Olfactory Impairment in Adults
The Beaver Dam Experience
Carla R. Schubert,aKaren J. Cruickshanks,aClaire Murphy,b Guan-Hua Huang,c Barbara E. K. Klein,a Ronald Klein,a
F. Javier Nieto,aJames S. Pankow,d and Ted S. Tweeda
aUniversity of Wisconsin, Madison, Wisconsin, USA bSan Diego State University, San Diego, California, USA cNational Chiao Tung University, Hsinchu, Taiwan, Republic of China
dUniversity of Minnesota, Minneapolis, Minnesota, USA
Olfactory function may be important for environmental and nutritional safety and enjoy-ment. Population-based epidemiological studies of olfaction are needed to understand the magnitude of the health burden, identify modifiable risk factors, and develop and test prevention and treatment strategies for olfactory impairment. However, measuring olfaction in large studies is challenging, requiring repeatable, efficient methods that can measure change over time. Two large cohort studies, the Epidemiology of Hearing Loss Study (EHLS) and the Beaver Dam Offspring Study (BOSS), included olfactory testing. In both studies, the San Diego Odor Identification Test (SDOIT) was used to measure olfaction. Subjects were asked to identify eight common household odors (such as coffee and chocolate). Olfactory impairment was defined as correctly identifying fewer than six out of eight odorants after two trials. The EHLS participants were age 53–95 years at the time of the first measurement (1998–2000), and participants in the BOSS were age 21–84 years. The prevalence of olfactory impairment in the EHLS was 25% overall, more common in men than women, and increased with age. Five years later olfaction was measured a second time and the majority (84%) of the EHLS participants were classified the same. Among those with impairment at the base line nearly one-third (31%) improved to unimpaired. This heterogeneity in olfactory impairment has unique implications for data analyses and predicting outcomes and associations. Preliminary data from the BOSS suggest the prevalence of olfactory impairment may be lower in younger generations. All these factors point to a continuing need for epidemiological studies of olfaction.
Key words: olfaction; epidemiology; population-based; EHLS; BOSS
Introduction
Olfactory function may be important for en-vironmental and nutritional safety, quality of life, and enjoyment. In one study, participants with olfactory disorders reported more prob-lems with paid employment, housework, social, and family life than controls, and in another,
Address for correspondence: Carla Schubert, MS, 610 Walnut Street, Rm 1087 WARF bldg, Madison, WI 53726-2336. Voice: 608-265-3722; fax: 608-265-2148. Schubert@episense.wisc.edu
participants with olfactory loss reported feel-ing impaired in their ability to detect smoke, gas leaks, or spoiled food.1,2Previous studies3,4 have found that olfactory impairment is more common in older age groups and men, but until recently the prevalence of olfactory impairment in the general population was not known.5 Similarly, olfactory impairment among peo-ple seeking diagnosis and treatment at clinics was associated with certain conditions and dis-eases, but little was known about risk factors for impairment in the general population.6,7
International Symposium on Olfaction and Taste: Ann. N.Y. Acad. Sci. 1170: 531–536 (2009). doi: 10.1111/j.1749-6632.2009.04102.x c 2009 New York Academy of Sciences.
Therefore, population-based studies of olfac-tion are important to understand the magni-tude of the health problem in the population in order to estimate the public health burden and to identify modifiable risk factors that could lead to prevention and treatment strategies.
