Les évolutions actuelles en santé publique amènent les autorités qualifiées à proposer, dans leurs recommandations de bonne pratique pour l’évaluation et le diagnostic des troubles du spectre autistique (TSA), des procédures qui s’appuient sur l’utilisation de l’Autism Diagnostic Interview-Revised (ADI-R) et l’Autism Diagnostic Observation Schedule (ADOS) considérés comme un « gold standard ». Cependant, la problématique du diagnostic différentiel reste prégnante dans la pratique clinique.

Notre objectif vise à souligner et nuancer les qualités métrologiques de l’ADI-R et de l’ADOS concernant le diagnostic différentiel des troubles autistiques, en proposant une synthèse des études récentes pour soutenir l’intérêt de maintenir ouverte une réflexion sur les modalités diagnostiques.

Notre revue sélective de la littérature récente met en avant les études qui confrontent l’ADI-R et l’ADOS, utilisés indépendamment et en combinaison, à diverses problématiques différentielles en clinique adulte et infantile de l’autisme, afin d’en souligner les qualités et les limites.

Les données issues de ces outils sont utiles dans le processus diagnostic des troubles du spectre autistique et constituent des preuves diagnostiques solides, à la condition d’une association avec un regard clinique critique. Cependant, ces éléments demeurent insuffisants car ces outils manquent de finesse pour l’évaluation des formes autistiques hétérogènes et des tableaux cliniques voisins.

La problématique de diagnostic différentiel reste au premier plan dans la clinique de l’autisme, dès lors une formalisation des démarches diagnostiques doit pouvoir rester ouverte et accompagnée d’une démarche clinique créative, en particulier dans les situations complexes peu solubles par les outils diagnostiques classiques.

Current developments in international public health are leading qualified authorities to release clinical practice guidelines for Autism Spectrum Disorders (ASD) assessment and diagnosis. Such documents incorporate procedures that rely on the Autism Diagnostic Interview-Revised (ADI-R) and the Autism Diagnostic Observation Schedule (ADOS) which are considered to be the “gold standard” assessment measures in the evaluation of ASD. Although these tools do prove their effectiveness in the evaluation of autistic symptomatology, they nevertheless stumble whenever the situation becomes more complex and reveal differential diagnostic issues in infantile and adult autism cases. This differential diagnostic issue remains significant in the clinical practice of daily life and has strong implications for the course of therapeutic treatment.

Our objective is to underline and nuance the metrological qualities of the ADI-R and the ADOS in the differential diagnosis of autistic disorders by presenting a synthesis of recent studies, thus supporting the interest of maintaining an open debate on diagnostic practices.

Our selective review of the recent literature focuses on studies that confront the ADI-R and the ADOS — used either independently or in a combination — with various differential issues in adulthood and childhood autism in order to highlight their qualities and limits.

The ADI-R is a semi-structured interview applied by trained examiners and applied to relatives and/or caregivers who collect developmental information about the patient's first years of life on a variety of behaviors and skills. It is therefore relatively dependent on the availability and personal bias of the interviewed third parties. Metric features highlighted by the reviewed studies straightforwardly reveal a certain effectiveness of the tool and a good discrimination of childhood disorders. At the same time, the tool's discriminatory capacity seems insufficient when applied to toddlers, very young children, and adults alike. The latter seems particularly true when it comes to differentiating between autistic and schizophrenic spectrum disorders. The ADOS is a semi-structured standardized observation assessment tool that has experienced several successive developments, such as optimizations concerning the sensitivity of its cutoffs. Many works have contributed to building such a tool with reliable and solid metric qualities which nevertheless retain important biases such as the subjectivity of the caregiver or the evaluator during the scoring process. For assessments of autistic children the tool still has a good diagnostic validity but seems to retain cases of incorrect diagnosis of ASD (false positives). In other words, disorders or developmental disabilities of some children and adolescents could not be distinguished from ASD when relying on this test alone. The ADOS Module 4, designed for the diagnosis of adolescents and adults with fluent speech, has undergone less updating. This revisited algorithm has metrological qualities useful for clinicians and remains one of the few available tools for this population. Unfortunately, its diagnostic accuracy is lower when applied to women, the elderly, people with personality disorders or higher intellectual abilities, or for the discrimination between ASD and schizophrenia. Overall, scores from these two instruments bring strong evidence of their usefulness in the diagnostic process of ASD, provided that they are used with caution and a critical clinical perspective, and only as a secondary technical support. Their use in combination is effective since they are complementary and compensate for each other's limitations. However, their globalized hegemony as “gold-standard” tools constitutes a setback insofar as it constrains the diagnosis of ASD to a set of stereotyped items. The latter in turn sets a normative model of autism that excludes other phenotypic forms, especially in the case of women and the elderly. Finally, the discrimination between autism and psychosis for children seems to remain an insoluble task even for the ADI-R/ADOS combination.

The problematics of differential diagnosis remain critical for clinical approaches to autism. Therefore, formalizations of the diagnostic procedures must be able to remain open-minded and accompanied by a creative clinical approach, especially in the case of complex situations that are not soluble by means of conventional diagnostic tools. One possibility may lie in the deepening of the phenomenological approach to autism as an attempt to model the subjective phenomena of autistic subjects and thus operationalize elements that serve the diagnostic process. In the same way, a psychodynamic epistemology could help clinicians to go beyond the consideration of observable behaviors and scores, introducing a psychoanalytic point of view that interfaces objective behaviors with the individual's dynamic intrapsychic functioning. This project could be articulated with projective methodologies — notably the Rorschach test — which respects the needs for standardization and quantification of conventionally used diagnostic tools.