An early diagnosis of autism spectrum disorder (ASD) is a critical need in psychology due to the significant role it plays in early intervention of ASD. Although the prevalence of ASD is currently estimated at around 1.5% of the general population, the average age of diagnosis in Europe is not until 3.5 years old (Developmental Disabilities Monitoring Network Surveillance Year 2010 Principal Investigators, & Centers for Disease Control and Prevention, 2014; Salomone, Charman, McConachie, & Warreyn, Reference Salomone, Charman, McConachie and Warreyn2016). Given that an early diagnosis allows for early intervention and therefore, may lead to better outcome, there are continual efforts to provide earlier diagnosis. Diagnosing ASD is challenging and the current diagnostic criteria are related to impairments in social interaction and communication, and restricted interests and repetitive behaviors (American Psychiatric Association, APA, 2013). These symptoms are usually not detected until the first two years of life. The purpose of this study was to examine the association between parents’ early concerns related to ASD’s criteria and age at diagnosis and to estimate to what extent socio-demographic variables could predict and/or mediate the time lag between early concerns and diagnosis.

In the absence of early biological markers for ASD, the early diagnosis of ASD is closely related to the recognition of the behavioral signs and symptoms of ASD. These signs and symptoms are present early in life and usually are detected by parents (Barbaro & Dissanayake, Reference Barbaro and Dissanayake2009; Guinchat et al., Reference Guinchat, Chamak, Bonniau, Bodeau, Perisse, Cohen and Danion2012). In Spain, parents are the first to see signs in up to 79 % of children with ASD at 18 months (Fortea-Sevilla, Escandell-Bermúdez, & Castro-Sánchez, Reference Fortea-Sevilla, Escandell-Bermúdez and Castro-Sánchez2013), but these early concerns are not specific to ASD.

The first signs of ASD generally become visible prior to 36 months of life (Guinchat et al., Reference Guinchat, Chamak, Bonniau, Bodeau, Perisse, Cohen and Danion2012); furthermore, some studies found that between 30–50% of early concerns related to ASD are expressed around 12 months of life, and 80–90% by 24 months (Volkmar, Chawarska, & Klin, Reference Volkmar, Chawarska, Klin, Chawarska, Klin and Volkmar2008). In Spain, although parents may be aware of some signs early, often their child does not receive a diagnosis until 37 months of life (Díaz-Atienza, García-Pablos, & Martín-Romera, Reference Díaz-Atienza, García-Pablos and Martín-Romera2004; Salomone et al., Reference Salomone, Charman, McConachie and Warreyn2016), and the time lag between early concerns and a formal ASD diagnosis is close to 13 months (Fortea-Sevilla et al., Reference Fortea-Sevilla, Escandell-Bermúdez and Castro-Sánchez2013).

The time that it takes for children to receive a diagnosis after initial concerns are expressed by parents can be influenced by the manner in which pediatricians elicit and answer parents’ developmental concerns (Zuckerman, Lindly, & Sinche, Reference Zuckerman, Lindly and Sinche2015). Parents and pediatricians may find it difficult to differentiate ASD from other developmental disorders at an early stage (Monteiro et al., Reference Monteiro, Spinks-Franklin, Treadwell-Deering, Berry, Sellers-Vinson, Smith and Voigt2015). This is due to the similarity of symptoms of ASD to symptoms of other developmental disorders (e.g., speech and language disorders or general developmental delay). The wide heterogeneity of ASD and the similarity of symptoms between different disorders could generate a delay in diagnosis of ASD because pediatricians have a wait-and-see attitude (Daniels & Mandell, Reference Daniels and Mandell2014; Klin, Klaiman, & Jones, Reference Klin, Klaiman and Jones2015).

Language delay and communication difficulties are two of the initial ASD-related concerns frequently reported by parents (Meek, Robinson, & Jahromi, Reference Meek, Robinson and Jahromi2012). These symptoms have been associated with early detection, but also with a later diagnosis (Volkmar et al., Reference Volkmar, Chawarska, Klin, Chawarska, Klin and Volkmar2008; Zwaigenbaum, Bryson, Rogers, Roberts, Brian, & Szatmari, Reference Zwaigenbaum, Bryson, Rogers, Roberts, Brian and Szatmari2005). Other features directly related to ASD diagnosis have been delays in reaching developmental milestones (e.g. speech and language delay), slower rate of development, loss of previously acquired skills (e.g. speech regression) (Kishore & Basu, Reference Kishore and Basu2011; Mishaal, Ben-Itzchak, & Zachor, Reference Mishaal, Ben-Itzchak and Zachor2014), less frequent instances of smiling, orienting, and vocalizing to people (Maestro et al., Reference Maestro, Muratori, Cesari, Cavallaro, Paziente, Pecini and Sommario2005), lack of joint attention (Kishore & Basu, Reference Kishore and Basu2011), poor eye contact, lack of play and interaction, and hand flapping and toe walking (Mandell, Novak, & Zubritsky, Reference Mandell, Novak and Zubritsky2005). Concerns related to repetitive behaviors often emerge later, and are often reported for children with other developmental disorders too (Mooney, Gray, & Tonge, Reference Mooney, Gray and Tonge2006).

