Using the Griffiths Mental Development Scales to Evaluate a Developmental Profile of Children with Autism Spectrum Disorder and Their Symptomatologic Severity

We implemented a longitudinal observational cohort study. For both assessment times (T0 and T1), the same measures were used, 12 months apart: GMDS and ADOS-2. The investigation was approved by the Ethics Committee of the NHS Local Health Unit and was conducted according to the principles established by the Declaration of Helsinki. All subjects and caregivers were informed about the purpose of the study and the confidentiality of data processing. The subjects’ parents or tutors-by-law provided written informed consent to participate in the study.

A multidisciplinary team from the Regional Centre for Autism of the Abruzzo Region, Italy, with experience in ASD diagnosis, performed the neuropsychiatric assessment based on standardised measures and clinical observations according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; 1). Evaluation of ASD symptoms and the child’s functional development was conducted by a trained clinical psychologist (please refer to the ‘Measures’ section). After the evaluation, caregivers were provided with assessment outcomes, including diagnosis and recommendations for intervention.

The inclusion criteria were patients aged between 18 and 48 months at their first clinical consultation, the availability of the GMDS at both baseline and follow-up (i.e. T0 and T1) and confirmation of the diagnosis if they were considered at risk in the first evaluation. Children who presented other significant medical conditions (e.g. other significant genetic disorders, other neurological disorders, epilepsy, significant hearing and visual sensory deficits, traumatic brain injury) were excluded. A total of 79 patients were recruited for the study and followed for ASD symptoms between June 2019 and November 2020. Four participants were excluded because the diagnosis was not confirmed at T1, and eight participants dropped out for various reasons. The final research sample was 67 children with a mean age of 33.08 months (SD 8.20 months), composed of 45 males (67.2%) and 22 females (32.8%). The GMDS Practical Reasoning scale was not administered to all children at T0 as it cannot be administered at an age below 24 months, as specified in the Measures section. Furthermore, because the scale is complex, some children were not able to take it and were not included in the measurement. Thus, the data from this scale are available for 38 subjects from the total sample at T0. The assessment of the scale is therefore available for the total sample at T1 (after 12 months). Demographic and clinical data of the participants for each timepoint are reported in Table 1.

The GMDS [30, 31] was developed as a child development assessment tool for children aged 0–24 months. The extended form can be used up to the age of 8 years. Hence, the instrument consists of two scales: the 0–2 scale and the 2–8 scale. The GMDS 0–2 and the extended form have been revised, leading to the development of the version for children up to 8 years of age: GMDS-ER [43, 44]. These scales measure a child’s abilities through reference to the following six sub-scales: sub-scale A, Locomotor, measuring postural development, deambulation and higher-order motor skills; sub-scale B, Personal-Social, concerning the subject’s ability to adapt, personal autonomy and social interaction skills; sub-scale C, Hearing and Language, tracing the main stages of linguistic and communicative development and assessing attention to sounds, production of vocalisations and early lexical development; sub-scale D, Eye and Hand Coordination, assessing visual control and fine-motor dexterity skills; sub-scale E, Performance, concerning visual perception awareness, including working speed and accuracy; and sub-scale F, Practical Reasoning, from the age of 2 years, it consists of items that assess the ability to use knowledge learned from the environment to solve problems and to understand mathematical concepts and moral problems.

A repeated measures t-test was performed to determine whether there was a difference between T0 and T1 samples of the GMDS sub-scales and GQ to understand whether any of the abilities had worsened during the time interval considered. Cohen’s d was calculated to assess the effect size. The alpha level was set at 5% multiple comparisons were adjusted by Bonferroni method. Accordingly, the p-values associated with the statistical tests carried out have been inflated by the factor 7, representing the number of comparisons.

To explore the association between GDMS and gravity of the symptoms, we performed multivariate regression analysis. Specifically, multivariate regression allows one to assess whether more than one independent variable (X_i) could predict a dependent variable (Y). The advantage of using multivariate regression is that it could account synchronically for the variation in the independent variables in the dependent variable, thus allowing a better estimation than considering independent variables individually. Specifically, we assessed whether the calibrated SA and RRB scores of the ADOS-2, at T1 could be affected by the GMDS measured at the same time. Multicollinearity was detected by using the variance inflation factor (VIF), where values < 10 were considered acceptable [45].

