Flawed strategies undermine examine on undiagnosed autism and suicide | Spectrum
Associate teaching professorStanford University
A study published last month in the British Journal of Psychiatry claimed that people who die by suicide have an elevated likelihood of being autistic or having undiagnosed autism, based on an analysis of coroner records and family interviews. Spectrum asked statistics expertKristin Sainani for her thoughts on the methods and results.
A new study purports to show evidence that people with elevated autism traits and undiagnosed autism are at an increased risk of suicide. It’s not surprising that autism may increase the risk of suicidebecause it is known to co-occur with other psychiatric conditions, such as depressionthat also increase this risk. But this study’s claims are unconvincing because of the lack of an appropriate control group.
The researchers, led by Simon Baron Cohen at Cambridge University in the United Kingdom, reviewed coroner records from 372 deaths in England that involved suicide or self-harm. They used data from these records, which include the coroner’s interviews with family members, to identify at least some evidence of autism in 40 of these individuals, or 10.8 percent. They compared this with the 1.1 percent autism prevalence in the UK general population.
But this comparison is misleading, as the two studies measured different things. The earlier study counted individuals as autistic only if they met the full criteria for a clinical diagnosis of autism, even if they were previously undiagnosed. By contrast, the new study flagged anyone who had even “possible” signs of autism.
In fact, of the 40 individuals flagged, only 2 had a documented diagnosis of autism. The remaining 38 had what the researchers called “possible” or “strong evidence” of autism. The researchers didn’t disclose how many were in each of these categories, but it’s fair to guess that most were in the “possible” category.
The researchers suggest that the “possible” category may represent undiagnosed autism. But their own data contradict this idea. They interviewed 29 families of the deceased, including the families of 12 individuals classified as having “possible” autism. But, according to the authors, “none of the 12 met the threshold for autism” using a validated diagnostic instrument.
Further undermining the study’s main argument, the authors found little correlation between the designation of “possible” autism and scores on a validated questionnaire that measures autism traits. This suggests that what researchers flagged as evidence of autism in the coroner reports is not a good proxy for undiagnosed autism or elevated autism traits.
Without a relevant control group, it’s unclear whether 10.8 percent is a high or low number. It’s possible that 10.8 percent of people in general would meet the authors’ criteria for “possible” autism. A good control here would have been to use the same methods to review the death records of people of similar age and gender who died from causes other than suicide. Without such a control, the study does not add much to our understanding of the role of autism in suicide.
Kristin Sainani is associate teaching professor of epidemiology and population health at Stanford University in California.
If you or someone you know is having suicidal thoughts, help is available. Here is a worldwide directory of resources and hotlines that you can call for support.