Association Between Olfactory Dysfunction and Mortality in US Adults

Key Points

Question  Is self-reported and/or objectively measured olfactory dysfunction associated with mortality when accounting for relevant factors among adults in the United States?

Findings  In this nationally representative cohort study of 3503 adults 40 years or older, objectively measured olfactory dysfunction was significantly associated with increased all-cause 5-year mortality among older adults independent of demographics, cardiovascular comorbidities, depression, and cognition (18% increased risk per 1-point decrease in Pocket Smell Test score [score range, 0-8]). Self-reported olfactory dysfunction was not associated with mortality.

Meaning  These findings suggest that objectively measured olfactory dysfunction is robustly associated with 5-year mortality among older US adults, and olfaction measured by objective smell test may be a useful indicator of health status in older adults.

Importance  A study of olfactory dysfunction and mortality in a large national cohort will aid in better understanding their association when accounting for multiple relevant factors and possible underlying mechanisms.

Objective  To investigate the association of olfactory dysfunction with all-cause 5-year mortality in US adults.

Design, Setting, and Participants  This cohort study included participants 40 years or older from the 2013-2014 National Health and Nutritional Examination Survey who had data on olfaction and mortality (n = 3503). Olfaction was assessed by self-report and objective test (8-odor Pocket Smell Test). Mortality was determined by linking with the National Death Index through February 24, 2019. Data were analyzed from July 1 to September 30, 2019.

Main Outcomes and Measures  Olfaction and 5-year mortality. Cox proportional regression models were used to examine the associations between olfaction and mortality while adjusting for demographics and medical comorbidities. Multivariate models were further adjusted for depression and cognitive assessments.

Results  Among the 3503 participants (1831 women [52.3%]; mean [SD] age, 59.0 [12.0] years), the prevalence of olfactory dysfunction was 13.5% (95% CI, 11.0%-16.0%) based on results of an objective smell test and 21.6% (95% CI, 18.9%-24.2%) based on self-report. Risk of mortality increased by 18% (95% CI, 7%-29%) per 1-point decrease in smell test score in a multivariate model. The association was significant among adults 65 years or older in association with binary (hazard ratio [HR], 1.95; 95% CI, 1.19-3.21) and linear (HR, 1.19; 95% CI, 1.08-1.31) measures of objective olfactory dysfunction, but not among adults aged 40 to 64 years. There was no association between self-reported olfactory dysfunction and mortality. The association between objective olfactory dysfunction and mortality remained after further adjusting for cognitive assessment battery and depression among older adults (HR, 1.18; 95% CI, 1.01-1.37).

Conclusions and Relevance  These findings suggest that objective olfactory dysfunction is associated with increased mortality among older adults. In addition to its effect on quality of life, the association of olfactory dysfunction with mortality has implications for physical and cognitive health.

Poor olfactory function has been directly implicated in malnutrition,¹ decreased safety,² and overall worse quality of life.³ It is responsible for more than 200 000 physician visits per year, representing a significant public health burden.^4,5 Emerging evidence suggests that olfactory dysfunction is significantly associated with increased all-cause mortality among older adults.^4,6-11 Most notably, a recent study by Liu et al¹² examining community-dwelling older adults aged 71 to 82 years showed clear evidence that poor olfaction alone explains higher long-term mortality, particularly in individuals with excellent to good health at baseline. Liu et al¹² found the elevated risk of mortality of patients with poor olfaction was only partially explained by neurodegenerative disease, cardiovascular disease, and weight loss. Olfaction is emerging as an early indicator of brain aging that can be objectively measured with a relatively simple smell test in the clinical setting.

Choi et al¹³ previously used the National Health and Nutrition Examination Survey (NHANES) to demonstrate that objectively measured olfactory dysfunction is associated with cognitive impairment independently of demographics and cardiovascular factors. Herein we further investigate the associations of olfactory dysfunction (measured by both objective smell test and self-report) with all-cause 5-year mortality in US adults 40 years or older, independently of cardiovascular factors, cognition, and depression.

