At-risk subgroups for misdiagnosis and the growing burden of undiagnosed ASD

Three distinct developmental phenotypes of ASD are likely to be missed or remain undiagnosed- females with ASD, individuals with ASD who possess higher cognitive abilities, and ethnic minority groups. The current diagnostic and screening measures, such as the Modified Checklist for Autism in Toddlers (M-CHAT), Autism Diagnostic Observation Scale (ADOS), and Autism Diagnostic Interview (ADI), have primarily been developed and validated utilizing samples that are mainly composed of males. This poses a significant challenge as these measures may lack sensitivity in detecting ASD in females, particularly those with higher cognitive and linguistic abilities.

To counteract gender bias, the second edition of the Autism Diagnostic Observation Scale (ADOS-2) has been revised to enhance its sensitivity in assessing females. A study aimed at examining the differential item and test functioning (DIF and DTF) of ADOS-2 items across sex and race revealed that, overall, there was no measurement bias across sex and race, except for hand mannerisms among female children and repetitive interests among African American children. In other words, ADOS-2 was reported to be equally accurate among males and females, indicating improved inclusivity in its application.Reference Kalb, Singh and Hong ⁶ These findings emphasize the need for ongoing efforts to develop and refine diagnostic tools that consider the diverse manifestations of ASD across different demographics in order to ensure accurate identification and support for all individuals irrespective of gender, cognitive abilities, and ethnic background.

Historically, ASD has shown higher prevalence in males, with estimates indicating a frequency 3–4 times higher among boys.Reference Messinger, Young and Webb ⁷ This observed gender difference might be attributed to females exhibiting a tendency to compensate for the deficits associated with ASD. Early findings from a study examining the Toddler Module of the Autism Diagnostic Observation Scale-2 (ADOS-2) suggested that this diagnostic instrument measures social and communication milestones similarly for both male and female subjects. While the initial results suggested a lack of significant gender differences in how the ADOS-2 (Toddler Module) assesses social and communication milestones, the authors acknowledge potential confounding factors. Significantly, biases in the sample, developmental factors, or item selection could influence the observed results. The author emphasizes the importance of exercising caution and recognizes that these factors might obscure the interpretation of the results. Because of these confounds, the study highlights the importance of future follow-up research on the same sample. If subsequent investigations reveal gender differences over time, it could lend support to the notion of compensation in females with ASD.Reference Ronkin, Tully and Branum-Martin ⁸ This study underscores the complexity of understanding ASD across genders and the ongoing efforts to refine diagnostic tools for accurate assessments.

Individuals with a high Intellectual Quotient (IQ) in the context of autism spectrum disorder (ASD) may exhibit a unique pattern of compensation for core deficits. For instance, they might effectively acquire social skills, mitigating challenges such as difficulty with eye contact or theory of mind. However, this compensation often comes at the cost of heightened levels of anxiety.Reference Livingston, Colvert, Bolton and Happé ⁹ The nuanced nature of their presentation implies that routine screenings and standard diagnostic tools may not adequately capture the complexity of ASD in individuals with a high IQ.

Cultural variations further contribute to the diversity in ASD symptoms. In Japan, for instance, cultural norms encourage children not to maintain direct eye contact with adults as a sign of respect. This cultural difference can pose challenges in diagnosing high-functioning individuals with ASD, as their behaviors align with cultural norms rather than being readily recognized as potential symptoms.Reference Harrison, Long, Tommet and Jones ¹⁰

Rural communities face unique challenges in accessing care for ASD. The Autism and Developmental Disabilities Monitoring (ADDM) data highlights significant disparities, with urban states recording evaluations at an earlier median age compared to rural states. Utah, Tennessee, and Missouri have the lowest percentage of evaluations recorded by the age of 36 months. This concerning underdiagnosis of ASD in rural United States emphasizes the need for targeted interventions and increased accessibility to diagnostic resources in these communities.Reference Maenner ¹

Recent studies have identified significant limitations in the sensitivity and positive predictive value (PPV) of the Modified Checklist for Autism in Toddlers (MCHAT), the most used screening tool for autism spectrum disorder (ASD). These studies report sensitivity ranging from 33.1% to 38.8% and PPV from 14.6% to 17.8%.Reference Kuhn, Levinson and Udhnani ¹¹ The findings suggest a need for a paradigm shift in ASD screening methodologies due to the current tools’ insufficient sensitivity and high rate of false negatives.

