Between 2002 and 2004, the researchers conducted behavioral tests—including the Mini-Mental State Examination and a modified version of Trails B—and an at-home sleep study of 461 women (mean age 82.3 years) from Minneapolis, Minnesota, and the Monongahela Valley, near Pittsburgh, Pennsylvania. None of these women had dementia at baseline, or used SDB therapy, such as continuous positive airway pressure (CPAP) devices or supplemental oxygen therapy. During sleep testing, a team overseen by Susan Redline of Harvard Medical School measured variables such as the number of arousals per hour and the number of airflow cessations (apnea) or reductions (hypopnea) lasting 10 seconds or longer with an associated dip in blood oxygen saturation of 3 percent or more per hour. Among the 298 women who met the criteria for inclusion in the final sample, 35.2 percent were classified as having SDB.

At the time of follow-up about five years later, the women underwent more extensive cognitive tests than they did at baseline. Of the women who did not have SDB at baseline, 31.1 developed MCI or dementia. Of the women who did have SDB, 44.8 percent received one of the diagnoses. After adjusting for demographic and physiological factors, as well as disease and medication use, the researchers found that the odds of developing MCI or dementia were 85 percent higher for women with SDB. Adjusting for baseline cognitive test scores increased the odds ratio to 2.36.