Prevalence of autism in preterm infants meta-analysed

"The prevalence of ASD [autism spectrum disorder] is significantly high in the preterm population. Adequate resources are needed to improve the outcomes of these children."

So said the findings reported by Sachin Agrawal and colleagues [1] who present the results of yet another meta-analysis with autism in mind; yet again on a topic that has filled quite a few peer-reviewed science column inches (see here and see here for examples).

'Preterm' typically refers to babies who are born alive but before completing the usual 37 weeks of gestation. It is further sub-categorised into extremely (less than 28 weeks), very (born between 28 and 32 weeks gestation) and moderate-to-late (born between 32 and 37 weeks gestation) preterm. The risk of a wide variety of 'adverse' outcomes seems to be significantly heightened following preterm birth, covering both physical (somatic) and psychological/developmental domains. Outside of the very final outcome that is early mortality, longer-term issues, particularly associated with brain development occurring outside of the womb, have been a feature of several results. Autism has also been mentioned in this context (see here).

Agrawal et al set about analysing the data pertinent to autism risk and preterm birth. They talk about the scenario of a 'perfect storm' whereby "whatever initiates the preterm birth process might also initiate abnormal pathways of brain development" as being potentially pertinent to autism. Covering the research literature up to May 2017, some 18 studies including over 3300 preterm infants were included for analysis. They specifically included studies that utilised known diagnostic tests for autism or ASD (ADOS, ADI, DAWBA) rather than those that used only "ASD screening tools" and looked up the cumulative prevalence rate for autism.

Results: 7%. That was the overall autism prevalence rate among preterm infants included in the various studies analysed. There was quite a bit of variation across the studies and the prevalence figure fluctuated to some degree as and when children with disabilities were included or not, but the 7% figure seemed to be an accurate one. Authors also put the 7% figure into some 'real world' perspective too: "This equates to ∼900 000 additional children each year who will develop ASD given that globally ∼15 million infants are born preterm (before 37 weeks’ gestation), of whom 13 million survive." Kinda takes your breath away doesn't it?

A couple of other important details are worthwhile mentioning too. So: "Our meta-regression analysis revealed no significant association between gestational age, birth weight, and prevalence of ASD in preterm infants." This doesn't totally rule out such factors as exerting an effect, but...

What else to say? Well preferential screening for autism or ASD in cases of preterm birth could be indicated. The allocation of further research resources into how and why preterm birth occurs could also be useful, focusing on how and why the infant brain seems to be so sensitive to maturation outside of mum's body. And then in relation to all those 'we don't know what causes autism' sentiments that continue to be expressed, well, add prematurity to the growing list of strong possibilities (see here and see here for other examples).

Oh and minus any medical or clinical advice being given or intended, it appears that nutrition, particularly omega-3 fatty acid levels, *might* play 'some' role in some preterm births [2] with some intriguing possibilities for supplementation (minus sweeping generalisations)...

----------

[1] Agrawal S. et al. Prevalence of Autism Spectrum Disorder in Preterm Infants: A Meta-analysis. Pediatrics. 2018 Aug 3. pii: e20180134.

[2] Olsen SF. et al. Plasma Concentrations of Long Chain N-3 Fatty Acids in Early and Mid-Pregnancy and Risk of Early Preterm Birth. EBioMedicine. 2018 Aug 2. pii: S2352-3964(18)30252-4.

----------

1 in 5 children "met criteria for low language at 7 years"

Although primarily looking at the potential predictors of language outcome, the study results published by Cristina McKean and colleagues [1] revealed the rather important title heading this blog entry: "Almost 19% of children (22/1204;18.9%) met criteria for low language at 7 years."

