Online diagnosis challenges
Diagnosing autism in adults via video meetings carries specific risks, particularly for late-diagnosed individuals who have developed sophisticated coping strategies such as masking. These strategies, combined with overlapping conditions that mimic autistic traits, make accurate diagnosis more difficult - especially in remote settings.
Distorted social observation
Online assessments risk distorting the observation of social communication - the core domain of autism diagnosis. Factors such as altered eye contact, camera-mediated interactions, and inconsistent technology use can mask or mimic autistic traits, undermining diagnostic accuracy. Subtle non-verbal cues, including facial expressions and body language, might be missed or misinterpreted during video calls, reducing the sensitivity and specificity of the assessment.
This study explicitly focuses on telehealth (i.e., online) assessments and reports that clinicians struggled to observe non-verbal cues such as eye contact, facial expressions, and body language during video calls. Environmental distractions, poor internet, and smartphone use limited visibility. Over 77% of clinicians reported being unable to reach a diagnostic conclusion in some cases, often requiring in-person follow-up.
Clinicians reported that telehealth autism assessments became common during the COVID-19 pandemic, offering flexibility and maintaining service delivery despite restrictions, but highlighted significant barriers such as inconsistent IT access, lack of standardised procedures, and increased diagnostic uncertainty, particularly for complex cases. Tools originally validated for in-person use, such as ADOS-2, were adapted for remote assessments without formal validation, raising concerns about the accuracy and reliability of diagnostic outcomes. Overall, while some clinicians see value in a hybrid model, challenges in assessing subtle social communication, increased healthcare disparities, and insufficient post-diagnostic support remain unresolved.
A survey of 52 autism diagnostic professionals across the UK and internationally found that while telehealth assessments became widespread during the pandemic, views on their validity were polarised, with 77% reporting they were unable to reach a diagnostic conclusion remotely in some cases due to challenges in observing subtle social cues and the lack of validated tools for online use. Many clinicians felt that telehealth worked well for straightforward cases but was inadequate for complex presentations, particularly where differential diagnosis was needed or when patients masked their traits. The study concluded that while telehealth offers flexibility and improved access for some, a hybrid model with in-person assessment remains essential for reliable diagnosis, especially for adults and those with complex presentation.
Masking in late-diagnosed adults
Many adults, especially women, engage in camouflaging - consciously or unconsciously - to fit in socially. This behaviour is often sustained in clinical settings and can be amplified in online environments, where individuals have greater control over their presentation (e.g., camera angle, scripting). As a result, core autistic traits may remain hidden, increasing the risk of missed or delayed diagnosis.
‘Camouflage and masking behavior in adult autism’
This review confirms that autistic adults, especially women, engage in camouflaging to blend in socially. The review notes that this behaviour is often sustained in clinical settings, making it harder to detect core traits - particularly when observation is limited to video.
‘Masking, social context and perceived stress in autistic adults’
The article shows that masking increases under stress - such as during a clinical assessment - and is context-dependent. In online settings, where environmental control is higher (e.g., camera angle, scripting), individuals may mask more effectively, further reducing diagnostic visibility.
Differential diagnosis
Autism shares features with several other conditions, including ADHD, social anxiety, borderline personality disorder, and CPTSD. These often present with similar social difficulties or rigid behaviours, which can be misattributed in remote assessments - particularly when collateral history and broad clinical context are not thoroughly explored. Without comprehensive input, symptoms may be incorrectly assigned to another condition, or autism may be overlooked entirely.
This study demonstrates that using both ADOS-2 and ADI-R together improves diagnostic accuracy when differentiating autism from other conditions. When only one tool is used - or when online assessments skip one due to format limitations - misdiagnosis rates increase. This supports the concern that online assessments, which often lack full tool implementation or collateral input, are more likely to misattribute symptoms.
Lack of validated tools for online use
The gold-standard observational tool for autism diagnosis, ADOS-2, was designed for in-person administration and has not been formally validated for telehealth use. As a result, clinicians conducting online assessments may need to rely on informal or adapted methods, which lack standardisation and reliability. This might compromise diagnostic integrity, especially in complex or atypical cases.
This review explicitly states that ADOS-2 was designed for face-to-face administration and has not been formally validated for telehealth use, and warns that adapting it for online settings without validation compromises diagnostic integrity, particularly in complex or atypical cases.
‘Are online autism tests accurate? Evaluating the validity of digital autism screening methods’
The article notes that while telehealth assessments show 80-91% accuracy compared to in-person, they are not equivalent and should not replace in-person evaluation, especially when standardised tools are adapted without validation.