In a British population-based cohort study, researchers discovered that children living in poverty and those with higher intelligence often miss initial screenings.Reference Hosozawa, Sacker, Mandy, Midouhas, Flouri and Cable ¹² Consequently, it is crucial to focus ASD testing efforts on preschoolers from low-income families and minority groups. The employment of multilingual staff may enhance screening accuracy, given the MCHAT’s notably poor PPV among ethnic minorities. There exists a distinct and subtle ASD phenotype that lacks symptoms in the early stages, complicating detection before the age of four due to the heterogeneous nature of the disorder. These findings underscore the importance of longitudinal follow-up beyond 36 months.Reference Gupta, Gupta and Moll ²

A significant disparity exists in ASD prevalence between states with urban populations (e.g., California and New Jersey) and those with rural populations (e.g., Missouri and Wisconsin). This disparity is likely attributable to changes in diagnostic criteria and higher healthcare literacy in urban areas. Additionally, state public health awareness programs play a crucial role in enhancing parental awareness of ASD.Reference Gupta, Gupta and Moll ² Addressing these multifaceted challenges is crucial for ensuring accurate and timely identification of ASD across diverse populations and settings.

Enhancing diagnostics: strategies for improvement

ASD continues to be primarily diagnosed through clinical assessment, utilizing a range of screening and diagnostic tools to aid clinicians in the diagnostic process. Nevertheless, clinicians commonly exhibit reluctance to heavily rely on clinical judgment, a tendency often linked to an overemphasis on testing. This hesitancy to diagnose ASD clinically contributes to gaps in ASD-specific care, including challenges in conducting thorough suicidal risk assessments.Reference Cervantes, Li, Sullivan, Seag, Baroni and Horwitz ¹³

Currently, there is no widely agreed-upon standardized approach or universally accepted standard of care for autism diagnosis. Different centers employ various screening and diagnostic tools (Table 1), including the Autism Diagnostic Observation Scale-2 (ADOS-2), Autism Diagnostic Interview-Revised (ADI-R), Childhood Autism Rating Scale (CARS), Screening Tool for Autism in Toddlers and Young Children (STAT), Modified Checklist for Autism in Toddlers (M-CHAT), Social Responsiveness Scale-second edition (SRS-2), Autism Spectrum Rating Scale (ASRS), Gilliam Autism Rating Scale (GARS), and others. The choice of tool often relies on the resources available and the clinician’s level of training.Reference Krafta ³

Table 1.

Comparison between Diagnostic and Screening Tools for ASD

ADOS: Autism Diagnostic Observation Scale; ADI-R: Autism Diagnostic Interview-Revised; CARS: Childhood Autism Rating Scale; STAT: Screening Tool for Autism in Toddlers and Young Children; MCHAT: Modified Checklist for Autism in Toddlers; SRS-2: Social Responsiveness Scale-second edition; ASRS: Autism Spectrum Rating Scale; GARS: Gilliam Autism Rating Scale.

This variability in diagnostic tools underscores the need for ongoing research and standardization efforts within the field of ASD diagnosis. Establishing a consensus on best practices and enhancing clinician training can contribute to more accurate and consistent diagnoses, ultimately improving the quality of care for individuals with ASD.

The Autism Diagnostic Observation Scale-2 (ADOS-2) is often considered the gold standard for diagnosing autism spectrum disorder (ASD), demonstrating high sensitivity and specificity in research settings, ranging from .89 to .92 and .81 to .85, respectively.Reference Lebersfeld, Swanson, Clesi and O’Kelley ²¹ However, in clinical settings, while ADOS-2 maintains high sensitivity, specificity tends to be variable.Reference Lebersfeld, Swanson, Clesi and O’Kelley ²¹ Overall, ADOS-2’s sensitivity (0.94) and specificity (0.80) are comparable to other tools such as the Autism Diagnostic Interview-Revised (ADI-R) and Childhood Autism Rating Scale (CARS).Reference Randall, Egberts and Samtani ¹⁹ Despite being the most sensitive among these tools, ADOS-2 exhibits modest specificity, leading to a higher likelihood of false positives.

