TCS Daily

The New Prometheans

By Jay Bryant - October 29, 2002 12:00 AM

Something really important is underway. It has the potential to improve the quality of life for millions of people, make vast fortunes for a few, stimulate the world economy, change fundamental human attitudes and create enormous passion, controversy, fear and loathing.

In recent days and weeks, financial commitments have been made. The race is on, and no one knows just where it will end.

Over the next five years, hundreds of millions of dollars will be spent to identify those most at risk of chronic degenerative diseases, including cancer, diabetes and heart disease.

The race is to develop the best algorithm for predicting who will get these diseases and when. The great hope for humanity is that by developing and refining statistical procedures so that a reliable risk profile can be created, timely proactive intervention by and on behalf of the outliers - those most at risk - can prevent the disease and enrich and lengthen their lives.

The great hope of the people putting up the money is that they can obviate the need to spend billions of dollars currently consumed in the extraordinarily expensive treatments accorded the patients these high-risk outliers eventually become.

The research involved in all of this is not so much medical as it is statistical. The goal is to identify what information truly predicts whether or not a person will be stricken with a given disease. What tests can we give you today that will best predict whether or not you will get cancer in the next two years, five years, ten years?

Certain health insurers, including many of the Blue Cross-Blue Shield companies, are already using early versions of the algorithm to seek out the at-risk population, but their success rate is limited. Right now, the best cancer algorithm around has a statistical reliability of only a little more than 25%, with a two-year horizon. For some other diseases, the two-year predictability can get above 50%, but that's about as high as it goes.

The people behind the new initiatives are betting that within the next five years they can get the numbers up into the 80% range for all chronic degenerative diseases. If they are right, it will soon be possible to predict your future health care cost.

The question is, will that information be used against you?

Take heart disease, for example. Current lipid profile (cholesterol) tests are sufficiently reliable that the National Institutes of Health has revised its guidelines. "Studies have shown that people at high risk for [coronary heart disease] should have more aggressive treatment to lower their levels of...LDL cholesterol (the 'bad' cholesterol) than was previously recommended," NIH says.
But the risk factors are still imprecise, and envelope-pushing doctors believe a less-often-performed test, which can measure the aggregate size of cholesterol particles, is the real key to understanding how cholesterol works to cause heart disease, and thus why some people with high overall cholesterol never have heart attacks, while others with low readings do.

The statisticians will find out if these doctors are right. They will run dozens, maybe hundreds and thousands of other hypotheses through the system as well. The data collected will get richer and richer as more and more cases are added. This is called longitudinal testing, and its power comes from the quantity of data. The more subjects whose tests results and subsequent health records are entered, the greater the statistical accuracy becomes.

Armed with the new information, doctors can intervene earlier, with drugs and other treatments. Patients can be instructed as to lifestyle changes that can prevent the onset of the disease. People will live longer and healthier, and spend less on health care.

But suppose, knowing that your cholesterol particle size is large (or small), and knowing the correlation between that and your risk of heart disease, your insurance company raises your rate, like your automobile insurance company does when it concludes you are not a good driver. What if you can't get health (or life) insurance at all because of your statistical profile?

What if, warned of the risk, and advised by your doctor to avoid, say, hot dogs, you nonetheless continue to eat hot dogs? Should the universe of other insurance premium payers have to subsidize your willful intransigence?

We will soon be entering the period of discussion, debate and demagoguery on these questions.

In Greek legend, the God Prometheus brought fire to the human race. Good news, bad news. The statisticians and their backers are new Prometheans, and the gift they are about to bring us may provide warmth against the chill of death itself. Careful, though - you might get burned.

The author is publisher of The Optimate.

TCS Daily Archives