A repeated measures t-test revealed significant reductions in GMDS abilities between T0 and T1 (Table 1). Specifically, Locomotor abilities decreased significantly from T0 (M₀ = 76.78, SD₀ 20.57) to T1 (M₁ = 66.85, SD₁ 19.87; t(66) = 4.84, p < 0.001, d = 0.59); Personal and Social abilities decreased significantly (t(66) = 3.23, p = 0.014, d = 0.39) between T0 (M₀ = 69.07, SD₀ 19.53) and T1 (M₁ = 61.93, SD₁ 23.27); Hearing and Language abilities decreased significantly (t(66) = 3.99, p < 0.001, d = 0.49) between T0 (M₀ = 57.07, SD₀ 21.26) and T1(M₁ = 48.45, SD₁ 16.12); Eye and Hand Coordination decreased significantly (t(66) = 3.26, p = 0.014, d = 0.40) between T0 (M₀ = 67.69, SD₀ 19.92) and T1 (M₁ = 60.67, SD₁ 19.94); and Practical Reasoning decreased significantly (t(37) = 3.85, p < 0.001, d = 0.63) between T0 (M₀ = 70.47, SD₀ 12.35) and T1 (M₁ = 57.66, SD₁ 21.34). Performance abilities did not change between T0 and T1. The GQ measured from the GMDS showed a significant decrease from T0 (M₀ = 67.13, SD₀ 16.82) to T1 (M₁ = 59.78; SD₁ 17.97; t(66) = 3.98, p < 0.001, d = 0.50). Results are shown in Fig. 1.

The regression model showed that the scores of the GMDS scales at T1 significantly predict the calibrated SA score of the ADOS-2 measured at the same evaluation, showing a good fit of the data and explaining 47% of the variance (F(5, 62) = 8.93, p < 0.001, R² = 0.47, f² = 0.88). Among the independent variables, we found Personal and Social skills (β =  − 0.04, SE = 0.02, t =  − 2.21, p = 0.031) and Practical Reasoning skills (β =  − 0.07, SE = 0.02, t =  − 3.80, p < 0.001) as significant predictors. Thus, keeping the other variables constant, these results indicate that for each sub-quotient point increase in Personal and Social skills we could expect a decrease of 0.04 in the calibrated SA score, and for each sub-quotient point increase in practical reasoning skills we could expect a decrease of 0.07 in the calibrated SA score.

The regression model also significantly predicts the calibrated RRB score of the ADOS-2 measured at the same evaluation, explaining 27% of the variance (F(5,62) = 3.54, p = 0.005, R² = 0.27, f² = 0.37), where Hearing and Language skills (β = 0.15, SE = 0.004, t = 3.61, p = 0.001) represent significant predictors of the RRB score. These results indicate that for each unit increase in hearing and language skills we could expect an increase of 0.15 in the RRB score. The results are shown in Figs. 2 and 3.

The Personal-Social scale assesses the child’s abilities in activities of daily living and the ability to interact with other children. We found that the Personal-Social skills measured by the GMDS affect ADOS-2 outcomes. Specifically, an increase of 1 point in this sub-scale could lead to a decrease in the calibrated SA score. The result obtained for this scale is not surprising, as a deficit in social skills is the core symptom in ASD [13, 18]. In particular, the Personal-Social scale includes items (e.g. ‘smiles’, ‘follows moving people with his eyes’, ‘claps in imitation’) that assess social learning and the acquisition of social habits, indicative of this stage of personal-social adaptation. Social experiences contribute to the development of a set of cognitive processes that enable individuals to interact appropriately with their environment [52].

From birth, infants appear to be predisposed to develop various social skills, such as the ability to identify social agents, mutual affiliation and a preference for social over non-social models [45]. The set of cognitive processes that drive social skills are grouped under the name social cognition. Many researchers [13, 53, 54] have shown that these skills, including theory of mind, are evolutive skills that develop along a continuum, following specific stages. In ASD, there is an alteration in the timing of the acquisition of functions and precursors underlying the formation of social cognitive skills [13, 55]. Although detecting social deficits in the first year of life can be challenging, there is growing evidence that children subsequently diagnosed with ASD show a reduction in attention to social scenes as early as 6 months [56]. Other indicators have also been reported as early signs of social problems that are present in early childhood, such as social interest, shared affection, looking at faces and eyes and responding to one’s name [14].

