Stat counter

View My Stats

Saturday, February 23, 2013

Being held to different standards

The recent spate of reported mass killings perpetrated by maladjusted young men has raised all sorts of issues and various constituencies are clamoring for some sort of action. The action called for and the wisdom of any given intervention hinges on this whole issue of causality. Obviously, guns are the root cause these tragedies; we must ban guns. No, guns do not cause these events; there is a breakdown of morals or the family or desensitization to violence mediated by the overall coarsening of the media. We must address these issues.... and so on. It all hinges on understanding causality and when events are low frequency events, causality becomes almost impossible to dissect.

I saw a patient in the hospital today who was admitted because of a severe adverse drug reaction to a common sulfa based antibiotic. This particular antibiotic is perhaps the most widely prescribed antibiotic in the outpatient setting. Side effects of this type occur very infrequently, but since it is widely prescribed, such rare events happen. In this case no one contests that the reactions are real, even though it is only a vanishing small fraction of people who take this drug who develop this severe and potentially life threatening complication.

There are also drug related events where bad outcomes happen in association with particular medications, and it is hotly contested whether these outcomes are causally related to the medication taken. The way the FDA regulations are applied is that if a novel life threatening event occurs when a patient takes a particular new drug, the FDA is empowered to change the drug labeling. They are within their powers to require inclusion of a block box warning, when the causality can be shown or not. Drugs widely prescribed to sick populations can be virtually guaranteed to get linked to spectacularly bad outcomes and their package inserts will, over time, become littered with all sorts of warnings, whether causally linked or not. Even if large scale studies are completed and call particular links to drug induced outcomes into question, the labels rarely if ever get changed, much to the delight of trial lawyers. Absence of proof is never proof of absence and even the perception of causality can be leveraged into big bucks. In the world of health you do not need proof of harm to act, only plausible perception.

We need also to touch upon the effect of the coarsening of the media  and its possible effects on people undertaking acts of violence. This question has been raised with every form of media at some point in the past; books, film, music, TV. Most recently the age old debate is raging around the effects of computer games. I am of the age where my children were among the first generation of those who had computer games from the earliest points of their lives. I saw the gradual evolution of these games from crude graphics (in space invaders) to the evolution of very graphic, very real appearing creatures and every day life circumstances.

We had our share of children and ultimately teenagers in our basement shooting things up on the big screen. When it started it was the Mario brothers and an odd menagerie of villains. This evolved into more people-like characters in war like games. At that time the on screen entities were more creatures than people and it did not occur to me that our children could be getting conditioned to commit violent acts against actual people. It was a gradual evolution. From what I have previewed online, some games are remarkably life like and the violence committed on screen is gratuitous and aimed at the innocent and defenseless.

It only makes sense that viewing violent acts chronically is not healthy. It has long been appreciated that living in a violent household teaches children to use violence to achieve their ends. Why those immersed in a virtual violent world should be immune to these influences is hard to fathom. It has been suggested that violent videos might facilitate a form of release and prevent violence. Based upon that logic, we should encourage violence in the home so as to prevent violence outside the household. I don't think it works that way.

Much in the same way we in medicine observe low frequency events which appear to be causally linked to particular medication exposures, our present experiences with violent mass killers appears to associated preferentially (if not exclusively) with young men who have played lots of violent first person shooter games. Yes, many young men play violent first person shooter games and do not go on to be mass murderers. However, millions of people take Bactrim and Tegretol and never have any problems with these drugs. Should that lead us to believe that the innocuous experience of the many should convince us of no linkage to the catastrophic effects on the rare few? Just because one million people can take these drugs with no harm does not negate the fact that one in a million will be devastated.

I am sure that the media companies who produce such games will rally around the first amendment for protection. Just because they may be legally protected from state restrictions does not mean their product does not cause harm. This brings to mind the denial of the tobacco companies, who maintained the safety of their product despite similar correlational evidence linking it to harm. The actual epidemiology evidence linking smoking to health harms followed only initial anecdotal observations of harm. The similarities are there. Both exposures required chronic and excessive exposure over long periods of time. Not all who were exposed were harmed.

I have no desire to parley the possibility of negative influence of first person shooter games to argue for some sort of state regulation. That basically never works and the unintended consequences of using blunt legal tools to restrict access targeting only a tiny minority far outweighs whatever good that comes of such attempts. That said, we cannot stay in denial that our brains are heavily influenced by repeated exposure to violent images and virtual world violence. It may not have profound effects on most people who are exposed but, in my opinion, there is enough empiric evidence from early formal studies and recent events to encourage caution, particularly in parents ( Ultimately social stigma and informal restriction from the mainstream may be more effective for control than any legal mandates. However, there must be recognition that these games are not healthy and that is is not unreasonable to have the concern that intense and long term exposure to virtual violence may influence select people to do terrible acts. This is not a popular message from the usual media outlets since there is a huge conflict of interest. They are loathe to draw any attention to this possibility.

