General findings
Several changes have occurred in ASD diagnostic criteria over the last decades. Originally considered a prerogative of child psychiatry, only in relatively recent years clinicians and researchers have acknowledged the lifelong persistence of the condition. In 2013, the publication of the DSM-5 has represented a substantial revolution, thanks to the introduction of the concept of “autism spectrum” and the possibility to diagnose the condition later in life [4]. Along with changes in psychiatric nosography, the awareness of ASD among the general population, parents, and mental health professionals has rapidly grown, leading to a dramatic increase in the estimated prevalence worldwide [1]. Early detection has significantly contributed to this rise in diagnoses; however, part of the growing number of cases of autism could be attributed to the identification of the so-called “lost generation”, who has started to seek a first formal diagnosis after the release of the DSM-5 [5].
The present study has been specifically designed to summarize the characteristics of people whose neurodiversity has never been identified or has been misdiagnosed with other psychiatric diseases. Our findings show that individuals who received the first diagnosis in adulthood (≥ 18 years) had generally average or above-average cognitive abilities and needed low or medium levels of support. One important issue to discuss is the long time elapsed between the first clinical evaluation and the definitive diagnosis in this specific group of people (median 11 years). This finding reflects the difficulties encountered by both child and adult psychiatrists in identifying ASD, at least in Italy. Nevertheless, our data are in line with that of Geurts and Jansen [15], who reported a median time of 12 years between the first contact with mental health services and the assessment for ASD in The Netherlands.
The proportion of psychiatric comorbidities detected during the assessment was lower compared to literature findings: for instance, anxiety disorders were experienced only by 6.2% of our sample, in contrast to 20% reported by a recent meta-analysis [14]; analogously, 9.9% of our sample had depression, in contrast to 11% found by Lai et al. [14]. Importantly, the prevalence of reported lifetime co-occurring conditions is even higher in other studies involving only participants with IQ in the normal range [26, 27].
The relatively low rate of psychiatric comorbidities found in our sample might represent one possible explanation for the late ASD diagnosis. We could hypothesize, in fact, that in the absence of the additional impairments caused by mental health disorders, individuals might have been less prone to contact our centers earlier. Another potential reason is that the limited time available for psychiatric assessment in routine clinical practice was not enough to fully understand the complex clinical pictures. Additionally, standardized tools were not systematically administered, and mainly used for differential diagnosis purposes. It is also worth mentioning that 20.5% of participants had comorbid ID, and it is well known that identifying comorbid psychopathology in people with low IQ is quite difficult, even for expert clinicians [28].
In this regard, many participants included in our sample had received a diagnosis of ID in the past, even if a comorbid diagnosis of ASD had not been performed. This is not surprising, because of the overlapping symptomatology between the two conditions: for instance, both neurodevelopmental disorders may lead to significant language impairments and may present with routine and stereotypic behaviors. Moreover, we could hypothesize that clinicians not familiar with autism (especially, those working in community mental health services) might not feel sufficiently self-confident to diagnose ASD in the presence of severe or profound ID. Interestingly, nine individuals who had received a diagnosis of ID in the past were collocated in the average or above-average range of intelligence after administering proper standardized tools. This could be due to the widespread malpractice of attributing low IQ to individuals who show excellent cognitive abilities in several non-verbal areas, despite presenting with severe socio-communication difficulties.
Adults with ASD had also received a high rate of diagnoses of psychotic disorders (16.1%) and personality disorders (PD) (14.9%), particularly “not-otherwise specified”. Like ASD, psychotic disorders can be characterized by social isolation, socially inappropriate behaviors, and low social insight. Additionally, thought disorders and the use of an atypical or nonsensical language (e.g., tangentiality, circumstantiality, neologisms) are common to both psychoses and ASD [29, 30]. PD can also have similar manifestations, such as odd behaviors (schizotypal PD), social withdrawal (schizoid PD), emotional dysregulation and self-injurious behaviors (borderline PD), low empathic traits (antisocial PD), social avoidance (avoidant PD) or sameness (obsessive–compulsive PD) [31]. Also, it is interesting to mention that past diagnoses of psychoses and PD were mainly “not-otherwise specified”. This may highlight the difficulties encountered by clinicians in framing people within specific diagnostic categories, due to heterogeneous phenotypical presentations.
In 13.7% of participants, a diagnosis of depression had been previously performed, probably for the social withdrawal [32]. Other frequent past diagnoses were anxiety disorders and OCD (7.5% each). Anxiety disorders, such as social phobia, generalized anxiety disorder (GAD), or agoraphobia, for instance, could significantly impact on the social functioning of an individual [33]. Also, OCD shares some features with ASD, such as the presence of rituals, rigid and stereotyped behaviors, or restricted interests. However, while in OCD the repetitive behaviors generally represent a means of calming obsessions and anxiety, in ASD stereotypies are usually not associated with obsessive thoughts [34]. Moreover, contrariwise to obsessions and compulsions, restricted and repetitive behaviors are not egodystonic and can play a positive function of pleasure and enjoyment, calming, or stimming, which may represent a coping strategy for people with ASD [35].
