Suicide risk and autism: data from "Utah over a 20-year period"

The findings reported by Anne Kirby and colleagues [1] are not for the faint-hearted but continue a vitally important theme in autism research and practice circles looking at suicide risk and autism (see here and see here). The 'value-added' bit to the Kirby results to distinguish them from other study in this area was the reliance on data across a 20-year period from a place that has some important autism research history: Utah in the United States (see here).

So: "Four sources of existing data were utilized for this study: URADD statewide autism surveillance data, statewide suicide surveillance data collected by the Utah Office of the Medical Examiner (OME), the UPDB, and Utah's Indicator‐Based Information System for Public Health (IBIS‐PH)." From such data sources, researchers accessed information for nearly 17,000 people diagnosed with an autism spectrum disorder (ASD) "alive at the beginning of 1998 and at least 5 years of age in 2013." Most were male and most were white. They similarly determined that nearly 9,000 people in their total population - not just those diagnosed with ASD - died by suicide between 1998 and 2017. Again, most were male and most were white. The data were combined and interrogated "to calculate the incidence (in 5‐year intervals) of suicide deaths in people with ASD over a 20‐year period (1998 to 2017) in total, as well as by sex, and compared suicide risk in people with versus without ASD." Researchers also looked at other important variables such as "sex, race, death age, occupational status, marital status, and manner of death" across the groups.

Results: "In the first 15 years of the study (1998–2012), we did not observe differences in suicide cumulative incidences between the ASD and non‐ASD populations." This means that when the groups were compared as a function of death by suicide, the figures for those with autism were not significantly different from those without autism for this time period. By saying that I don't want to belittle the fact that between 1998 and 2012 for example, 2 people out of 5,202 autistic people died by suicide or that 1,671 out of 1,928,484 non-autistic people died. Each of these figures was a person with a life and with a family, and that's something that should never ever be forgotten.

The pattern however changed when researchers looked at the period between 2013 and 2017: "For the most recent time interval (2013–2017), the cumulative incidence of suicide death in the ASD population was 0.17%, which is significantly higher than the non‐ASD population cumulative incidence of 0.11%." This percentage (0.17%) represented 28 deaths from an autistic population of 16,907 and 2,791 deaths from a non-autistic population of 2,630,221. Although a sideline point, I'll also bring to your attention how the autistic population numbers changed over the 5-year blocks of study in the Kirby paper: 1998-2002: 5,202 people; 2003-2007: 8,722 people; 2008-2012: 13,890 people; 2013-2017: 16,907 people.

A few other details were observed by Kirby et al: "In comparison with non‐ASD + suicide cases, ASD + suicide cases had significantly younger average death ages (32.4 years vs. 41.8 years; t = −3.8, P < 0.001)." Also: "Combined, 73% of the ASD + suicide cases used methods for suicide considered to be violent; the remaining 26% used nonviolent methods." This again, is important information.

Researchers also mention how across the 2013-2017 period, another important trend was observed: "suicide risk in females with ASD was over three times higher than in females without ASD (relative risk (RR): 3.42; P < 0.01)." They contrast this with the finding that "there were no documented cases of suicide death among females with ASD during the first 15 years of the surveillance period" and what this could mean when it comes to possible explanations of suicide risk in relation to autism.

There's quite a bit to learn from the Kirby findings. Although there are limitations attached to the study design - "inadequate data on intellectual ability was available to examine the influence ID may have on suicide risk in individuals with ASD" - the study was a good one because of its population-wide focus and the pretty good autism-related resources that Utah has (and has had for many years). It demonstrates once again that the difficult topic of suicide and autism should remain a research priority in order to identify who might be most at risk and why, alongside the ways and means that such risk *might* be mitigated (see here for one example).

And for those who might need someone to text / email / talk to, there are always options (see here for services in the UK or see here for those elsewhere).

----------

[1] Kirby AV. et al. A 20-year study of suicide death in a statewide autism population. Autism Res. 2019 Jan 21.

----------

Statistically significant group differences in self-reported repetitive movements between diagnosed and self-identifying autism

One needs to remember that, at the time of writing this post, the article posted by Joost Wiskerke and colleagues [1] suggesting that "camouflaging of RMs [repetitive movements] may contribute to under diagnosis of autism, at least in females and transgender people" is just a preprint ("a version of a scholarly or scientific paper that precedes publication in a peer-reviewed scholarly or scientific journal"). It has not been through formal peer-review yet, but rather the authors have bravely 'put it out there' for old farts like me to pore over.