With this scale the examiner is able to assess the child’s receptive and expressive language skills. It includes items such as naming colours and objects, repeating sentences and answering questions that test the child’s comprehension. The relation between this sub-scale measured at T1 and the calibrated RRB score of the ADOS-2 is of particular interest. The regression analysis showed that acquisition of language skills also leads to an increase in the RRB score, thus influencing symptom severity. Although this finding could be considered a counterintuitive outcome, this result can be traced back to the characteristics of the ADOS-2 diagnostic tool. In fact, during the ADOS-2 assessment, the examiner selects the most appropriate module on the basis of the level of expressive language. In particular, children in our sample were assessed using Module 1, which is intended for children aged 31 months and over who possess verbal abilities ranging from no verbal language to simple sentences [40]. Within the assessment, the presence of restricted and repetitive behaviour in children with ASD is also evidenced by the stereotyped use of words or phrases, as well as by the intonations of vocalisations and verbalisations. Coding for this includes the use of deferred echolalia or highly repetitive expressions, such as words or phrases that are understandable and may be appropriate to the conversation, focusing on the stereotypical quality of the sentence or the unusual use of words. This makes it clear that the assessment of language use is possible when the child begins to express themselves verbally. Another possible consideration concerns the coding and scoring of this characteristic. In the instrument, there is a coding of 8 (i.e. not applicable), which is converted to 0 when the examiner has to count the final score. This coding, in reality, represents a lack of information and influences the total RRB score; thus, for children with very low language skills for whom the items regarding repetitive language are not applicable, their RRB score is likely to be reduced by the inclusion of a score of 0. There are also studies in the literature showing that the expressive language skills of children with ASD increase significantly from infancy to primary school; however, difficulties remain in pragmatic and semantic features [57]. These difficulties are reflected in children’s attempts to initiate social interactions; to tell stories, events or experiences; and by the occurrence of certain unusual features (e.g. immediate echolalia, stereotyped language). The non-spontaneous and functional use of language adds important information for the development of specific intervention programs for children with ASD.

The Practical Reasoning scale is administered from 24 months onwards. It investigates the child’s ability to solve practical problems and incorporates mathematical concepts as well as ethical and moral issues that require greater linguistic comprehensibility. Our results show a relationship between the acquisition of these skills and a lower score for communication and social interaction difficulties investigated by the calibrated SA score of the ADOS-2. This scale requires the acquisition of those skills and general cognitive functioning, such as reasoning, attention, learning and memory. These skills represent a set of higher-order processes that influence social competence [59] and can be encapsulated under the umbrella term of executive functions. Executive functions can have an impact on social competence by negatively affecting processes that involve theory of mind skills and social recognition [58‐62]. Thus, as our results show, working on the acquisition of those cognitive skills that allow the child to solve practical problems consequently influences the skills needed for social engagement. Interventions on social skills in children with ASD should be aimed at identifying and enhancing key components of executive functioning that influence social competence in children with ASD.

Previous studies [6, 26, 58, 63] have shown an association between cognitive skills and autistic symptoms; however, in this study, our results show that it is the development of these skills that is associated with ASD symptoms. Knowing how and when a child acquires these skills means being able to intervene early to modify atypical developmental trajectories and improve prognosis. Compared to previous studies, our research focus takes a dynamic approach—that is, we expect changes in acquired abilities depending on the severity of the symptomatology or the use of specific, individualised interventions. Development is a dynamic process, so the study of the development of skills in children cannot be limited to a static view (i.e. to a single assessment of a moment), but to monitoring over time.

We think that the most important limitation in our study concerns T0, because in this first assessment we not only included subjects at risk, but also children with a certain diagnosis. For this reason, in the future we think it is necessary to consider children under 30 months of age who do not yet have a diagnosis and to determine if indeed the GMDS evaluation could affect the diagnosis of ASD. Moreover, at present the updated version of the GMDS is version III, but at the time of our data collection we did not have this version of the test available.

Another limitation regards RRB scores of ADOS-2. Although in our study we use Module Toddler and Module 1, Kuhfeld and Sturm [64] have demonstrated that RRB measured with the 3 and 4 Modules are an unreliable measure of this construct. Thus, we suggest having some caution in interpreting results related to RRB. We also used the calibrated severity score of two domains of ADOS-2 (AS and RRB); the calibrated severity score is not a continuous indicators of autism symptom severity, and this represents another limitation of this study.

In addition, a procedural limitation regards the limited developmental time period over which children were assessed (1 year).

It is unclear from our study, if the observed uniformity of development delay in GMSD domains in ASD, indicates a true decrease in ability, or it is artifact of the measure itself. Consequently, further investigations are requested. Our future perspective is to overcome these limits to identify a typical developmental profile for ASD. In addition, we would like to evaluate how cognitive abilities, in a larger sample ASD, change over time to make possible differential diagnosis with other neurodevelopmental disorders.