Translation, Cross-Cultural Adaptation, and Validation of a Dutch Version of the Actions and Feelings Questionnaire in Autistic and Neurotypical Adults

Twenty-five autistic individuals (15 female; 23 right-handed) and twenty-five neurotypical individuals (14 female; 21 right-handed) participated in the study. No participants were excluded from analysis. Autistic participants were recruited from the Radboud University Medical Centre (UMC), Nijmegen. Patients were recruited via two routes. In the first route, patients were contacted by their psychiatrist at the Radboud UMC with general, global information about the study and asked if they agree to being approached by researchers. In the second route, a message was posted private, organization specific social network, where Radboud UMC psychiatrists have message board style contact with past patients. All participants were clinically diagnosed with Autism Spectrum Condition according to the criteria defined in the DSM-5 (American Psychiatric Association, 2013). The neurotypical control group was recruited via the Radboud University participant recruitment system (SONA), which allows for pre-signup screening of several participant characteristics. For both groups, potential participants were excluded if they had a history of any other (neuro-)psychiatric disorders, brain surgery or brain trauma, or use of anti-psychotic medication. Participants were required to be proficient in Dutch and have normal or corrected-to-normal vision. By starting recruitment of the ASC group first, we were able to pre-screen our control group in an attempt to match age and gender between the two groups. We additionally collected data on education and handedness for further group matching. The study was approved by a local ethics committee (CMO Arnhem-Nijmegen). All participants provided informed consent regarding the procedure and purpose of the study.

The translation and cross-cultural adaptation was done in four steps: (1) independent forward-translation, (2) synthesis, (3) back-translation, and (4) expert check (face validity). This process was based on established guidelines for translation and cross-cultural adaptation of questionnaires (Beaton et al., 2000).

To assess content validity, we checked the face validity of the AFQ-NL by ensuring that this adaptation reflected the construct that the tool is meant to measure (Mokkink et al., 2010, 2020). This was done by an expert clinician in the field of ASC [author CC], who compared the back-translation (Dutch to English) of the AFQ-NL to the original English version and assessed with the back-translation was comparable in content to the original AFQ, and that the Dutch version accurately reflected the same constructs as the original and back-translated English version.

To assess construct validity, we tested whether the AFQ total score correlates with autism quotient (AQ) (Hoekstra et al., 2008) score. This test of convergent validity is based on Huggins et al. (2019) finding that AFQ negatively correlates with broad autism phenotypes in neurotypical adults (Huggins et al., 2019), and the suggestion that difficulties in recognizing emotions in oneself and in others are common in ASC (Bird & Cook, 2013; Kinnaird et al., 2019). This was tested using a linear model, with AFQ as dependent variable and AQ as independent variable. As a test of divergent validity, we similarly tested for a correlation between total AFQ score and IQ score, as estimated by the WASI-IV short form (Wechsler, 2011). This was done as there are no indications that the AFQ should be correlated with general intelligence. All tests were carried out in the statistical program R (R Core Team, 2013).

The internal consistency of the AFQ-NL was assessed in order to ensure that this adapted version also provides consistent responses. Utilizing the R package psych (Revelle, 2020), we calculated Cronbach’s alpha to assess overall consistency (total scale), as well as for each of the subscales described in (Williams & Cameron, 2017). In short, these subscales emerged based on confirmatory factor analysis, and consisted of three subscales: Feelings, Imagery, and Animation. Adequate consistency is considered to be an alpha between 0.7 and 0.9 (Nunnally & Bernstein, 1994).

During the expert comparison of the synthesized (T12), back-translation, and original version, one small adjustment was made to the wording. In Question 9, the word “normale” (“normal”) was changed to “gewone” (“common”), as “normale” was considered to carry more of a value judgment. Besides this, the synthesized version was considered to be an adequate translation and adaptation of the original. The main resulting translation steps are provided in Supplementary File 1. The final version (AFQ-NL) can found in Supplementary File 2.

Both the independent native Dutch speaker and the clinical expert approved of the final version.

For structural validity, we assessed convergent validity by testing whether the AFQ correlated with the AQ. As expected, we found a significant negative correlation (r = − 0.667, t = − 5.863, p < 0.001). However, when splitting this analysis by group, we found that this correlation was only present in the neurotypical group (r = − 0.454, t = − 2.223, p = 0.039), but not the ASC group (r = − 0.122, t = − 0.536, p = 0.598). See Fig. 1a. We assessed divergent validity by testing whether the AFQ correlated with estimated IQ. We found no evidence for such a correlation within our pooled sample, r = − 0.139, t = − 0.907, p = 0.369 (see Fig. 1b).

Cronbach’s alpha for the total scale was 0.81, for the Feelings subscale 0.82, for the imagery subscale 0.27, and for the animation subscale 0.67.

Accumulating evidence suggests that many of the social and communicative difficulties experienced by ASC individuals are linked to the motor system (Cook, 2016), and thus to motor cognition. However, testing for motoric differences may require more specialized experimental and analytical techniques, which is not practical for clinicians. Availability of the AFQ in Dutch therefore provides clinicians in the Netherlands with an easy to collect assessment tool that can provide insights into one’s social-motor skills. Additionally, the AFQ can provide novel information for researchers, both for those working with ASC and those working with neurotypical populations. These insights can help to bridge our understanding of how the motor system, as an integrated aspect of human multimodal communication, relates to social functioning.

The main limitation of this study is the relatively small sample size when compared to many questionnaire-based studies. This is due to the fact that this validation and adaptation was part of a larger study involving behavioral and brain-imaging experiments, which limits the feasibility of obtaining large sample sizes. However, the COSMIN guideline for assessing content validity (Mokkink et al., 2020) suggests a minimum sample size of 50, while the guideline of Beaton et al. (2000) recommends a sample size of at least 30–40 participants. For both guidelines, our sample size is adequate. Additionally, while we showed convergent validity for the AFQ-NL in the neurotypical population, no correlation between AFQ and AQ was found in the ASC population. The originally reported correlation between AFQ and autistic traits was found in neurotypicals (Huggins et al., 2019) and we expected that correlation to extend to ASC individuals. However, AQ may not have been the best variable to calculate convergent validity for the ASC population. We are still confident, however, that the AFQ-NL is a valid adaptation also for ASC individuals, given the strong correspondence in AFQ scores between the current study and the values reported in Williams and Cameron (2017).