Table 1 presents demographic estimates of the population subdivided into age cohorts. Children with ASD were predominantly male (82.1%) and came from two parent households (70.6%) and households where the highest educated household member had at least graduated from high school (76.9%). There were few significant demographic differences found between age cohorts. Children in the younger cohort had an earlier age of reported parental concerns (p<.01), service initiation (p<.001) and ASD diagnosis (p<.001) than children in the older cohort. Approximately half (49.2%) of all children received services before their diagnosis (children could receive services for ASD symptoms or developmental delays before a formal diagnosis).

Before children reached 3 years of age, 63.8% had parents who had reported a concern to their child's doctor, 18.5% had received an ASD diagnosis, and 32.3% were receiving services. Additional differences between age cohorts are presented in Figure 1. Children in the younger cohort were more likely than children in the older cohort to have had a parent report a concern to their child's doctor (65.3% vs. 62.0%), received an ASD diagnosis (23.5% vs. 12.5%), and started receiving services (37.3% vs. 26.1%) before turning three years old.

Table 2 describes the behavioral and developmental characteristics of the population. Behavioral (87.9%) and nonverbal communication concerns (88.8%) were the most commonly endorsed concerns among parents who reported concerns to their child's doctor. Differences were examined between age cohort at the time of the survey, with parents of younger children being more likely to have reported a concern about their child's verbal communication (p<.05) than parents of older children.

This study found several child characteristics and parental developmental concerns associated with the reported age when parents first spoke to their child's doctor or other health care provider about a developmental concern, the age the child received an ASD diagnosis, and the age the child started receiving services.

Children diagnosed with ASD whose parents reported initial concerns about verbal communication were younger when their parents reported these concerns, when they received an ASD diagnosis and when they began to receive services. Children whose parents noted specific concerns about their child's nonverbal communication and the presence of unusual gestures and movements were diagnosed with ASD at a younger age, findings that are consistent with those in another national sample of children with ASD (Maenner et al., 2013). It has been reported elsewhere that children diagnosed with speech delays and language deficits or with unusual gestures and movements also receive earlier diagnoses (Bickel et al., 2015; De Giacomo and Fombonne, 1998; Mandell et al., 2005), which may additionally be related to the greater availability of services (e.g. speech, occupational, or physical therapy) at a younger age.

The study also revealed a later age of reported parental concern among children whose parents first had noticed behavior difficulties. It has been suggested that a parent may not report a behavioral concern to their child's doctor until after the behaviors are no longer seen as developmentally appropriate. It may be particularly challenging for a parent to recognize behaviors specific to ASD when they occur during early, otherwise, typical development (Herlihy et al., 2015; Werner et al., 2005). This may be compounded by the fact that many children with ASD present with other developmental delays that have a high level of symptom overlap (e.g. ADHD) (Rao & Landa, 2014). For these reasons, Herlihy and colleagues (2015) emphasized the importance of encouraging parents to report any concerns to their child's doctor as early as possible, while similarly encouraging clinicians to probe parents for any concerns, particularly about their older children. Intriguingly, later ages when parents reported behavioral concerns to the doctor did not translate into later ages of first treatment, suggesting despite potential reporting delays, children are receiving important interventions.

Interestingly, there was no relationship found between concerns about motor problems and any of the age outcomes. Motor function has received considerable attention given the notable delays in gross and fine motor skills among children diagnosed with ASD (Fournier et al., 2010). It is possible that much like behavioral problems, parents do not express concern about their child's motor functions until the problem persists past a developmentally appropriate age, but perhaps not to the extent of the previously mentioned behavioral problems.

There were few demographic characteristics associated with the age outcomes of interest. Consistently, younger children had earlier ages for all outcomes, which is consistent with a decline in age of diagnosis in recent cohorts (Fountain et al., 2011), a recent increased recognition of ASD symptoms, and a growing awareness of the need for early identification over the past decade (Matson and Kozlowski, 2011). Study findings may reflect the earliest efforts to lower the age at which children are screened for potential developmental disabilities in the younger cohort (Daniel et al., 2009). A recent examination of children aged four years of age in 2010 and children aged four years of age in 2006 using data from the Autism and Developmental Disabilities Monitoring Network also revealed significant decreases in the age of first diagnosis and age of first comprehensive evaluation in a younger cohort (Christensen et al., 2015).

Children who have an older sibling may have had an earlier age of reported parental concern because they have parents who are more attuned to their child's development, or are more aware of deviations from typical development. Other researchers have found that children with older siblings who were also diagnosed with ASD have parents who report earlier concerns to their child's doctor (Herlihy et al., 2015), and that birth order is a predictor of age of ASD diagnosis (Bickel et al., 2015).