Having glanced through the abstract first and discovered the use of the words "self-identified as autistic" my brow furrowed somewhat. Upon further reading of the paper in its entirety, the furrowing brow furrowed a little less as I understood why the authors mentioned use of a cohort that was "56% formally diagnosed participants and 44% who self-identified as autistic." But that's not to say that I was completely enamoured with the study write-up as it currently stands...

Before continuing and in order to repel any accusations that come my way, I'll mention that I don't have any particular issue with people without a formal diagnosis of autism thinking/believing they are on the autism spectrum. It's their right to do so and indeed, such self-realisation is often an important step towards getting a professional assessment for autism. For some people, their hunch about being autistic eventually turns out to be correct as per them reaching defined clinical cut-off points for autism on the various instruments of assessment that are available during said professional assessment. But that's not to say that every person who self-identifies as being autistic/having autism is always going to be right. As we are starting to realise from the wealth of peer-reviewed science on this topic, the label autism does not seemingly have any exclusive rights to the presentation of autistic traits (see here and see here for examples). And it is for that reason why we have so many good and knowledgeable diagnosticians when it comes to such professional assessments, combined with an important focus on: "Symptoms caus[ing] clinically significant impairment in social, occupational, or other important areas of current functioning" in order to receive a diagnosis. Access to such autism assessments is another issue, but shouldn't be used as an excuse...

Onward:

Wiskerke et al started from the premise that repetitive movements (RMs) are common in autism but there are some potentially important differences in the expression of such issues between the genders. This follows other independent findings in this area (see here) detailing how stereotyped and repetitive behaviours may be less frequently observed in females [2] minus any sweeping generalisations. Researchers then moved on to suggesting that one of the reasons why such RMs may be less frequently observed is because they are actively being masked or camouflaged, and this could eventually lead to an under-diagnosis of autism among certain groups. 'Masking' or camouflaging in the context of autism is a bit of a hot [social media] topic at the moment (see here and see here).

"In this exploratory study, we took a first step... by using self-report measures from a large number of female and transgender adults, using recruitment on social media to an on line questionnaire." Ah yes, the on line questionnaire, which seems to becoming more and more popular in certain autism research circles [3] (see here also for another example). There's nothing wrong with using such tools for asking about opinions and the like, but they are typically open to anyone and everyone (who has access to the internet!), and not exactly great for authenticating complicated things like clinical diagnoses, behaviours and comorbidity for example. As for the incorporation of data from transgender adults, well, again this follows a theme in autism research recently (see here) where 'autistic identity' and sexual identity seem to be converging for some people/groups (see here).

"We assessed current RMs using a combination of visual analog scales for specific behaviors and textboxes for free-text responses." There's mention of using some of the DSM-5 criteria for autism in this section. Authors also talk about testing for camouflaging "using the matrix multiple-choice question “Did/do you hide these behaviors from others…” 1) “…as a child?”, 2) “…as an adolescent?”, 3) “…as an adult?”" Yet again (see here) a research priority needs to be the formulation of a valid and reliable questionnaire pertinent to 'measuring' masking/camouflaging with autism in mind. And finally there was the use of an old favourite  - the autism spectrum quotient (AQ) - by the authors, but with some added caveats...

Results:  "We found high rates of RMs in both diagnosed and self-identifying participants, and a striking prevalence of camouflaging" was one of the headlines. But... there were also some other important details observed too. So: "Higher scores in the diagnosed group were found for object fidgeting, repetitive hand movements, rocking, object spinning and hand flapping." I think most people would agree that such behaviours represent some of the more 'classical' manifestations included in the RM categorisation of autism. By contrast, other RMs such as scratching/rubbing skin, walking in circles and 'banging head' were not different between the diagnosed and self-identifying groups.

Based also on the strength of the camouflaging data presented by Wiskerke et al I have to say that I'm not yet altogether convinced by the evidence presented. It's not that I don't believe that 'self-inhibition' plays a role in such masking, nor that: "There were many references (47 participants) to having been bullied or disciplined for childhood RMs." It's just that methodologically speaking, offering up one question on such a complicated issue does not make for a scientifically compelling argument. As I previously said, autism science needs to invest in some high quality research on how to assess masking in the context of autism; perhaps including some important measure of self-monitoring for example. And one also needs to control for things like intellectual ability too and perhaps be open to other reasons why the presentation of autism between the genders/sexes might be subtly different.