The ADOS-2 comprises 5 modules tailored to the developmental and language capabilities of the individual being assessed. Module 1 is designed for children exhibiting minimal or no phrase speech, Module 2 is intended for those with phrase speech but not yet fluent, Module 3 is suitable for verbally fluent children and adolescents, and Module 4 is utilized for verbally fluent older adolescents and adults.

One study found an overall accuracy across modules of 70.4%, with a sensitivity of 90.9% and specificity of 66.0%. The accuracy decreases as the module number increases, with Modules 3 and 4 showing lower overall accuracy.Reference Hutchins ²²

The assessment of E-scores, which indicate emotional and behavioral issues in ADOS-2, includes symptoms like overactivity suggestive of attention deficit hyperactivity disorder (ADHD), disruptive behavior suggestive of mood disorders, and anxiety. These symptoms are common comorbidities, and research suggests that higher E-scores are associated with lower sensitivity and specificity.Reference Havdahl, Hus Bal and Huerta ²³ Therefore, in clinical samples characterized by a higher prevalence of comorbidity, ADOS-2 may not be as effective as a gold standard. A study administering ADOS-2 Modules 3 and 4 among patients aged 9–18 admitted to a child and adolescent psychiatric inpatient unit found lower sensitivity (Module 3: 58.3%; Module 4: 55.6%) and specificity (Module 3: 56.5%; Module 4: 59.5%) in this clinical context.Reference Colombi, Fish and Ghaziuddin ²⁴ This highlights the importance of considering the specific characteristics of the population being assessed when evaluating the effectiveness of diagnostic tools for ASD.

In a clinical sample of children and adolescents without autism spectrum disorder (ASD), a study reported a notable 34% false positives when ADOS-2 was administered. The identified group exhibited characteristics of high anxiety and low Restricted and Repetitive Behavior (RRB) symptoms. As a result, the study recommended against using ADOS-2 in isolation for ASD diagnosis and suggested incorporating structured parental interviews, which typically demonstrate better specificity. However, this approach has the potential to extend testing times, contributing to the existing issue of long wait times for assessments.Reference Greene, Vasile, Bradbury, Olsen and Duvall ²⁵

In comparison to ADOS-2, other diagnostic tests for ASD are generally more cost-effective, quicker, easier to administer, and more readily available. However, these alternates are often underutilized. Despite their advantages, these tests exhibit significant variability in interrater reliability, influencing their sensitivity and specificity. Even ADOS-2, considered a more semi-structured test, demonstrates interrater reliability for diagnostic classification ranging between 64%–82%.Reference Zander, Willfors and Berggren ²⁶ It’s crucial to note that the published data on various tools often originate from research settings, and their performance in clinical samples may be lower. This emphasizes the need for ongoing research and efforts to improve the reliability and applicability of diagnostic tools for ASD in real-world clinical scenarios.

Various tests for autism spectrum disorder (ASD) exhibit variability in sensitivity and specificity, with clinical performance sometimes falling short of their research-derived values. In a recent meta-analysis focused on the M-CHAT-Revised with Follow-up (M-CHAT-R/F), the pooled sensitivity and specificity were notably high at 0.83 and 0.94, respectively, in screening for ASD among toddlers.Reference Wieckowski, Williams, Rando, Lyall and Robins ¹⁴ Given that M-CHAT is one of the most readily available and free parent screening questionnaires for ASD in toddlers, these results are promising. However, it’s important to note that, like other tools, translated versions of M-CHAT in different languages often exhibit lower sensitivity and specificity.

Diagnosing ASD can be particularly challenging, especially for individuals with high levels of functioning and comorbid mental illnesses, who are more likely to yield false negatives in testing. In such cases, a structured interview and a comprehensive assessment of comorbidities can be valuable tools to support the clinical diagnosis of ASD, especially when suspicion is high.