I also will not be surprised if the violent video game industry moves into the cross hairs of the trial attorneys. The data showing violent first person shooter games change behavior is only now developing. I would not be at all surprised if grieving families who have lost loved ones to violence facilitated by immersion in violent media. While the mass shootings gain the spotlight, I suspect that there is a much larger pool of violent acts where violent games play a role. We do not have the epidemiological tools to appreciate them yet. It is just a matter of time before the trial bar lays claim to their ill gained earnings. They will be held to a similar standard as those within the health care field.

Sunday, February 17, 2013

Risk Arbitrage

When an asset is valued differently in two different markets, shrewd investors can take advantage of such situations by buying in one market and selling in another and pocketing the spread. Someone will become very wealthy if they can figure out how to exploit similar phenomenon in the realm of other risks.

Earlier this week there were two stories juxtaposed in the WSJ which highlighted how irrational we are when we express fears of specific risks. The stories were on what at first appears to be very disparate topics. The first was on the risks of headers in soccer players. In this study, investigators at the Imperial College in London measured the impact of a regular soccer ball inflated at normal pressures travelling at speeds readily attainable in non-professional play contexts. What they determined was that the impact of an adult soccer ball to the head was roughly equivalent to what would be experienced when being punched in the head. Based upon what is now known about head injuries this sounds like a real issue and a real threat, particularly to children and young adults who play soccer in droves.

The second story was on a meta analysis of studies of bis-phenol A (BPA), which is used an additive in a variety of plastics, particularly items holding food or water. The study was funded by the EPA and results presented at the American Association for the Advancement of Science (AAAS). It included data from 30,000 people in 19 countries, including women and infants. The conclusion of the study are the BPA exposure occurs at levels thousands of times lower than you see any effects in animals.

The stories presented two distinct perspectives on the respective risks. The head injury study is so obvious that it borders proving common sense. Do I expect the the results of this study to have any particular immediate effect on our behavior? No, not likely. I raised my children with a healthy dose of competitive soccer, as well as a host of competitive sports. I was constantly amazed at the acceptable levels of trauma and risks in these realms. Injuries were the norm, not the exception, and even routine occurrences which resulted in the requirement for emergency visits, surgery, general anesthesia, and prolonged convalescence (associated with substantial financial impact) failed to lessen enthusiasm for the competitive athletic culture.

If one does a web search using the terms bisphenol A health risks you get 167,000 hits. If one does a similar search using the search terms soccer headers brain damage you get on 25,000 hits. It is part of a greater trend. Raise the possibility that there some sort of "chemical" within the same zip code as our children and our risk tolerance changes dramatically. 21st century Luddites rail against the risks of dangerous "chemicals" including pesticides, sunscreens, non-stick cookware,personal care products, household cleaners, and virtually any other synthetic enhancement. These chemicals represent the witches of our times and they are viewed as much more dangerous than plain old blunt trauma, real evidence and data not withstanding. Better to just burn the witches as part of the precautionary principle. Better to be safe than sorry.

We are much more concerned about the essentially non-existent threat of BPA than the real risk of brain damage. This is just stupid. Within the financial markets, you can get rich when wagering against dumb investors and it tends to get rid of the dumb investors or at least the particular stupid investment decisions. It is too bad there is no similar arbitrage available to bet against other ill informed and superstitious players.

Sunday, February 3, 2013

Cognitive Errors, Blind Spots and Priorities

I recently completed reading Daniel Kahneman's book "Thinking Fast and Slow". It is a tour de force. Perhaps my only criticism is that despite being well written in a very readable style, it is still very dense because it is so filled with important observations and insights. I had to read and re-read most sections. My copy of the book is dog-eared and marked with mini-Post-it notes.

Toward the end to the book Kahneman highlights a common perceptual error which I think is extraordinarily important and touches upon almost every element of human life and decision making. In short, our ability to understand the impact of low frequency events is fundamentally flawed. I sort of appreciated this prior to reading the book but Kahneman highlighted aspects of this I had not previously taken into consideration. The low frequency events can be either events we can benefit from or be hurt by.

This has huge implications within the realm of medical decision making. In the realm of dealing with well patients where we push interventions to avoid possible bad outcomes in the future, both physicians and patients vastly over estimate their likelihood of benefiting from screening or preventative therapeutic interventions. Simultaneously, we tend to vastly over estimate the likelihood from harm from measures vastly more likely to help us.

The issue is how we perceive possibility of events at the extremes. We are easily convinced to buy lottery tickets despite the fact that no one in their right mind and an understanding of the odds should buy a ticket for anything other than the entertainment value. Similarly, both our recommendations to patients and patient assessments of what they are likely to benefit from or be harmed by are riddled with the same biases. No where in our training are these issues consistently addressed. I find it hard to imagine we can even begin to address shared decision making and real informed consent without recognizing that decision making by patients and providers is driven by these biases.

Are entering into a post-truth age of medicine?

I read an interesting essay by Ronald Bailey at entitled "Do we live in a Post-Truth Era?" (

I have to admit I was initially perplexed by such a title, but I read the essay nonetheless. Bailey touches upon the work of Horst Rittel and Melvin Webber who forty years ago made a very insightful distinction between lame and wicked social problems. This is work which I have previously highlighted almost four years ago. (// Wicked problems often involve high stakes domains such as health care. It does not get much more high stakes than life or death, health or sickness, function or dysfunction. However, because there are no right or wrong solutions to wicked problems, only potentially better or worse approaches (and even those distinctions are murky depending upon the outcomes measured and the time points examined), addressing wicked problems is invariably fraught with contention and disagreements, most of which can never be settled by finding the "truth".

