Childhood Diagnoses in Individuals Diagnosed With Autism in Adulthood

Autism is a developmental condition, where symptoms are expected to occur in childhood, but a signicant number of individuals are diagnosed with autism for the rst time in adulthood. Here we use the National Danish Patient Registry to investigate diagnoses given in childhood among those that are diagnosed with autism in adulthood (N = 2199). We found that most childhood diagnoses were given after the age of 12, and attention-decit hyperactivity disorder, affective disorders, anxiety, and stress disorders were the most prevalent childhood diagnoses. However, 69% of males and 61% of females with adult autism diagnoses had not received any of the included diagnoses before the age of 18. In most cases, the late autism diagnosis is therefore unlikely to be explained by either misdiagnosis or diagnostic overshadowing. This result is at odds with the prevailing notion that autistic symptoms tend to diminish with age. Therefore, further research is warranted to examine how early signs of autism may have manifested among these individuals, and how similar they are to autistic people diagnosed earlier in their development. Milder to moderate cases of psychiatric conditions that have been solely treated by family physician or school psychologists may not be fully included in our dataset.


Introduction
According to diagnostic criteria(1), symptoms of autism must appear in childhood, but there are also individuals who are diagnosed with autism for the rst time during adulthood. This has led researchers to hypothesize a "lost generation" of autistic individuals whose autism symptoms were not identi ed in childhood (2). It is relevant to question why the autism condition, which should have manifested across contexts during childhood, might have been missed until adulthood. It has been suggested that factors such as high IQ, female gender and camou aging behaviours may hinder an early identi cation of the condition (3,4).
Studies of autism diagnosed in adulthood generally report a high frequency of comorbid psychiatric diagnoses (5)(6)(7)(8)(9). Another hypothesis therefore is that autism symptoms in childhood were mistaken as symptoms of other conditions (5,10) such as attention-de cit hyperactivity disorder (ADHD) (11,12) or obsessive-compulsive disorder (OCD)(13) due to symptom overlap. It is also possible that both conditions were present, but due to diagnostic overshadowing, the autism symptoms were attributed to the other condition and no autism diagnosis was given (14,15).
Although numerous studies have investigated previous psychiatric diagnoses among individuals diagnosed with autism as adults, most of these studies did not investigate when previous diagnoses were given, i.e. in childhood or relatively shortly before the autism diagnoses well into adulthood. Here, we speci cally investigate childhood deviation in individuals who were diagnosed with autism in adulthood. We used a Danish national health registry to investigate which diagnoses these individuals were given in childhood, and thus whether misdiagnosis and diagnostic overshadowing might explain why autism can have gone undiagnosed throughout childhood.

Method
Data was retrieved from the Danish National Patient Registry (DNPR), which contains information about all diagnoses given within the Danish hospital sector (in-patient and out-patient) from 1994 to 2018. The DNPR is connected to the Danish Central Person Registry, allowing diagnoses given to the same individual at different times to be linked. Access to the registries were obtained through the Danish Health Data Authority. To ensure that the available data covered all relevant diagnoses given in childhood, we excluded individuals born before January 1, 1993. Our autism population thus consisted of individuals born between January 1, 1993 and December 31, 2000 who were diagnosed with autism (ICD-10 codes F84.0, F84.1, F84.5, F84.8, F84.9) at age 18 or later (1312 males, 887 females). We calculated the percentage of males and females that had received one of the psychiatric or neurological diagnoses shown in Figure 1 before the age of 12, as well as before the age of 18. The ICD-10 codes used for each diagnosis are shown in Supplementary Table S2. The diagnoses were chosen based on what previous studies have found in populations diagnosed with autism in adulthood, as well as diagnoses as which autism might be misclassi ed due to symptom overlaps. The data was extracted on August 24, 2020.

Results
The childhood prevalence (0-17 years of age) of each selected diagnosis among those diagnosed with autism in adulthood is shown in Figure 1a. 69% of males and 61% of females had not been registered with any of the selected diagnoses before their 18 th birthday. Since autism symptoms would be expected to occur in early childhood, we also investigated how often the selected diagnoses were given before the age of 12. This is shown in Figure 1b. 84% of males and 91% of females had not been registered with any of the selected diagnoses before their 12 th birthday.

