Factors Associated With Phantom Odor Perception Among US Adults: Findings From the National Health and Nutrition Examination Survey

Key Points

Question  How does the prevalence of phantom odor perception vary by age, sex, socioeconomic position, health status, health behaviors, smell function, and oral and sinonasal symptoms among US adults?

Findings  In this cross-sectional study of 7417 adults, the prevalence of phantom odor perception was 6.5% (n = 534) and was greater among women, younger age groups, and those of lower socioeconomic position. Phantom odor perception was more common among those with poorer health, a history of head injury, or dry mouth symptoms.

Meaning  Epidemiologic characterization may provide clues to cause and alert clinicians to the importance of this disorder.

Importance  Phantom odor perception can be a debilitating condition. Factors associated with phantom odor perception have not been reported using population-based epidemiologic data.

Objective  To estimate the prevalence of phantom odor perception among US adults 40 years and older and identify factors associated with this condition.

Design, Setting, and Participants  In this cross-sectional study with complex sampling design, 7417 adults 40 years and older made up a nationally representative sample from data collected in 2011 through 2014 as part of the National Health and Nutrition Examination Survey.

Exposures  Sociodemographic characteristics, cigarette and alcohol use, head injury, persistent dry mouth, smell function, and general health status.

Main Outcomes and Measures  Phantom odor perception ascertained as report of unpleasant, bad, or burning odor when no actual odor exists.

Results  Of the 7417 participants in the study, 52.8% (3862) were women, the mean (SD) age was 58 (12) years, and the prevalence of phantom odor perception occurred in 534 participants, which was 6.5% of the population (95% CI, 5.7%-7.5%). Phantom odor prevalence varied considerably by age and sex. Women 60 years and older reported phantom odors less commonly (7.5% [n = 935] and 5.5% [n = 937] among women aged 60-69 years and 70 years and older, respectively) than younger women (9.6% [n = 1028] and 10.1% [n = 962] among those aged 40-49 years and 50-59 years, respectively). The prevalence among men varied from 2.5% (n = 846) among men 70 years and older to 5.3% (n = 913) among men 60 to 69 years old. Phantom odor perception was 60% (n = 1602) to 65% (n = 2521) more likely among those with an income-to-poverty ratio of less than 3 compared with those in the highest income-to-poverty ratio group (odds ratio [OR], 1.65; 95% CI, 1.06-2.56; and OR, 1.60; 95% CI, 1.01-2.54 for income-to-poverty ratio <1.5 and 1.5-2.9, respectively). Health conditions associated with phantom odor perception included persistent dry mouth (OR, 3.03; 95% CI, 2.17-4.24) and history of head injury (OR, 1.74; 95% CI, 1.20-2.51).

Conclusions and Relevance  An age-related decline in the prevalence of phantom odor perception is observed in women but not in men. Only 11% (n = 64) of people who report phantom odor perception have discussed a taste or smell problem with a clinician. Associations of phantom odor perception with poorer health and persistent dry mouth point to medication use as a potential explanation. Prevention of serious head injuries could have the added benefit of reducing phantom odor perception.

The perception of phantom odors is a qualitative olfactory dysfunction whereby affected individuals perceive odors in the absence of an external stimulus. The sensation is typically unpleasant and described with terms such as “foul,” “rotten,” or “chemical.” There are few epidemiological studies of phantom odor perception. Prevalence estimates are in the range of 5% to 6%^1,2 and may be greater among women. Phantom odor perception has been correlated with depression^3,4 and has been described in relation to migraine aura,⁵ radiation therapy,⁶ sinonasal disease, post–upper respiratory tract viral infection,⁷ and head trauma,⁸ although most evidence has been limited to case reports and small clinical studies. The biological mechanism of phantom odor perception is thought to arise from aberrant peripheral olfactory sensory neurons⁹ signaling perception centers in the brain, or overactive brain cells creating the perception.^9,10 Reports of phantom odors have not been correlated with objective measures of olfactory loss.^2,11 Phantom odor symptoms may disappear, improve, or worsen over time,¹² and treatments are not reliably effective.^9,13

To our knowledge, the epidemiology of phantom odor perception in the United States has never been described. The objective of the present study was to use recent national survey data to evaluate associations between phantom odor perception and sociodemographic factors, certain health behaviors, and health conditions among middle-aged and older US adults.