NHANES is a database collected by the US Centers for Disease Control and Prevention to assess the nutritional and health status of the noninstitutionalized, civilian population in the United States. Each cohort in NHANES uses a complex sampling design with selective oversampling of low-income individuals and racial minorities.¹⁴ Analyses accounting for the stratified, multistage probability sampling design yield results that are representative of the entire US population.¹⁵ The analytic cohort for the present study consisted of 3503 individuals 40 years or older who had complete data on olfaction and mortality in the 2013-2014 NHANES. Olfaction was assessed by both an objective olfactory test and self-report. The study was exempt from institutional review board approval because the data had already been deidentified and are publicly available. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Objective olfactory testing was assessed using the NHANES Pocket Smell Test (Sensonics International), which includes an 8-item scratch-and-sniff test. The 8 odorants include onion, soap, leather, smoke, grape, strawberry, chocolate, and natural gas. Participants were asked to identify each odorant from 4 alternative names. Normal olfaction was defined as being able to correctly identify at least 6 odors (score range, 0-8) as in previous literature.¹⁶ Olfactory dysfunction was defined as having a score of 5 or less. Good test-retest reliability was observed from the NHANES olfaction protocol.¹⁷

Questionnaires on subjective olfaction were collected by the interviewer using the computer-assisted personal interviewing system. Self-reported olfactory dysfunction was defined as reporting a problem with smell in the past 12 months, worse sense of smell since 25 years of age, or phantosmia as previously defined.¹⁸

The cognitive function battery administered as per NHANES protocol consisted of the Animal Fluency Test,¹⁹ assessments from the Consortium to Establish a Registry for Alzheimer Disease (CERAD),^20-22 and the Digit Symbol Substitution Test.²³ Depressive disorders were evaluated using the Patient Health Questionnaire.²⁴ Participants were categorized as having major depressive disorder if either of the initial questions addressing depressed mood was answered as “more than half the days” or “nearly every day” and the Patient Health Questionnaire score was at least 9 (possible range, 0-27, with higher scores indicating greater depressive symptoms). Details of each assessment are available in previously published studies examining the associations of olfactory dysfunction with cognition¹³ and depression.²⁵

The National Center for Health Statistics has linked data collected from the NHANES surveys with death certificate records from the National Death Index.²⁶ Mortality was determined by probabilistic matching between NHANES data and death certificates through February 24, 2019.²⁷ For the NHANES survey, follow-up time has been calculated using person-months from the date of interview to the date of death or the end of the mortality period.

Demographic data and medical history were obtained during interviews. Self-reported race and ethnicity were grouped as White, Black, Hispanic, Asian, or other. Household income was collapsed into 5 categories and educational level into 3 categories as in Table 1. Medical history variables included hypertension, stroke, diabetes, smoking status, and cardiovascular disease (myocardial infarction, congestive heart failure, angina pectoris, or coronary artery disease). Other olfaction-related medical history data included ever having a broken nose or other serious injury to the face or the skull, persistent cold/flu for longer than 1 month in the past 12 months, ever having 2 or more sinus infections, and ever having a loss of consciousness because of a head injury.

Data were analyzed from July 1 to September 30, 2019. Sample weights were used to account for the complex sampling design according to the NHANES analytic guidelines.¹⁵ Baseline characteristics of the study participants were compared using the 2-tailed t test and Pearson χ² test. The association between various measures of olfactory dysfunction (binary self-reported, binary objectively measured olfactory dysfunction as defined above, and linear objectively measured olfactory dysfunction by a 1-point decrease in Pocket Smell Test score) and mortality was investigated using Cox proportional hazards regression models. Multivariate models were sequentially adjusted for age, demographics, cardiovascular risk factors, and olfaction-related medical history. In a subgroup of adults younger than 65 years with available information on cognitive function and depression (n = 1022), a multivariate model was additionally adjusted for the cognitive assessment battery and major depressive disorder. STATA, version 16 (StataCorp LLC) was used for all analyses.