The lack of racial or ethnic differences in the age of diagnosis adds to the currently mixed literature on the subject (Daniels and Mandell, 2013). A later age of service use among Hispanic children relative to Non-Hispanic White children could be the result of the overall difficulties minority populations encounter in attempting to access services (Liptak et al., 2008; Thomas et al., 2007). Hispanic children with ASDs receive fewer specialty services and procedures while having higher unmet services needs when compared to their non-Hispanic peers (Broder-Fingert et al., 2013; Magaña et al., 2013).

Children with ASD who were also diagnosed with an intellectual disability had an earlier age of parental reported concern and service initiation. An intellectual disability may prompt earlier parental concerns about development generally which can result in earlier recognition and intervention by clinicians. In addition, intellectual disability may present as part of a syndrome where genetic or other physical markers are present during the pre- or perinatal period. This combination can also lead to earlier treatment and diagnosis of ASD (De Giacomo and Fombonne, 1998).

A number of study limitations should be considered. First, the study data come from a probability-based national sample of children with special health care needs who were previously diagnosed with ASD who were initially identified in NS-CSHCN and then successfully re-contacted for Pathways. Consequently, the response rate is the product of two survey response rates. Estimates based on telephone surveys with low response rates may be unreliable due to selection biases resulting from sampling, nonresponse, lack of coverage of households without telephones, and respondent classification and reporting errors. The impact of these potential biases on the results presented in this study is unknowable. Therefore, the representativeness of the estimates cannot be confirmed because biases may remain that are not related to known demographics. However, sample weights were calculated in accordance with best practices for sample surveys which included adjustments for known demographic correlates of nonresponse that have been shown to minimize the potential for non-response bias (CDC, 2011). The composition of the analytic sample of children in the current study was compared to the population of children currently diagnosed with ASD from the 2011-2012 National Survey of Children's Health (NSCH), a nationally representative sample of children in the United States. Although children did not differ on the average age of ASD diagnosis, nor on several key demographics (sex, race, household income, family type), children in the current study come from higher educated households than children in the NSCH. This may limit the generalizability of findings, as age outcomes and parental concerns could look differently in children from households with less educated parents.

In addition, Pathways is a cross-sectional survey. Data about first concerns and diagnostic history were collected at only one point in time for children whose ages ranged from 6 to 17 years of age. All parents were asked to reflect over a time period that extended back almost the entirety of their child's life as it related to their child's developmental history and diagnostic experiences. For this reason, parents of the oldest children needed to reflect over the longest reference periods, which could introduce the possibility for differential recall biases as it relates to the accuracy of the parent's report. Additionally, older children had a longer time period than younger children to have experienced all three age-outcomes which translated to later outcomes in the older cohort of children. However, we used interactions terms to explore the difference between cohorts and specific concerns and demographic characteristics and found few differences. We also adjusted appropriate models by the child's age and the age of the first parental concern in the hopes of minimizing this potential bias. Finally, the survey uses parent reports as the sole basis of developmental history and diagnostic data, lacking clinical validation. Children were required to have an ASD diagnosis to be included in the study, but there are likely children in the population who qualify for an ASD diagnosis but simply have not yet been diagnosed. These children would be missed by parental report.

Future studies would benefit from a longitudinal cohort design which would minimize the risk of recall bias, allow for children to be followed for an extended and equal period of time, and accommodate survival analyses which could provide additional insight into the relationship between parental concern, age of ASD diagnosis and age of first service.

As the number of children diagnosed with an ASD has increased in the past decade and efforts to identify and intervene with children early are heightened, it is important to examine the experiences of families raising children with developmental disabilities in the general population. Guidelines offered by the Council on Children with Disabilities of the American Academy of Pediatrics (AAP) (Johnson and Myers, 2007) underscore the importance of early identification, through the use of standardized developmental screening instruments (rather than narrowing attention to just parents' pressing concerns (e.g. behavioral problems)), as well as the importance of referrals to developmental and educational programs for positive outcomes. The lowered ages found for parental reported concern, ASD diagnosis, and service usage for the younger cohort of the present study are consistent with recommendations. However, despite these improvements, overall fewer than 1 in 5 children with ASD had received their ASD diagnosis before age 3, and children whose parents reported concerns to their child's doctor or other health care provider at a later age received a delayed ASD diagnosis, and perhaps more importantly were delayed in entering support services.