by Craig Klugman, Ph.D.

In the 1979 novel Sophie’s Choice by William Styron, the reader meets a Holocaust survivor who was forced in the camps to choose which of her two children would die immediately. Making the choice led to a life of alcoholism, depression, and self-destructive behavior. One interpretation of this novel, later made into an Academy Award winning film (1982), is that having choose whether a loved one lives or dies is a spirit-wrenching decision.

And yet, everyday, health care providers and bioethicists ask legally appointed health care power of attorneys and other designated surrogates to decide whether an incapacitated patient has surgery, receives a feeding tube, is resuscitated, or is intubated. One metanalysis of 40 studies, showed that over one-third of surrogate decision-makers experienced a “negative emotional burden as the result of making treatment decisions.”

Not only can surrogate decision-making harm the surrogate, but it also turns out that they are not very good at choosing what the patient would want. To test whether patients and surrogates would make the same choices, several studies have conducted interview studies surrounding specific scenarios. The “patient” and “surrogate” are given a scenario and questioned separately. Their answers are correlated to see if there is agreement. Obviously this is a very artificial set up, but asking incapacitate patients who need a surrogate about agreement in choices is fairly difficult (if not impossible) to do. One study showed that there was good correlation in stroke treatment but bad correlation on research participation. “The degree of surrogate agreement is dependent on the type of decision being made.” Other studies found that surrogates incorrectly predict patient choices in one-third of cases. “These data undermine the claim that reliance on surrogates is justified by their ability to predict incapacitate patient’s treatment preferences.” Some studies even show that prior conversations between patients and surrogates do not improve accuracy of decision-making.

However, in these scenarios there is good correlation between patient preferences and physician treatment decisions. These findings beg the question, what’s the best way to make decisions for the incapacitated?

(1) Last week’s Institute of Medicine Report advocated for greater focus on education and designating a surrogate. The goal would be to make sure more advance care planning and conversations between patients-surrogates-physicians occurs.

(2) Perhaps we should drop the idea of surrogacy and allow physicians to take the lead on decision-making. Such substituted interest models have been advocated by Daniel Sulmasy and Ezekiel and Linda Emanuel.

(3) Take out the human equation. Maybe decision-making should be left to the machines. This is not like having a random number generator spit out an answer. Consider the increase of health information technology that is part of the Affordable Care Act. Huge databases are being amassed that indicate everything about a patient’s background, demographics, medical condition, choices, and ultimate course of illness. It should be possible in a few years to use informatics techniques to build a database of patient types and the decisions that they would make. In ten years, an algorithm could exist that would permit entering a patient’s prognosis, age, ethnicity, socioeconomic status, disease, and a few other factors and the machine calculates the statistically likely decision such a person would make, based on an enormous database of what has happened in the past. This approach removes any concern about guilt, haunting emotional burden, lack of surrogate education, and subjectivity. Combining known patient preferences and medically likely outcomes, a clear answer can be calculated. This may strike some as “cold” or “lacking humanity” but it could turn out to be more accurate and easier to work with. A database does not need days of education and time to reflect. Give it the facts and it gives you a response.

The future of surrogacy might be in substituting a machine for the human. If this seems unlikely, consider that a very small proof-of-concept study found that such an approach may be highly accurate and effective: Patients and physicians-using-a computerized decision-aid correlated their choices 84 percent of the time.