The overall study cohort consisted of 3503 adults 40 years or older who participated in the 2013-2014 NHANES (1831 women [52.3%] and 1672 men [47.7%]; mean [SD] age, 59.0 [12.0] years). Unweighted participant characteristics are summarized in Table 1. The estimated prevalence of olfactory dysfunction was 21.6% (95% CI, 18.9%-24.2%) based on self-report and 13.5% (95% CI, 11.0%-16.0%) based on objective smell test results.

As of February 24, 2019, 105 individuals were determined to have died from the study cohort. The estimated 5-year mortality rate for adults aged 40 to 64 years was 1.2% (95% CI, 0.7%-2.0%), whereas for adults 65 years or older, it was 6.0% (95% CI, 4.5%-7.8%). Mortality rates were comparable between those who reported subjective olfactory dysfunction (2.5%; 95% CI, 2.0%-3.1%) and no subjective olfactory dysfunction (2.8%; 95% CI, 1.7%-4.8%) (Table 2). In contrast, those who had objectively measured olfactory dysfunction were found to have a higher mortality rate at 5.8% (95% CI, 3.9%-8.7%) compared with those who had normal olfactory function at 2.1% (95% CI, 1.5%-2.8%).

In the age-adjusted model, there was no association between self-reported olfactory dysfunction and risk of mortality (hazard ratio [HR], 1.12; 95% CI, 0.68-1.84) (Table 3). The binary measure of objective olfactory dysfunction was associated with 53% increased risk of mortality (HR, 1.53; 95% CI, 1.02-2.30). When considering the NHANES Pocket Smell Test scores as a linear variable (score range, 0-8), a 1-point decrease in score was associated with a 19% increased risk of mortality (HR, 1.19; 95% CI, 1.08-1.30). In a multivariate model adjusting for demographics, comorbidities, and olfaction-related medical history, mortality risk was associated with linear measures of olfactory dysfunction (HR, 1.18; 95% CI, 1.07-1.29).

Subgroup analyses were performed by age group (middle-aged vs older adults). There was no association between mortality risk and subjective or objective measures of olfactory dysfunction among middle-aged adults. Among older adults, increased risk of mortality was observed in association with both binary measures (HR, 1.95; 95% CI, 1.19-3.21) and linear measures (HR, 1.19; 95% CI, 1.08-1.31) of objective olfactory dysfunction after adjusting for demographics, comorbidities, and olfaction-related medical history (Table 3).

Within a group of older adults who completed depression and cognitive assessments (n = 1022), an additional model including major depressive disorder (based on Patient Health Questionnaire score) and the cognitive assessment battery (including the Digit Symbol Substitution Test, Animal Fluency Test, and CERAD assessment) was constructed to examine whether the corresponding variables account for the association between olfactory dysfunction and mortality. The binary measure of objective olfactory dysfunction was associated with an estimated 61% increased risk of mortality (HR, 1.61; 95% CI, 0.98-2.66). A 1-point decrease in smell test score was associated with an estimated 18% increased risk of mortality (95% CI, 7%-29%) in older adults (HR, 1.18; 95% CI, 1.01-1.37) in this model with additional adjustment for depression and cognitive assessments.

Objectively measured olfactory dysfunction was independently associated with increased risk of mortality at 5-year follow-up in this representative sample of US adults 40 years and older. Subgroup analysis by age demonstrated this association among adults 65 years or older but not among adults aged 40 to 64 years. These results were robust to analysis adjusting for other covariates, including demographics, cardiovascular disease, olfaction-related medical history, depression, and cognitive function. There was no association between self-reported olfactory dysfunction and mortality.