Data were collected in 2011 through 2014 as part of the National Health and Nutrition Examination Survey conducted by the National Center for Health Statistics. The cross-sectional survey used a stratified, multistage, probability cluster design that resulted in a nationally representative sample of the noninstitutionalized, civilian US population. Study participants were interviewed in their homes and underwent a physical examination at a mobile examination center. Adults 40 years and older were eligible for the interview question on phantom odors. Of 7418 eligible adults, 7417 had nonmissing data on phantom odor perception. The demographic and other characteristics are detailed in Table 1. Overall response rates for the interviewed sample were 72.6% in the 2011-2012 period and 71.0% in the 2013-2014 period. Survey protocols were approved by the National Center for Health Statistics Research Ethics Review Board. Study participants provided written informed consent.

Phantom odor perception was measured with a single question: “Do you sometimes smell an unpleasant, bad, or burning odor when nothing is there?” Sociodemographic characteristics, including age, sex, and educational attainment, were obtained during in-home interviews. Race/ethnic background information was collected routinely as part of ongoing national surveillance. Participants reported their own race and ethnicity and were allowed to select more than 1 category. Income-to-poverty ratio was defined as the ratio of reported total family income to the US Census Bureau’s poverty threshold, which varies by size of family and age of family members. A person with an income-to-poverty ratio of 1.5, for example, belongs to a family with an income that is 50% above the poverty threshold. Income-to-poverty ratio has a defined maximum value of 5; persons whose income-to-poverty ratio exceeds that are assigned a value of 5.

People who reported never having smoked 100 cigarettes in their lifetime were classified as having never smoked. Among the others, smoking status was defined as currently smoke every day, currently smoke some days, or former smoker. Alcohol use was classified using a threshold of 12 alcoholic beverages in the past year. Frequency of alcohol use was categorized as 0, 1 to 2 days/week, and 3 to 7 days/week, based on self report.

Participants assessed their general health status as excellent, very good, good, fair, or poor. Participants also reported on specific physical health conditions experienced over the past 12 months, including nasal congestion from allergies, cold or flu lasting longer than a month, and persistent dry mouth. Lifetime history of lost consciousness due to head injury, broken nose, or other serious face or skull injury, and greater than 2 sinus infections were assessed from interview questions. For survey years 2013 through 2014, reduced smell function was assessed for 3519 participants with an 8-item odor identification test in a mobile examination center. Odorants presented were strawberry, smoke, soap, chocolate, natural gas, leather, grape, and onion. Reduced smell function was defined as fewer than 6 correctly identified odorants.¹⁴

Frequency distributions of sociodemographic and certain health factors were computed for those who did and did not report phantom odors. For each level of each factor, we computed the difference (95% CI) in the proportion exhibiting that characteristic between those with phantom odor perception and those without. We estimated the unadjusted prevalence (95% CI) of phantom odor perception overall, stratified by age and sex, and stratified by various health conditions. The 95% CIs serve to determine the range within which point estimates would fall on repeated sampling. Independent correlates of phantom odor perception were identified using a parsimonious multiple logistic regression model. All variables that were examined in the bivariable analysis were also examined in the multivariable analysis. With the bivariable analysis as our guide, we fit variables sequentially into the model, allowing significant variables to remain in the multivariable model until entering additional variables identified them as not independently associated.¹⁵ We report effect sizes as odds ratio (OR) (95% CI). An age-by-sex interaction was entered to test whether sex modified the observed age association. Analyses were performed using SAS, version 14.2 (SAS Institute Inc) and SUDAAN, version 11.0.1 (RTI International), incorporating 4-year sample interview or examination weights. The prevalence of phantom odor perception stratified by reduced smell function was estimated using 2-year sample examination weights.

