if you’re not asking specific questions about development, attention, playing, then I think you’ll miss it.

Some participants raised concern and/or uncertainty as to whether there is currently adequate assessment capacity to identify ASD particularly within busy practices, rural practices and family physician practices, and suggested a screening tool may be useful:

If you’re in a small rural center, you can’t take an hour for autism because you’ve got all of your inpatients, newborns, acute problems, so that could slip by the wayside.

Formal screening was also believed to be useful for family physicians’ office because they may have less training in paediatrics:

I feel like because they get only two months of [paediatric] training […] family docs are not as comfortable with it. If the family is not concerned, it might not get missed in the total diagnosis, but it might get picked up later.

Beyond addressing a perceived gap in detecting signs of ASD among some general paediatricians, formal screening tools were believed to offer a proactive opportunity to educate families on typical child development and help to outline challenges their physician may have identified:

[For] the families who think I’m blindsiding them a little bit; [the screening tool] helps to get the family on your side, gives them some things to think about, particular things that we look to see at a certain age, as opposed to just, “I’m concerned”.

Needs not addressed by ASD screening

Most participants referred patients to a developmental specialist without using any ASD screening tools. They relied on a tertiary centre for formal testing and diagnosis, and used the developmental specialist and tertiary team to help families to accept the diagnosis and provide further support and resources:

I always use [the local autism centre] to confirm […] because there’s no objective tests as such, I would like another opinion with an expert.

I still often refer them, because it’s a diagnosis of such impact that I think it’s beneficial to the family to hear it from more than one source. […] Counselling with that multidisciplinary team is beneficial.

Participants also identified the need for education, including normal child development, and accessibility of community resources:

Screening’s great, but from a training point of view, the trainees […] need more exposure to normal growth and development patterns, […] which screening tools, doctors, services do they implement, how closely do they follow this kid. […] There’s not a single number to call or single resource to go to, which is why I just refer to everybody because sometimes someone actually has that information.

Elements that foster screening tool utility

Participants identified screening tool characteristics that they considered to be important, which related to the administration, length, language, and validity, reliability, sensitivity and specificity of the tool. Most participants saw it being completed by a parent in the waiting room or implemented through public health. Brevity was valued and linked to a higher likelihood of completion:

[The implementation of screening] has to be relatively fast. In a primary care setting, […] 25 [items] is too long; particularly when it’s not a form you can hand to families, and when you need to add them up and questions depend on each other.

Some participants suggested including ASD screening in a more general developmental screening tool:

The thing is instead of having a million tests if you can just add a few more specific [elements] for autism to this it would make more sense.

Given substantial diversity and the multilingual context of Canada, simple language and availability of the tool in different languages were also perceived to be important:

So our clinic: low income, high density, mental health problems, socioeconomic problems and families whose first language isn’t English or whose parents have limited education […] Even if the screening tool had a lower specificity and sensitivity but is in a format that the parents can understand, read and fill out with reasonable accuracy in a timely way, I would go with that tool because I will get a higher yield in the population.

Concern was raised about cultural biases that would be important to avoid in a screening tool administered in a Canadian or other multicultural context. For instance, a participant referred to the M-CHAT (18) in identifying the following concern:

Some of this would be hard especially in refugee and immigrant families […] peekaboo and hide and seek they may not be familiar with.

Beyond recommendations for screening tool content, participants argued for the inclusion of visual items on a questionnaire, particularly if self-administered. Graphic appeal was believed to provide more clarity.

Elements that limit screening tool utility

Many participants saw their role as the frontline ‘first screen’ for ASD and are, therefore, already conducting an initial screen:

General paediatricians I see as being kind of the first screen […] for suspecting and screening for autism. So in the regular checkups, [I look for] any kind of behavioural delays or speech delays.

Some participants were confident in their ability to identify a child with a possible ASD. They generally argued that a paediatrician should be able to identify these children during routine visits, especially when there is adequate time spent with the child:

You know when you see a kid in the ER with fever, whoa, something’s just not right, the vital signs are not good, the nostrils are flaring, sats are 90%, you know within seconds; “this needs work”. I see that the same way for the kid that comes in, and you see abnormal behaviour, you would know right away.

Beyond the perspective that paediatricians should detect anomalies of function and behaviour associated with ASD, participants raised concerns over the feasibility of administering and scoring a screening test. Some doubted that every paediatrician would be able to conduct or interpret a formal screening test; hence, referral for specialized assessment was needed regardless of a screening process:

The thing is I don’t think that everyone is qualified to screen or to manage so I’m happy to get the red flags out there and go oh, this is out of my league and send off to the right person who can screen appropriately or diagnose appropriately.

Some participants identified the risk for over-relying on a screening mechanism. They argued that a screening tool was only one of many factors that affect a paediatrician’s index of suspicion for ASD, and using tools may detract from the ‘whole picture’:

I worry about losing “the art of medicine”. I have no problems telling residents that they’re smarter than me, and many of them are. But they’re not better doctors than me yet. They know more, have better memory, and are up to date. The difference is the art of medicine and experience.

Finally, participants raised concerns about the practicality of screening in a busy practice. Developmental screening was recognized as time consuming, and the inability to bill for time spent in some of the screening process was considered to be a potential barrier for some physicians.