I was also a little dismayed that the discussion of results by the authors did not seem to fully 'tally' with their findings. So: "The striking similarities between diagnosed and undiagnosed participants are consistent with a clinically relevant prevalence of autism in the undiagnosed group." As I've mentioned, the picture of 'striking similarities' was far from consistent when comparing RMs across the groups and those statistically significant differences reported on between diagnosed and self-identifying autism. Added to the fact that authors zoomed in on RMs without too much clinical focus on the other elements to autism, and the 'clinically relevant' picture is also far from complete based on this study alone. Indeed, other text in the discussion provide further clues as to the probable inclination of the authors: "We believe that this study in part reached the “lost generation” of autistic adults... many of whom appear to have turned to social media for support and kinship, sometimes after many disappointing encounters with clinicians and scientists." Emotive language such as 'Lost generation' really shouldn't be in a science paper without [strong] corresponding evidence.

What else? Well the authors did acknowledge some shortcomings in their study: "the on line format limited our ability to ascertain that robust diagnostic procedures had been used in all cases" and: "The self-report format also makes it possible that some participants erroneously reported an official diagnosis" (erroneously?) but perhaps more is needed to be said about participants ("the vast majority were indeed very cognitively able") and onward the applicability of results to other parts of the autism spectrum. It would also have been useful to include a few other (self-report) measures of other conditions/labels where autistic characteristics overlap, just to see...

I do think Wiskerke et al have tapped into a increasingly important research 'need' for some of those on the autism spectrum - masking - and how such behaviour does seem to impact on quality of life for quite a few people. We need a lot more research on this topic, and yes, there also needs to be some further analysis of what social accommodations could be made too, bearing in mind such understanding needs to come from lots of different quarters of society (see here). I'd also like to acknowledge the fact that one of the authors on the Wiskerke took the time to converse (on Twitter) with me about their paper, which was rather encouraging.

But still, I can't shake the idea that the methodological shortcomings of this study and the sweeping interpretations imply caution before any generalisations are made as a result. The fact also remains that self-identifying as being autistic/having autism is not the same as receiving a formal diagnosis of autism, however much people might want this to be true or find 'kinship' with the autism spectrum...

----------

[1] Wiskerke J. et al. Camouflaging of repetitive movements in autistic female and transgender adults. bioRxiv. 2018. Sept 10.

[2] Mandy W. et al. Sex differences in autism spectrum disorder: evidence from a large sample of children and adolescents. J Autism Dev Disord. 2012 Jul;42(7):1304-13.

[3] Kupferstein H. Evidence of increased PTSD symptoms in autistics exposed to applied behavior analysis. Advances in Autism. 2018; 4: 19-29.

----------

"ASD characteristics in adulthood are differently perceived across age, sex, and informants"

The research tag-team that is Anne Lever and Hilde Geurts have provided peer-reviewed fodder for this blog before (see here and see here). Without wishing to cajole their research interests into a specific box, quite a bit of their time seems to be taken up by looking at autism in the context of ageing, alongside how the label of autism does not seem to have a monopoly on the presentation of certain autistic traits.

A recent paper published by this team [1] provides yet more blogging material, specifically focused on testing "the association between age and ASD [autism spectrum disorder] characteristics, including empathy and sensory sensitivity, in adults aged 19–79 years." A scientific hat-tip is offered to other research in this area [2] that previously observed that: "older age was associated with higher ratings of ASD traits and better cognitive performance."

Drawing on data derived from a participant group numbering above 400 (N=237 with autism and N=198 without autism) spanning the age ranges, a variety of self-report and informant-report (family members, friends, other significant others) were utilised. I should point out that the autism participant group seemed to be represented by the 'more able' part of the autism spectrum (I don't use the term 'functioning') insofar as most either being diagnosed with Asperger syndrome or PDD-NOS (pervasive developmental disorder - not otherwise specified), being predominantly independent or living with a partner or housemate in residential status terms and also with that reliance on self-report used throughout the study kept in mind. This is worth knowing given other discussions on under-represented groups when it comes to scientific study (see here) and how representative autism research may or may not be to the entire spectrum.

