by Michael D. Anestis, M.S.
They've done it again...Matt Nock and his colleagues have designed, written-up, and published a study that leaves me feeling like a scientific minor leaguer. This morning, a colleague of mine - April Smith - forwarded along this document, which I immediately recognized as the type of information that should be passed along to PBB readers and hopefully many others who do not fall within that umbrella.
The study, published in Psychological Science, takes a look at implicit thoughts about death and their ability to predict both past and future suicidal behavior. Nock and his colleagues (2010) gave a thorough review of the literature on suicidal behavior and noted that, because most studies rely upon self-report, they are burdened by a fairly substantial weakness: individuals are not always aware of their own suicidal thoughts and, when they are, they are not always honest about reporting them. They noted that this is evidenced by the fact that one study found that 78% of patients who died by suicide deny experiencing suicidal thoughts during their last verbal communication prior to taking their own lives (Busch, Fawcett, & Jacobs, 2003) and that suicide risk is amplified in the days immediately following release from inpatient hospitalization, which theoretically must have been precipitated by the patient indicating that he or she did not have any suicidal desire or intent (Qin & Nordentoft, 2005).
This is not to say that reports of suicidal ideation are meaningless. As much of the research we have discussed on the site notes, when an individual reports thoughts of suicide, this is very meaningful information. The point instead is simply that reporting a lack of suicidal ideation does not necessarily mean that the thoughts are not really there and the individual is not at risk. To adjust for this type of situation, researchers have developed what are termed "implicit association tasks." These tasks present participants with a number of stimuli, sometimes images, sometimes words and ask them to press a particular key indicating the category in which the stimulus fits. In this particular task, participants were presented with words on a computer screen representing "death" (e.g., die, dead, suicide, lifeless), "life" (e.g., alive, survive, thrive, breathing), "me" (e.g., I, myself, mine, self), and "not me" (e.g., they, them, their, other). During one part of the task, participants were told to hit one key if the word presented represented "life" or "me" and another key if the word represented "death" or "not me." During the other part of the task, participants were told to hit one key if the word represented "death" or "me" and another key if the word represented "life" or "not me." The order of those two sets of instructions was randomized across participants. When participants responded more quickly to the "death/me" blocks relative to the "life/me" blocks, they were said to have displayed a stronger implicit association between death and self. The idea here is that the individual is better able to quickly recognize, categorize, and respond to stimuli that fit well together in their mind.
The participants in this study were 157 adults who had presented to a psychiatric emergency department in a northeastern hospital. 43 of those individuals reported having attempted suicide within the past week. Once the patients had been stabilized and given consent to participate, they were administered the implicit associations task, as well as a number of other questionnaires. These included a measure of past suicidal behavior and non-suicidal self-injury, the participants' and clinician's predictions regarding the likelihood that the individual will attempt suicide within the next six months, and the presence of depressive symptoms. Participants were approached again six months later, at which point they were assessed for suicidal behavior and non-suicidal self-injury since the initial experiment time and hospital records were checked to determine if the participant had been admitted due to suicidal behavior.
The authors found a number of interesting results. First, individuals who had been admitted to the hospital due to a suicide attempt demonstrated a stronger implicit association between self and death. In other words, it is not simply that they reported having such thoughts, but rather that they demonstrated this tendency in a task that moves too quickly for them to be able to manipulate their performance. Taking this even further, this finding was specific to suicidal behavior. In other words, individuals who engaged in self-injurious behavior without the intent to die did not show a stronger implicit association between self and death than did individuals admitted to the hospital for other psychiatric emergencies.
As impressive as these findings are, they do not quite compare to the follow-up results. 14 individuals in the study attempted suicide in the six months between the initial experiment time and the follow-up appointment. As it turns out, individuals who demonstrated a positive association between death and self (they responded more quickly in that condition than in the condition in which words representing life and self were paired together) also demonstrated a six-fold increase in risk for suicide attempts in the six months immediately following the experiment. The implicit measure made this prediction above and beyond the effects of other well-known predictors of suicide risk, including depressive disorders and previous suicidal behavior. Participants' predictions about the likelihood of making a future attempt significantly predicted future suicidal behavior but was not a stronger predictor than the implicit measure.
The final finding I wanted to discuss was the prediction of future suicidal behavior by clinicians. In this particular study, clinicians predictions were not significantly related to outcome. Clinicians were asked to base their predictions on clinical intuition and all that they knew about the patients. It is not entirely clear how each clinician weighted intuition versus data (e.g., number of previous attempts, presence of plans and preparation, access to means), but the non-significant finding is an important parallel to the work of Dawes, Faust, and Meehl (1989) who demonstrated so clearly that clinical judgment simply does not outperform actuarial methods. Given the importance of being able to predict and thus prevent suicidal behavior, this is a particularly important consideration with respect to how we as clinicians conduct proper risk assessments. Fortunately, we do have empirically tested methods for assessing risk, so hopefully findings like this will encourage clinicians to consider utilizing actuarial methods in an effort to mitigate risk.
************
If you would like to learn more about suicide or other topics discussed on PBB, we recommend that you consult our online store of scientifically-based psychological resources.
Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University and an incoming resident at the University of Mississippi Medical Center
Subscribe in a reader
