Autism Spectrum Disorder (ASD) is a lifelong neurodevelopmental condition characterized by a range of differences in social communication, behavior, and sensory processing (1). An accurate diagnosis of autism is paramount for several reasons. Individuals who are formally diagnosed with autism can receive specific support that significantly improves their quality of life (2). For instance, once diagnosed, individuals can secure tailored educational plans that cater to their learning styles and needs, and obtain workplace accommodations that enhance their job performance and job satisfaction (3). This support directly addresses the unique challenges faced by autistic individuals, promoting better educational and occupational outcomes. Furthermore, a diagnosis can foster a sense of community and belonging. Studies have shown that autistic individuals who connect with others in the autistic community experience lower levels of loneliness and increased emotional support (3, 4).

Beyond post diagnostic support and establishing a community, an accurate diagnosis can validate the struggles and experiences that individuals have faced, sometimes for many years (3). This validation contributes to enhanced self-acceptance and self-esteem, as individuals no longer feel as though they are struggling in isolation or grappling with unrecognized difficulties (5, 6). Research by Powell and Acker (7) indicates that a diagnosis can help individuals make sense of their experiences and behaviors, fostering a positive self-identity and reducing feelings of isolation or being misunderstood.

Overall, the above highlights the comprehensive benefits of receiving an autism diagnosis. However, the diagnostic process, especially in adulthood, is difficult. Autism’s heterogeneity means that autistic features can present very differently across individuals, making standardized diagnostic tools less effective (8). Furthermore, clinicians often struggle with limited resources and institutional pressures, which complicate the diagnostic process and can lead to missed or misdiagnoses (1). The complexity is compounded for adults who may have developed coping mechanisms, such as camouflaging their symptoms, making them appear non-autistic to others, even professionals (9).

A further significant challenge in diagnosing autism is the fact that other psychiatric disorders can mask or mimic autism symptoms. Many autistic individuals also experience additional psychiatric comorbidities, such as depression, anxiety, and attention deficit hyperactivity disorder (ADHD) (10–12). Jadav and Bal (13) found that psychiatric conditions are more frequently endorsed by adults diagnosed with autism in adulthood compared to those diagnosed in childhood. These comorbidities can complicate the diagnostic process, leading to misdiagnosis or delayed diagnosis (10).

Diagnosing autism in women presents another layer of difficulty. Research indicates that females are often diagnosed later than males and are more likely to be misdiagnosed initially (14). This delay and misdiagnosis can be attributed to a higher likelihood of women to camouflage their features, employing strategies to hide their autistic traits and better fit into social norms (9). This camouflaging behavior not only makes it harder for clinicians to identify autism but is also associated with worse mental health outcomes, as individuals exert significant effort to mask their true selves (15).

Additionally, alexithymia, a condition characterized by difficulty in identifying and describing emotions, is prevalent among autistic individuals and complicates the diagnostic process (16). Hobson et al. (17) found that alexithymia can affect the outcomes of autism diagnostic assessments, as socio-emotional abilities, which are critical for these assessments, are impaired. This can lead to an underestimation of autistic symptoms or even a failure to diagnose autism altogether.

Overall, while receiving an autism diagnosis is important and beneficial for autistic individuals, the diagnostic process is accompanied by many struggles, resulting in misdiagnoses that complicate prevalence data, affect research outcomes and may misdirect individuals from therapies that best suit their needs (18–20). To overcome these issues, it is essential to implement robust and comprehensive diagnostic practices that ensure individuals receive the appropriate support and intervention. One reason for these difficulties is the lack of well-validated measures that accurately capture the real-life social difficulties autistic individuals experience in daily interactions with non-autistic individuals. This highlights the critical need to safeguard research integrity and optimize resource allocation by ensuring diagnostic tools are reliable, valid, and reflective of the complexities of autism. Addressing this diagnostic gap requires the development and use of more nuanced tools that can improve diagnostic accuracy and reduce the risk of misdiagnosis.

One promising tool that has emerged recently is the Short Story Task (SST) (21). In the SST, participants read ‘The End of Something’ by Ernest Hemingway and answer a series of comprehension, and mentalizing questions. This story-like design closely mirrors real-life experiences and assesses the difficulties autistic individuals face in interpreting the actions of non-autistic individuals. The SST has demonstrated discriminatory power when differentiating between autistic and non-autistic individuals (22), as well as distinguishing autistic individuals from those with depression (23).

While these initial findings are promising, they primarily involve comparisons with healthy non-autistic individuals or those with isolated psychiatric conditions. They do not fully reflect the complexity of real-world clinical settings, where mentalization impairments are often observed across a range of psychiatric disorders (24). In outpatient services for autism diagnostics, patients are typically not compared to individuals without any mental health problems but rather to those experiencing various forms of psychological distress and social difficulties (25). These individuals seek help for challenges they face in daily life, often suspecting autism as a possible underlying cause. This creates a more nuanced and challenging diagnostic landscape, where tools like the SST must demonstrate their reliability and specificity in distinguishing autism-related difficulties from those associated with other conditions.

To explore whether the SST can meet these challenges, we conducted a retrospective study involving 211 outpatients who presented at a specialized outpatient autism service at a tertiary referral center (Clinic and Polyclinic for Psychiatry and Psychotherapy at the University of Regensburg). Our objective was to evaluate the SST’s discriminatory power within this heterogeneous clinical population, focusing on its ability to differentiate between those ultimately diagnosed with autism and those who were not. We hypothesized that the SST would show sufficient discriminatory ability in this setting, providing a valuable tool for improving the accuracy and reliability of autism diagnostics in outpatient services.

Descriptive statistics for the total sample, as well as for the ASC and no-ASC groups, are presented in Table 1 . There were no significant differences between the groups in sex, age at first contact, verbal IQ, non-verbal IQ, level of education, or total alexithymia score (see Table 1 ). However, the ASC group scored significantly higher on the EOT alexithymia subscale compared to the no-ASC group, indicating a higher level of externally oriented thinking. Furthermore, the groups did differ significantly in their scores on the AAA, AQ, EQ, and SI difference measure (see Table 1 ). Specifically, the ASC group exhibited higher scores on the AAA, AQ, and SI, but lower scores on the EQ compared to the non-ASC group. Regarding SST performance, no significant differences were found between the ASC and non-ASC groups in either comprehension or mentalizing (see Table 1 ). Finally, there was a significant difference in the number of books read, with the ASC group reporting fewer books read. The distribution of diagnostic categories and other psychiatric diagnoses across the two groups is depicted in Figure 1 . See Figure 2 for a graphical depiction of SST performance across groups.

The aim of this study was to evaluate the discriminative power of the Short-Story Task (SST) in a complex clinical setting, specifically among 211 outpatients undergoing autism diagnostics at a clinic and polyclinic for psychiatry and psychotherapy. We hypothesized that the SST would effectively differentiate between autistic and non-autistic individuals, thereby enhancing the accuracy and reliability of autism diagnostics in outpatient services.

Our study revealed significant group differences in established autism measures, such as the Adult Asperger Assessment (AAA), Autism Quotient (AQ), Empathy Quotient (EQ), and Sensory Inventory (SI). While these tools provide valuable insights, it is important to interpret the results with care. For instance, research shows that self-report questionnaires like the EQ rely strongly on mental imagery (38), which can be reduced in autistic individuals (39) and potentially resemble low empathy regardless of actual empathic abilities. Similarly, the AQ, which relies on self-perception, may introduce discrepancies between self-reported traits and actual behaviors (40). To ensure diagnostic accuracy, a comprehensive assessment process was employed, incorporating psychological testing, structured interviews, and external anamnesis, followed by multidisciplinary team discussions.

No significant differences were observed in SST performance between the groups. Importantly, the SST was not used as a diagnostic tool in this study, which avoided potential biases in group assignment. While this approach allowed for an unbiased comparison of SST performance, it also limited our ability to directly compare the SST with other diagnostic measures, as these measures, along with structured and unstructured diagnostic interviews, determined group membership. The lack of significant differences in SST performance suggests that the SST may not be sensitive enough to identify autism based solely on mentalizing performance scores. This conclusion is supported by the SST’s moderate internal consistency and its poor discriminative ability between ASC and non-ASC groups, as indicated by ROC analyses.

Although the SST has previously demonstrated good discriminatory ability in distinguishing between autistic and non-autistic individuals (22), and between autistic and depressed individuals (23), its effectiveness appears diminished in a sample with various comorbidities (10, 13). Individuals in our sample may have developed camouflaging techniques and exhibited higher alexithymia, both of which are associated with complications during the diagnostics process (9, 17). Moreover, although the significant correlation between the externally oriented thinking scale of alexithymia and SST mentalizing performance did not hold after correction, it may still indicate a weak relationship between reduced attention to one’s own emotions and the ability to infer feelings and intentions of non-autistic others or characters within a story.

Regression analysis revealed that group membership (ASC vs. non-ASC) was a significant predictor of SST mentalizing performance. This suggests that, while the SST may not strongly differentiate between ASC and non-ASC groups overall, certain elements of the task do capture underlying differences in mentalizing abilities associated with autism, particularly when mentalizing is applied to non-autistic individuals and characters. However, SST comprehension and mentalizing scores were not effective in directly predicting group membership, highlighting a limitation in their diagnostic utility.

