Autism epidemic? More likely we're just better at diagnosis

From: Ars Technica

In the developed world, rates of autism spectrum diagnoses have skyrocketed in recent years, raising the specter that a new environmental factor has been altering the developmental trajectory of the youngest children. Searches for putative environmental influences, however, have generally come up empty, even as researchers have identified very strong genetic influence on the disorders. The disparate rates of progress provide some support for an alternate interpretation: autism has always been around at roughly this level; we've just gotten much better at diagnosing it.

A study in the UK now provides a bit more support for this argument. Researchers have performed a large-scale survey of adults, looking for indications of autism spectrum disorders where none had ever been diagnosed. And they found cases, enough to suggest that the rates haven't changed considerably.

The report starts out by describing the evidence for a rising tide of autism spectrum disorders (ASDs). Surveys prior to the 1990s had indicated an incidence of about 4.4 cases per 10,000 individuals. By the '90s, that rate had risen to 12.7, and more recent surveys suggest an incidence of 10 per every 1,000, or roughly one percent. This startling rise, if it's real, suggests a recent uptick in an environmental exposure is altering neural development in early childhood.

Based on that suggestion, the British team started with a few simple hypotheses. If the prevalence of ASDs has been increasing, then the youngest adults would have rates similar to the ones seen in the oldest children, with the rates dropping with age. They also hypothesized that any adults they did find would share features with the other ASD individuals, including poor social integration; like the current generation, they'd be more likely to be male.


They then set about an enormous undertaking: diagnosing a representative population of UK adults. As a first pass, they selected a population based on postal codes and employment statistics. With over 14,000 households identified, interviewers fanned out, ultimately completing over 7,000 interviews that provided a first level of screening for ASD diagnoses. From that pool, 850 individuals were chosen for a second, more in-depth interview; 630 people completed that, and the authors adjusted their data to take a failure to respond into account, leaving them with a population that was largely representative of the English population over 16 years of age.

As the authors note, however, this population excludes those who have been institutionalized or have severe disabilities, so it's not fully representative. There was, however, no indication that ASD issues made anyone less likely to participate in the survey.

Based on the scores obtained using the Autism Diagnostic Observation Schedule (a standard diagnostic test), the authors identified 19 adults who qualify as autistic. That's a rate of 9.8 per 1,000, or roughly the same rate as its appearance in children. Unfortunately, because of the small sample size, the 95 percent confidence interval is pretty broad, ranging from three to 17.

The small sample size also makes it difficult to say much about the population as a whole. Males are far more likely to have ASD symptoms, and the group as a whole was less likely to have completed university education or have gotten married, and more likely to be living in government-run housing. Beyond that, none of the measures were statistically significant.

This lack of statistical robustness also keeps the authors from evaluating one of their hypotheses, that apparent rates would drop with age. There is a slight drop of ASD diagnoses in older individuals, but the study is so small that the trend is statistically consistent with rates actually increasing. Nevertheless, the authors note that the trend is in no way consistent with a sudden upswing in ASD prevalence.

Overall, this looks like a very solid preliminary study. From a practical point of view, it suggests that there is a substantial population of economically and educationally disadvantaged people with ASD who are undiagnosed. Better recognition of and support for this group would seem to be a significant public health concern, given its apparent size.

The rate of apparent diagnosis also suggests that autism has been historically under-diagnosed, and may have been present at roughly its current rates for a long time. The small size of the population in the study, however, limits the confidence in these results. Unfortunately, given the effort and expense that were required to generate this small survey—two rounds of in-person interviews—we're not likely to see a larger study that's equally thorough.

Archives of General Psychoatry, 2011. DOI: 10