The source of the finding was a cohort of some 1900 infants "recruited at age 8 to 10 months" who were followed until aged 7 years old and subject to quite a bit of research inspection looking at "early life factors", maternal factors and "child language ability" at various points through the childhood years. I believe this was part of the The Early Language in Victoria Study (ELVS) initiative; something that has previously created a bit of stir in speech and language circles. The authors reported that alongside the quite high percentage of children who met 'low language' criteria (based on standardized receptive or expressive language scores "≥1.25 SD below the mean"): "Child language ability at 4 years more accurately predicted low language at 7 than a range of early child, family, and environmental factors." Said low language abilities at 7 years old were also "associated with a higher prevalence of co-occurring difficulties."

Aside from pointing out that language ability at 4 years old might be quite important to language ability at 7 years old, the question that should be in most people's minds is 'why?' Why are nearly 1 in 5 children presenting with low language ability at 7 years old (and presumably at 4 years old too)? Yes, there are variables such as adverse early life factors (prematurity, birth weight, coming from a non-English speaking background, etc) that will no doubt influence various aspects of language ability, but the authors note that such factors only account for roughly 15% at most of the variation in language scores seen in their cohort (not including "child language scores at ages 2 and 4"). Ergo, there are other factors involved with regards to these findings.

In light of the McKean findings, I'm also going to draw your attention back to another occasion when language ability has been discussed on this blog (see here) and specifically: "At school entry, approximately two children in every class of 30 pupils will experience language disorder severe enough to hinder academic progress." [2] Low language (ability) is not necessarily the same as a diagnosed language disorder, and probably accounts for the variation between the studies (1 in 5 vs. 1 in 15). But in amongst a spectrum of language ability (disorder?) the questions about 'why?' still very much remain (and please, no sweeping generalisations about us 'just being better at diagnosing').

----------

[1] McKean C. et al. Language Outcomes at 7 Years: Early Predictors and Co-Occurring Difficulties. Pediatrics. 2017 Feb 8. pii: e20161684.

[2] Norbury CF. et al. The impact of nonverbal ability on prevalence and clinical presentation of language disorder: evidence from a population study. Journal of Child Psychology and Psychiatry. 2016. May 16.

----------

McKean C, Reilly S, Bavin EL, Bretherton L, Cini E, Conway L, Cook F, Eadie P, Prior M, Wake M, & Mensah F (2017). Language Outcomes at 7 Years: Early Predictors and Co-Occurring Difficulties. Pediatrics PMID: 28179482

Gestational age at delivery and special educational need

An article by Boyle and colleagues* (full-text) published in the BMJ provided some food for thought recently with the conclusion that even babies born at what would be considered term (37-38 weeks) may suffer from poorer health than those born at 39 - 41 weeks. To me this seems like a very narrow window when it comes to something like birth timing and future health outcomes bearing in mind what risk means and the myriad of potential confounding variables.

Autism is not specifically mentioned in the Boyle article but nevertheless there is some opinion on a possible relationship between gestational time and associated risk. Indeed, I don't know about you but to me it seems like almost every week a new study is published suggesting that adverse conditions at birth might place a person at elevated risk of developing an autism spectrum condition. OK perhaps I over-exaggerate with the 'every week' bit but certainly this is a recurring theme. As if to prove a point take a look at this paper recently published by Movsas and Paneth** on gestational age and symptom severity in autism.

There are other examples suggesting that low birth weight increases the risk, breech presentation, planned caesarean section, hyperbilirubinema... the list goes on. Indeed so numerous are the factors, studies and accompanying media reports that I very often don't even read them in their entirety anymore given the wealth of evidence that is being built up around our entrance into the world.

On the one hand, in a world full of questions about autism - its aetiologies (plural) and natures - such reports have contributed to a valuable area of research with regards to risk and also potentially mechanisms pertinent/contributory to at least some cases. With a condition which has by far raised more questions than science has currently answered, these kinds of 'pointers' might turn out to be important ones.

On the other hand, there does seem to be a tendency to present this relationship slightly out of context; in that autism receives all the attention in relation to something like premature/pre-term babies yet at the expense of a much larger relationship between something like birth dates and special educational needs (SEN) as a whole.