Finally, hyperactivity and behavioral problems might be interpreted as symptoms of ADHD or conduct disorders rather than expressions of the autistic condition per se, as in a small proportion of our sample. In fact, ADHD symptoms seem common in ASD; on the contrary, autistic symptoms are not common in ADHD [36]. Shared features may be, for instance, difficulties in social interaction and over-reactivity, meltdowns, or aggressive behaviors. Contrariwise, symptoms such as unusual fascination for specific interests, repetitive movements, regression of language, and special abilities are not typical of individuals with ADHD [36]. It is important to take into account that ASD and ADHD frequently co-occur [37] and this may be relevant for prognosis, since this subgroup is more likely to have quantitatively more and more severe psychiatric difficulties [38]. However, this comorbidity was not found in the sample included in this study. Our result could be partially justified by the fact that the intensity of ADHD symptoms, especially hyperactivity and impulsivity, tend to decrease during the life course [39, 40].
Sex differences
Our findings showed that females usually obtained a more delayed diagnosis than males (26 years vs. 22 years). This is in line with the notion of the existence of a “female autism phenotype”, which has been thoroughly discussed by experts over the last few years [8, 41, 42]. The “female autism phenotype” consists in a slightly different presentation of the core and associated autistic characteristics, which may not be fully explained by the diagnostic criteria and tools, which are based on the typical male features [42]. This peculiar phenotype may partially explain the male-to-female ratio in ASD [12]. It has been widely reported that women with ASD, especially those with high cognitive abilities (such as women included in our sample), develop more effective strategies to “camouflage” difficulties in social situations [8, 43] and usually present with less pronounced symptoms, encountering the risk of going undiagnosed. Particularly, camouflaging of the autistic core deficits in females with ASD has been related to better language and social mimicry skills [44], more “active but odd” interactional behavior, less challenging behaviors or hyperactivity in the school environment, and less eccentric special interests [42].
Also, women on the autism spectrum present more frequently uncontrollable mood and interpersonal problems, as well as higher borderline and passive–aggressive traits [45, 46]. Compared to their male counterparts, they are more vulnerable to suffer from “internalizing” problems, such as anxiety, depression and eating disorders, and are less likely to present “externalizing” behaviors, such as hyperactivity, impulsivity, and conduct problems [8, 47]. This last consideration perfectly fits with our findings: indeed, women had more frequently received diagnoses of depressive, anxiety, or personality disorders that typically manifest with internalizing symptoms; conversely, males were more frequently identified as having ADHD, psychoses, or conduct disorders. However, these observations were based only on the visual inspection of Fig. 1, as we did not statistically compare the female and male data due to the small sample sizes.
Of note, while observing the general characteristics of the sample, we can notice that the ADOS-2 scores (i.e., direct observation of the patient) of the Social Interaction, Communication + Social Interaction, and RRB domains were significantly lower in the female group. We also found that 96.9% of males scored above the proposed cutoff of the ADOS-2, in contrast to only 73.8% of females, further confirming that symptoms are less pronounced in adult women. This finding is not surprising since diagnostic and screening tests, such as the ADOS-2, have been developed based on the typical male phenotype of ASD, excluding some of the features of girls with autism [41, 48]. Indeed, in one of the first studies focusing on sex-specific profiles of core symptoms in ASD diagnosed in adulthood, Lai et al. [49] found milder symptoms in females in all core areas, as assessed with the ADOS-2. The same differences were found by subsequent research [41].
Looking at the ADI-R scores (i.e., a semi-structured interview which mainly refers to childhood symptomatology), a significant difference could be detected only in the RRB domain, even if the difference disappeared after correcting for FDR. Even if a cautious interpretation is needed, our findings may support the notion that females with ASD manifest less repetitive and stereotyped behavior (RRB) than male peers, even in the presence of equal socio-communication impairments [47, 50]. The lack of associated sex-specific differences in social communication reported for early childhood, as determined by the ADI-R, may reflect the higher capability of females with ASD to develop adaptive compensation strategies beyond childhood and adolescence rather than innate behavioral differences. Indeed, both clinical observations and self-reports have suggested that females are particularly motivated and/or skilled in “camouflaging” their social difficulties [51,52,53,54].