Beside these early difficulties, several factors have been shown to influence age of early concerns specifically related to ASD. For instance, having another child without ASD has been associated with an early detection and diagnosis, as parents could compare the development of the child at risk of ASD with their child with typical development (TD) (Fountain, King, & Bearman, Reference Fountain, King and Bearman2011; Herlihy, Knoch, Vibert, & Fein Reference Herlihy, Knoch, Vibert and Fein2015; Mishaal et al. Reference Mishaal, Ben-Itzchak and Zachor2014; Rosenberg, Landa, Law, Stuart, & Law (Reference Rosenberg, Landa, Law, Stuart and Law2011). In general, higher maternal education would be associated with more knowledge of child development. Shattuck et al. (Reference Shattuck, Durkin, Maenner, Newschaffer, Mandell, Wiggins and Cuniff2009) showed that a lower level of maternal education and younger maternal age were associated with older children’s ages for early concerns related to ASD. However, Mishaal et al. (Reference Mishaal, Ben-Itzchak and Zachor2014) did not find significant correlations between parental age and educational attainment and age of ASD diagnosis. Furthermore, the gender of children with ASD has been another variable of scientific interest that might have an impact on the age in which children are diagnosed. Some studies have found no differences between gender and the age of diagnosis (Begeer et al., Reference Begeer, Mandell, Wijnker-Holmes, Venderbosch, Rem, Stekelenburg and Koot2013; Fountain et al., Reference Fountain, King and Bearman2011). Nevertheless, parents may expect more social behavior from girls, and encourage them to act more socially, which could delay detection.

There is no clear consensus about which features are more important in the earlier detection of ASD and which variables have a significant impact on the time lag between type of parents’ early concerns and the diagnosis of ASD. It is important to consider features that improve clinical practice in eliciting these concerns and as a result, help to improve the early diagnosis of ASD. In addition, an early and accurate diagnosis helps parents cope with the stress related to the disorder (Pozo & Sarria, Reference Pozo and Sarria2014). Thus, this retrospective study investigates the association between parents’ early concerns related to ASD symptom criteria and age at diagnosis. In particular, we investigated to what extent socio-demographic variables (i.e., type of parents’ early concerns, parental level of education, parental age, child’s age, sibling status, and child’s gender) predict the time lag between early concerns and diagnosis.

Method

Participants

A total of 48 families from Granada (Spain) of a child with ASD (range: 4.1–10.1 years old) were enrolled in this study. All parents signed the informed written consent before participation, and The Ethics Committee of the University of Granada approved the methodology of the study. Inclusion criterion for each child was having an ASD diagnosis based on Diagnostic and statistical manual-text revision (DSM-TR-IV; APA, 2000) and the Autism Diagnostic Interview-Revised (ADI-R; Le Couteur, Lord, & Rutter, Reference Le Couteur, Lord and Rutter2003) or Autism Diagnostic Observation Schedule-WPS (ADOS-G; Lord, Rutter, DiLavore, & Risi; Reference Lord, Rutter, DiLavore and Risi2002). The diagnosis was confirmed by an independent tester with the Guilliam Autism Rating Scales (GARS; Guilliam, Reference Guilliam2004). In addition, one of our purposes was to investigate whether having an older sibling with typical development could have delayed or accelerated parents’ doubts. As a result, an additional inclusion criterion was that for the children with older siblings, the older siblings must be deemed typically developing in order to be included in this analysis.

Parents were interviewed and they completed an open-ended questionnaire measuring several aspects related to detection of ASD and diagnosis. This kind of questionnaire enabled us to obtain more spontaneous information and to run both qualitative and quantitative analyses (Chamak, Bonniau, Oudaya, & Ehrenberg, Reference Chamak, Bonniau, Oudaya and Ehrenberg2011; Guinchat et al., Reference Guinchat, Chamak, Bonniau, Bodeau, Perisse, Cohen and Danion2012). Families were asked (1) about early concerns that made them think that their child could have ASD, (2) what was their child’s age when these signs were detected, and (3) when the child received the diagnosis of ASD.

Coding procedure and analysis

We focused on parents’ early concerns about ASD. The coding procedure consisted of breaking down early concerns into 2 fields related to two domains from DSM–5 (APA, 2013): (1) Social communication (SC), and (2) restrictive and repetitive behaviors and interests (RRBI). Moreover, because many parents usually reported concerns that were not specific to autism criteria (Guinchat et al., Reference Guinchat, Chamak, Bonniau, Bodeau, Perisse, Cohen and Danion2012), we included a third category (3) other concerns non-specific to ASD (OC). The operational definitions of what variables were considered as SC, RRBI, or OC are mentioned in Table 1. Two independent coders identified the three different types of concerns. Mean percentage agreement (based on Cohen’s Kappa inter-rater agreement measurement) was over 95% for all symptoms. Disagreements were resolved through discussion.

Table 1.