Among the measures included for analysis we have an old favourite - the Autism-Spectrum Quotient (AQ) - as well as the Interpersonal Reactivity Index (IRI) (measuring various aspects of empathy) and the Sensory Sensitivity Questionnaire (SSQ) (examining sensory hyper- or hyposensitivity). Obtained results were collated and subjected to quite a few statistical analyses.

Results: noted as a 'group difference' findings were reported observing that: "Adults with ASD reported higher scores on the SSQ and on all subscales of the AQ than adults without ASD." This is pretty much what would be expected, despite any qualms I might have about what the AQ actually measures (see here for more of them). Sensory issues being reported as being greater in those with a diagnosis of autism also ties in well with their inclusion in the DSM-5 schedule for diagnosing autism or ASD (see here).

Then: "Within the ASD group, age-related differences were observed in self-reported ASD traits and sensory sensitivity, with a peak among middle-aged adults." Alongside that previously 'hat-tipped' study, the authors conclude that "ASD characteristics are more heavily experienced in middle adulthood than in younger or older adults." This is interesting from quite a few perspectives; bearing in mind that such findings say nothing about the expression of autism in childhood and early adulthood. Appreciating that the self-report of autism or sensory traits may not be the same as everyday autism expression, such findings *might* have some relevance to various issues such as the rise and rise of adult autism diagnoses being given and indeed, how a diagnosis of autism for some (a few?) might not be a 'lifelong' issue (see here). I know that last point raises blood pressure in some quarters as words like 'masking' are banded around (with the need for far greater study), but there is good reason to think that like many other conditions/labels, autistic behaviours and/or traits ebb and flow according to environment and perhaps other factors, such as the presentation of comorbidity for example (as per the author's other work [3] already mentioned). In short, the presentation of autism is probably dynamic and fluidic, rather than just a static thing.

Onward: "we replicated earlier findings that females with ASD had more sensory issues and reported more ASD characteristics than males... whereas females without ASD manifested fewer ASD traits than non-ASD males.". Females with ASD (74 females vs. 163 males) 'reported more ASD characteristics than males'? Intriguing - "ASD females reported higher scores than ASD males on the AQ total score" - to say the least in light of other [childhood] findings [4] but not without cautions. I go back once again the question of what the AQ is actually measuring and whether for example, the typically higher rate of mood disorders generally noted in females could be a potential confounder when it comes to AQ scores in the context of that previous reference [3] from the authors. This area requires further investigation.

Finally: "Overall, the current results show poor to fair agreement between self- and other-reports of well-known proxies, even though the agreement of the overall group was similar to those previously reported for social responsiveness." Again, some potentially important lessons to be learned here insofar as the 'meaning of autism' to a person and those around them. Indeed I note the authors' offer one explanation: "the self may be more accurate about traits that describe unobservable thoughts and feelings due to privileged access (e.g. feelings of empathy and sensory sensitivity), whereas an informant would be more accurate about observable behavior (e.g., ASD traits)." Again, jumping back into 'hot potato' territory, such a finding may have implications for the whole 'self-diagnosed vs. formal diagnosis' debate that still continues at a pace (see here).

There is quite a bit more to take in from this latest paper from Lever & Geurts and definitely some food for thought. I'll leave you however with one final quote from the authors that is perhaps one of the more important take-away points from their study: "it is important to repeatedly assess self-reported ASD characteristics during adulthood." Who would argue with that in terms of getting wants, wishes, needs and requirements accurate and up-to-date?

----------

[1] Lever AG. & Geurts HM. Is Older Age Associated with Higher Self- and Other-Rated ASD Characteristics? Journal of Autism & Developmental Disorders. 2018. Jan 18.

[2] Happé FG. et al. Demographic and Cognitive Profile of Individuals Seeking a Diagnosis of Autism Spectrum Disorder in Adulthood. J Autism Dev Disord. 2016 Nov;46(11):3469-3480.

[3] Geurts HM. et al. Autism Characteristics in Older Adults with Depressive Disorders. The American Journal of Geriatric Psychiatry. 2016; 24: 164-169.

[4] Øien RA. et al. Sex-Differences in Children Referred for Assessment: An Exploratory Analysis of the Autism Mental Status Exam (AMSE). J Autism Dev Disord. 2018. Feb 8.

