Background
Autism spectrum disorder (ASD), a neurodevelopmental disorder defined by impairment in social communication and the presence of restricted repetitive behaviors [
1], has steadily increased in reported prevalence over the past decade [
2]. Pediatricians are frequently involved in ASD diagnosis in pre-school age children in Canada [
3]. Numerous guidelines have been published for diagnostic assessment for ASD, with varied recommendations for personnel and tools in the assessment [
4‐
9]. A concern about models for diagnostic assessment is how they may extend the waiting period to receive intervention. Waiting for an ASD diagnostic assessment occurs during a critical period of brain development [
10] and an extended wait time may delay receipt of intervention and reduce effectiveness [
11]. To date, little work has been done looking specifically at wait times for ASD diagnosis, and how diagnostic practices influence wait times.
Age at diagnosis has often been used as a proxy to understand wait times; however, other factors beyond diagnostic demand and supply influence this metric. Milder ASD subtypes have been associated with a later age at diagnosis [
12]. These children may not show significant impairment associated with ASD until their skills are exceeded by social demands, which will occur later for more mildly affected children. Additional studies have shown that severe language impairment/regression, unusual mannerisms, and toe walking were features of the clinical presentation associated with younger age at diagnosis [
13,
14]. Co-occurring or alternative diagnoses such as attention deficit hyperactivity disorder may delay diagnostic evaluation for ASD [
15]. Factors external to the child also decrease the age at diagnosis, including older maternal age [
15] and having relatives with ASD [
14], both of which may be indicative of caregivers who are more aware of the possibility of ASD. Lower socioeconomic status, being a visible minority, and living in a rural setting are all associated with an older age at diagnosis [
13,
14,
16]. The number of potential confounders makes it difficult to isolate the impact of wait times for assessment on the diagnosis age.
Few studies have reported current ASD diagnostic practices and their association with wait times. One USA study evaluated factors related to wait times for diagnosis, which was 13 months on average in their sample [
17]. Reported associations in this study were mostly between wait times and patient demographic factors, but there was no association between the use of a standardized diagnostic tool and mean age of first ASD diagnosis. One chart review of 70 children’s cases from Scotland found that receiving more information prior to the assessment, such as contextual information and results of other assessments, reduced the number of assessment visits needed and decreased the total wait for diagnosis [
18]. This sample was taken from only eight children’s services, and therefore provides little insight on between-provider variability in practice and its impact on wait times.
Further information on diagnostic assessment and wait times from diverse clinical practices is needed to inform ASD service planning in constrained health care systems. The study objectives were to (1) document the diversity of practices of Canadian pediatricians with regard to their diagnostic assessment for ASD; and (2) to identify the elements of clinical diagnostic practice that are associated within a longer wait times for ASD diagnosis.
Discussion
This study is the first to examine detailed self-reported practice patterns and wait times for ASD diagnosis among pediatricians across Canada. There was wide variation in reported practices for the diagnosis of ASD, including personnel and tools used in the assessment. There were no two identically composed MDTs across the country, which may reflect the lack of uniformity in guidance documents regarding the necessary personnel for ASD diagnostic assessment [
4,
5,
9,
24].
A longer time spent on the assessment was significantly associated with longer total wait from referral to diagnosis, indicating that clinical decisions regarding necessary assessment elements have an important impact on wait times for families. Physicians with longer assessment times will likely have fewer available clinic slots and as a result will see fewer patients, lengthening their queue for ASD assessment. Use of a diagnostic tool was not significantly associated with total wait time, though statistical power may have limited our ability to detect a significant association, making this question worthy of further study. A wide range of reported wait times between the first clinic visit and the completion of the assessment was also observed. This period may represent a particularly stressful time for families as they likely know their child is being assessed for ASD but do not have the diagnosis required (in most jurisdictions) to access intervention. Each component of diagnostic delay may put the child at risk for suboptimal developmental outcomes [
11].