To illustrate this point, my attention turned to quite a powerful study published a few years back by MacKay and colleagues** (full-text) which looked at the rate of SEN based on population registry data where detailed birth data was also present. I should at this point describe a few details about how things worked.

The study was conducted in Scotland which, at the moment, is still part of the United Kingdom (UK). For those unfamiliar with UK healthcare system, we have the National Health Service (NHS) which is paid for through direct taxation and currently means that all citizens, irrespective of income or status, have a right to free healthcare at the point of need. From cradle to grave, each person has a unique NHS number, which alongside some quite copious amounts of note-taking, provides an impressive bank of health information about a person. Health services often overlap with other areas such as social and educational services (most of the time) which can form quite a detailed picture of a citizen as evidence by the Child Health Profiles reports for example which are accessible to all and provide a wealth of information. Having experienced first-hand how much information is gathered at birth and the early years, I can vouch for this being a pretty good system.

In addition, I quote from the study: "Under the Special Educational Needs and Disability Act of 2001, both schools and local education authorities in the United Kingdom have a statutory duty to identify, assess, and make provision for children with SEN". SEN covers quite a lot of diagnostic ground including learning disability, autism spectrum conditions, ADHD, dyslexia and dyspraxia.

So with these details in mind, the study:

• Based on a large school-aged population covering 19 Scottish local authority areas, details of 514,188 children above 4 years and below 19 years of age were included for study.
• School census data showing SEN status was linked to the Scottish Morbidity Record (SM2) via birth certificate data.
• Complete data was available for 362,688 children of whom 17,784 (4.9%) had a record of SEN.
• In amongst the large amount of results obtained, a few key points were noted including: low birth weight (<2500 g) was associated with an increased risk of subsequent SEN (unadjusted odds ratio [OR] 2.22). This finding was linked to the primary finding that preterm delivery also correlated with risk of SEN, with extreme prematurity (24-27 weeks) carrying the greatest risk (OR = 6.92) steadily declining as children were born closer to their due date but then increasing for infants born overtime after their due date (42 weeks).
• Importantly, whilst preterm births (<37 weeks) increased the risk of SEN, only a relatively small proportion of SEN was linked to prematurity (5.3%) compared with SEN with a history of 39 weeks gestation (1.7%). 
• As per the description of SEN, autism as a diagnosis is included. Having said that, no specific data is provided on the distribution of particular diagnoses in the presented dataset so we can't readily ascertain any specific relationship between autism diagnoses and preterm birth from the current paper.

If you have a bit of time free and are interested in this kind of study, I would encourage you to have a look through the MacKay article in its entirety. For me, the important point to take from this study is that autism, some cases of autism, may very well be linked to preterm birth and onward to factors such as birth weight but such a relationship does not appear to be exclusive to autism and might not necessarily take into account other SEN diagnoses working as comorbidities. Perhaps a smaller detail is that with the current lack of a national autism register here in the UK, SEN status with a focus on autism might be a rough-and-ready way to provide a figure on autism rates at least in the UK pediatric population?

To finish, my car journeys are currently filled with the sound of the Beautiful South and the line - She's a PhD in “I told you so,” you've a knighthood in “I'm not listening” as featured in the song Don't marry her (the clean version).

* Boyle EM. et al. Effects of gestational age at birth on health outcomes at 3 and 5 years of age: population based cohort study. BMJ. March 2012
DOI: 10.1136/bmj.e896

** Movsas TZ. & Paneth N. The effect of gestational age on symptom severity in children with autism spectrum disorder. JADD. March 2012.

*** MacKay DF. et al. Gestational Age at Delivery and Special Educational Need: Retrospective Cohort Study of 407,503 Schoolchildren. PLoS ONE. 2010; 7: e1000289.
DOI: 10.1371/journal.pmed.1000289