Our results are generally consistent with results from previous studies demonstrating association of olfactory dysfunction with increased mortality risk in older adults,^7-12,28-30 although the study population demographics, type of olfactory assessment test, and duration of follow-up vary substantially by report. Seven previous studies^7-12,28 have explored the association between objective olfactory dysfunction and mortality. Although most previous studies were recruited from regional community-dwelling older adults, the present study included a nationally representative sample of US adults with the largest sample size. One previous study⁷ based on a nationally representative US sample from the National Social Life, Health and Aging Project (age range, 57-85 years) similarly demonstrated increased mortality risk among older adults with olfactory dysfunction. Our study had additional data on a validated measure of depression (Patient Health Questionnaire) and diverse measures of cognitive function and further included adults aged 40 to 65 years. One longitudinal cohort study from a city in northern Sweden⁹ included a subgroup of middle-aged adults defined as 40 to 70 years. In that study, there was a significant association between objectively measured olfactory dysfunction (measured with the 13-item Scandinavian Odor Identification Test) and increased risk of mortality.⁹ The discrepancy observed in our study is likely due to longer duration of follow-up (mean of 9.9 years) in the Swedish study compared with the 5-year follow-up data in the present study. Sensitivity analysis with adjustment of the age cutoff of 40 to 70 years did not show any association in our cohort. These findings suggest that olfactory function is not associated with 5-year mortality among middle-aged adults in the United States, but there may be an association with longer follow-up. Future longitudinal studies are needed to examine the role of olfactory dysfunction as a predictor of mortality at follow-up longer than 5 years.

Cognitive function was assessed as a potential mediator of the association between olfaction and mortality in a subgroup of older adults. Objective olfactory dysfunction is known to be associated with lower scores across various domains of cognitive function, including attention and executive function (Digit Symbol Substitution Test), verbal fluency (Animal Fluency Test), and memory (CERAD assessment).¹³ In a multivariate model additionally adjusting for the cognitive assessment scores, a 1-point decrease in the smell test continued to be associated with higher risks of mortality similar to the findings from previous studies that have included a measure of cognitive function as a covariate (eg, the Mini-Mental State Examination,^9,10,12,28 the short Portable Mental Status Questionnaire,⁷ or reported clinical diagnosis of dementia^8,28). One previous study from a suburban Australian cohort¹⁰ demonstrated that the link between olfaction and mortality was no longer significant after adjusting for the Mini-Mental State Examination score. Decline in cognitive function and its associated neurodegenerative diseases are likely one of several pathological processes that may account for the link between olfaction and mortality. Olfactory dysfunction has been identified to precede full clinical emergence of neurodegenerative diseases such as Alzheimer disease and Parkinson disease.^31-34 Previous pathologic studies^32,35,36 have shown that neurodegenerative markers in the olfactory tract (ie, α-synuclein, β-amyloid, and tau) are potentially involved in the early disease process. Another longitudinal cohort study in older adults¹² found that neurodegenerative diseases explain 22% of the increased risk of 10-year mortality among individuals with poor olfaction. Our findings demonstrating olfaction as associated with mortality independently of cognitive functioning imply that additional mechanisms underlie the association.

Several additional mechanisms have been proposed in the literature. First, a direct insult to the olfactory nerve, the only cranial nerve exposed to the environment, from a virus, bacteria, or toxins may be associated with injuries to other systemic organs, such as the central nervous, pulmonary, and cardiovascular systems, causing increased risk of mortality. Second, olfactory dysfunction may be a marker of advanced physiological aging, because poor recovery of olfactory functioning after environmental or internal insults is an indicator of poor cellular regeneration.^37-40 A recent study²⁹ has demonstrated increased interleukin 6 serum levels among older adults with olfactory dysfunction and suggested the increased inflammation as a common pathophysiological pathway that may be associated with hyposmia, frailty, and mortality. Olfactory dysfunction can also lead to malnutrition, unsafe food choices, and increased risks of accidents due to gas leaks and smoke.¹²

