Olfactory identification deficits and increased mortality in the community

Abstract

Objective: To examine the association between odor identification deficits and future mortality in a multiethnic community cohort of older adults.

Methods: Participants were evaluated with the 40-item University of Pennsylvania Smell Identification Test (UPSIT). Follow-up occurred at 2-year intervals with information on death obtained from informant interviews and the National Death Index.

Results: During follow-up (mean = 4.1 years, standard deviation = 2.6), 349 of 1,169 (29.9%) participants died. Participants who died were more likely to be older (p < 0.001), be male (p < 0.001), have lower UPSIT scores (p < 0.001), and have a diagnosis of dementia (p < 0.001). In a Cox model, the association between lower UPSIT score and mortality (hazard ratio [HR] = 1.07 per point interval, 95% confidence interval [CI] = 1.05-1.08, p < 0.001) persisted after controlling for age, gender, education, ethnicity, language, modified Charlson medical comorbidity index, dementia, depression, alcohol abuse, head injury, smoking, body mass index, and vision and hearing impairment (HR = 1.05, 95% CI = 1.03-1.07, p < 0.001). Compared to the fourth quartile with the highest UPSIT scores, HRs for mortality for the first, second, and third quartiles of UPSIT scores were 3.81 (95% CI = 2.71-5.34), 1.75 (95% CI = 1.23-2.50), and 1.58 (95% CI = 1.09-2.30), respectively. Participant mortality rate was 45% in the lowest quartile of UPSIT scores (anosmia) and 18% in the highest quartile of UPSIT scores.

Interpretation: Impaired odor identification, particularly in the anosmic range, is associated with increased mortality in older adults even after controlling for dementia and medical comorbidity.

Figure 1

Mortality rates during follow-up by baseline UPSIT score quartiles. UPSIT (range 0–40) quartile scores: [0–20], [20–26], [26–31] and [31–40]. For maximum available follow-up, the UPSIT score was significantly associated with mortality (chisq=54.25, p < .001). For study participants, mortality rates were 45.36% in the highest quartile, 28.75% and 20.63% in the next two quartiles, respectively, and 18.39% in participants with the lowest quartile of UPSIT scores as shown in the four bar charts in the figure.

Figure 2

Cumulative Hazard Function for mortality during follow-up for participants with baseline UPSIT scores at the 10^th and 90^th percentiles in the sample (total n=1169). Cumulative hazard function from Cox proportional hazard regression analyses for baseline UPSIT scores at the 10^th percentile (UPSIT score=15, solid line) and the 90^th percentile (UPSIT score=34, dotted line) with death as the outcome, adjusted for the following covariates: age, gender, education, race/ethnicity, language of test administration, Charlson comorbidity index, dementia status, depression, head injury, alcohol use, BMI, smoking history. The gray shaded areas are the 95% confidence intervals of the cumulative hazard functions.