Process of application

In general, participants’ current practice did not include the use of specific screening tools for ASD or systematic means for ASD detection; for example:

I don’t follow a particular checklist […] I tend to let it kind of free flow […] so that I can observe and watch them […] I wouldn’t say I do a very structured job of it.

Potential methods of implementing ASD screening were recommended, including having parents complete a questionnaire while in the paediatrician’s waiting room. Other participants suggested screening through routine community monitoring. For instance, in this jurisdiction, ASD screening could occur during routine immunization visits with the public health nurse, although for some, this generated opposing concerns about erroneously raising parental anxiety.

Alternative strategies identified by participants included training certain general paediatricians to conduct more formal screening:

I guess one of my concerns is if there isn’t screening and the screening is all done by [a developmental paediatrician]. He’s [or she’s] going to get overwhelmed and […] his [or her] resources should be spent on kids who have autism so I think that screening by [a general paediatrician with specific ASD training] is appropriate.

In further identifying strategies for early ASD detection, several participants identified the potential role of other fields of practice. Speech-language pathologists were believed to be effective in detecting signs of risk related to language acquisition and early child development.

Similarly, early child educators were identified as potential detectors of early signs of ASD risk. However, participants raised concern that child care workers currently may lack training and, hence, acumen for ASD detection and/or screening tool administration:

It’s a nice idea, but I don’t think it’s reasonable, because you need a high level of education and experience.

Finally, participants identified concern that systems be enacted to ensure fidelity and standard application of testing, and for accuracy of ASD detection. Standardization of care through guidelines was discussed as an important part of the screening process, as illustrated by the following participants’ desire for clinical care guidelines:

As long as we get guidelines we’ll do stuff. But if don’t have any guidelines, we’ll just do whatever we think is best, and whatever we think is best is not always the best.

DISCUSSION

A range of perspectives and factors were identified that influenced paediatricians’ use of ASD screening tools, as well as an intricate interplay of physician, patient, family and systemic factors. Participants articulated potential benefits of ASD screening, but also emphasized that implementation of screening is confronted with many challenges. These include time limitations, lack of familiarity with screening tools and ease of use, as well as a perceived redundancy between developmental surveillance and formal screening. Many also preferred to refer children to a specialized assessment team rather than add the intermediate step of screening once concerns were identified. Participants also emphasized that multidisciplinary support to families to help them navigate specialized assessment and treatment services is also essential.

These challenges are consistent with other studies investigating the implementation of screening using formal tools. For instance, King et al (14) reported that screening was often not performed if developmental surveillance suggested concerns, even in the context of an AAP-supported pilot study designed specifically to implement that approach. Practices reportedly also had difficulties making and tracking referrals (14). A recent systematic review on early detection approaches for ASD also identified time and resource constraints as well as difficulties in follow-up after a positive screen (19). Despite these concerns, there is growing evidence that developmental screening programs can be successfully implemented, particularly if there is a community physician ‘championing’ the program and resources and training available for support staff, with increased referral rates of at risk children for assessment and potential earlier intervention (20).

The study design allowed for exploration of themes affecting ASD screening by community paediatricians. While findings offer important insight about facilitators and barriers to ASD screening, the conclusions reflect interviews only within a single geographical milieu. While the participants served varying populations and a broader outlying geographical region, findings may not be generalizable to other centres, at which screening and diagnosis of ASD may be structured differently.

CONCLUSION

The present study identified factors that influence paediatricians’ perspectives and role in formal ASD screening. As screening tools improve, future research should focus on the complex issues that surround implementation, and should be responsive to physician and community needs, to implement changes in practice and improve outcomes for children.

APPENDIX. Targeted question guide and follow-up prompts

1. What is your role in diagnosis and follow-up of children with autism? How do you see your role?

   1. How were these children diagnosed?
   2. Is it from family history? Observation? Structured measures?
2. How do you determine if a child may be at risk for autism?

   1. What are the strengths of your approach?
   2. What might be possible drawbacks to your approach?
   3. What would be other methods of determining if a child is at risk for autism?
3. What would improve early detection of children with autism?

   1. What are the current barriers to early detection of children with autism?
   2. What could you do in your own office?
   3. What can be done throughout the city? Throughout the province?
   4. What are your views regarding earlier detection of children with autism?
4. What are your thoughts on autism screening?

   1. How might screening be helpful?
   2. How do you judge a good screening tool?
   3. Sensitivity? Specificity? Use in Canada?
   4. What are some of the downfalls to autism screening?
   5. What would be your goal in diagnosing a child with autism?
   6. Therapy? Diagnosis? Programs?
5. What are your thoughts on using autism screening instruments in the general paediatricians’ office?

   1. Role in primary care versus consultant paediatrician?
   2. What would encourage you to use one?
   3. What deters you from using one?
   4. What do you personally look for in a good screening tool that you would use in your office?
6. What screening tools are you aware of?

   1. Do you use any of these tools? If so, why have you chosen this tool? If not, why not?
   2. Interviewer may present a copy of the M-CHAT and ask participants if they would consider a tool such as the M-CHAT in their office.
7. Are you aware of any practice guidelines regarding autism screening?

   1. Do you agree with these guidelines? What in the guidelines to you agree or disagree with?
   2. What about the guidelines can be improved?

8. How could early screening be implemented in your practice?

9. How could early screening be implemented throughout the province?