Overshadowed clues: clinical screening strategies

There is a wide range of distinct comorbid conditions closely associated with individuals with milder symptoms of ASD, often overshadowing core symptoms.Reference Gupta, Gupta, Gupta and Gupta ²⁷ With the ontogenic unfolding of complex human development, widespread variability in the onset and heterotopic continuity of these symptoms is observed during its different stages. Firstly, in infancy and toddlerhood, the subtle signs and symptoms of ASD are often overlooked as normative, even though speech delays, restrictive interests, atypical sleep patterns, and sensory processing problems could be present at these stages.Reference Thomas, de Marchena and Wieckowski ²⁸ Many factors not limited to misinterpretation of screening tools, access to mental health, clinical training, clinician’s bias, and stigma contribute to missed diagnostics.

In preschoolers and school-age groups, ADHD remains a more frequent diagnosis among ASD youths, increasing the burden of undiagnosed ASD phenotypes even though there are suboptimal responses to first-line stimulants. Until 2015, before the inception of DSM-5, ASD, and ADHD were mutually exclusive diagnoses, usually leading to an overemphasis on ADHD due to the stigma and clinicians’ hesitancy to diagnose ASD. With widespread misinformation suggesting neuropsychological testing tools are required for ASD diagnosis, clinicians’ hesitancy is often amplified by policies (managed care, school, etc.) that are not supported by empirical evidence.Reference Joshi, Wilens, Firmin, Hoskova and Biederman ²⁹

In late school-age children, social anxiety disorder frequently overshadows ASD diagnoses due to lagging social skills,Reference Kleberg, Högström, Nord, Bölte, Serlachius and Falck-Ytter ³⁰ and likewise, obsessive-compulsive disorder due to restrictive interests being misinterpreted as compulsions.Reference Martin, Jassi, Cullen, Broadbent, Downs and Krebs ³¹ If an ASD diagnosis is missed until adolescence, individuals struggling with social skills often camouflage their symptoms while feeling pressured to fit in.Reference Alaghband-rad, Hajikarim-Hamedani and Motamed ³² These myriad clinical presentations in early adolescence are complicated by self-harm, often mimicking symptoms of borderline personality disorder.Reference Iversen and Kildahl ³³ Though self-harm is contextual and phenomenologically different in ASD, it is frequently misunderstood and overshadowed by BPD. It’s not uncommon for ASD individuals to continue to have BPD diagnosis during transitional age years until adulthood.Reference Powell, Parker, Kitson and Rogalewski ³⁴

Similarly, psychotic symptoms are not uncommon in ASD, potentially leading to serious adulthood trajectories that resemble schizophrenia, even though first-rank symptoms are absent. The notable rise in ASD diagnoses, partly due to the inclusion of milder forms has led to challenges in distinguishing ASD from psychotic disorders (PD). Individuals with ASD may show ‘pseudopsychotic’ symptoms like those of psychosis. As a result, ASD is often misdiagnosed as PD. Conversely, new psychotic disorders in those with ASD can be missed because the core symptoms of autism overshadow them. Studies report that the rate of PD in adults with ASD is at least 10 times higher than in the general population, underscoring the importance of accurate diagnosis.Reference Gesi, Giacovelli, Reibman and Dell’Osso ³⁵

In recent years, Avoidant/restrictive food intake disorder has been frequently associated with ASDReference Keski-Rahkonen and Ruusunen ³⁶; however, emerging evidence suggests that adults with Anorexia Nervosa score higher on the ADOS-2 algorithms.Reference Sedgewick, Kerr-Gaffney, Leppanen and Tchanturia ³⁷ Restrictive eating behaviors should be carefully evaluated as a potential indicator of ASD, particularly in adults with minimal observable impairments.Reference Casati, Prodi, Vedani, Caruso, Gesi and Dell’Osso ³⁸

This paradigm underscores the importance of detailed clinical assessments and contextual correlation to ensure accurate diagnoses for individuals with ASD, particularly those with milder subtle symptoms as often cooccurring disorders overshadow its early identification.

