Prognostic factors and predictors of outcome in children with autism spectrum disorder: the role of the paediatrician

Autism spectrum disorder (ASD) is considered a complex condition with different levels of severity that is characterized by impairments in the social, communicative, affective, and adaptive spheres. The characteristics of clinical symptoms can be very heterogeneous in terms of both complexity and severity. Furthermore, individuals with ASD often have different neurological, psychiatric, and medical comorbidities. International epidemiological studies report a generalized increase in the prevalence of ASD: currently, the median value of the prevalence is 6.2/1000 (1: 160; range: 1: 333–1:86) in Europe and 6.5/1000 (1: 154; range: 1: 769–1:91) in the USA, with great variability in estimates between and within geographical areas. In the latest report published by the Centers for Disease Control and Prevention [1], it is estimated that approximately 1 in 68 (14.6 per 1000) school-aged children have an autism spectrum disorder with a clear male prevalence (4.5: 1).

No differences emerge in the prevalence of ASD between America, the Western Pacific area and Europe [2], and there are no official data referring to the African continent.

There are no national prevalence estimates in Italy, and the only prevalence estimates available in Italy refer to the Emilia-Romagna and Piemonte regions. The prevalence in Piemonte is 5% in the 7–11 age group, and in Emilia-Romagna, it is 3.9% in the 0- to 17-year age group (distribution by age group: 0–2 years: 2.9%; 3–5 years: 5.5%; 6–10 years: 4.3%; 11–17 years: 3%).

In recent years, research inquired into the presence of any predictive factors of optimal outcome (OO), defined as a positive trajectory in children with a prior diagnosis of ASD who, following a therapeutic path, showed not only a significant reduction in the “core” symptoms of autism but also a positive evolution of social, adaptive, affective, and relational skills. As described by Fein et al. [3], the first definition of “best outcome” or “recovery from autism” dates to the late 1980s, when Lovaas [4] found that 47% of children diagnosed with autism, after being involved in intensive behavioural intervention, were able to attend primary school regularly and achieved a normal level of cognitive functioning. The definition of OO that Fein and colleagues suggest [3] foresees the absence of the typical symptoms of autism (deficits in the social, communicative, and affective spheres and restricted and repetitive behaviours and interests) and the presence of an average IQ, although other vulnerabilities may still be present, for example, in executive functioning or in anxious or depressive symptoms.

Over the years, there has been much debate over the role of adequate linguistic and intellectual functioning in autism. For some authors, the IQ does not represent a predictor of possible diagnostic changes from the autism classification, as reported in a systematic review [5], in which the findings showed that in only 4 out of 11 randomized studies analyzed the initial IQ assessed was associated with OO.

Kelley et al. [6] studied children between 5 and 9 years of age with a prior diagnosis of autism who were no longer showing significant symptoms and compared them to a group of typically developing children matched by age, sex and productive vocabulary. The results showed that the children in the two groups achieved the same level of language skills, but the children with a prior diagnosis of autism continued to present difficulties in the semantic and pragmatic aspects of language (theory of mind, construction of narratives, etc.).

In a recent study of highly functioning adults with autism, Otsuka et al. [7] investigated the combination of verbal and emotional competences as a predictor of good social and adaptive coping skills and found that emotional competence represents an important “mediator” of the relationship between verbal skills and positive outcomes.

Therefore, from a clinical perspective, it would be interesting to acquire more knowledge not only on cognitive factors as predictors of positive outcomes but also on the emotional and relational aspects of children at the time of the first assessment. In this way, the research results could increasingly support the planning of targeted and personalized treatment and identify the therapeutic approaches that could be most useful and effective for children with specific characteristics at the time of treatment beginning.

A growing number of therapeutic approaches exists for the management of autism, in particular to lessen the impact of symptoms on children’s functioning [8, 9]. For example, Early intensive behavioral interventions for young children with ASD (e.g. Lovaas approach, Early Start Denver Model; the TEACHH, the Applied Behaviour Analysis) currently represent the most well-known and studied methods for ASD management [8, 10]. There is also another group of research on naturalistic and psycho-developmental approaches, such as DIR-Floortime model (Developmental, Individual-Difference, Relationship-based model [11];), or DERBBI model (Developmental, Emotional Regulation and Body-Based Intervention [12];); these models presuppose that the communicative, cognitive, emotional and social skills are acquired through meaningful relationships and often through body-mediated interactions.