Notably, exploratory analyses uncovered that the ASC group performed significantly worse on a specific comprehension item involving the understanding of a romantic relationship’s end within the story. We are aware that exploratory findings have to be interpreted with caution. Nevertheless, this finding may suggest that certain narrative elements or themes may be particularly challenging for individuals with ASC, potentially due to difficulties in processing social and emotional cues, especially when they are not prompted (41). A subsequent logistic regression using this specific item as a predictor effectively distinguished between ASC and non-ASC groups, suggesting that individual replies to SST items, especially those tapping into complex social or emotional understanding, may offer greater discriminatory power than the overall SST scores.

While one might consider the possibility that autistic individuals may have deficits in linguistic comprehension, we found no significant differences in the overall comprehension score or in the other comprehension items between the two groups. Additionally, there were no differences in verbal or non-verbal IQ between the ASC and non-ASC groups, further suggesting that purely linguistic factors are unlikely to explain these findings.

Interestingly, although the item focusing on the story’s theme (a break-up) was not originally classified as a mentalizing item, its effectiveness in distinguishing between groups highlights a broader issue. If nuanced aspects of the story’s break-up are not fully recognized, the comprehension question may shift from measuring detailed understanding to evaluating a broader grasp of subtext. This suggests that comprehension items might measure overall interpretive ability rather than just specific comprehension skills.

These insights suggest that while traditional diagnostic tools still rely on specific cut-off scores, there may be value in considering a shift from focusing solely on these thresholds to evaluating the reasoning behind individual responses in the SST. By examining how individuals interpret social narratives and characters’ actions, this approach could provide a more detailed understanding of cognitive and emotional processing. Rather than emphasizing predetermined thresholds, it allows for a more nuanced view of how individuals articulate their understanding, which could offer insight into their cognitive and emotional experiences. Additionally, since the SST is not a self-report measure but assesses actual interpretive responses, it could provide a more direct reflection of real-world understanding, particularly in social contexts. While the current analysis follows the SST manual (21), future research could explore whether a more narrative-focused analysis of individual responses might complement traditional diagnostic approaches, potentially enhancing the SST’s diagnostic utility. This approach aligns with a broader trend in autism assessment, emphasizing a move towards personalized, individual profiling that accounts for cognitive diversity within the spectrum (42). Traditional diagnostic tools often rely on scoring thresholds that may not fully capture the complexity of autistic experiences and may inadvertently exclude subpopulations, such as autistic females (43).

While this approach shows promise for enhancing understanding and guiding psychosocial interventions, we acknowledge that further research is needed to establish its potential impact on diagnostic accuracy and utility. This shift toward a more personalized approach should be viewed as complementary to, rather than a replacement for, existing diagnostic practices.

The study has several strengths, including a large sample of individuals seeking autism diagnostics from a specialized outpatient clinic and the SST’s objective assessment format, which is not used for group assignment. The SST’s question-answer approach also provides valuable insights into real-life scenarios and social-emotional processing. However, there are notable limitations.

Firstly, the sample comes from a psychiatric clinic and polyclinic, which means it includes more complex cases with comorbidities. While this sample includes patients that are particularly difficult to diagnose, it could limit how well the findings apply to other groups. Additionally, the high levels of alexithymia observed in the sample, which are associated with difficulties in emotional processing, might have influenced the results. Elevated alexithymia scores could confound SST performance and complicate its interpretation, suggesting that its utility may vary based on the emotional and cognitive characteristics of the sample.

Lastly, research indicates that autistic individuals often benefit from frequent social learning experiences through reading (44), and while our sample’s mentalizing performance was not influenced by the number of fiction books read, the measure used may be too simplistic to capture the full range of reading experiences. This limitation underscores the need for more nuanced, research to better understand how social and emotional processing evolves over time in autism.

In conclusion, this study evaluated the discriminative power of the Short-Story Task (SST) in a complex clinical setting with individuals seeking autism diagnostics. While the SST did not differentiate between autistic and non-autistic groups, our findings suggest that its diagnostic sensitivity may be limited in populations with various comorbidities.

However, certain narrative elements, such as social and emotional cues, showed potential for distinguishing between groups. This suggests that a narrative-focused approach, which is not inherent to the SST but could be explored further, may offer more valuable insights into individual cognitive and emotional processing. This aligns with a broader trend in autism diagnostics toward moving beyond rigid thresholds and considering individual profiles.

While the SST is not designed to specifically evaluate narrative responses, integrating a more narrative-focused analysis could complement traditional assessments and provide a more nuanced understanding of autism. Further research is needed to explore how this approach might enhance the SST’s diagnostic utility, particularly in complex clinical populations.