People who have mild memory problems but who can correctly identify the source of certain smells are probably not in the first stages of Alzheimer’s disease, suggesting that smell tests could offer a new technique for early diagnosis, researchers reported Monday.

As part of the test, people with cognitive difficulties completed a “scratch and sniff” test, in which they scratched off 40 different odors and used multiple choice to identify the origin of each smell.

Study author Dr. D.P. Devanand of Columbia University in New York explained that all of the people in the study had visited a memory clinic because they were having problems, but none of the traditional evaluations could determine clearly whether they were in the early stages of Alzheimer’s disease.

People who never developed Alzheimer’s disease tended to correctly identify around 35 scents, while people who eventually were diagnosed with the brain disorder could pinpoint only about 20 to 30 scents correctly, Devanand explained.

Looking at individual scents, the researchers found that people’s ability to identify key odors such as menthol, lemon, lilac and leather were most predictive of whether or not they would develop the memory-robbing disease.

In an interview, Devanand, who also works at the New York State Psychiatric Institute, explained that researchers have known for years that smell memory is one of the first things to go in someone with early Alzheimer’s disease. People can detect smells perfectly well, he said, but cannot identify them.

“You have to recognize (the smell) against your memory of smells of that type,” Devanand told Reuters Health. “And that’s where the problem is.”

Devanand and his colleagues asked 150 people with mild memory impairments and 63 others without impairments to identify 40 different smells. The researchers followed them for an average of five years, noting who developed Alzheimer’s disease.

The researchers then looked at which smell memories were most often impaired in people with early Alzheimer’s disease, and reviewed smell data from people known to have Alzheimer’s disease, to pick which smells were most predictive of a person’s chance of getting the disease.

Ultimately, Devanand and his team narrowed the list of 40 smells down to 10, which also included clove, strawberry, pineapple, smoke, soap and natural gas. “These 10 items predicted as well or better than the entire 40-item test,” Devanand said, and a shorter list takes much less time to complete.

The smell test detected Alzheimer’s disease better than brain imaging and genetic testing did, and worked just as well as an extensive memory test that can be very frustrating and unpleasant, Devanand noted.

The next step, he said, is to continue testing the smell test to make sure it can truly diagnose the disease. He predicted that it will likely be a “couple of years” before it is available to people with mild memory problems.

Devanand and his team presented the findings Monday at the American College of Neuropsychopharmacology annual meeting in San Juan, Puerto Rico.