A novel screening instrument for the assessment of autism in German language: validation of the German version of the RAADS-R, the RADS-R

During the past years we can observe an increase of the diagnosis of autism spectrum disorder (ASD). While the prevalence was estimated at four in 10,000 people in 1989 [5], more recent studies in 2011 estimate a prevalence of one in 100 people [6, 7]. Whether this is due to an actual increase in the disorder, due to subtle changes in the cut-off criteria, or due to a shift in the focus of mental health professionals has been the subject of many debates. However, it is clear that the skillset required to diagnose a patient with ASD is not sufficiently trained in relevant medical professions. A survey carried out among service users in the United Kingdom highlighted the lack of knowledge by general practitioners as one of the barriers to diagnosis [8]. Yet, a diagnosis is essential to obtain social support as well as treatment within the healthcare system. A formal diagnosis allows people with ASD to apply for the formal recognition of disability, adaptations of their working conditions, and specialized training. Specialized treatment for adults with ASD is not widely available, but programs such as the FASTER (“Freiburger Asperger-Spezifische Therapie für ERwachsene” = Freiburg Asperger-specific therapy for adults) and EVA&SCOTT (“Social Cognition Training Tool”, and its enhancement “Emotionen Verstehen und Ausdruecken” = understanding and expressing emotions) have been developed and validated in phase I/II studies and are currently under examination in a multicenter, controlled phase III trial [9‐13]. Other programs, such as the GATE manual, have also been created for group settings [14]. Data from the Expectations to Psychotherapy Autism Spectrum (EPAS) questionnaire, which was developed to assess the requirements and needs of persons with ASD with regards to psychotherapy, highlighted the need to tailor psychotherapeutic concepts to autism-related deficits [15]. Autistic traits, already a diagnostic challenge themselves, can lead to the development of comorbid axis I symptoms such as social anxiety [16]. Therefore, people with ASD present with a higher rate of co-occurring mental health conditions such as ADHD (28%), anxiety disorders (20%), depressive disorders (9%), and schizophrenia (4%) compared to the general population [17]. This further complicates ASD diagnosis, as comorbidities do not automatically rule out the diagnosis of ASD, while overlapping symptoms can make the evaluation of ASD-related symptoms more difficult.

Efforts to identify objective biological markers have shown s an autism-associated increase in the albumin quotient in the cerebrospinal fluid [18], but other biological parameters, such as rate of intermittent rhythmic delta or theta activity (IRDA/IRTA) before HV in an EEG, have not shown difference between autistic and non-autistic participants [19]. Currently, no biological markers have sufficient sensitivity and specificity to be used as diagnostic tools in everyday practice. Thus, a clear-cut biological indicator with sufficient universality (i.e., applicable to the whole ASD population and not just a sub-group of for example a certain genetic disorder) has yet to be identified. Therefore, the current diagnostic procedure has to rely solely on clinical interviews by trained professionals. To standardize clinical interviews, the Developmental, Dimensional and Diagnostic Interview—Adult Version (3Di-Adult) was developed [20]. For testing individuals without a learning disability, the UK guidelines of the National Institute for Health and Care Excellence (NICE) recommends to apply the Ritvo Autism Asperger Diagnostic Scale—Revised (RAADS-R) as a screening instrument and the Adult Asperger Assessment (AAA) as a structured interview which is an interview essentially based on the Autism Spectrum Quotient (AQ) and Empathy Quotient (EQ) [1, 2, 21]. In terms of standardized tests NICE recommends the Autism Diagnostic Observation Schedule—Generic (ADOS-G) as well as the Asperger Syndrome (and High-Functioning Autism) Diagnostic Interview (ASDI) [2, 22, 23]. Apart from the RAADS all tools are available in the German language. Locally other questionnaires such as the Freiburg Questionnaire of linguistic pragmatics (FQLP) or the Bermond–Vorst Alexithymia Questionnaire (BVAQ) may also be useful as additional tools [24, 25]. It has to be noted that even carefully developed instruments such as the ADOS-G upon further inspection do have a higher proportion of false negative diagnosis in females than in males [26]. Recognizing that females might experience and express their symptoms differently the Girls Questionnaire for Autism Spectrum Condition (GQ-ASC) was modified for adults [27]. To adjust for compensation effects in the sense of camouflaging, a trait more common in females with ASD, the Camouflaging Autistic Traits Questionnaire (CAT-Q) was developed [28]. Since ASD is a neurodevelopmental disorder and its characteristics can overlap with other mental disorders, such as schizophrenia and depression, verifying the onset of symptoms in childhood is crucial. For adults, who were not assessed in their childhood or adolescence, this is often achieved by obtaining a thorough developmental history (e.g. Autism Diagnostic Interview Revised (ADI-R) [29]). Psychometric instruments such as the “Australian Scale for Asperger’s Syndrome (ASAS)” [30] may provide additional information. Further childhood documents like school reports or photographs may be helpful [2]. The currently very high numbers of requests for diagnoses and the time-consuming diagnostic process leads to longer waiting times for individuals to be assessed for ASD. However, earlier diagnosis is highly desirable as it allows for earlier allocation of specific care. It also helps the patients, who are not suffering from ASD to allow to explore possible other treatable diseases. Taken together, the development of a potent screening instrument remains an important research goal.