Finally there is an increasing number of studies have begun to examine the beneficial effects of the inclusion of animals in both recreational and therapeutic interventions [Animal Assisted Interventions, which demonstrate the usefulness of these therapies included in multidisciplinary treatments [13, 14]. However, despite the large number of therapeutic approaches, at present neither proven therapies nor preventive measures exist for the universal treatment of autism.

In this study we retrospectively investigated the predictive value of emotional-relational factors (emotional contagion and understanding of the intentions of others) in children with a diagnosis of autism who showed OO after at least 2 years of therapy, thus by achieving the lack of impairments in social and communication skills and in repetitive behaviours.

The purpose of the study is to verify the presence of emotional-relational factors that predict OO, including children’s level of functional or symbolic play since this aspect is related to executive functioning [15] and constitutes a learning factor associated with greater cognitive and communicative development [16, 17].

Our data show that 15 out of 40 children, enrolled in this study and subjected to the same model of autism therapy at least for 2 years, no longer fell into the diagnostic ASD category based on the ADOS-2, DSM-5 and clinical criteria.

Children under 30 months of chronological age who passed the clinical cut-off for the ADOS-2 Toddler Module (children at risk for ASD) and children over 30 months who passed the ADOS-2 Module 1 clinical cut-off (children with ASD) were included in the study. Research over the past two decades has revealed that an ASD diagnosis can be considered very reliable at the age of 2 years [28‐30], although many children do not receive a final diagnosis until they are older. This “latency to diagnosis” could pose a risk in that children may not receive timely help. In fact, the American Academy of Paediatrics [31] recommends that children undergo specific screening for ASD during regular medical visits at 18 and 24 months.

When studying samples under 30 months, the risk of misdiagnosis could increase; in the present research we tried to control this risk, including only children for an autism spectrum disorder clinical suspicious, from the local services; immediately afterwards, the children were assessed both directly and indirectly by a specialized team, as required by the DSM-5. Several studies have found that ASD can be reliably diagnosed in children under 3 years of ageby experienced, highly trained clinicians in specialty clinic and research settings [32, 33] and that the greatest accuracy in diagnosis of young children is achieved when using a standardized parent interview and a standardized observational measure in combination with clinical judgment [34].

In the present study, among the children under 30 months who fell into the ASD risk category at T0, after 2 years, 8 (out of 15) had a continuing diagnosis of ASD, while 7 achieved optimal outcomes (ASD-OO); thus, they lost the symptomatic characteristics (in terms of both ADOS-2 scores and clinical evaluation) that initially caused them to fall within a framework of high risk of ASD. This finding is consistent with the study of Helt et colleagues [35] who concluded that between 3 and 25% of individuals with ASD eventually lost their diagnosis. From a therapeutic perspective, all the children continued the treatment and monitoring path in the following years, as suggested by the main paediatric associations [31].

At T0, the children in the ASD-ASD and ASD-OO groups were homogeneous for chronological age, sex, and family features and were all characterized by a poor linguistic vocabulary. For this reason, language skills were not included among the predictive variables investigated in this research.

The most relevant result that emerged from the research is that the two groups instead differed from the intake (T0), in all socio-cognitive variables measured; indeed the children in the ASD-OO group initially had a higher IQ than those in the ASD-ASD group, lower severity of autistic symptoms, greater understanding of intentions, more emotional contagion, and better quality of play. Therefore, these factors have been investigated as probable predictors of optimal outcome.

Regarding the IQ, although the ASD-OO children had higher initial scores than the ASD-ASD group, the children in both groups showed improvements over time in cognitive functions (IQ) regardless of the chronological age factor and the initial autism severity score. This increase can be interpreted as the emergence of previously unexpressed potential [36]; meaning that, when symptomatology decreases, planning and organization skills can emerge more easily [25]. It should also be noted that the children included in this study followed a rehabilitation path based on a developmental approach [12] that focuses its intervention on mainly psycho-physical rather than cognitive activities, so the improvement in cognitive functioning is not a consequence of specific enhancements.

These data seem to be in line with the results of Sutera and colleagues [37], who studied children who no longer had a diagnosis of autism after 2 years of treatment (the authors do not specify which treatment) and found adequate profiles in adaptive behaviour, visual reception, motor skills, language and IQ.

Regarding the socio-affective variables, and in particular to emotional contagion, the results of this research suggest that from the moment of the first diagnosis, children who undergo a positive symptomatic evolution (ASD-OO) have a better capacity for emotional contagion, a skill that improves over time until they reach normal levels.

Additionally, a recent study [38] highlighted the presence of emotional contagion in children with autism, especially in conditions of familiarity with the interlocutor. The authors note that while children with typical development can show emotional contagion in all conditions, children with ASD are strongly influenced by the familiarity of the stimulus. If they are exposed to familiar figures (for example, parents or teachers), they are able to show a quality of emotional contagion similar to that of their peers with typical development.

In the literature, however, there are still few studies on social and affective factors as predictors of OO. For this reason, one of the variables that we considered was the presence of emotional contagion; perceiving the emotions of others generally leads to an empathic concern based on both emotional and cognitive processes. The precursor of empathy is emotional contagion, the phenomenon whereby a person’s emotions and related behaviours directly trigger similar emotions and behaviours in other people. Emotional contagion is important for personal relationships because it fosters emotional synchrony between individuals [39]. Reduced empathy has long been regarded as a contributing factor to social difficulties in individuals with ASD. In contrast to these definitions, a recent study [40] conducted with high-functioning adolescents with ASD who were given a functional MRI while watching short video clips of people suffering pain argued that there were no significant differences in brain activation between individuals with ASD and typically developing individuals. The authors therefore suggested that the mechanisms involved in emotional empathy may be preserved in high-functioning ASD individuals.

As for the ability to understand the intentions of others, which is usually present in typically developing children from 18 months of age [41, 42], it was considered as another precursor of cognitive empathy. The data of the present research describe the presence, from the first diagnosis, of these abilities in children in the ASD-OO group but not in children in the ASD-ASD group.

Understanding intentions involves recognizing that physical actions depend on the goals and intentions of the actor. Studies that have investigated this component in ASD children often have not found impairments in the ability to understand intentions in children between the ages of 2 and 5. Despite this, children’s performance was lower when the understanding of intentions also involved social sharing, as in the case of shared attention [43, 44].

These findings offer a new perspective on whether emotional contagion, the ability to understand the intentions of others and the prerequisite of empathy in general, is neither deficient nor intact in children with ASD but whether they can demonstrate it depending on the contexts and relationships in which they are involved [38].

With respect to the quality of the play, and in particular to the presence of symbolic play, in which the child was required to treat an object or a situation as if it was something else (eg, using a banana as a telephone) the results of our study suggest that at T0, the children of the ASD-OO group had a qualitatively better play (mostly functional or symbolic) than children of the ASD-ASD group (mostly stereotyped or cause-and-effect). Two years later, significant improvements emerged in the children of both groups, regardless of chronological age. In literature, symbolic play has been described as an autism-specific deficit as children with other developmental disorders tend to show comparable amount of play with typically developing children [45], but some authors have found that symbolic play was not significantly associated with verbal or non verbal ability and weakly correlated with executive functions measures, while it was associated with theory of mind measures [46]. They argued that the observed deficit in symbolic play is due to deficit in cognitive mechanism rather than the inability to understand verbal instructions and produce verbal responses in the play trails. A better understanding of the underpinning impairments associated with symbolic play thus contributes to the development of effective intervention as well as play development in children with typical development.