Based on the findings of this study, we estimated the prevalence of phantom odor perception as 6.5% (n = 534) (95% CI, 5.7%-7.5%) among US adults 40 years and older (Table 2). Only 11.1% (n = 64) of people who reported phantom odors have discussed a taste or smell problem of any kind with a clinician. The frequency distribution of sociodemographic and health behavior characteristics is summarized in Table 1, along with the difference (95% CIs) between the absolute percentage among people who reported phantom odors and those who did not. People who reported phantom odors were less likely to be older than 70 years than those who did not report phantom odors (12.4% [n = 85] vs 19.7% [n = 1698]). A greater proportion of those who reported phantom odors were non-Hispanic black (14.4% [n = 151] vs 10.5% [n = 1626] among those who did not experience phantom odors). Non-Hispanic white individuals (n = 2852 [71.7%]) and non-Hispanic Asian individuals (n = 863 [4.9%]) were represented in greater proportions among those who did not report phantom odors compared with those who did (n=194 [63.5%] and n=31 [2.7%] among non-Hispanic white individuals and non-Hispanic Asian individuals, respectively). Women made up 52.8% (n = 3862) of the national sample among adults 40 years and older. Two-thirds (68.0% [n = 356]) of adults who reported phantom odors were women. People who reported phantom odors were less likely to have more than a high school education than those who did not report phantom odors (48.0% [n = 218] vs 62.2% [n = 3609]). Moreover, two-thirds (n = 762) of people who reported phantom odors had an income-to-poverty ratio of less than 3 (or 300% of poverty level) compared with 46.1% (n = 3749) of those who did not report phantom odors.

Almost 20% (n = 1309) of US adults reported smoking every day or on some days. Of those who reported phantom odors, 25.3% (n = 130) had a history of current or former cigarette smoking compared with 17.8% (n = 1179) among those who did not report phantom odors. Overall use of alcohol did not vary between those who reported phantom odors and those who did not, but people who reported phantom odors were less likely to consume alcohol at least 3 days per week (11.3% [n = 49] vs 21.1% [n = 936]).

The Figure displays the prevalence of phantom odor perception by age group and sex in the study population. Women had about 2 times greater prevalence of phantom odor perception than men at every age group. Nonoverlapping CIs in the 40 to 49 and 50 to 59 years age groups indicated statistically significant sex differences. We noted a declining prevalence of phantom odors by age, which was more pronounced in women than it was in men.

The frequency distribution of potentially correlated health conditions and the corresponding prevalence of phantom odor perception are summarized in Table 2. Adults in fair or poor health represented 20.4% (n = 1793) of the population. More than 13% (n = 227) of people in fair or poor health reported phantom odors compared with 5.8% (n = 169) among those in good health and 4.2% (n = 84) among those in very good or excellent health. People who reported persistent dry mouth had more than 3 times the prevalence of reporting phantom odors as those who did not (17.4% [n = 204] vs 4.9% [n = 330]). More than 10% (n = 112) of people who had lost consciousness from a head injury reported phantom odors compared with 5.7% (n = 422) of those without a history of head injury. Almost one-fifth of the population (n = 1037) reported a history of injury to their nose, face, or skull but no greater prevalence of phantom odor perception. Reduced smell function or suboptimal performance on an odor identification test was identified in 13.5% (n = 630) of the population. Individuals with reduced smell function had somewhat, but not significantly, lower prevalence of phantom odor perception.

In Table 3, we detail estimates for the strength of independent associations of phantom odors with demographic factors and health conditions, adjusted for all variables remaining in the model. We observed that the strength of the association between age and phantom odor perception varies markedly by sex. Men in younger age groups (40-49, 50-59, and 60-69 years) were twice as likely to report phantom odors as men 70 years and older, but this association was statistically significant for only the men aged 40 to 49 years (adjusted OR [aOR], 2.32; 95% CI, 1.25-4.28). Among women, however, we observed a strong graded inverse correlation between the likelihood of phantom odor perception and age. Compared with men 70 years and older, women in this age group had no greater likelihood of phantom odor perception, but women in the younger age groups (40-49, 50-59, and 60-69 years) had 5 (aOR, 5.12; 95% CI, 2.77-9.44), 4 (aOR, 4.51; 95% CI, 2.03-10.02), and 3 (aOR, 3.18; 95% CI, 1.61-6.27) times the odds of phantom odor perception, respectively. Compared with people in the highest socioeconomic category (income-to-poverty ratio of 5), those with an income-to-poverty ratio of less than 3 had 60% to 65% greater likelihood of experiencing phantom odors (aOR, 1.60; 95% CI, 1.01-2.54; and aOR, 1.65; 95% CI, 1.06-2.56). Adults in fair/poor health were more than twice as likely to report phantom odors than those in excellent or very good health (aOR, 2.27; 95% CI, 1.66-3.08). People with persistent dry mouth had 3 times the odds of reporting phantom odors (aOR, 3.03; 95% CI, 2.17-4.24). History of head injury with loss of consciousness was associated with 74% greater odds of phantom odor perception (aOR, 1.74; 95% CI, 1.20-2.51).

