Stat counter


View My Stats

Saturday, January 15, 2011

Responses to spectacularly bad outliers

I like the Numbers Guy who writes a column in the WSJ (Carl Bialik). His piece today in on the perils of trying to count and define mental illness. This piece was spurred by the groundswell of support for the notion that gun purchases should be prohibited for those with "mental Illness". That is all well and good but you still need to define what mental illness is and also define thresholds for when to invoke specific prohibitions. This it turns out is not so easy and in fact, it is almost impossible.

http://online.wsj.com/article/SB10001424052748703959104576081920430619618.html?mod=ITP_pageone_2#articleTabs%3Darticle

This particular has moved front and center because recent shootings by the deranged young man in Arizona. It is yet one more tragic episode in a long line of periodic tragic episodes involving troubled young men. There is something about men of this age who reach crossroads in their lives. Many make really bad choices which can result in individual and personal tragedies. The rare ones make choices which result in high profile and public tragedies.

It is the latter circumstances that create a public cry for some sort of early intervention. In the grip of such public tragedies our gut tells us this call seems reasonable. We need to act before things like this happen. We need to identify who is at risk and act to prevent such acts before they occur. However, when someone like the Numbers Guy starts to raise basic questions about how, our more reflective brains point to the reality that the best responses may be to change little.

The good news is such events are really rare. This is not to suggest that a lower frequency is not desirable. However, we are all too aware of the numerators but not at all cognizant of the denominators. You have to be aware of the latter in order to know how and when to act in terms of prevention. There are about 50 million men between the ages of 15 and 30 in the US. That appears to be the at risk group. While my recollection of history is incomplete and not exhaustive, public high profile events appear to occur every few years (Columbine, VA Tech, University of Texas (both this year and 1966)).  How are we doing by the numbers. Let us assume one public event per year done by one young male. That is one in 50 million. How low a frequency would be viewed as acceptable?

The behaviors exhibited by the perpetrators prior to their actions are always portrayed after the fact as strange, suggesting that someone (or some entity) could have acted ahead of time to prevent such events. The problem is the math. Odd behavior in this age group is not rare. Given the size of the group, even a relatively low frequency of odd behavior translates to big numbers. No matter how you cut it, predicting dangerousness is not a science. There are rules of thumb as opposed to definitive predictors. How many interventions would be required and of what type? Who would or could authorize them? How many simply eccentric people would be raked over the coals? What would constitute success?

The ignorant of the denominator phenomena is not unique to the mental health field. It permeates the entirety of medical practice. Our actions are driven by the outliers. One bad outcome and we feel compelled to propose re-engineering how we approach particular clinical problems. In terms of risk factors, much of the low hanging fruit has been harvested has been dealt with in the US. Unlike much of the developing world, we have few issues with sanitation and water issues, widespread infant mortality, and workplace deaths. Among selected populations, we still have problems with high risk behaviors resulting in morbidity and death (tobacco use, high risk sex, illicit drugs, and alcohol). While medical breakthroughs may mitigate particular consequences of these high risk activities, these behaviors will always be risky. However, cultivating anxiety in these populations does not tend to change their behaviors. Populations likely to benefit from action are generally not inclined to adopt changes

We have moved to address a host of risk factors for people living in relative affluence and spend an increasing amount of our time catering to the anxieties of basically healthy people. Despite their relatively low risk statuses, this population is very responsive to anxiety provoking marketing approaches. It is a great business model since it involves receiving regular payments from patient who are basically well. Well people rarely need unscheduled care and are more often than not gainfully employed. Odds are the problems we work to prevent were not going to happen anyway and are not likely to happen on our watch. However if physicians and their patients were cognizant of the actual numbers involved in these preventative measures, there would be much less enthusiasm. Yes, low dose aspirin in the right populations decreases the numbers of MIs, but only modestly more that the additional bleeding episodes.  Intervene with huge populations and only a handful of people might benefit. However, these interventions are targeted to populations which are more inclined to adopt changes whether they help them or not. They are more interested in the illusion of control than any actual results.

Maybe that is the key. Actual results do not really matter in the short term. It is only the illusion of control or the act of doing something that is essential. That is a sufficient end in itself, or at least until the next tragedy occurs. There is no limit to the variety of useless gestures which are at our disposal. We can only hope that the responses can limited to being simply useless and do not become destructive in and of themselves.

No comments:

Post a Comment