Two large studies, the Epidemiology of Hearing Loss Study (EHLS, 1993 to present), a longitudinal, population-based study of sen-sory loss and aging in Beaver Dam, Wisconsin, and the Beaver Dam Offspring Study (BOSS) (2005 to present), a study of the offspring of the EHLS participants, have included olfac-tory testing as part of their examinations, and the methodological challenges from these stud-ies and results are presented here.5,8,9
Methodological Considerations
There are some factors unique to olfac-tory testing in large population studies that need to be considered. First, the test needs to be acceptable to the participants. Subjects in population-based studies are invited to partici-pate because they are residing within the area defined as the sampling frame for the study, and, unlike clinic-based studies or studies in se-lected populations, participants are not iden-tified because they have an underlying con-dition motivating them to participate. Thus, uncomfortable or extensive testing may dis-courage many from participating in the study. Second, population-based studies often involve many modalities, and each one is limited in the amount of time available for testing. There-fore, in these studies it is important to keep participant burden low to maximize participa-tion and maintain interest in participating in subsequent follow-up examinations. Third, the olfactory test selected must be time and cost efficient for a large number of participants. Above all, the reliability and repeatability of the olfactory test must be high so any change measured in olfactory function over time is likely true change and not measurement error. In the EHLS and the BOSS, the San Diego Odor Identification Test (SDOIT) was used
to measure olfaction. The SDOIT consists of eight common household odorants that are wrapped in gauze and placed in opaque jars. The odorants are presented in a random or-der to participants, who are asked to close their eyes before the odorant is placed under their nose. Participants are able to smell the odor-ant as long as they wodor-ant, usually about 3–5 s. Participants are then told they may open their eyes and identify the odor using an array that contains pictures of the eight odorants plus 12 distracters. There is a 45-s delay between odor-ant presentations to allow for adaptation. Par-ticipants who misidentify an odorant are told the correct identification of the odor and the odorant is presented again later in the testing process. The score for the test is the number of correctly identified odorants after two trials. Olfactory impairment is defined as identify-ing fewer than six of the eight odorants cor-rectly. These methods have been published else-where.5,10
The SDOIT is an efficient test, both in terms of time and cost and is well received by partic-ipants. In the EHLS and the BOSS, the av-erage administration time for the SDOIT was about 15 min including the time between odors, which can be used for the administration of questionnaires. The SDOIT is a very reliable and repeatable odor-identification test. In pre-vious research, children and Down syndrome patients tested a mean of 5.4 days apart showed a test–retest reliability of 0.86 and more re-cently, in a study of adults, the SDOIT had a concordance correlation coefficient of 0.85 in people age 50–70, with a mean time between tests of 3 weeks.10,11
Results and Discussion
Olfaction was measured at the 5- and 10-year examinations of the EHLS (n= 2800, age 53–95 years in 1998–2000 and n = 2395 in 2003–2005, respectively).9 The prevalence of
olfactory impairment in the EHLS population was 24.5% overall, increased with age, and was more common in men than women (Fig. 1).5
Figure 1. Prevalence of olfactory impairment at the baseline examination, by age and sex: EHLS 1998–2000. (Figure based on data from Ref. 5.)
Factors associated with the prevalence of ol-factory impairment were older age, male sex, being a current smoker, nasal congestion or up-per respiratory infection during the week pre-ceding the testing, and a history of epilepsy or stroke.5 The EHLS is the only population-based study in the United States to measure olfaction, and it provides much-needed data on the prevalence of olfactory impairment, as well as factors associated with impairment, in a general population of older adults.
In addition, in the EHLS, olfactory impair-ment was associated with the 5-year incidence of cognitive impairment. Compared with par-ticipants without olfactory impairment, partici-pants with an olfactory impairment at the base-line olfactory examination were more likely to have developed cognitive impairment at the ex-amination 5 years later (odds ratio (OR)= 3.33, 95% confidence interval (CI)= 2.04, 5.42).12
However, despite this strong association, most of the participants with olfactory impairment (84%) at the baseline examination did not de-velop cognitive impairment during the 5-year period, and the sensitivity of the SDOIT was only 55.1% for cognitive impairment.12
Now, for the first time, there is information on the change in olfaction in a general pop-ulation of older adults followed for 5 years; 1901 participants in the EHLS had olfaction measured at two examinations 5 years apart. The majority of these participants (84%) had no change in their olfactory classification at the 5-year follow-up examination.
An interesting finding of this study was that 31% of participants who had olfactory impair-ment at the baseline examination were unim-paired at the 5-year follow-up examination. The percent of olfactory improvement is shown in Table 1 by age and sex. There was no overall difference between men and women in the per-cent improving, but there was less improvement in the oldest age group.
In Table 2, the distribution of olfac-tory improvement is shown by the baseline SDOIT score. The majority of participants who changed classification did so by increasing their score by 1–3 units between the baseline SDOIT and follow-up, whereas only 11.2% in-creased their score by 4 or 5 units.