To our knowledge, the present study is the first population-based epidemiologic analysis of phantom odor perception using nationally representative data from the United States. We estimated the prevalence to be 6.5% among adults 40 years and older. This estimate is somewhat greater than the 4.9% prevalence observed in 1 Swedish community, although adults in the Swedish study were 60 years and older.² When we limited our sample to include only US adults 60 years and older, we corroborated the prevalence of phantom odor perception to be 5.4% (95% CI, 4.3%-6.8%). We observed that younger age and lower socioeconomic class were independently and positively associated with phantom odor perception. Two-thirds of people who experienced phantom odors were women, which is consistent with the sex ratio observed in the Swedish population-based data.² People reporting fair to poor health were more likely to report phantom odor perception than those in good health. Adults with persistent dry mouth or history of serious head injury were more likely to report phantom odor perception independent of reported health status.

Phantom odor perception is an olfactory disturbance in which individuals perceive an odor in the absence of a stimulus. The cause of this condition is unknown. People with phantom odor perception may have a neural signaling imbalance that allows nonsense olfactory signals to reach the central nervous system. The sensation may originate in the peripheral nervous system at the level of the olfactory sensory neurons, may reflect damage to the olfactory nerve, or may originate in the central brain.¹⁶ We found phantom odor perception to be inversely associated with age predominantly among women. If phantom odors are the result of aberrant peripheral neurons, an age-related loss of olfactory sensory neurons may explain an age-related decline in phantom odor perception consistent with an age-related decline in odor sensitivity.^17,18 Reasons for the sex difference or the age-related decline are unclear.

Black and Hispanic people are overrepresented among people who report phantom odors. However, when we accounted for the greater prevalence of phantom odors among people of lower socioeconomic position, we observed no difference across race/ethnic groups. The greater prevalence of phantom odor perception among people of lower socioeconomic position may reflect greater exposure to environmental toxins.¹⁹ Socioeconomic indices such as family income are inversely associated with exposure to fine particulate matter and other ambient air pollutants, and poor indoor air quality.^19,20 Lower socioeconomic status could also be a marker for a number of health conditions understood to be associated with phantom odors, such as migraine or primary headache.²¹ Nonfatal injury rates also vary inversely with socioeconomic status.^22,23 Given the association we observed between phantom odors and severe head trauma, it is possible that our measure of low income may be capturing injuries that have damaged the olfactory neurons, the olfactory bulb, or the cerebral cortex, despite no loss of consciousness. Alternatively, the inverse association between phantom odors and income may reflect differential reporting of nonfatal head injuries by social class.²⁴

Evidence from a recent review suggests that current but not former smoking is related to subjective measures of olfactory dysfunction. Phantom odors were not considered among the olfactory conditions in the review.²⁵ We observed greater prevalence of phantom odor perception among people who currently smoke, compared with former or never smokers. However, after adjustment for cigarette smoking being more common among younger people and those of lower socioeconomic status, we no longer observed an association between smoking and phantom odor perception. People who drink alcohol were not more likely than people who abstain to report phantom odors. However, when alcohol use was broken down by drinking days per week, we found that people who drink more frequently have fewer reports of phantom odors, but this association was not robust to adjustment for dry mouth symptoms. Recent pathobiologic evidence suggested no decreased neurogenesis in the olfactory bulb when comparing a group with a 30-year history of high-volume alcohol use with a group with modest alcohol consumption.²⁶