It is important to note that autism can also be confused with bipolar disorderReference Fusar-Poli, Brondino, Politi and Aguglia ⁴⁰ and borderline personality disorder,Reference Iversen and Kildahl ³³ both of which manifest traits resembling autism and ruminative thinking. In many cases, individuals on the spectrum are diagnosed with bipolar disorder, especially when they experience heightened activation or irritability on SSRI. Aside from improving access to diagnostic and screening tools, there is a need for clinician education and improvement in comorbidity assessment.Reference Gupta and Gupta ³⁹

ASD is frequently overlooked, underdiagnosed, and sometimes misdiagnosed, influenced by various contributing factors. Initial signs and symptoms are frequently perceived as normative development patterns in many instances, leading to a reluctance for reevaluation when improvement is not observed after a period of watchful waiting.Reference Davidovitch, Levit-Binnun, Golan and Manning-Courtney ⁴¹ Symptoms such as increased irritability heightened susceptibility to anger, poor eye contact, communication delays, and the presence of sensory issues manifest across a spectrum of variable intensity in the dimensional realm of ASD symptoms.

A thorough examination often necessitates a detailed developmental history, including inquiries about potential complications during and after pregnancy. Concerns related to frequent crying, difficulties with gastrointestinal symptoms, poor eye contact, language developmental challenges, struggles in forming friendships, and notable aggression during childcare interactions, may raise suspicions warranting ASD screening. While the M-CHAT tool is integrated into pediatric clinics and child development assessments, it may overlook subtle signs crucial for ASD identification. Recognizing this limitation, various software-based face and voice recognition applications are under development.Reference Franco, Oliveira and Portolese ⁴² Given the elusive nature of these subtle symptoms for such tools, direct questioning remains a clinically sound approach. Additionally, exploring alternative biomarkers through unconventional technology could offer valuable insights beyond conventional measures.

In clinical settings, a distinctive developmental history unfolds among children diagnosed with ASD. These children encounter challenges in forming friendships with peers of their age and frequently exhibit a strong inclination toward technology, particularly video games, and a keen interest in music, puzzles, games, collectible cards, dinosaurs, and Lego sets. Notably, reports indicate that children on the spectrum often show an affinity for Japanese culture, particularly through an understanding of anime, and they display curiosity about different languages.

Children with ASD also commonly experience sensory issues, clinically manifested as heightened sensitivity to various textures, colors, and smells. They may resist activities such as haircuts and prolonged showers, and exhibit aversions to tight clothing or touch-related sensory stimuli, often cutting tags from their T-shirts.Reference Kirby, Bilder and Wiggins ⁴³ Additionally, these children often have comfort items and unique interests, such as a fascination with marine biology, and forensic science, a penchant for horror movies and crime thrillers, a specific liking for certain types of food, and difficulties in engaging in reciprocal peer interactions.

Many of these symptoms may not become apparent until the child reaches the 5th to 6th grade, typically between 11 and 12 years old, when distinct gender differences emerge. Females on the spectrum often display more eye contact and employ compensatory behaviors, including language use and camouflaging. The hypothesis posits that the lack of reciprocal feedback from peers, coupled with challenges in interacting with same-age peers, and the cognitive exhaustion from constant camouflaging to fit in, can contribute to mental health crises in this population.Reference Cook, Hull, Crane and Mandy ⁴⁴

Children and adolescents with ASD, initially presenting in mental health settings, may exhibit symptoms resembling social anxiety disorder or generalized anxiety disorder. However, their challenges stem from deficits in theory of mind, shaping their unique clinical phenotypes, rather than a fear of negative scrutiny.Reference Iversen and Kildahl ³³ Non-suicidal self-injury is not uncommon among this demographic, and they may be mistakenly diagnosed with borderline personality disorder due to defiance and interpersonal difficulties, despite the absence of chronic feelings of emptiness.Reference Moseley, Gregory, Smith, Allison and Baron-Cohen ⁴⁵ It’s crucial to recognize that ASD typically emerges at an earlier age, often showing symptoms around three to 4 years old. In contrast, higher-functioning ASD symptoms may become apparent later, during school age and pre-puberty. Recent empirical data indicate that borderline personality disorder tends to manifest around the age of 11 or 12, and a DSM-5 diagnosis can be made if symptoms persist consistently over 1 year. Additionally, some children with ASD may be misdiagnosed with serious and enduring mental health disorders like schizophrenia or intellectual disability, as psychotic symptoms are not uncommon in this population (Figure 2).Reference Gupta, Gupta, Gupta and Gupta ²⁷

Figure 2.

