That includes people with alcoholism, epilepsy or uncontrolled high blood pressure, researchers said.

But one downside, the study found, is that people were less likely to go back to doctors who had given them a medical warning and reported it to the government.

Nonetheless, lead researcher Dr. Donald Redelmeier of the Sunnybrook Health Sciences Centre in Toronto told Reuters Health the findings should convince doctors to warn patients about driving risks.

"It isn't just that it can save your patient's life, not just that (not reporting) imposes risks on others in the community as well," he said. "It's also that crashes are such a widespread cause of property damage that everybody pays, either through insurance premiums, or congested roadways, or the price of consumer goods."

If the results from the Canadian study, conducted by the Institute for Clinical Evaluative Sciences, were extrapolated throughout the United States, he said the annual savings might total $160 million.

Smaller studies have offered conflicting findings about the effectiveness of such warnings.

The new report, published in The New England Journal of Medicine, looked at the risk of getting in a crash and needing emergency room care in 100,075 people who were advised by their doctor not to drive.

Since 2006, doctors in Ontario have been paid $36.25 per case to encourage reporting of such problems. Patients subsequently receive a warning letter from the government.

The reasons for getting a warning letter in this study ranged from diagnoses of uncontrolled high blood pressure and diabetes to epilepsy and stroke.

"The vast majority of patients in our study had more than one of those disorders," said Redelmeier - including 21 percent who had five or more conditions on the worrisome-driving list.

Among the people reported under the program, 10 to 30 percent ultimately had their license suspended, "so everybody gets cautioned, but relatively few people get suspended," he said.

AN ‘IMMEDIATE, SUBSTANTIAL' EFFECT

During the three years prior to the warning, 1 in 210 people in the study were involved in a crash that resulted in an ER trip each year - about double the rate in the general population.

One year after the warning was issued, the risk to those patients had fallen to 1 in 367, a 45-percent reduction.

"The reduction in risk was immediate, substantial, and sustained," the researchers said. "Together, these data suggest that warning patients who are medically unfit to drive may reduce the risk of road crashes."

"It turns out that medical warnings for unfit drivers are substantially more effective than physician advice about smoking cessation or weight loss or regular exercise," Redelmeier said.

Eight U.S. states have mandatory reporting of medically impaired drivers: California, Delaware, Georgia, Maine, Nevada, New Jersey, New Mexico and Pennsylvania.

The researchers didn't have data on how often patients drove in general after receiving the warnings.

Warnings had no effect on the rate of accident-related ER visits when the patient wasn't the driver.

However, their risk of showing up to be treated for depression rose from 1 in 52 annually before the intervention to 1 in 42 after.

And although patients didn't have fewer visits to any doctor after the warning, they were less likely to see the physician who reported them.

Dr. Alice Pomidor, a geriatrician at Florida State University College of Medicine in Tallahassee, who was not connected to the study, said the decline in the accident rate was "quite impressive," but conditions in the U.S. might discourage that type of reporting.

Under the Canadian system, patients can't avoid the warning by going to another doctor. It follows them in their records.

"It also doesn't really have any financial impact on the Canadian physician if the patient 'takes their business elsewhere,'" she told Reuters Health in an email.

"Here, the threat to never come back, and to tell all their friends so that no one else will go to the reporting physician, has potentially a much greater impact on an individual physician/practice."

In addition, "Physicians are concerned that if they 'rat out' the patient, he or she may also become afraid to tell the physician the truth about their medical conditions, to the detriment of their health," Pomidor said. "So there needs to be a really compelling personal reason involving the medical safety of an individual patient to get a physician to report" to a government authority.

Redelmeier said the reduction in accidents seen in the study "is about two times larger than the combined effects of modern trauma systems in saving patients' lives. This is a very large effect size traceable to a preventative program, as opposed to a resuscitation program."

SOURCE: bit.ly/QETDM4 The New England Journal of Medicine, online September 26, 2012.

• A Permanent Economic Emergency