Dr. Alexander K. Smith is a brave man.

It has taken physicians a very long time to accept the need to level with patients and their families when they have terminal illnesses and death is near — and we know that many times those kinds of honest, exploratory conversations still don’t take place.

Now Dr. Smith, a palliative care specialist at the University of California, San Francisco, who also practices at the San Francisco Veterans Affairs Medical Center, and two co-authors are urging another change, one they acknowledge would “radically alter” the way health care professionals communicate with their very old patients.

In a recent article in The New England Journal of Medicine, they suggested offering to discuss “overall prognosis,” doctorspeak for probable life expectancy and the likelihood of death, with patients who don’t have terminal illnesses. The researchers favor broaching the subject with anyone who has a life expectancy of less than 10 years or has reached age 85.

“Advanced age itself is the greatest predictor of poor prognosis,” Dr. Smith told me in an interview.

By age 85, the article points out, the average remaining life expectancy for Americans is six years. An 85-year-old has a 75 percent chance of living another three years, but only a one in four chance of surviving for 10. Which category a particular old person falls into has much to do with the medical problems he or she has, or doesn’t have, and with his or her ability to function.

When the odds are that they have only a few remaining years, should doctors discuss that with them?

Dr. Smith and his co-authors, Dr. Brie Williams and Dr. Bernard Lo — a geriatrician and an internist, respectively — vote yes. “This is about empowering patients to make informed choices and encouraging individual decision-making,” he said.

Sadly, it takes guts to propose this when mention of the D word to patients still raises alarms. The Obama administration had to cancel plans for Medicare to reimburse doctors when they discuss end-of-life care with their patients. Death panels! Rationing!

But to Dr. Smith, understanding how much time remains could help his older patients make the most of those years and help them ward off interventions, tests and treatments whose benefits, if any, are years away but whose harms could be immediate.

A “substantial minority” of older patients won’t want to have this discussion, Dr. Smith acknowledged. “It’s important to offer the information, not force it on people,” he said.

But in his experience, it’s the protective family caregivers who object to talking about prognosis, more than their older relatives. “A lot of very elderly patients realize they’re in their final years,” he said. “This doesn’t come as a surprise to them. My friends in their 90s are already thinking about it.”

He cited a study he and colleagues published in The Journal of General Internal Medicine, based on interviews with 60 older people with disabilities, their average age 78 — an admittedly small but ethnically diverse sample. About two-thirds told researchers they’d want their doctors to tell them if they had less than five years to live. (Readers here had even stronger opinions: see this post about public access to longevity indexes.)

And when they do think about it, Dr. Smith continued, “they want to get their finances in order, plan for long-term care, spend time with children and friends.” They may be able to take fewer medications and undergo fewer procedures, with the emphasis on quality of life, or otherwise shift priorities.

“This is a challenge to people,” Dr. Smith acknowledged. “I’ve had reactions from ‘This is terrific; I’ve been arguing for this for years,’ to a mentor at U.C.S.F. who said, ‘This is ridiculous; my patients don’t want to hear this, and there is no way to predict life expectancy anyway.’”

But while it’s true that no one can foretell a particular individual’s death with any certainty — and health care workers should be clear about those limitations, Dr. Smith emphasized — a number of geriatric calculators do provide reasonably good projections, based on several health factors, age, cognitive status and functional abilities and sometimes laboratory test results. An index developed by Dr. Smith’s U.C.S.F. colleague Sei Lee, for example, can correctly predict mortality within four years about 75 percent of the time.

Still, it’s O.K. with Dr. Smith if professionals and patients have strong reactions, pro and con. “The point of the article is to get a national conversation started about this,” he said. It’s a conversation you’re invited to join in the comments section below.