Depression is another potential mediator significantly associated with olfactory dysfunction and mortality. Olfactory bulb ablation in an animal model has been found to cause alterations in chemical and behavioral states that are similar to a depressed state.⁴¹ On the other hand, depression can potentially cause elevated levels of inflammatory cytokines and glucocorticoids inhibiting neurogenesis of the olfactory system.^42-44 A large body of literature^45-48 has demonstrated the effect of depression on increased risk of mortality across various patient populations via worsening disease severity and development of additional comorbidities. A recent randomized clinical trial from Germany demonstrated that older adults with olfactory dysfunction who completed olfactory training for 5 months were found to report improved depressive symptoms.⁴⁹ These findings warrant future research to understand the effect of olfactory training on mortality in addition to quality of life and mental health. In our study, major depressive disorder was significantly associated with higher mortality risk, but olfactory dysfunction remained an independent risk factor associated with mortality in a multivariate model.

Our study findings suggest olfactory dysfunction as independently associated with mortality has clinical implications for physical, mental, and cognitive health, especially among older adults. Adults with olfactory dysfunction are expected to be prone to malnutrition, because these individuals may have decreased appetite and ability to enjoy food, thereby leading to poor food intake.^1,50 Olfactory dysfunction may also prevent adults from recognizing life-threatening situations, such as a gas leak or a fire.⁵ Olfactory dysfunction is known to be associated with poorer quality of life and higher prevalence of depressive symptoms.^3,25,50 In addition to olfactory dysfunction being suggestive of accelerated brain aging, it has been found to be an early factor associated with development of Alzheimer and Parkinson disease.^51,52 Detection of olfactory dysfunction, especially among older adults, suggests that further workup for malnutrition, depression, and neurodegenerative disease may be needed. Adults with known olfactory dysfunction should be more cautious of life-threatening situations because they are unable to smell danger signals in the household environment.⁵³

There are limitations to this study. First, various definitions of subjective and objective olfactory dysfunction and tests exist. We defined olfactory dysfunction a priori based on previous NHANES studies examining olfaction. The results may vary based on the definitions adopted. Second, despite adjusting for multiple major confounders and possible mediators, residual confounding by other environmental or medical factors cannot be completely ruled out. For example, the covariates obtained from cross-sectional data in this study limit assessment of potentially relevant information, such as changes in olfactory dysfunction (temporary vs permanent causes) or weight loss. In addition, the relatively smaller size of the subgroup analysis among older adults may have limited the power to detect significant associations in multivariate models. The specific causes of mortality were also not available in this study, which could have allowed for further analysis of a possible mechanism underlying the link between olfaction and mortality. Future studies are required to explore the basis of the association between olfactory dysfunction and mortality in a longitudinal study cohort at longer follow-up.

Objectively measured olfactory dysfunction is associated with an increased risk of 5-year all-cause mortality among older (≥65 years) but not middle-aged (40-64 years) US adults. Olfactory dysfunction was identified as independently associated with mortality after accounting for demographics, medical comorbidities, depression, and cognitive functioning.

Accepted for Publication: August 14, 2020.

Corresponding Author: Janet S. Choi, MD, MPH, Caruso Department of Otolaryngology–Head and Neck Surgery, Keck School of Medicine of the University of Southern California, 1540 Alcazar St, Ste 204M, Los Angeles, CA 90033 (janet.choi@med.usc.edu).

Published Online: October 22, 2020. doi:10.1001/jamaoto.2020.3502

Author Contributions: Drs Choi and Wrobel had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Choi, Kim, Hur, Ference, Wrobel.

Acquisition, analysis, or interpretation of data: Choi, Jang, Ference, Wrobel.

Drafting of the manuscript: Choi, Jang, Kim.

Critical revision of the manuscript for important intellectual content: Choi, Jang, Hur, Ference, Wrobel.

Statistical analysis: Choi, Jang.

Administrative, technical, or material support: Choi, Kim, Ference.

Supervision: Hur, Ference, Wrobel.

Conflict of Interest Disclosures: None reported.

Meeting Presentation: This paper was presented at the Triological Society Sections Meeting; January 23, 2020; Coronado, California.