Recent history is replete with examples where people and groups takes sides on particular issues, obtaining funding from like minded foundations (or the government), and hiring their own form of expert mercenaries. These efforts are generally not devoted to find any form of truth but instead are directed to affirm what they already believe. As Bailey notes in his essay:

Progressives who believe that corporations are unfairly denying workers a living wage can point to research by analysts at Institute for Research on Labor and Employment to argue that higher minimum wages do not increase unemployment. Free marketeers can turn to the Employment Policies Institute for evidence that boosting minimum wages increases unemployment among the youthful and poor. The pro-immigrant Migration Policy Institute can report that Washington "spends more on its immigration enforcement agencies than on all its other principal criminal federal law enforcement agencies combined." The Center for Immigration Studies, which favors strict immigration enforcement, can denounce the study as "bogus" and "riddled with false statements, cherry-picked statistics, and inappropriate comparisons." Climatologists at the University of Alabama in Huntsville can assert that the atmosphere "has not warmed noticeably since the major El NiƱo of 1997–98—giving us about a decade and a half of generally stable temperatures." Researchers associated with the Potsdam Institute for Climate Impact Research can report that the warming rate has been "steady" since 1979.

Bailey cites another recent essay "Wicked Polarization" by Michael Shellenberger and Ted Nordhaus, also worth reading. In this piece the authors also draw upon Rittel and Webber's original work noting:
The result, Rittel and Webber suggested, was neither the end of ideology nor the end of expertise but rather the continuation of ideological battles on new, more expert terrain. Criminologists might agree that the crime rate had gone up or down but would disagree over whether crime is caused by poverty, racism, the prohibition of drugs, the weakening of traditional moral values, or too few police officers. Any and all of those arguments can be supported empirically. With crime, as with so many other issues, myriad overlapping influences confound simplistic efforts to define causality.

I was struck by the how applicable their observations are to the world of medicine and health care. Health care is absolutely riddled with wicked problems. No other realm is so populated by experts touting the latest truth de jour. Does mammography, PSA tests, or total body skin exams benefit patients? What about treatment of hyperlipidemia or modest hypertension? What about the usefulness of high imaging, cardiac stents, anti-depressant use, to raise but a few. The answers you receive will depend upon what experts you ask what what biases they have, what end points you look at, and in whom you examine these measures. It also depends upon what patients ultimately want from the industry we designate at "health care", which is increasingly integrated into every other industry and service which address human needs and wants. It appears very unlikely that data in the form of outcomes research and and experts will provide unambiguous answers. You need only to look at the controversy triggered by the US Preventative Services Task force in 2009 by their recommendations regarding screening mammography.

I have to admit that up to this point I viewed finding truth in medicine as difficult but possible, limited only by the ability to collect sufficient information on enough individuals over time. I viewed these challenges as perhaps unlikely to be accomplished but essentially possible. I am not so sure anymore and I now suspect that we can never get to actual "truth" in regards to many of the issues we are grappling with in health care. I now have begun to believe that we are entering into a post truth age in medicine.
The challenge we face is how to adapt to such a world. Shellenberger and Nordhaus note:
The problem is not that we are in a post-truth age but rather that we have not learned to adapt to it. Perhaps a good place to begin is by recognizing our own biases, perspectives, and agendas and attempting to hold them more lightly.

One of the major limits to adapting is the increasingly partisan nature of the environment where these issues are addressed. Nowhere is this any more apparent than in the world of health care. Disagreements in the marketplace are settled by consumer choices and these decisions are immediately constrained by resources which are available to individuals or voluntary associations. This provides a reasonable stopping mechanism when addressing a wicked problem, providing at least some assurance that resources will no longer be directed to a particular problem or problems when the investment has reached a point of diminishing returns. When one appeals to people free to invest or spend their personal resources and it does not really matter how partisan the disagreements might be. However, move these issues into the political realm and unlike the market based world, the end game results in a more difficult environment to have expression of individual preferences.

As health care financing has moved more and more into the public realm and the sums of money involved have become more and more staggering, the discussions have become hyper-polarized and partisan, and our prospects of reaching any form of truth ever more remote. Furthermore, as the financing of public goods is less and less immediately constrained by financial resources through the illusory magic of borrowing from future generations of yet to be born, spending on wicked problems, particularly health care wicked problems, runs the real risk of consuming our entire economy. Such wasteful diversion of resources is less likely to happen if private entities are responsible for making such investment decision. When private entities make bad investment decisions they tend to quickly disappear. The same can be said for governments but the time lines may be be very much prolonged and the impacts of failure much worse.

Then how do we adapt to a post-truth age in health care? I don't really know the answer to that question but I suspect that it depends upon who (or what) controls resources which are directed at wicked problems and whether decisions to allocate resources are made at the local (individual) levels via individual decisions or are made via governmental entities through political processes. Stay tuned for more...