We observed a strong association between phantom odor perception and symptoms of persistent dry mouth. Our measure of persistent dry mouth may have captured people with Sjogren syndrome, an autoimmune disorder that has been associated with reduced smell function^27,28 but not, to our knowledge, with phantom odors. With 13% of adults reporting dry mouth symptoms, this measure may also be an indicator of medication use. A variety of medications have been implicated in olfactory dysfunction, including chemotherapeutic agents, antihypertensive agents, antibiotics, and antidepressants.²⁹ Self-reported general health is an indicator of individual health that is strongly correlated with mortality, particularly among people who have indicators of circulatory disease.³⁰ Sjölund et al² observed a modest association between a nonspecific indicator of cardiovascular risk burden and phantom odor perception using data collected as part of a community-based study in Sweden. Again, medication use could explain correlations with fair or poor general health in our data and with cardiovascular risk in the Swedish data. Assessing associations with medication use was beyond the scope of the current study. Finally, we observed no association between reduced smell function based on an odor identification test, which corroborates evidence from the Swedish study.²

We acknowledge limitations to our study. It is possible that our measure of phantom odor perception misclassified people who have burning mouth syndrome or oral phantom sensations. However, when we conducted a sensitivity analysis reclassifying 13 people who also reported burning or other taste phantoms, there was no difference in the results. We do not have information on intensity, duration, or periodicity of the phantom odor perception. We have no data regarding health conditions, including seizures, primary headache, migraine headache, and serious mental illness, which could serve to explain the age, sex, and income patterns we describe. Data were not collected on adults younger than 40 years. We might expect some misclassification in the reporting of medical conditions. The misclassification of severe head trauma, in particular, could be differential, which would bias our point estimate in an unpredictable direction. However, the positive association we report is biologically plausible and consistent with earlier reports of olfactory dysfunction, including phantom odors, among people with head trauma.³¹These cross-sectional data are not able to discern the temporal relationship of phantom odors with any of the correlated factors. For example, living with and finding no relief from phantom odors could lead people to report fair or poor health. Finally, the sample design does not allow for adults who are institutionalized or serving in the military. People hospitalized with head trauma and frail elderly people living in nursing facilities are not represented.

To our knowledge, this is the first observational study of phantom odor perception using nationally representative survey data and the largest epidemiological investigation of this condition to date. We estimate the prevalence of phantom odor perception at 6.5% among US adults 40 years and older. Among US adults 40 years and older, phantom odor perception is more common among women, younger age groups, and those of lower socioeconomic position. Phantom odors are reported more commonly among those in poorer health and among those who have persistent dry mouth. Head injury is strongly associated with phantom odor perception, and rates of traumatic brain injury seen in emergency departments have been steadily increasing.³² Response to treatment for loss of odor sensitivity has been shown to depend on duration of symptoms.³³ Only 11% of people who report phantom odors have discussed a taste or smell problem of any kind with a clinician. Increased awareness of phantom odor perception as a clinical problem and the risk factors associated with this condition may contribute to more affected individuals seeking guidance or treatment for this condition.

Corresponding Author: Kathleen E. Bainbridge, PhD, MPH, National Institute on Deafness and Other Communication Disorders, 6001 Executive Blvd, MSC 9670, Bethesda, MD 20892-7180 (bainbridgek@mail.nih.gov).

Accepted for Publication: May 7, 2018.

Published Online: August 16, 2018. doi:10.1001/jamaoto.2018.1446

Author Contributions: Dr Bainbridge had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Bainbridge.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Bainbridge.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Bainbridge, Byrd-Clark.

Administrative, technical, or material support: Byrd-Clark.

Study supervision: Bainbridge.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support:The National Institute on Deafness and Other Communication Disorders provided funding support for National Health and Nutrition Examination Survey chemosensory data collection via interagency agreements with the National Center for Health Statistics of the Centers for Disease Control and Prevention and the University of Connecticut. Support for the statistical analysis was provided by Social & Scientific Systems, Inc under Contract No. GS-00F-173CA, Task Order HHSN275201700074U.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed in this report represent those of the authors and do not necessarily represent the agencies or institutions for which they work.