How to ask narrow questions to screen for ASD in complex phenotypic presentation.

Therefore, a thorough chronological assessment of symptom ontology, considering distinct symptom clusters emerging at an early age, deficits in theory of mind, sensory issues, social challenges, a preference for routine, and atypical language development, is essential for accurate diagnostics (Tables 2 and 3).

Table 2.

A Symptom Checklist Based on Domains of Impairments in Individuals with ASD

Table 3.

Effective Strategies for Incorporating Empirical Data Gaps into Clinical Practice and Decision-Making.

The clinical ramifications of increasing burden

The efficacy of treatment for individuals with ASD relies heavily on early accurate diagnostics. When co-occurring conditions are treated without considering the underlying ASD, responses can be suboptimal, with potential paradoxical reactions or outcomes.Reference Gupta, Gupta, Gupta and Gupta ²⁷ Children and adolescents with ASD may have subdued responses when treated with serotonin reuptake inhibitors or stimulant medications without recognizing the additional diagnosis of ASD.Reference Joshi, Wilens, Firmin, Hoskova and Biederman ^{46 ,} Reference Williams, Brignell, Randall, Silove and Hazell ⁴⁷ The lack of integration with multi-modal interventions such as social skills training and speech therapy can lead to increased morbidity and poor overall outcomes. When conditions become refractory to these treatments, it places additional strain on caregivers and fosters a negative view of mental health among patients and families.Reference Alibekova, Kai Chan and Crape ⁴⁸ Therefore, recognizing and accurately diagnosing ASD before initiating treatment is critical. Many treatments require appropriate modifications, including the use of therapeutics and the addition of multi-modal interventions. The modifications while considering medication dosage, psychological interventions, and the treatment of other co-occurring mental health and medical conditions are essential to achieving the desired outcomes.Reference Gupta and Gupta ³⁹

Early diagnosis of autism spectrum disorder (ASD) is crucial to provide timely and appropriate care. Applied Behavior Analysis (ABA) therapy is considered evidence-based for severe symptoms, but its cost-effectiveness remains a concern. A study estimates that around 37% of individuals with ASD do not receive any therapy, and 30% do not receive any treatment at all.Reference Xu, Strathearn and Liu ⁴⁹ Some individuals may be unaware of their ASD condition, and this lack of support is alarming, especially in light of the association between ASD and other mental health conditions such as depression, anxiety, and suicidality.Reference Kõlves, Fitzgerald, Nordentoft, Wood and Erlangsen ^{50 ,} Reference Dow, Morgan and Hooker ⁵¹ A study conducted in the New York University Pediatric Psychiatry Emergency Room found that less than half of clinicians believe autism is an independent risk factor for suicide.Reference Cervantes, Li, Sullivan, Seag, Baroni and Horwitz ¹³

Medication use, including polypharmacy, is widespread in the ASD population. However, there is a diverse response to medications, mirroring the heterogeneity of autism itself. Selective Serotonin Reuptake Inhibitors (SSRIs), commonly used for mental health conditions, may not be as effective in individuals with ASD and could potentially lead to harm due to behavioral activation.Reference Williams, Brignell, Randall, Silove and Hazell ⁵² Stimulants may not exhibit a linear dose–response curve in individuals with ASD.Reference Joshi, Wilens, Firmin, Hoskova and Biederman ²⁹ Given the sensory-perceptual disturbances associated with autism, individuals with ASD are more susceptible to side effects from medications. It’s noteworthy that almost all medications used in ASD, except for aripiprazole and risperidone, are considered off-label.Reference Gupta and Gupta ³⁹ Addressing the challenges in providing appropriate therapies and medications for individuals with ASD is crucial for improving their overall well-being and mental health outcomes.