They've finally arrived. The results of the English Adult Psychiatric Morbidity Survey 2014 have been published by NHS Digital (yes, our Nation's healthcare services has a digital arm) and when it comes to autism (adult autism 18 years+), some rather peculiar statistics have been produced.
OK, for those who want/need a quick heads-up on all-things Adult Psychiatric Morbidity Survey (APMS), I'll refer you to a previous post I wrote covering this prevalence survey with autism in mind (see here). APMS provides estimates of the numbers of various mental health diagnoses among adults living in private households in England.
The 2014 data report some key facts, not least that: "One in three adults aged 16-74 (37 per cent) with conditions such as anxiety or depression, surveyed in England, were accessing mental health treatment, in 2014." This figure is an increase on the 2007 APMS data (24%). There are also some other important data derived from the 2014 survey too with regards to sex differences in relation to "common mental disorder (CMD)" in diagnosis and in symptoms. Lessons need to be learned.
It is however with autism in mind (see here for the section covering autism), that I'm concentrating on in this post and the observation that: "The estimated prevalence of autism in 2014, using the threshold of a score of 10 on the ADOS [autism diagnostic observation schedule] to indicate a positive case, was 0.7% of the adult population in England (equivalent to a rate of 7 per thousand). The estimated prevalence of autism in the 2007 data (1.0%) was similar to the 2014 estimate; with largely overlapping confidence intervals." A separate 'additional notes' document accompanies the APMS 2014 autism findings (see here).
1% in 2007 and 0.7% in 2014? Accepting that when it comes to prevalence estimates there is always a degree of 'error' expected (as per the comment on 'overlapping confidence intervals') I'm a little bit puzzled by this latest data and the idea that the figures are described as 'similar'. Puzzled because as well as suggesting that adult autism prevalence estimates might have actually dipped between the years, the authors note that their search of 3 quite populous areas of England ("Leicestershire, Lambeth and Sheffield") across both the 2007 and 2014 data combined only found "31 participants identified with autism."
So what's going on with the APMS and autism?
A question indeed and I assume that the 'combining' of the 2007 and 2014 datasets reveals quite a bit more than just the very small number of participants identified in the studies. I have to say that my brow is furrowing a little at the sight of the Autism Spectrum Quotient (AQ) as retaining an 'autism screener' role in the APMS 2014. If I've learned anything about the AQ in recent times it is that whilst measuring something, it may not be a particularly great exclusive screen for autism (see here). Even the authors attached to the APMS 2014 autism data have said so : "The AQ-20 was only a weak predictor of ADOS-4 cases." I've also mentioned about the ADOS module situation and the whys and wherefores with APMS in mind in that previous post on the topic (see here again).
So we're left with a quandary. The much heralded '1% of adults may have autism' statistic is replaced by a lower value (with appropriate caveats on confidence intervals) of 0.8% when the APMS 2007 and 2014 data are combined. Is this a true reflection of adult autism in England in recent times? How does this tally with the suggestion that child and adolescent rates of autism are on the increase as per that seen in other parts of the UK (see here)?
Or, are the processes pertinent to estimating adult autism used by the APMS not really cutting the statistical/methodological mustard?
Which one is it?
 Brugha TS. et al. Validating two survey methods for identifying cases of autism spectrum disorder among adults in the community. Psychol Med. 2012 Mar;42(3):647-56.
Brugha TS, McManus S, Smith J, Scott FJ, Meltzer H, Purdon S, Berney T, Tantam D, Robinson J, Radley J, & Bankart J (2012). Validating two survey methods for identifying cases of autism spectrum disorder among adults in the community. Psychological medicine, 42 (3), 647-56 PMID: 21798110