To address the current disparity between the number of adults seeking autism assessment and the actual number of available assessment appointments, the research groups of are not suffering from ASD to allow to explore possible of Cologne and Freiburg sought to explore the relevance of the existing Ritvo Autism Asperger Diagnostic Scale—Revised (RAADS-R) based on the recommendation provided in the German S3-guidelines on ASD diagnosis over life time [31].Accordingly the aim of this project was to create and validate a German version of the RAADS-R.

The original questionnaire was proposed in English and tested in Canada, UK, USA, and Australia [1]. Estimating an information gap in the questionnaire tools for autism at that time, the RAADS-R was validated and published in 2010 by Ritvo and colleagues. In addition to information on social relatedness, circumscribed interests, and language, which is also provided by other questionnaires, the RAADS-R includes twenty questions on sensory and motor characteristics [1]. These symptoms are relevant and entered diagnostic criteria of DSM-5 in 2013 and the ICD-11 in 2022 [3]. The multicenter study by Ritvo et al. showed good results at a cut-off of 65 with a sensitivity of 97%, a specificity of 100%, and a test–retest reliability of r = 0.987. This led to the questionnaire being included in the recommendation of current NICE guidelines [2]. Unlike many other questionnaires, the RAADS-R was validated with a clinician present to guide the participant through the questionnaire [1].

Further studies applying the RAADS-R have shown a wide range of results. A British study reported a sensitivity of 95% and a specificity of 71% [32]. The validation of a Swedish version of the questionnaire showed a sensitivity of 91% and a specificity of 93% in 272 participants [33]. A Dutch version of the RAADS-R presented a sensitivity of 73% and a specificity of 58% for the cut-off at 98 [34]. The translation of the questionnaire into French showed a sensitivity of 99% and a specificity of 67.8% in the validation study with 307 test participants. In this study, the false positive rate for patients with psychiatric disorders was 55.6% [35].

In our study, we aimed to validate a newly translated German version of the RAADS-R referred to in the following as RADS-R. Our goal was to establish a useful screening tool that could help in the decision making process for a valid diagnosis and possibly speed up the diagnostic process. Therefore, we used it as a classical self-rating instrument without presence or assistance of a clinician. Even though the original name suggests the tool to be used only for Asperger’s syndrome, following latest developments in DSM-5 and ICD-11, we employed it as a potential tool in the broaden concept of ASD. We therefore propose to change the name to RADS-R (Ritvo Autism Diagnostic Scale—Revised), a change approved by the original author of the RAADS-R, Ariella Ritvo.

The German version of the RADS-R replicated excellent result, with a high sensitivity of 92.5% and a high specificity of 93.6%, comparable to the RAADS-R. While these values are remarkably high, they do not reach the original version’s sensitivity of 97% and a specificity of 100%. Although this could theoretically be due to systematic differences in diagnostic procedures and norms between the different subsamples, this is unlikely, given that participants of the study of the English version were recruited from three different continents. The most plausible explanation for this slight difference in sensitivity and specificity is that in our German validation study, there were no clinicians present to assist in completing the questionnaire. However, the absence of a clinician was an intended since we aimed to validate the instrument as a self-rating tool. In clinical practice, it is more practical if patients fill out the questionnaire at home prior to the clinical assessment. Thus, we believe the validation of the instrument as a self-rating tool outweighs the minor reduction of sensitivity and specificity, putatively due to the absence of a clinician.