Discussion
The presence of childhood diagnoses of psychiatric or neurological disorders may explain why autism has not been diagnosed throughout childhood, e.g. due to misdiagnosis or diagnostic overshadowing.
The present data showed that only 31% of males and 39% of females were registered as having received any of the diagnoses shown in Figure 1 before the age of 18, and even fewer were registered with any of the diagnoses before the age of 12 (16% of males and 9% of females). In these cases, the late autism diagnosis may be attributed to misdiagnosis or overshadowing. Similarly, there may have been a recategorization of diagnoses, such that symptoms that are now classi ed as autism would previously have been categorized as another condition, e.g. intellectual disability (16). However, the remaining majority of late autism diagnoses likely have other explanations. Some individuals may not have exhibited signs of autism to a degree that has caused a thorough diagnostic evaluation to be initiated. Recent evidence suggests that the threshold for receiving an autism diagnosis has been lowered through the last decades (17), which may also explain why a person that did not meet the criteria in childhood can get a diagnosis later in life.
Some researchers have hypothesized the concept of camou aging, describing coping strategies used to hide autism symptoms and maintain an appearance of normal social functioning(4). Studies of camou aging have suggested that the coping behaviour requires an intense effort(4), which may eventually cause distress to a degree where a psychiatric evaluation is initiated, leading to a late autism diagnosis (18). Alternatively, it is possible that stressors experienced later in life (such as increased social and executive requirements during the transition from adolescence to adulthood) may exacerbate latent subclinical traits and trigger a late onset of a phenotype that resembles autism. There have been concerns about the validity of the camou aging construct and suggestions that further research is needed to operationalize it and differentiate it from other constructs (19). Speci cally, it is relevant to strengthen the validity of the distinction between camou aged symptoms and absence of symptoms, in order to elucidate to which extent camou aging contributes to individuals with an adult autism diagnosis not being diagnosed in childhood.
We found high incidences of ADHD, mood disorders, and anxiety before the age of 18, which is consistent with previous studies of comorbidity in individuals diagnosed with autism in adulthood (5)(6)(7)(8)(9)20). Incidences before the age of 12 were considerably lower, showing that most of the pre-autism diagnoses were given in adolescence. Sex differences of childhood prevalence were generally in the same direction as sex differences among the general population (Supplementary Table S1). We found a high incidence of stress disorders before the age of 18, particularly among females, whereas previous studies have either not reported rates of stress disorders(5-7) or only reported rates speci cally for post-traumatic stress disorder (8,9).
The fraction of individuals who did not receive any of the investigated diagnoses in childhood was slightly higher than what has previously been reported. Rydén & Bejerot (2008) (8) found that 53% of those diagnosed with autism as adults had received psychiatric care in childhood. This difference might be attributed to changes in the autism population over time, or differences in which diagnoses were included. It is also possible that the difference can be ascribed to the fact that the DNPR only contains diagnoses given in the hospital system, while some individuals may have received childhood psychiatric care in the primary healthcare sector, e.g. for conditions such as depression or anxiety (21).

Limitations
The present results are based on records of autism diagnoses and cannot inform about how the diagnoses correspond to a "true" autism condition. Because our data only contains diagnoses given within the hospital sector, some individuals with an adult autism diagnosis may not be included if they were diagnosed and treated exclusively by their primary physician or a psychiatrist in private practice.
However, since diagnostic assessment for autism in adults is complicated by several factors(2), diagnoses not conducted by specialist teams might have insu cient validity and could introduce falsepositive diagnoses if included. Investigated childhood diagnoses also do not include those given by school psychologists or primary physicians. However, problems that are managed solely by a primary physician or school psychologist (e.g. milder cases of anxiety or depression) are likely less pronounced than the cross-context challenges that would be expected due to autism.

Conclusion
Most individuals receiving an autism diagnosis in adulthood did not receive any of the investigated psychiatric diagnoses. In such cases, the late autism diagnosis cannot likely be explained by misdiagnosis or overshadowing. Further research is warranted to investigate how autism manifested in childhood in these individuals, whether they might have bene ted from access to support, and whether cases of autism that only reach a clinical threshold in adulthood are biologically distinct from their childhood counterparts. Competing interests KWM declares having received consultancy fees from Lundbeck A/S and Janssen-Cilag A/S in the past three years. EMR, KJ, and LM declare no con icts of interest. Funding EMR was supported by grants from the Ivan Nielsen Foundation and the So e Foundation. KWM is supported by the Lundbeck Foundation and the Weimann Foundation. LM was supported by a grant from Brain Canada (MIRI2015-3736). The funding bodies were not involved in designing or conducting the study, nor in writing the manuscript.

Authors' contributions
All authors contributed to the design of the study. EMR and KJ performed the analyses and drafted the manuscript. KWM and LM critically revised the manuscript. All authors read and approved the nal manuscript. Childhood prevalence (0-17 years) of diagnoses among males and females diagnosed with autism as adults. a) Diagnoses given before the age of 18. b) Diagnoses given before the age of 12.