THE VIRGINIAN-PILOT
Copyright (c) 1994, Landmark Communications, Inc.
DATE: Sunday, November 27, 1994 TAG: 9411270056
SECTION: FRONT PAGE: A4 EDITION: FINAL
SOURCE: KNIGHT-RIDDER NEWS SERVICE
DATELINE: WASHINGTON LENGTH: Medium: 90 lines
A patient in intensive care looks up from a hospital bed and asks the poignant question: ``What are my chances, doc?''
Traditionally, physicians have answered as best they could, based on their training and personal experience.
Now, medical researchers have developed a number of computer systems they say predict - more accurately than any human - whether a critically ill person will live or die.
The systems, which draw on the results of hundreds of thousands of previous cases, give a doctor a much broader foundation on which to base decisions in life-or-death situations.
Advocates say these new systems offer a reasonable, fair and humane way to spare a dying person needless suffering and expense.
But critics fear that government agencies or insurance companies in the future will use computer-based mortality predictions to justify refusing to pay for treatment for patients whose condition is hopeless.
This has not happened yet, but some health-care providers are recommending it begin soon. Their ideas are provoking a heated debate in medical circles.
Under intense pressure to cut expenses, the health-care industry is struggling to cut back services that are judged to be medically unnecessary or of little benefit. Care for patients expected to die in a few days or weeks is a ripe target for cost-cutters.
``With the rising cost of health care approaching $1 trillion per year in the United States, payers - and indeed society - are demanding new and better ways of defining medical necessity,'' Dr. David Hadorn, an emergency care specialist with the Rand Corp. in Santa Monica, Calif., told the American College of Critical Care Medicine.
The key question, said Hadorn, is ``who will decide who gets these benefits - me, some bureaucrat or some computer?''
The leading computerized mortality prediction system, known as APACHE, is already in use in more than 400 hospitals, according to its inventor, Dr. William Knaus, of George Washington University Medical Center.
APACHE estimates a patient's short-term probability of dying by comparing his or her symptoms - such as age, blood pressure and breathing difficulty - to a computer analysis of 200,000 previous cases stored in its memory bank.
An electronic prognosticator like APACHE is often ``better than Dr. Welby,'' said Kenneth Goodman, director of the Forum for Bioethics and Philosophy at the University of Miami.
APACHE and similar systems have been in existence for about 10 years. So far, they have been used mostly to study the death rates among groups of patients - not individuals - in order to evaluate the performance of different hospitals and the effectiveness of various treatment methods.
Now, pressure is building to use APACHE-type systems to help decide what to do with individual patients, a far more controversial application of computer technology.
Last April, the Society for Critical Care Medicine, an organization of intensive care specialists, published guidelines for deciding who should or should not be admitted to a hospital's intensive care unit.
According to the guidelines, ``patients with very poor prognoses and little likelihood of benefit should not be admitted.'' In determining admissions policy, health care providers ``should utilize predictive instruments (like APACHE) with a full understanding of their strengths and limitations.''
Two committee members who drafted the guidelines, Dr. H. Tristram Engelhardt Jr., of Baylor College of Medicine in Houston, and Dr. Michael Rie, of the University of Kentucky Medical Center, have been pushing for years for society's right to refuse to pay for expensive care that appears to be futile.
``Government and third-party insurance programs should explicitly state that funds will not be made available (to extend) the process of dying . . Association.
Other doctors and bioethicists reject the application of computer predictions to individual cases, especially by third-party payers like government agencies or private insurance companies.
``It will be tempting for payers to simply substitute clinical decision rules - based on statistical analysis of past outcomes - for clinical judgment,'' Hadorn, the Rand specialist, wrote in the October issue of the Journal of Critical Care Medicine.
``If this possibility becomes the practice, something important may be gained in terms of accuracy and overall benefit, but something terribly important will have been lost in terms of the human factor.''
Even the inventors of computer-based systems urge extreme caution in applying them to individuals.
``There is simply not enough information contained in any one model to make it useful as a guide to individual patient decisions,'' said Stanley Lemeshow, a professor of biostatistics at the University of Massachusetts. by CNB