----------

Autism and the 'female camouflage effect'

Two papers provide some blogging fodder today. The first is from Agnieszka Rynkiewicz and colleagues [1] (open-access available here) who introduces a concept that many people with an interest in autism might have considered: a 'female camouflage effect' in autism. The second paper is by C Ellie Wilson and colleagues [2] and continues the idea that sex/gender differences present in autism might have some important implications for diagnostic evaluation.

Both these papers entertain the idea that although the history of autism has been very 'male-dominated' in light of the sex ratio for the label for example (see here), science and clinical practice is coming round to the idea that there may be important sex/gender differences in the presentation of core issues (see here) that might mean quite a few females have been 'over-looked' or 'under-identified' when it comes to autism. It's of increasing interest.

Rynkiewicz et al set about to present "an innovative computerized technique to objectively evaluate the non-verbal modality of communication (gestures) during two demonstration tasks of ADOS-2 (Autism Diagnostic Observation Schedule, Second Edition)." ADOS, by the way, is one of the gold-standard assessment instruments designed with autism in mind (see here). As part of their really interesting report on "automated coding of non-verbal mode of communication (gestures)" and computation of a "Gesture Index (GI)" they compared results from boys and girls ("high-functioning" as they describe them) and whether "females with autism had a higher GI compared to males with autism."

After data were crunched and the like, the authors concluded that: "High-functioning females with autism might present better on non-verbal (gestures) mode of communication than boys with autism." Further: "This may be because they are effective at camouflaging other diagnostic features." Potentially very important information indeed.

The second paper by Wilson et al continues the theme that sex/gender might have a part to play on the manifestation of autism and onwards diagnostic evaluation. With the aim of reporting "sex differences in clinical outcomes for 1244 adults (935 males and 309 females) referred for autism spectrum disorder assessment", researchers found among other things that: "Males had significantly more repetitive behaviours/restricted interests than females." Onwards: "The sexes may present with different manifestations of the autism spectrum disorder phenotype and differences vary by diagnostic subtype."

Variation in the presentation of repetitive and/or restricted behaviours/interests by sex has been talked about in the peer-reviewed research before [3]. As part of a wider suggestion of a specific female phenotype of autism potentially emerging (see here) (being careful with those generalisations) I'd like to think that a greater awareness of such issues could notably enhance the whole diagnostic system especially for females. Indeed, one area that I think would substantially benefit from a little more inspection of a potential 'female camouflage effect' in autism is that of the 'diagnosis' (although not formally noted in standardised texts) of pathological demand avoidance (PDA). At least one person has talked about female autism and the overlap with PDA (see here) and I'm minded to suggest that they might be on to something rather important.

Oh, and since I'm on the topic of sex/gender and autism, I'm minded to bring in the paper by Katarzyna Chawarska and colleagues [4] talking about infant 'at-risk for autism' girls and social attention potentially pertinent to discussions on any female camouflage effect...

----------

[1] Rynkiewicz A. et al. An investigation of the 'female camouflage effect' in autism using a computerized ADOS-2 and a test of sex/gender differences. Mol Autism. 2016 Jan 21;7:10.

[2] Wilson CE. et al. Does sex influence the diagnostic evaluation of autism spectrum disorder in adults? Autism. 2016. Jan 22.

[3] Van Wijngaarden-Cremers PJ. et al. Gender and age differences in the core triad of impairments in autism spectrum disorders: a systematic review and meta-analysis. J Autism Dev Disord. 2014 Mar;44(3):627-35.

[4] Chawarska K. et al. Enhanced Social Attention in Female Infant Siblings at Risk for Autism. Journal of the American Academy of Child & Adolescent Psychiatry. 2015. Dec 17.

----------

Rynkiewicz, A., Schuller, B., Marchi, E., Piana, S., Camurri, A., Lassalle, A., & Baron-Cohen, S. (2016). An investigation of the ‘female camouflage effect’ in autism using a computerized ADOS-2 and a test of sex/gender differences Molecular Autism, 7 (1) DOI: 10.1186/s13229-016-0073-0




Wilson CE, Murphy CM, McAlonan G, Robertson DM, Spain D, Hayward H, Woodhouse E, Deeley PQ, Gillan N, Ohlsen JC, Zinkstok J, Stoencheva V, Faulkner J, Yildiran H, Bell V, Hammond N, Craig MC, & Murphy DG (2016). Does sex influence the diagnostic evaluation of autism spectrum disorder in adults? Autism : the international journal of research and practice PMID: 26802113