Researchers have been studying suicide for a long time, for obvious reasons: It’s a tragic, universal human phenomenon that seems to be intimately linked to certain feelings of hopelessness and despair. Given all this research, one would think there had been a steady forward march of progress on understanding which factors most put people at risk for suicidal thoughts and behaviors, or STBs, as researchers call them. Unfortunately, that’s not the case at all. To a distressing extent, suicide researchers are still feeling around in the dark.
That’s the key takeaway from a big new meta-analysis in Psychological Bulletin. A team led by Joseph C. Franklin, a psychology researcher at Florida State, examined 50 years of suicide research to try to determine whether and to what extent researchers have homed in on risk factors that usefully predict STBs. And the short answer is that they just haven’t. Despite all those decades of research, and all those studies, we simply aren’t yet at a point where we can meaningfully predict who is at the greatest risk of STBs.
Now, that doesn’t mean researchers haven’t been able to draw some conclusions. According to Franklin and his colleagues, some factors are, in fact, correlated with significantly increased rates of STBs: If you have prior STBs, you are about 2.4 times more likely than a general member of the population, all else being equal, to think about or attempt suicide. If you have a family history of STBs, you’re about 1.5 times more likely. Generally speaking, the authors concluded, researchers do ask the right sorts of questions when conducting suicide-risk assessments.
These numbers are less useful than they sound in a practical context, though, simply because of how probability works. A tragic number of people commits suicide each year — as the authors point out, suicide accounts for an “estimated one million annual deaths across the globe,” which is “more annual deaths than homicide, AIDS, car accidents, and war.” On top of that, there are 25 million suicide attempts. So from a public-health perspective, this is a major cause of death.
But the average likelihood of any individual committing suicide is very low. As Franklin and his colleagues point out, in 2013 “0.013 of every 100 people in the United States died by suicide.” Even a risk factor that tripled the odds of death by suicide within a year — and the authors didn’t find any that even reached this level — would only raise this number to “0.039 per 100 people,” meaning an individual would still have a “near-zero” risk of suicide. When the base probability of the event is so low, risk factors that multiply it by three or four or five don’t get you all that far, in terms of predictive capabilities. In absolute terms, a 0.039-per-100-people number “is still a near-zero risk of dying by suicide that year.”
It isn’t all bad news — it just turns out that suicide is really, really complicated and may require some fancy math and technology to better unpack and predict:
Consistent with recent research (Kessler et al., 2015) and many hypotheses and theories about suicide, accurate STB prediction will likely require a complex combination of a large number of factors (i.e., 50), many of which are timevarying. Moreover, it is likely that there are many different paths to STBs. [A] single one-size-fits-all algorithm for STB prediction is unlikely; researchers will likely need to develop separate algorithms different populations (e.g., adolescents; the elderly; inpatients; soldiers/veterans; prisoners; mood disorder patients; psychotic disorder patients; etc.).
In terms of less complicated tweaks to how suicide resarch is done, Franklin and his colleagues recommend more of a focus on short-term outcomes, rather than long-term ones. If researchers conducted more studies examining which factors are correlated with STBs in the immediate future, rather than over the course of (for example) the following year, different, more clinically useful patterns might emerge. So while suicide researchers have a ways to go, there are at least some signposts pointing them in the right directions.