The rising diagnosis of autism spectrum disorders has been recognised in the US and Europe over the past 20 years. Though, measures of diagnosis are different, which limits comparability across the globe (Boyle et al., 2011). In the UK, Taylor et al (2013) found a recorded autism diagnosis to be stable at age eight over the course of six years (in eight-year-old children who participated between 2004 to 2010), whereas US studies show that increased awareness, as well as inclusion of high-functioning and cognitively able individuals has increased the number of diagnoses (Keyes et al., 2012). While a single Swedish study reports that autism symptoms in the population have remained constant over ten years, the Swedish patient register shows larger increases in recorded diagnoses (Lundström et al., 2015). Lundström et al (2018) argue that one requires less autism symptoms to receive an autism diagnosis nowadays, and that there is no parallel increase in autistic symptoms in the population (Lundström et al., 2015).
Consequently, contradicting findings beg the question, how the trend of new autism diagnoses has developed in the UK. Therefore, Russel et al. (2021) aimed to establish the new incident time trend in autism diagnosis in the UK over 20 years (1998-2018) using the Clinical Practice Research Datalink (CPRD). Around 98% of the UK population are registered on the CPRD and this will therefore help to expand the previous study by Taylor et al (2013) by including multiple developmental stages, gender, and diagnostic subtypes of autism diagnoses. The study is the first to establish developmental trends of autism diagnoses in the UK, with an expected increase in incidence of autism diagnoses over the 20-year period.
In the UK, have incidences of autism diagnoses increased over the past 20 years?
Methods
Clinical Practice Research Datalink (CPRD): The database CPRD Aurum contains data from 738 GP practices in England and Northern Ireland. Routinely collected data of over 19 million patients is registered, such as diagnoses, prescriptions, symptoms, referrals and tests. In 2018, seven million patients contributed to the data. The data from 1990 – 2015 was validated by comparing the CPRD data to original medical records in a subsample of patients, which showed a positive predictive value of the data (Hagberg & Jick, 2017). The practices were found to be representative of the broader population, as well as gender and age.
Diagnostic codes: The codes infantile autism, autism, childhood autism, autistic disorder, Asperger’s syndrome, pervasive developmental disorder (PDD), atypical autism and autism spectrum disorder were recorded in the data. After the DSM-5 revision in 2013, the Asperger’s code was no longer used, and was integrated into the autism spectrum diagnosis variable. Lastly, an additional category to differ between ‘broad autism’ (BA) and ‘more severe autism’ (SA) was generated.
Gender was recorded for each patient, and patients were coded into developmental age bands as recommended by UK NICE guidelines (2017): preschool and infancy (0-5years), childhood (6-11 years), adolescence (12-19 years). An extra group was generated for adult diagnoses (over 19 years).
In 1998, the sample comprised 6,786,212 individuals. In 2018, the sample consisted of 9,595,598 individuals. Incidence for each age band, gender, BA and SA were calculated using the index number to assess change in incidence over time. Multivariable regression analyses were carried out to evaluate the relationship between year and index number with gender and age band. An interaction term between predictor and putative moderators (age band and gender) was added. A separate model included an interaction between gender and age band to see whether female diagnoses in adulthood contributed more to incidence pattern versus childhood patterns observed.
Results
- A higher number of males (81.6% in 1998) had a recorded autism diagnosis at baseline
- A larger proportion of younger age bands were assigned a diagnosis at baseline (91.4% in 1998 were 19 or younger, 80.6% were under 12)
- An increase of 787% in recorded autism diagnosis was seen over the course of 20 years
- The mean age of diagnosis M=9.6 years in 1998 rose to M=14.5 in 2018 overall and within each age band, however no change in infant and preschool groups was observed
- In 1998, 1 in 100,000 adults had an autism diagnosis, 20 in 100,000 adults had a recorded autism diagnosis in 2018
- Gender was found to be a significant moderator, as female gender significantly predicted greater increase in incidence over time
- Until the DSM-5 revision in 2013, ‘broad autism’ (BA) diagnoses increased more than in the ‘more severe autism’ (SA) group.
The mean age of autism diagnosis rose from 9.6 years in 1998 to 14.5 years in 2018.
Conclusions
The findings show an exponential increase of autism diagnosis over the past 20 years. As this was the first study to detect increasing reporting of diagnosis across developmental age bands and gender (especially among adults and females), replication of these findings is needed.
There was a 787% increase in recorded autism diagnosis between 1998 and 2018.
