Abstract
People with autism think about killing themselves and die from suicide at a horrifying rate. Small studies find that 20%–40% of adults with autism have considered killing themselves (Hedley et al., 2017) and 15% report making at least one attempt (Balfe and Tantam, 2010). A large Swedish cohort study found that adults with autism are nine times more likely to die from suicide than other adults (Hirvikoski et al., 2016). Studies of children with autism find that 11% have suicidal ideation and 4%–7% have made suicide attempts (Mayes et al., 2013), and the risk of making a suicide attempt is six times that of other children (Moses, 2017).
This public health concern rightly has gained considerable attention of late. In the United Kingdom, the largest autism research charity, Autistica, put out a compelling report detailing the magnitude of the problem and calling for governmental and industry efforts to reduce suicide among people with autism (Personal Tragedies, Public Crisis, 2017). In the United States, the Interagency Autism Coordinating Committee, which advises the Secretary of Health and Human Services, brought attention to this issue through two sets of presentations at its public meetings. There still are very few studies on this issue; almost all describe the prevalence of suicide, attempts, and ideation. To our knowledge, only one funded study currently is assessing the best way to screen for suicide risk; none addresses the processes that lead to suicide or the best ways to prevent suicide among people with autism (Cassidy and Rodgers, 2017).
As we think about the urgent task of reducing premature mortality in people with autism, it is worth revisiting some passages from the Autistica report that deserve as much public attention as those describing suicide risk. The risk of early mortality from all causes among people with autism is nearly twice that of the general population. Those with autism and no accompanying learning disability die an average of 16 years earlier. Those with autism and intellectual or learning disabilities die an average of 30 years earlier (Hirvikoski et al., 2016). The study by Hirvikoski and colleagues represents the largest and most careful study of mortality and autism to date. They found that 12% of deaths were due to suicide; 22% were due to diseases of the circulatory system, which include heart disease and stroke; and 12% were due to neoplasms, which include cancers. Diseases of the nervous system, which include epilepsy, accounted for 9% of deaths. These numbers vary greatly between individuals with and without intellectual disability. Suicides accounted for 14% of deaths in autistic adults without intellectual disability and 4% of deaths in those with intellectual disability. Death from nervous system diseases accounted for 4% of deaths in those without and 19% of those with intellectual disability.
It is probably worth a small digression at this point to talk about the difference between “risk” and “attributable fraction” in mortality studies. Risk, often quantified as an odds ratio, answers the question of how much more likely one group is to die from a particular cause than another group. Attributable fraction quantifies the proportion of premature mortality relative to the general population that could be eliminated if we stopped all deaths from a particular condition. Odds ratios get a lot of attention. It is startling and begs a call to action when we find that people with autism are nine times more likely than the general population to die from suicide. But if the base rate—that is, the number of suicide deaths in the general population—is low, odds ratios can give an inflated sense of the proportion of premature deaths in autism associated with suicide. Attributable fractions require more data to calculate, but from a public health perspective, represent a more critical number if our ultimate goal is to reduce the total amount of premature mortality.
Four years ago, Croen et al. (2014) published a rigorous, large-scale study of the prevalence of various health conditions in adults with autism. They also found an increased risk of suicide attempts relative to the general population (1.8% vs 3% of the sample), with an odds ratio of 5, meaning that adults with autism were five times as likely to attempt suicide as the general population. Much more common were cardiovascular disease (37% of people with autism were diagnosed with 1; odds ratio of 2.5), epilepsy (12%; odds ratio of 16), and diabetes (7.6%; odds ratio of 2.2). Although the odds ratios associated with cardiovascular disease and diabetes were much smaller than those associated with suicide and epilepsy, they may ultimately be associated with more death than suicide.
By no means am I suggesting that we divert attention or resources from determining the best ways to reduce suicide risk in people with autism, but we must not lose sight of our primary goal, which should be to reduce all premature mortality and increase quality of life among people with autism. It would be a great service if future studies of mortality in autism provided attributable fractions, with the goal of directing public health efforts. Studies to date suggest that there is no single cause or even small set of causes that accounts for all the excess mortality in autism. Until we know exactly where to direct our efforts, improving primary and preventive care for people with autism (Nicolaidis and Raymaker, 2013), and addressing the healthcare needs that are common to all adults but exacerbated in adults with autism, may result in better outcomes than specialty programs that address one risk factor at a time.