To further explore the factors associated with olfactory improvement, multivariate logistic re-gression analyses were done. These analyses found the risk for olfactory improvement de-creased with increasing age (OR = 0.80 per
TABLE 1. Percent of Participants with Olfactory Impairment at Base Line Who Were Unimpaired 5 Years Later (Improvers)
Olfaction not impaired at 5 years Olfaction impaired at base line
Baseline age (years) N at risk Men (%) Women (%) All (%)
53–59 27 25.0 45.5 33.3
60–69 116 42.9 35.9 40.5
70–79 125 26.4 39.6 32.0
80–97 77 19.4 10.9 14.3
TABLE 2. Distribution of Olfactory Improvement by Baseline San Diego Odor Identification Test (SDOIT) Score
SDOIT score Olfaction impaired Remained impaired at Improved to unimpaired at base line∗ at base line, N 5 years (score 0–5), n (%) at 5 years, n (%)
0 14 14 (100) 0 1 19 19 (100) 0 2 36 31 (86.1) 5 (13.9) 3 49 44 (89.8) 5 (10.2) 4 87 63 (72.4) 24 (27.6) 5 140 67 (47.9) 73 (52.1) Total 345 238 (69.0) 107 (31.0)
∗Number of correctly identified odorants; olfactory impairment is defined as <6 of 8 odorants correctly identified.
5 years of age, 95% CI = 0.68, 0.92) and in-creased with the use of nasal steroids at the baseline olfactory examination (OR = 3.19, 95% CI = 1.15, 8.90) in a model adjusted for impaired cognition, anti-anxiety medica-tion use, history of a cold, stuffy nose, or sinus problems in the 7 days before the exam, or a history of a doctor-diagnosed sinus infection or chronic sinusitis (Table 3). Therefore, older par-ticipants were much less likely to experience im-provement in their olfactory ability, while those who reported using nasal steroids at baseline were more likely to improve. Nasal steroids have been shown to have some success in improving olfactory function when prescribed for treat-ment of sinus or nasal disease.13,14
Analytical Considerations
Whenever a cut-point is assigned to a scale or score for purposes of classification, it must
be considered whether any change in classifica-tion detected with repeat testing is due only to movement around the cut-point, which could be considered measurement error, or if it is due to true change in function. To ensure the risk factors associated with olfactory improvement in these analyses were not due to bias around the cut-point, the multivariate logistic regres-sion analysis was repeated excluding those par-ticipants whose impairment classification im-proved to unimpaired by only one point (from a score of 5 to a score of 6). The results of this sensitivity subset analysis (Table 3, third col-umn) were similar to the original analysis and indicate the associations found using the full data set were not driven by measurement er-ror. Additionally, in a test–retest study (n= 90), the total agreement of the SDOIT classification was 96%, suggesting any measurement error would be minimal.11
Additionally, it is important to consider the impact of short-term versus long-term olfactory
TABLE 3. Multivariate Logistic Regression Estimates for Olfactory Improvement
Using full data set (n= 330) Using sensitivity subset∗(n= 300) Baseline covariate odds ratio†(95% confidence interval) odds ratio†(95% confidence interval) Age, effect of each 5 years older 0.79 (0.68, 0.92) 0.80 (0.68, 0.95)
Men vs. women 0.86 (0.52, 1.41) 1.00 (0.57, 1.77)
Nasal steroid use 3.19 (1.15, 8.90) 3.85 (1.30, 11.39)
†Adjusted for: upper respiratory infection, stuffy nose, sinus problems in week prior to testing; history of sinus infection or chronic sinusitis; impaired cognition; and taking anti-anxiety medications.
∗Sensitivity subset excludes participants who correctly identified five odorants at base line and six odorants at
impairment and how it may affect analyses when evaluating etiological factors related to impairment. As demonstrated in this study, in a general population there is heterogene-ity in olfactory function over time: some people stay impaired and others improve. This high-lights the need to have olfactory testing at more than one time point to determine if olfactory impairment is chronic or temporary. Fluctua-tions in olfactory status may not affect preva-lence or incidence rates, but it could still di-minish the ability to predict other outcomes and associations with olfactory impairment.