Strengths and limitations
The study used a large dataset, broadly representative of the UK population based on geographical spread and deprivation, gender, and age. This was also the first study to include developmental age bands to investigate trends among different age groups.
However the data has considerable limitations:
- Considering the long waiting times in the NHS (NHS Digital, 2021), private diagnoses might have been missed in CPRD dataset, due to a lack of communication between private and NHS practices.
- Thus, research was only able to use a comparative approach to identify patterns and trends in the data. It is unknown whether autism was actually recognised in the first instance, which could potentially mean population prevalence was underreported.
- ‘Broad autism’ (BA) and ‘more severe autism’ (SA) were impossible to distinguish after the DSM-5 revision, which made it impossible to explore differences in BA/SA groups
- It is possible that there may be a real increase of diagnoses (more people who have autism symptoms) over time. This cannot be overlooked.
This dataset may have missed private diagnoses of autism, which can be common due to the long NHS waiting times for autism diagnosis.
Implications for practice
Replication of these findings is needed to establish trends of autism diagnosis in different data sets. In doing so, recognition of changes (e.g. rise in female and adult diagnoses) may raise awareness about female autism to increase clinical recognition. Considering that one cannot leave out the possibility that there is a real increase in cases over time, the demand for access to treatment may increase. Additionally, this may also change the training provided for professionals to identify autistic individuals earlier and to reduce discrepancies between service models and provision (McCarthy, Chaplin & Underwood, 2015). Furthermore, exploring datasets can also focus on different aspects such as accessibility to diagnosis, socioeconomic status, and ethnicity into consideration. Investigating whether diagnosis was followed by support may also open doors to improve clinical guidelines and access to support overall.
Given the increase in autism diagnosis in the past two decades, there is a need to ensure increased access to autism support services.
Statement of interests
N/A
Links
Primary paper
Russell, G., Stapley, S., Newlove-Delgado, T., Salmon, A., White, R., Warren, F., Pearson, A. & Ford, T. (2021). Time trends in autism diagnosis over 20 years: a UK population-based cohort study. Journal of Child Psychology and Psychiatry. 1-9. Doi: 10.1111/jcpp.13505
Other references
Arvidsson, O., Gillberg, C. Lichtenstein, P. & Lundstöm, S. (2018). Secular changes in the symptom level of clinically diagnosed autism. Journal of Child Psychology and Psychiatry, 59, 744-751.
Boyle, C. A., Boulet, S., Schieve, L. A., Cohen, R. A., Blumberg, S. J., Yeargin-Allsop, M. . . . & Kogan, M. D. (2011). Trends in the prevalence of developmental disabilities in US children, 1997-2008. Pediatricts, 127, 1034-1042. Doi: 10.1542.peds.2010-2989
Hagberg, K. W. & Jick, S. S. (2017). Validation of autism spectrum diagnoses recorded in the clinical practice research datalink, 1990-2014. Clinical Epidemiology, 9, 475-482.
Keyes, K. M., Susser, E., Cheslack-Postava, K., Fountain, C., Liu, K. & Bearman, P. S. (2012). Cohort effects explain the increase in autism diagnosis among children born from 1992 to 2003 in california. International Journal of Epidemiology, 41, 495-503. Doi: 10.1093/je/dyr193
Lundström, S. Reichenberg, A., Anckasarsäter, H., Lichtenseun, P. & Gillberg, C. (2015). Autism phenotype versus registered diagnosis in Swedish children: prevalence trends over 10 years in general population smaples. BMJ, 350.
McCarthy, J., Chaplin, E. & Underwood, L. (2015). An English perspective on policy for adults with autism. Advances in Autism, 1, 61-65. Doi: 10.1108/AIA-08-2015-0011
National Institute for Health and Care Excellence (2017). Autism spectrum disorder in under 19s: Recognition, referral and diagnosis.
NHS Digital (2021). Autism Waiting Time Statistics – Quarter 2 2019-20 to Quarter 1 2020-21 and Quarter 2 (July to September) 2020-21.
Taylor, B., Jick, H. & MacLaughlin, D. (2013). Prevalence and incidence rates of autism in the uk: time trend from 2004-2010 in children aged 8 years. British Medical Journal Open, 3 , e003219
Photo credits
- Photo by Annie Spratt on Unsplash
- Photo by Denes Kozma on Unsplash
- Photo by Anna Vander Stel on Unsplash
- Photo by Isaac Smith on Unsplash
- Photo by Levi Meir Clancy on Unsplash
- Photo by Caleb Woods on Unsplash
