By Austin J. Gallyer
Suicide is an international public health crisis. In the United States (U.S.), the suicide rate increased by 33% from 1999 to 2017 , and in Scotland, the suicide rate increased by about 15% from 2017 to 2018 . Because of this large public health burden, scientists have conducted research to improve the prediction and prevention of suicidal behavior. Unfortunately, a recent meta-analysis revealed that even the strongest predictors of suicidal behavior—including past suicidal behavior—are relatively ineffective at determining who will die by suicide . Moreover, there is mixed evidence whether prevention efforts, including psychotherapy and pharmacological treatments, are effective at reducing the number of individuals who die by suicide [4, 5].
Since Franklin and colleague’s meta-analysis, several solutions have been forwarded to address the field of suicidology’s lack of progress after over fifty years of research. These solutions include, but are not limited to, the use of machine learning , using ecological momentary assessment , applying ideation-to-action theories of suicide , and focusing on means safety methods, such as through firearm access . While these and other solutions have the potential to increase the community’s knowledge to decrease the number of suicides and provide relief to those suffering from suicidal thoughts, many, if not all, of these lines of inquiry are conducted by small independent groups of researchers.
This is troublesome for reasons that are largely due to the fact that suicide is a low base-rate phenomenon . Because suicide is a low base-rate behavior, due to the limited resources available to even the most established suicide researchers, any study conducted by a small group of researchers will likely have less than ideal statistical power, or have relatively poor generalizability to the population (that is, aside from large epidemiological studies, which provide descriptive information). Low statistical power increases the probability of both type I and type II errors, and poor generalizability implicates that our findings may not apply or replicate in other populations. Thus, what is a poor prediction crisis in suicidology, may be a replication crisis in suicidology.
A potential solution
So far, suicidology, and clinical psychology more broadly, has been relatively absent from the conversation about the replication crisis and the open science movement (though, this is starting to change) [11, 12]. This is not necessarily because the studies in clinical psychology (or suicidology) are more methodologically sound than those in areas like social psychology. Indeed, studies from a special issue in the Journal of Abnormal Psychology have shown that statistical power in clinical psychology is limited to detect many effects of interest to clinical psychologists, and that our evidence-based treatments may not be as well-supported as previously thought [13, 14]. Therefore, solutions to these issues proposed and tested in other areas of psychology may be effective at addressing the issues present in suicidology.
One solution that has emerged in social psychology is the Psychological Science Accelerator (PSA) . The PSA is a crowdsourced, large-scale network of labs around the world that investigates both novel research questions and attempts to replicate previous research in a large, diverse sample. A similar organization in suicidology would have the strong potential to conduct studies that are not possible for any one group of researchers to employ. Specifically, though not necessarily the approach such a PSA should take, a PSA in suicidology would have the ability to conduct rigorous, large scale experimental studies in diverse samples. For example, a PSA in suicidology could use a translational approach to examine causes of suicidal behavior by using a virtual reality self-injury paradigm  to directly test competing theories of suicidal behavior. Using a similar approach, a PSA in suicidology could also experimentally investigate which media guidelines for talking about suicide or showing images of self-harm, if any at all, are associated with increased risk for engaging in proxy suicidal behavior. A PSA in suicidology could also follow-up on these studies using more ecologically valid approaches. A PSA in suicidology could also conduct exploratory or qualitative research to inform research questions. For example, a recent study found that many youth do not find self-harm minimization strategies to be effective . This work could inform a study by a PSA in suicidology that investigates which self-harm minimization strategies are effective, or, more broadly, inquiring whether youth from a large, diverse sample also find self-harm minimization strategies to be ineffective.
A PSA in suicidology would not mean that all suicide researchers must use all of their time to study the same topic. On the contrary, it is important that individual scientists can think of and test their own ideas. However, some time and resources from each lab could be devoted to a study from the PSA in suicidology. To an extent, this principle is already being practiced. For example, the Military Suicide Research Consortium (MSRC), a U.S. based organization funded by the U.S. Department of Defense, requires studies funded by the MSRC to include the Common Data Elements, which are a set of suicide-relevant self-report questionnaires. This data is then merged across studies to create a large database of participants that can then be analyzed to answer questions with a larger, more diverse sample. This requires some sacrifice on the investigator’s part, since including the Common Data Elements may require the investigator to exclude other measures to limit participant burden. A PSA in suicidology would take this principle a step further, with participating labs devoting resources to contribute to a completely separate study.
Where do we go from here?
There are existing organizations that have the potential to become a PSA of suicidology, such as the MSRC and the Youth Suicide Research Consortium. However, the key to a PSA of suicidology will be to recruit from multiple sites around the world, and devoted lab time and resources among all consortium members to at least one study. Based on these—albeit vaguely defined—terms, a PSA of suicidology does not yet exist. Whether it be through one of these organizations, others that I haven’t mentioned, or a new organization entirely, I think a discussion surrounding the generalizability, replicability, and impact of the findings of our field needs to occur. I have offered one solution that I think would, at least in part, address these issues. The International Summit on Suicide Research, a premier international conference of suicide research, occurs in October. This may be a great venue to discuss these issues, the solution offered here, and other solutions the field may have. I hope that suicide researchers, those more established or those early in their career, will lead the way to discuss developing large collaborations with one another to move the field forward. Due to current incentive structures for university professors to conduct independent research to obtain tenure, funding mechanisms, and the fact that it is not the current norm, a PSA in suicidology will not be easy to develop. However, by increasing the generalizability, replicability, and impact of our research investigating suicidal thoughts and behaviors, such an organization has the potential to save lives.