Future Directions
Preliminary analyses comparing the BOSS cohort to the EHLS population indicate that the prevalence of olfactory impairment may be decreasing in more recent generations. The prevalence of olfactory impairment for those aged 60–69 years in 1998–2000 was 17.3% (EHLS cohort), whereas the prevalence was only 9.7% in those aged 60–69 years in 2005– 2008 (BOSS cohort). Therefore, in these two cohorts, people born approximately 10 years later had less olfactory impairment at the same age. Further explorations of these data will be done to determine if birth cohort differences in lifestyle, health, or environmental factors are associated with this difference.
Conclusion
As demonstrated, it is possible to test olfac-tion in large epidemiological studies and ob-tain reliable and important information about olfactory impairment in the general popula-tion. From the EHLS it has been shown that olfactory impairment is common in aging but there may be modifiable risk factors as well as treatment opportunities. Recent preliminary data from the BOSS study indicate a lower prevalence of olfactory impairment may exist in younger cohorts (generations), and further
analyses of these data may lead to new informa-tion on the etiology of olfactory impairment.
Acknowledgments
Funded by NIH Grants AG11099 (EHLS) and AG021917 (BOSS) to K.J.C. from the Na-tional Institute on Aging.
Conflicts of Interest
The authors declare no conflicts of interest.
References
1. Bramerson, A., S. Nordin & M. Bende. 2007. Clinical experience with patients with olfactory complaints, and their quality of life. Acta Oto-Laryngol.127: 167– 174.
2. Takaki, M., M. Furukawa, T. Tsukatani, et al. 2001. Impact of olfactory impairment on quality of life and disability. Arch. Otolaryngol. Head Neck Surg.127: 497– 503.
3. Ship, J.A. & J.M. Weiffenbach. 1993. Age, gen-der, medical treatment, and medication effects on smell identification. J. Gerontol. 48: M26– M32.
4. Doty, R.L., P. Shaman, S.L. Applebaum, et al. 1984. Smell identification ability: changes with age. Science 226: 1441–1443.
5. Murphy, C., C.R. Schubert, K.J. Cruickshanks, et al. 2002. Prevalence of olfactory impairment in older adults. JAMA288: 2307–2312.
6. Deems, D.A., R.L. Doty, R.G. Settle, et al. 1991. Smell and taste disorders, a study of 750 patients from the University of Pennsylvania Smell and Taste Center. Arch. Otolaryngol. Head Neck Surg.117: 519– 528.
7. Harris, R., T.M. Davidson, C. Murphy, et al. 2006. Clinical evaluation and symptoms of chemosensory impairment: 1000 consecutive cases from the Nasal Dysfunction Clinic in San Diego. Am. J. Rhinology20: 101–108.
8. Cruickshanks, K.J., T.L. Wiley, T.S. Tweed, et al. 1998. Prevalence of hearing loss in older adults in Beaver Dam, WI: The Epidemiology of Hearing Loss Study. Am. J. Epidemiol.148: 879–886.
9. Cruickshanks, K.J., T.S. Tweed, T.L. Wiley, et al. 2003. The five-year incidence and progression of hearing loss: The Epidemiology of Hearing Loss Study. Arch. Otolaryngol. Head Neck Surg.129: 1041– 1046.
10. Murphy, C., J.A. Anderson & S. Markison. 1994. Psy-chophysical assessment of chemosensory disorders in clinical populations. In ISOT XI . K. Kurihara, Ed.: 609–613. Springer Verlag. Tokyo, Japan.
11. Krantz, E.M., C.R. Schubert, D.S. Dalton, et al. 2008. Repeatability of the San Diego Odor Iden-tification Test (SDOIT) and Comparison with the Brief Smell Identification Test (B-SIT). International Symposium on Olfaction and Taste, San Francisco, CA, July 24.
12. Schubert, C.R., L.L. Carmichael, C. Murphy, et al.
2008. Olfaction and the 5-year incidence of cogni-tive impairment in an epidemiological study of older adults. JAGS56: 1517–1521.
13. Wolfensberger, M. & T. Hummel. 2002. Anti-inflammatory and surgical therapy of olfactory dis-orders related to sino-nasal disease. Chem. Senses27: 617–622.
14. Mott, A.E., W.S. Cain, D. Lafreniere, et al. 1997. Topical corticosteroid treatment of anosmia associ-ated with nasal and sinus disease. Arch. Otolaryngol.