Limitations
To our knowledge, this is the largest study specifically focused on the characteristics and the psychiatric history of adults who received a diagnosis of ASD in adulthood. However, some limitations should be acknowledged. First, since this was a naturalistic study, our sample was probably not sufficiently large to detect all shades of this relatively new field of research; however, we have planned to extend our sample and replicate our findings. Second, differently from previous similar studies [7, 15], we focused only on adults who did obtain a diagnosis of ASD after being referred to our centers; conversely, we did not report information about individuals who asked for an ASD assessment without obtaining a final diagnosis. Since this is a crucial topic, we have planned to discuss the theme of the correctness of self-diagnosis in future studies. We believe that the main strength of our paper relies on the inclusion of people who were diagnosed with ASD after a complex clinical assessment, using standardized tools, as suggested by international guidelines. Conversely, psychiatric comorbidities (excluding ID) and adaptive abilities were not systematically assessed using standardized tools, but only performed according to the DSM-5 criteria, and we acknowledge this point as a limitation of our study. Finally, our analysis was limited to the experience of two Italian university centers. Therefore, we cannot generalize our findings to other countries, in which the awareness for the condition might be greater, and early detection of autism might be widespread. However, up to date, very few research reports have addressed the issue of undiagnosed or misdiagnosed adults with ASD presenting data from clinical practice.
Clinical implications
Our data might be relevant for both child and adult psychiatrists, with several implications for diagnosis and treatment. On the one hand, child psychiatrists should avoid concerns in diagnosing ASD, with the risk of covering this condition under more general developmental delays. This is risky, since children with ASD may not receive specific treatments for the condition. It has in fact been demonstrated that early intervention improves both the short- and long-term outcome of ASD [5]. On the other hand, adult psychiatrists should be able to recognize the symptoms of ASD in complex psychiatric conditions, since ASD core symptoms may partially overlap those of some mental health issues (i.e., psychosis, personality disorders, ID) or be present in comorbidity. It is possible, in fact, that individuals with ASD may remain unrecognized until social demands exceed their socio-communication capacities, causing severe distress that may lead to a psychiatric consultation or hospitalization.
In case of suspected ASD, it is important to consult a specialized center for an exhaustive diagnostic assessment, which might be quite complex in adults. According to international guidelines [55] and also to our clinical experience [6], standardized tools should be administered also in adults with suspected ASD, but integrated in a complex clinical evaluation, with the consultation of caregivers, spouses, or other relatives when needed. Examining in detail the reports of past diagnoses can be useful to orientate the clinician through the diagnostic pathway.
Given the frequent misidentification of ASD with ID, also when ID is not actually present (as in nine individuals included in our sample), a standardized IQ test should be possibly administered, for several reasons. First, because it might help assessors understand whether the poor social abilities are consequences of low cognitive capacities or not, thus facilitating the differential diagnosis between ASD and ID. Second, tools examining different cognitive domains (i.e., the Weschler Scales, which compute both verbal and performance IQ) may be useful to detect some specific peculiarities of people with ASD, which might not be manifest during direct observations or interviews (e.g., deficits in fine motricity). In ASD individuals, in fact, a discrepancy between performance and verbal subtests is very common [13]. Also, people with ID usually show homogeneous impairments in cognitive profile, while ASD individuals, even those with an average intellectual functioning, tend to have scattered profiles, with areas of strengths (“islets of abilities”) and weaknesses [56]. The case of A., presented at the beginning of the paper, is quite explicative in this sense.
Apart from ID, while assessing ASD in adulthood it is important to evaluate the presence of co-occurrent conditions [57]. In fact, as reported above, the rates of psychiatric comorbidities are higher in people with ASD than the general population [14, 58]. For instance, the lifetime prevalence for adults with ASD has been estimated between 27 and 42% for anxiety disorders, and between 23 and 37% for depressive disorders [59], with peaks among individuals without ID [26, 27]. The assessment and follow-up of comorbidities have crucial implications for the outcome and the follow-up. For instance, the presence of psychiatric comorbidities may increase the risk of suicidal ideation and behaviors [60]. A recent review has reported that prevalence rates for suicidal ideation were between 11 and 66% and suicidal attempts were between 1 and 35% in ASD [61]. Also, Hirvikoski et al. [57] reported that 0.31% of premature deaths in ASD were due to suicide. Moreover, according to the study, suicide was 7.55 times higher in people with ASD than controls, and even 9 times higher in ASD people without associated ID, who often present with co-existing psychiatric disorders [57].
Getting to a diagnosis of ASD and a timely identification of comorbidities have also relevant implications for the choice of treatment. As suggested by guidelines, psychosocial and behavioral therapies should be preferred in people with ASD [55, 62]. Conversely, obsessive-like symptoms (i.e., repetitive behaviors or restricted interests), paranoia, or social withdrawal might not be responsive to common psychiatric psychopharmacological treatments, sometimes causing important side effects [63]. Moreover, as reported above, repetitive behaviors and restricted interests are typically egosyntonic and may represent fundamental coping strategies for people with ASD, having a function of calming or stimming to deal with under- or overstimulation [35]. Therefore, suppressing these behaviors might be deleterious for this group of people. On the contrary, other co-occurrent conditions (e.g., moderate to severe depression and anxiety, OCD, and psychoses above all) might be worthy of pharmacological treatment. Importantly, a diagnosis of ASD might also facilitate the achievement of specific support, such as disability benefits, or academic and job inclusion.