Finally, regarding the main objective of our study, namely, to identify a combination of predictive indicators of positive outcome, these indicators were IQ, emotional contagion, understanding of others’ intentions and level of play achieved. The results suggest that, taken individually, none of the factors considered are predictive of OO. The mere presence of one or two predictive indicators at the time of diagnosis is related to a rare possibility that the child may no longer show autistic symptoms after at least 2 years of therapy. To have a good chance (57%) of reaching an OO, at least 3 predictive indicators must be present. Only children in whom the coexistence of all 4 predictive factors was identified at the time of diagnosis showed OO in 100% of cases. If we consider that, with the exception of IQ, all the indicators have a social-affective nature, this seems to be in line with research in the neuroscientific, psychodynamic and cognitive fields [47‐51], which argues that at preschool age, a therapeutic path for autism spectrum disorders must be considered fundamental for intervening in the affective-relational components that represent the foundation for the expression of cognitive and, ultimately, communicative potentialities. The usefulness of early intervention has its roots in the concept of neuroplasticity, that is, the biological ability of the central nervous system to undergo maturation, change structurally and functionally in response to experience and adapt following an injury.

In the healthy developing brain, neuroplasticity exhibits a heterochronous cortex-specific developmental profile and is heightened during “critical and sensitive periods” of pre- and postnatal brain development that enable the construction and consolidation of experience-dependent structural and functional neural connections.

Professionals who deal with autism in both the diagnostic and intervention phases have long questioned the different developmental trajectories over time of children with ASD and ways of individualizing treatments and therapeutic projects. As our data suggest, it is important to pay attention in both in the diagnostic and prognostic phases to emotional-relational, communicative, and social factors, which could support clinicians in defining a more complete diagnostic framework and in planning a more personalized therapeutic path. Through a retrospective analysis, we found that during the first assessment, the coexistence of emotional contagion, understanding of intentions and functional and/or symbolic play can represent prognostically positive factors for OO. The instruments used in this study can be rapidly administered since they require 5 min (UOI and Play) to 10 min (TCE) and are therefore suitable for short observation sessions in paediatric patients. During the check-up that clinicians (paediatricians, neuropsychiatrists and psychologists) usually perform with infants and children during the first year of life, it is now common practice to pay attention to the neurobehavioural profile, such as sucking/swallowing reflexes, the rhythms of sleep, body postures and gaze patterns. Approximately 18 months later, further observations are added, such as the presence of gaze fixity, repetitiveness or dispersiveness during free activities and the absence of deictic gestures or joint attention. Therefore, relatively simple and increasingly refined tools that add information to the clinician’s observational picture represent an important opportunity for prognostic purposes. Currently, among the indirect assessment tools, for example, paediatricians have a parent-report screening tool such as the Modified Checklist for Autism in Toddlers, Revised (M-CHAT-R [52];), a useful for children aged 16 to 30 months of age.

Without prejudice to the mandatory nature of specific assessments for diagnostic purposes (for example, through the ADOS-2, the ADI and intellectual assessment scales), for which a second-level assessment is usually referred to the multidisciplinary team, we believe that it may be useful to integrate activities and proposals that highlight the presence of social-relational skills, such as the level of play achieved, the child’s ability to understand what others are about to do and sensitivity to the emotions manifested by others. A properly trained paediatric specialist can acquire the skills to conduct an important and qualified screening in a short time. Early diagnosis and therapeutic intervention, as demonstrated in this study, are essential for the sustainability of the system and for the optimization of treatment.

The sharing of the diagnostic path between specialists supports comparison not only of “concerns” with respect to the child’s development but also of the resources available to the child and alternatives to be offered to parents, especially during the first years of the child’s life, which represent a unique “window of opportunity” for the child’s “plastic” development. The possibility of early identification of indicators capable of guiding therapeutic paths and various prognostic opportunities represents a significant enhancement of the role of the paediatrician in providing appropriate diagnoses, the correct use of specialist services, and the ability to promptly modify potentially worsening evolutions and thus to ensure the sustainability of the health system.

A limitation of this research concerns the short period (2 years after the first diagnosis) of functional re-evaluation. Since all the children who participated are included in a four-year treatment plan, it will be interesting to monitor any changes after another 2 years.

In addition, it would be useful to increase the number of subjects of different age groups, in order to study with more accuracy the effect of early diagnosis (< 30 months) compared to a later diagnosis.