Stat counter

View My Stats

Sunday, September 19, 2010

We are all gamblers

I am reading a book by Peter Bernstein entitled "Against the Gods: The remarkable Story of Risk". As those of you who have read my previous blog pieces already know, I am fascinated by concepts of risk and prediction. I believe that understanding these concepts are absolutely essential to the practice of medicine. Unfortunately, the general public as well as the medical profession has a poor grasp of these concepts and it serves as an impediment to best decision making.

What is most intriguing about the history described by Bernstein is that gambling was the driver of gains in the statistical realm which ultimately catalyzed the quantum leaps in expanding the understanding of the statistical tools and concepts required to both understand and predict risks. One the concepts were understood as they related to dice and cards, the mathematical application to other domains outside typical gambling became apparent. The rest is history in the world of finance and insurance.

At the most fundamental "gut" level, I think all of us understand that we are constantly making bets and weighing the odds. I suspect that one of the attractions to the world of overt gambling is the rules are more explicit, the stakes are reasonably well defined, and the outcomes are tightly temporally linked. We do not need to wait a long time to discern whether we are winners or losers. Furthermore, it is easy to tell. The outcomes are not ambiguous. If there is anything we humans like it is certainty and immediate gratification.

The practice of medicine is laced with bets. In contrast to the clarity of the gaming tables, medicine is filled with uncertainty overlaid with ambiguity. Take a population of patients with mild hypertension. Intervene with some sort of medication in thousands of patients, follow for decades, and statistically someone is a winner but no one knows who. Similarly, screen tens of thousands of patients for breast, lung, or prostate cancer using whatever tool which is in vogue. In contrast to patients with hypertension, patients who participate in cancer screening get an almost immediate feedback with what appears to be a result which is as unambiguous as rolling snake eyes. The problem is that virtually no medical tests are that black and white. We would like to be able to predict with certainty the winners and losers (although in the end we are all losers) and most of the public and much of the medical community have virtually no appreciation of the limits of these tests.  They fail to understand that the diagnosis of cancer in many cases is simply a prediction about what might happen in the future with no way of knowing whether the prediction is correct 95%, 50%, 5%, or 0.5% of scenarios.

In my own practice I have solved this problem by focusing on the present as opposed to the future. I deal primarily with patients who are symptomatic and uncomfortable and my measure of success is whether I am able to fix that. I like to say that I deal with present problems as opposed to dealing with things that might happen in the future. I still need to be aware of what might happen in the future, particularly as it relates to what my interventions may cause in the future. When trying to address someone is is in pain, is short of breath, is itching from dermatitis, or is functionally impaired in some way, all possible fixes entail some risk of creating some new or worse problem.

The obvious question which any well informed patient should ask is how likely is that to happen? Alas, there lies the problem with even the most modern medicine. For the most part, we do not know. In the 16th century, gamblers were in similar straits. Men had been playing dice games for literally thousands of years but it was not until the works of Girolamo Cardano, a Renaissance man and inveterate gambler that assessment of risk and odds changed. His work revolutionized how the financial world dealt with risk, moving it from gut and intuition to statistical analysis. With his mathematical tools, he could precisely predict the odds of rolling a two or a seven. The same principles were applied to annuities, insuring commerce,  and finance.

For the most part, risk assessment in the practice of medicine still operates primarily at the gut level. The statistical tools exist but they require actual data to be applied. Absent the availability of relevant data, we are left with a world similar to those throwing dice prior to Cardano. We may understand that we are placing all types of bets but we don't have the tools to understand the odds.

No comments:

Post a Comment