A declining sense of smell in older people is a strong predictor of death within just five years, according to new research.
Thirty-nine percent of study subjects who failed a simple smelling test died during that period, compared to 19 percent of those with moderate smell loss and just 10 percent of those with a healthy sense of smell, the journal PLOS ONE reported on Wednesday.
The hazards of smell loss were "strikingly robust," according to researchers, who said that olfactory dysfunction was better at predicting mortality than a diagnosis of heart failure, cancer or lung disease.
Only severe liver damage was a more powerful predictor of death, they said.
"We think loss of the sense of smell is like the canary in the coal mine," said the study's lead author Jayant Pinto, an associate professor of surgery at the University of Chicago.
"It doesn't directly cause death, but it's a harbinger, an early warning that something has gone badly wrong, that damage has been done.
"Our findings could provide a useful clinical test, a quick and inexpensive way to identify patients most at risk."
Precisely how sense of smell loss relates to mortality is unclear.
"Obviously, people don't die just because their olfactory system is damaged," said Martha McClintock, the study's senior author.
Pinto added: "Of all human senses, smell is the most undervalued and underappreciated -- until it's gone."
The study was part of the National Social Life, Health and Aging Project (NSHAP), the first in-home study of social relationships and health in a large, nationally representative sample of men and women ages 57 to 85.
In the first wave of the NSHAP, conducted in 2005-2006, professional survey teams from the National Opinion Research Center at the University of Chicago used a well-validated test for a field survey of 3,005 participants.
It measured their ability to identify five distinct common odors.
In the second wave, during 2010-2011, the survey team confirmed which participants were still alive. During that five-year gap, 430 (12.5 percent) of the original 3,005 study subjects had died and 2,565 were still alive.