First Signs related to ASD Reported by Parents

Note: SC = Social communication signs; RRBI = Restrictive and repetitive behaviors and interest signs; OC = Other developmental concerns non-specific to ASD.

Data analyses

All statistical analyses were performed using Statistica 13 software. To evaluate relationships between the two outcome variables (i.e., child’s age at suspicion, and time lag between suspicion and diagnosis) and four predictor variables (i.e., sibling status, SC, RRBI, and OC) bivariate Pearson correlations were conducted. To determine the influence of each predictor, we run several multiple regression analyses. Furthermore, to examine potential meditational effects, path analyses were computed. Parental age, education, and child’s gender were entered as covariates. We consulted the bootstrapping method with bias-corrected confidence estimates to test our results. The 95% confidence interval of the indirect effect was obtained with 5000 bootstrap samples.

Results

Forty-eight families were enrolled in this study. Descriptive statistics for all study variables are presented in Table 2. From the whole sample, 12.5% (n = 6) reported child’s age at diagnosis less or equal to 24 months; 54.2% (n = 26) reported child’s age at diagnosis between 25 and 35 months; the remain 33.3% (n = 16) reported child’s age at diagnosis longer than 36 months.

Table 2.

Socio-demographic Variables from Families with a Child with ASD

Note: SC = Social communication signs; RRBI = Restrictive and repetitive behaviors and interest signs; OC = Other developmental concerns non-specific to ASD.

^a Months. ^b Years. ^c Post-secondary and university education. ^d Having an older sibling with typical development.

Bivariate Pearson correlations are presented in Table 3. Correlations between parents’ early concerns (i.e., SC, RRBI, and OC), parental level of education, parental age, child’s age at suspicion, time lag between suspicion and diagnosis, sibling status, and child’s gender were computed. We focused on two of these variables: Child’s age at suspicion and time lag between suspicion and diagnosis. Younger child’s age at suspicion was related to more number of SC reported by parents (p < .05), longer time lag between suspicion and diagnosis (p < .001), and having an older sibling with typical development (p < .001). Additionally, longer time lag between suspicion and diagnosis was positively related to having an older sibling with typical development (p < .05).

Table 3.

Correlations Coefficients Values between Symptoms related to ASD, and Variables related to Parents and Children

Note: SC = Number of social communication signs; RRBI = Number of restrictive and repetitive behaviors and interests signs; OC = Other developmental concerns non-specific to ASD.

*p < 0.05; **p < 0.01

We then conducted two multiple linear regression analyses with outcomes child’s age at suspicion, and time lag between suspicion and diagnosis respectively. The results of these regressions are presented in Table 4, including standardized regression coefficients for each predictor.

Table 4.

Regression Analyses to Determine the Influence of Each Predictor on Child’s Age at Suspicion and Time Lag between Suspicion and Diagnosis

Note: SC = Number of social communication signs; RRBI = Number of restrictive and repetitive behaviors and interests signs; OC= Other developmental concerns non-specific to ASD.

*p < .01, **p < .001

Child’s age at suspicion. This model accounted for 56% of the total variance, F(4, 42) = 15.612, p < .001. From all potential predictors, the number of SC and having an older sibling with typical development were the statistically significant variables (p < .005, and p < .001 respectively).

Time lag between suspicion and diagnosis. This model accounted for 35% of the total variance, F(5, 41) = 5.896, p < .001. In this model, from all potential predictors, child’s age at suspicion was the only statistically significant variable (p < .05).

Path analysis. To describe a comprehensive model, we carried out path analyses. We tested a path model that was based on the regression analyses and was consistent with the actual timeline of events (i.e., that having an older sibling with typical development influences the potential signs noticed by parents, which can in turn influence their suspicions of ASD, and finally the diagnosis).

The number of SC was related to child’s age at suspicion, suggesting that it could be a potential mediator of the relationship between sibling status and the child’s age at suspicion (see Figure 1). To test this possibility, we estimated indirect effect on number of SC via child’s age at suspicion (see Table 5). The results showed that there was a significant indirect effect of number of SC on child’s age at suspicion via sibling status (95% confidence interval excluding 0).

Figure 1.

Results from mediation analyses.

Note: Coefficients are standardized β; SC= Number or social communication signs; RRBI= Number of restrictive and repetitive behaviors and interest signs; OC= Other developmental concerns non-specific to ASD * p<.01; ** p<.001

Table 5.

Estimated Indirect Effects from Sibling Status Via Path Analysis

Note: SC = Number of social communication signs; RRBI = Number of restrictive and repetitive behaviors and interests signs; OC = Other developmental concerns non-specific to ASD.

Regarding the second model, child’s age at suspicion was related to time lag between suspicion and diagnosis, suggesting that it could be a potential mediator of the relationship between sibling status and the time lag between suspicion and diagnosis. To test this possibility, we estimated indirect effect on child’s age at suspicion via time lag between suspicion and diagnosis. The results showed that there was a significant indirect effect of child’s age at suspicion on time lag between suspicion and diagnosis through sibling status (95% confidence interval excluding 0). Moreover, results showed another significant indirect effect of SC on child’s age at suspicion through sibling status.