Given the increase in prevalence of ASD diagnoses [
2], demand for diagnostic assessments may exceed available resources, leading to wait times. As such, our results are relevant to all jurisdictions that provide publicly funded ASD diagnostic assessments. A study from the UK using data from 2001 to 2002 found that only 19% of assessments occurred within the recommended time frame of 30 weeks [
25]. The time period between referral and receipt of ASD diagnosis has been repeatedly described as a highly stressful time for families, increasing the impetus to provide timely access to diagnosis [
26,
27]. Parental stress and dissatisfaction are significantly associated with a higher number of professionals seen during the diagnostic process [
26,
28]. To meet diagnostic demand, clinical guidelines for ASD diagnosis have focused on the need to train more providers to perform ASD assessments [
4] and to fund more MDTs [
5]. Our results suggest that further work is needed to determine the optimal balance between accuracy, quality, and efficiency in ASD assessments, allowing a higher volume of assessments to be completed and reducing wait times.
Our analysis revealed intriguing findings regarding the types of clinicians involved in the diagnostic assessment. Many pediatricians conducted developmental consultations (with wait times and billing costs) but did not provide ASD diagnoses in their practices. Though general pediatricians had a trend toward a shorter time to first visit of the diagnostic assessment, they trended toward a longer overall time to completion of the assessment. General pediatricians who diagnose ASD may avert the need for an additional referral to a developmental pediatrician or MDT in the diagnostic journey for families. Further work is needed to ensure that a general pediatrician diagnosis of ASD is accurate, acceptable to families, and that it is completed in an expedient manner. A qualitative study of general pediatricians in Ontario, Canada has shown varying willingness to diagnose ASD in their practices [
29]. While many of those interviewed felt they were able to provide quality assessments that helped families access services faster than they would have if they waited for a subspecialist, there were identified barriers to conducting ASD diagnostic assessments, including uncertainty about the role of the general pediatrician in ASD diagnosis, inadequate training, and inadequate remuneration. In order to deal with diagnostic uncertainty, solo general pediatricians talked to other clinical staff in their office or reached out to subspecialist colleagues for “hallway consultations.” Similar to the results of this survey, participants reported differing use of diagnostic tools, including the ADOS, an abbreviated form of the ADOS, and using a screening tool such as the Modified Checklist for Autism in Toddlers [
30] to structure their diagnostic interview. Participants identified barriers beyond identifying ASD, including disclosure of the diagnosis to families and knowledge of available resources in a fragmented system; as such, efforts to increase diagnostic capacity in general pediatricians must consider aspects of diagnostic assessment beyond accuracy, including communication skills and availability of service navigation.
A number of study limitations were present. The low response rate and the voluntary nature of the survey increase the potential for volunteer bias with respondents more likely to have an interest in ASD. As such, caution should be taken in extrapolating the results to all Canadian pediatricians. The study sample did not include psychologists and psychiatrists, who may also be involved in the diagnosis of ASD. Wait times in the survey were self-reported by clinicians, leading to the potential for reporting bias. To minimize the possibility of recall bias, participants were asked to report their current wait time, as opposed to estimating an average over a previous interval. Participants were carefully instructed in the consent form that they would be asked for their current wait time for ASD diagnostic assessment and to have this information prepared. This information could not be verified and is an acknowledged limitation of the work. The total wait time in this analysis only considered the wait time for one referral; this will underestimate wait times for diagnostic journeys that include non-productive referrals to non-diagnosing practitioners at early stages of the assessment pathway. The use of the single wait time for ASD diagnosis was chosen so that the influence of diagnostic practices on wait time would not be diluted by adding the wait times for other physician referrals. The total number of respondents limited the statistical power, though the response rate is similar to other Canadian physician surveys [
31]. With a larger sample size, additional significant determinants of wait times may have been identified. The findings are nevertheless highly informative as they represent the first study looking at this critical question of capacity in the face of growing ASD prevalence.
Results of this study have identified an important association between the length of the ASD diagnostic assessment and wait times, although far more research is needed to determine the optimal balance between efficiency and comprehensiveness for a complex condition such as ASD. Further analysis is needed at the individual patient level, such as through health administrative or insurance databases, to determine the proportion of children/adults receiving their diagnoses from various providers/teams. This could be compared across jurisdictions with differing requirements for ASD diagnosis, including analysis of resource use and wait times. The variability in diagnostic assessment models reported in this study demands further evaluation of the accuracy of assessment types, such as MDT versus solo assessment and subspecialist versus general pediatrician assessment. Any system-wide strategy for improving efficiency of ASD diagnostic assessments should have accompanying qualitative examination related to uptake of new recommendations and requirements in all relevant stakeholders.