Our results are in line with the French and Swedish validation studies. Only the study by Jones and colleagues shows vastly different results with a specificity of 3,03% (13,15,17). This difference can be attributed to different sample definition. While the validation studies recruited persons with an established diagnosis of ASD or without, the study by Jones and colleagues recruited service users in the queue of undergoing ASD assessment [48]. This of course is a more realistic scenario to explore the value of the RADS-R as screening tool. At the same time, this help-seeking population might have different motivations for completing the questionnaire, possibly leading to bias. As mentioned earlier, an official diagnosis can provide certain benefits and justify support from the social and health care systems. Interestingly, in recent years, autism has become less stigmatized than other psychiatric diagnoses, such as schizophrenia or personality disorder, and has even become more attractive to many people due to popular movies, modern media and health awareness campaigns [49, 50]. The diagnosis not only opens the door to a possible support network, but also provides the person and their environment with an explanation of why certain behavioral traits exist and why certain biographical events occurred as they did. This could well lead to an aggravation bias in people expecting to receive the diagnosis. However, this bias does not only play in role in responding to questionnaires but also during clinical interviews. Furthermore, it was never the intention of the original authors of the RAADS-R to replace the clinical interview with a questionnaire, but rather to use it as a tool to assist in the diagnostic process [1].

We explored the RADS-R as a screening tool and in line with prior research found that the version with 14 items showed promising results, significantly correlating with the full 80-items version and demonstrating an even higher specificity [47]. Considering the number of questionnaires and paperwork patients have to fill out and the potential overload that come with it, it may be worth considering using the shortened version in further clinical research.

An unexpected finding of our study was the correlation between age and higher RADS-R scores. This warrants further considerations. One possible explanation could be that the older participants had more time in their life to think about their characteristic and possivly deviating personality properties. Another explanationrefers to our recruitment procedure. The contact for study participation was established either by persons who were still in treatment in the two different departments in Cologne and Freiburgor others who visited our website looking for new studies. Thereby the recruitment mainly took place amongst people who were still attached to the clinical departments. While less severely affected patients were more likely to lose contact to the diagnosing institution, those more severely impaired often stay in touch more often because they need substantially more support. Therefore, our population sample of older people with ASD could in fact have more severe symptoms than our younger group.

In this study, the German version of the RADS-R questionnaire was validated in three groups: people diagnosed with ASD, people diagnosed with a primary mental disorders, and people without any mental disorder. These groups obviously do not cover the heterogeneity of psychiatric diagnoses. As realized in ICD-11, there are boundaries not only to normality (broader autism phenotype) but also to many other psychiatric conditions not represented in our sample, e.g. different personality disorders and other developmental disorders (most notably ADHD). Identifying these comorbidities and differential diagnoses is part of the clinical diagnostic work-up. Questionnaires like the RADS-R are not developed for this purpose.

Additionally, a self-report questionnaire, in the way the RADS-R was used in our study, does bear the possibility of false representation of the symptoms. Underreporting could result from participants who do not understand or are unable to relate the questions to their own life (e.g., thinking in a too concrete manner and failing to transfer the question to their real life setting). Overreporting could result from a person aggravating existing peculiarities. This would most likely be the case with people with a preference for a diagnosis of ASD. As the study was carried out amongst volunteers and group II and III did not register for a diagnostic assessment, this effect is unlikely to have played a significant role in our study setting.

In this study, we were able to validate the German version of the RADS-R. It achieved a very high sensitivity of 92.5% and a high specificity of 93.6% using the RADS-R as a screening instrument filled out by the patient alone with no assistance by any other person (e.g. clinician). The data suggest that RADS-R is a feasible screening instrument for ASD in the German language. In contrast to most other competing instruments, it also includes questions in the area of sensorimotor symptoms as well as information about the longitudinal presence of symptoms. As such it has a high potential for clinical practice. Projects for future research should focus on the a validation of the RADS-R of a diagnosis-seeking population in a prospective manner.