Friday, August 1, 2014

Musings on Potential Predictors of Suicidal Behavior

I was recently asked for my thoughts about blood tests as a genetic predictor of suicide on Twitter.  I tweeted them out, but I thought I'd quickly organize them here as well.  It admittedly may be very flawed, but here's my logic:

1. Completing suicide is one isolated behavior that results from a complex myriad of genetic, biological, environmental, and situational factors.   As such, simply looking to genetics - the building blocks of life - to predict one isolated behavior seems far fetched.

2.  There's also a lot of noise introduced when your outcome is the one isolated behavior of a suicide attempt resulting in death.  Someone who may be prone to suicide, for instance, may die in an accident before they ever have the change to attempt suicide.   Further, people who may be prone to suicide otherwise may lead a charmed life & the thought may never cross their minds.  Alternatively, someone not genetically prone to suicide may have a terrible life, develop a substance abuse habit, and impulsively kill themselves one night while under the influence.  Because the environment and circumstance play such a large role in whether or not someone actually dies by suicide, looking to genetics intuitively doesn't feel right to me.  Studying epigenetic factors, however, may yield something very interesting, but my money isn't on it.  I don't think it will be the silver bullet we want it - or hype it - to be.  Think back to the human genome project. (P.S.  Psychologists will recognize this argument as part of the "low base rate behavior" issue.)

3.  It is true that suicidal behavior is linked with psychological disorders, but is perhaps more important to focus on this: the behavior of completing suicide is intimately tied to the circumstances in the immediate environment.  These immediate circumstances will have an immediate effect not a a person's genetics per se, but on their physiology.

For instance, on any given day, a woman with moderate depression may be at a certain point in her menstrual cycle, on substances, vulnerable from a myriad of factors (e.g. relationship stress, lack of sleep, health condition, etc.), and feeling particularly impulsive.  Lets just say that we add a life tragedy to this vulnerable mix of factors and she gets admitted to an inpatient unit for suicidal thoughts.  Eventually she reports her suicidal thoughts have remitted, she is discharged, and then our hypothetical woman unfortunately makes a suicide attempt.  Some of the aforementioned factors (e.g. menstrual cycle, etc.) have a specific physiological impact.  If we had measurements taken from the inpatient unit of this woman's estrogen/progesterone levels, oxytocin, waking levels of cortisol, toxicity, etc. it would result in a specific constellation of physiological factors.  Over time, scientists can examine which constellation(s) of physiological indicators most often co-occur with recent (serious) suicide attempts / completions.  Determining & using immediate constellations of physiological factors to predict suicidal behavior in the near future seems to me like it will be much more accurate (i.e. have "incremental predictive validity") than our current tools (e.g. self-report questionnaires).  The general idea is that in the future we may be able to divine what people are actually feeling based on physiological measures rather than self-report.

4.  One last thought about the utility (and sensitivity) of genetic testing for suicide:
Not only do I think that blood tests for genetic factors will not yield reliable predictors of suicide completion, they will likely not be of much practical utility.  Consider this: what course of action must a clinician take with the information that a person is at genetic risk for suicide?  If someone is admitted to a psychiatric inpatient unit because of depression and suicidal thoughts, at some point they report they are not imminently at risk.  At that point, we must then discharge them from the hospital.  We would not be able to detain them longer based on a genetic test.  Correct?  How much longer would we keep someone on an inpatient unit based on a genetic test?  Genetic markers are not what we call sensitive measures - they are not sensitive to change.  They are static.  We would, however, have a more firm basis for detaining someone in the hospital if we had evidence that certain physiological constellations were present in a person & evidence that this type of constellation is associated with imminent risk of suicide.  Over time, we would have an idea about how generally someone would be feeling based on the output of physiological tests. Of course, how accurate or difficult to manipulate these hypothetical physiological constellations may be is unknown.  But if I were a betting woman, I would put my money on the scientists examining circumstantial physiological compositions winning the race of determining the most accurate, useful, & sensitive predictor of suicidal behavior.

In short, a suicide attempt is a behavior, not a disorder.  I'm not so sure that treating suicide attempts like a disorder by examining genetic contributions will produce the best predictors of this behavior.  Exploring physiological factors that represent immediate shifts in a person's internal experience however, intuitively feels more promising to me.

P.S.  This post was in response to this CNN article: link.

No comments: