Professional Documents
Culture Documents
doi:10.1093/chemse/bjx070
Original Article
Advance Access publication November 8, 2017
Original Article
Correspondence to be sent to: Jason Chan, Department of Otorhinolaryngology, Head and Neck Surgery, Room 84026, 6/F Lui
Che Woo Clinical Sciences Building, Prince of Wales Hospital, Shatin, N.T., Hong Kong SAR. e-mail: jasonchan@ent.cuhk.edu.hk
Abstract
Diabetes is a significant chronic disease that in limited studies has been linked with olfactory
dysfunction. We investigated the cross-sectional association between diabetes and olfactory
dysfunction in 3151 adults aged ≥40 years who participated in US National Health and Nutrition
Examination Survey 2013–2014 with information on olfactory dysfunction and diabetes. Diabetes
was defined from fasting serum glucose ≥126 mg/dL, oral glucose tolerance test ≥200 mg/dL,
HbA1c levels ≥6.5%, physician-diagnosed diabetes, or current use of oral hypoglycemic agents
and/or insulin. Self-reported olfactory dysfunction was defined as a positive answer to any of the
following questions: 1) “Have you had problem with smell in the past 12 months?”; 2) “Have
you had a change in the ability to smell since age 25?”, or 3) “Do you have phantom smells?”.
Participants were considered to have severe hyposmia or anosmia if they had <5 correct answers in
the 8-item pocket smell test. Analyses were adjusted for the main confounders, including olfactory
dysfunction risk factors. Compared to non-diabetics, diabetics under insulin treatment showed a
higher prevalence of phantom odors [OR(95% CI): 2.42 (1.16; 5.06)] and a non-significant higher
prevalence of severe hyposmia/anosmia [OR(95% CI): 1.57 (0.89; 2.78)]. Amongst diabetics, there
was a significant trend to severe hyposmia/anosmia for those on more aggressive treatments
[OR (95% CI) including oral and insulin treatment compared to those who reported no use of drug
treatment, respectively: 1.33 (0.60; 2.96) and 2.86 (1.28; 6.40); P trend 0.01]. No association was
observed between diabetes duration and prevalence of olfactory dysfunction.
© The Author(s) 2017. Published by Oxford University Press. All rights reserved.
59
For permissions, please e-mail: journals.permissions@oup.com
that is of significant burden to both patients and society, is a pro- diabetics who reported no use of oral hypoglycemic agents and/or
cess that requires significant commitment from the patient including insulin, diabetics treated with blood glucose lowering drugs, and
lifestyle modifications, weight loss, and compliance with medication diabetics treated with insulin regardless of their use of blood glucose
(Inzucchi et al. 2015). lowering drugs. Participants with self-reported diagnosed diabetes
Olfactory dysfunction indicates abnormalities in the conduction were asked the age when first told they had diabetes.
of olfactory stimuli externally to the olfactory epithelium or the sen-
sory pathways arising from the olfactory epithelium. Manifestations Olfactory dysfunction
of olfactory dysfunction may include phantosmia—the identifica- Self-reported olfactory dysfunction was defined as a positive answer
tion of smells in the absence of an external stimuli, parosmia—the to any of the following 3 questions: 1) “Have you had problem
identification of a smell that is incongruent with the external stimuli, with smell in the past 12 months?”; 2) “Have you had a change in
hyposmia—a diminished ability to detect and identify smells, and the ability to smell since age 25?” and 3) “Do you have phantom
anosmia—the inability to detect or identify smells. Amongst patients smells?”( CDC 2013; Rawal et al. 2016)
with diabetes, data suggests that olfactory dysfunction, alterations of The 8-item pocket smell test was used to evaluate for the presence of
the sensitivity of smell, may be a degenerative complication of diabe- anosmia or severe hyposmia (Sensonics International). The pocket smell
tes (Le Floch et al. 1993; Weinstock et al. 1993; Palouzier-Paulignan test has been shown to be sensitive in identifying moderate to severe
et al. 2012; Varkonyi et al. 2014; Gouveri et al. 2014; Khil et al. 2015; olfactory dysfunction and have good test–retest reliability (Rawal et al.
Mehdizadeh Seraj et al. 2015; Duda-Sobczak et al. 2017), secondary 2015). According to this test, participants were considered to have
to the effects of this disease on the central nervous system (Biessels severe hyposmia or anosmia if they had less than 5 correct answers.
2013; Bruce et al. 2013). There may also be reduced olfactory function
in type 2 diabetics when compared with patients with type 1 diabe- Sociodemographic variables, lifestyle, obesity, and reported
tes (Palouzier-Paulignan et al. 2012; Gouveri et al. 2014). If olfactory comorbidity
dysfunction in diabetics does result from central nervous system dys- Sociodemographic variables included age, gender, ethnicity (non-
function, similar to the cognitive impairment and depression noted in Hispanic White, non-Hispanic Black, Mexican American, Other),
diabetes (Biessels 2013; Bruce et al. 2013), this suggests that olfactory and education level (< high school, high school or equivalent, > high
testing may be helpful in the early recognition of diabetic complica- school). Participants were classified as never smokers if they had
tions (Brady et al. 2013; Gascon et al. 2013; Sanke et al. 2014). not smoked ≥100 cigarettes during a lifetime, and as ever smokers
Recently, the US National Health and Nutrition Examination if they had done so but were not current smokers at the time of
Survey (NHANES) included a smell component in the 2011–2014 the interview. Participants were asked about their previous history
chemosensory protocol using a standardized rapid smell assess- of cardiovascular disease (congestive heart failure, coronary heart
ment and smell questionnaire, with the data from the 2013–2014 disease, heart attack, stroke), hypertension, head, or face injury,
cycle readily available. A previous study of the NHANES has shown xerostomia (persistent dry mouth in past 12 months) and sinonasal
increased risks for olfactory dysfunction dependent on age, eth- symptoms (persistent cold/flu/nasal congestion in past 12 months,
nicity, socioeconomic status, education and alcohol consumption presence of 2 or more sinus infections). Hypertension was defined
(Hoffman et al. 2016). Here, the aim of this study was to evaluate as a systolic blood pressure ≥140 mmHg, a diastolic blood pressure
the association between diabetes and olfactory dysfunction in the US ≥90 mmHg, or the current use of antihypertensive drugs. The body
general population using NHANES. mass index (BMI) was calculated as the weight in kg divided by the
square height in metres.
Methods
Statistics
Study design and population
All statistical analyses were performed in STATA version 11.2 sta-
Between 2013 and 2014, the NHANES examined a US nationally
tistical software (Stata Corp), using the survey (svy) command to
representative sample of 14 332 people by using a complex multi-
account for the complex sampling design and weights in NHANES.
stage sample design. Of those selected, 10 175 completed an inter-
The prevalence of olfactory dysfunction was calculated overall and
view. From these 10 175 participants, we excluded 6360 who were
by sex, age, ethnicity, education, tobacco consumption, BMI, pres-
younger than 40 years, 305 without complete information on olfac-
ence of sinonasal symptoms, xerostomia, previous history of head/
tory function, 147 with no information on diabetic status, and 212
face injury, hypertension, or cardiovascular disease.
with no information on potential confounders, leading to a sample
To evaluate the association between diabetes and olfactory dysfunc-
size of 3151 participants. The NHANES protocol was reviewed and
tion, odds ratios (OR) and 95% confidence intervals (95% CI) were
approved by the National Center for Health Statistics Institutional
estimated using logistic regression. Three models were built. In the basic
Review Board. Written informed consent was given by the partici-
model, we considered age, gender, and race. We then fitted 2 models
pants for their clinical records to be used in a research study.
by progressively adding: 1) potential confounders of the association
between diabetes and olfactory dysfunction (educational level, smoking
Study variables status, presence of sinonasal symptoms, xerostomia, previous history of
The study variables and names are available in Supplementary head, or face injury); and 2) potential mediators of diabetes microvascu-
Table 1. lar and macrovascular complications (hypertension, obesity, and cardio-
vascular disease). In sensitivity analyses, we also adjusted for additional
Diabetes mellitus sociodemographic factors (marital status, poverty to income ratio)
Participants were considered diabetic if they had fasting serum glu- and cardio-metabolic biomarkers (high density lipoprotein (HDL) and
cose ≥126 mg/dL, oral glucose tolerance test ≥200 mg/dL, measured low density lipoprotein (LDL) cholesterol), with similar findings. The
HbA1c levels ≥6.5%, self-reported diagnosis of diabetes, or cur- association between diabetes duration and olfactory dysfunction was
rent use of oral hypoglycemic agents or insulin. Study participants assessed using logistic regression models with the same confounders as
were categorized into 4 mutually exclusive groups: non-diabetics, mentioned above. Statistical significance was set at 2-sided P < 0.05.
Table 1. Odds ratios and 95% confidence intervals for the association between diabetes and olfactory dysfunction
Model 1: Logistic regression model adjusted for age (continuous), gender, and race. Model 2: As model 1 and additionally adjusted for educational level, smok-
ing status, sinonasal symptoms, xerostomia, and previous history of head or face injury. Model 3: As model 2 and additionally adjusted for potential mediators:
BMI, hypertension and cardiovascular disease. CI, confidence interval.
a
Self-reported olfactory dysfunction was defined as a positive answer to any of the following questions: 1) Have you had problem with smell in the past
12 months?; 2) Have you had a change in the ability to smell since age 25? or; 3) Do you have phantom smells?
b
Measured olfactory dysfunction (presence of severe hyposmia or anosmia) was defined as having less than 5 correct answers in the 8 item pocket smell test.
Table 2. Odds ratios and 95% confidence intervals for the association between diabetes treatment and olfactory dysfunction
Model 1: Logistic regression model adjusted for age (continuous), gender, race, and duration of diabetes. Model 2: As model 1 and additionally adjusted for
educational level, smoking status, sinonasal symptoms, xerostomia, and previous history of head or face injury. Model 3: As model 2 and additionally adjusted for
potential mediators: BMI, hypertension and cardiovascular disease. CI, confidence interval.
a
Self-reported olfactory dysfunction was defined as a positive answer to any of the following questions: 1) Have you had problem with smell in the past
12 months?; 2) Have you had a change in the ability to smell since age 25? or; 3) Do you have phantom smells?
b
Measured olfactory dysfunction (presence of hyposmia or anosmia) was defined as having less than 5 correct answers in the 8-item pocket smell test.
association observed between diabetes duration and prevalence of be associated with significant atrophy of cortical and subcortical
hyposmia/anosmia as shown in Supplementary Table 3). regions of the brain, accelerated cognitive decline (McCrimmon
In sensitivity analyses, additional adjustments for race, poverty et al. 2012; Musen et al. 2012), and neural connectivity abnormali-
income ratio, total serum cholesterol or serum high-density lipopro- ties (Cui et al. 2015). Phantosmia is thought to be related to reduced
tein cholesterol, did not substantially change our results (data not grey-matter volume at the insular cortex, cingulate cortex, hippo-
shown in tables). campus, and orbitofrontal cortex (Bitter et al. 2011), and so dia-
betes could increase its risk by also affecting the orbitofrontal cortex
(Ajilore et al. 2010). Second, macrovascular disease may explain the
Discussion increased prevalence of severe olfactory dysfunction. In this sense,
To our knowledge, this is the first nationally representative study to controlling for cardiovascular disease and hypertension attenuated
evaluate the association between diabetes and olfactory function in the effect of diabetes on phantosmia. However, macrovascular dis-
the United States. Results suggest that, compared to non-diabetics, ease did not explain the increased risk of severe olfactory dysfunction
diabetics on insulin medication have a higher prevalence of phantom noted in diabetics. Third, insulin resistance in the olfactory bulb and
odors, severe hyposmia, and anosmia. Additionally, they show that, orbitofrontal cortex potentially could also play a role in olfactory
amongst diabetics, there is a significant trend to severe hyposmia and dysfunction (De Felice et al. 2014). In this sense, there is evidence
anosmia for those on more aggressive treatments. However, they do that insulin regulates the olfactory mucosa physiology (Lacroix et al.
not suggest an association between duration of diabetes and devel- 2008) and that intranasal insulin may improve olfactory sensitiv-
opment of olfactory dysfunction. ity and discrimination (Schopf et al. 2015). This third mechanism is
Several mechanisms may explain the observed associations. First, supported by the fact that the odds of severe olfactory dysfunction
hyperglycaemia has been shown to result in increased cortical thin- was more likely in diabetics on insulin as opposed to those treated
ning of the orbitofrontal cortex, and type 2 diabetes has shown to conservatively or oral medication.
Lacroix MC, Badonnel K, Meunier N, Tan F, Schlegel-Le Poupon C, Durieux Rawal S, Hoffman HJ, Bainbridge KE, Huedo-Medina TB, Duffy VB. 2016.
D, Monnerie R, Baly C, Congar P, Salesse R, et al. 2008. Expression of Prevalence and risk factors of self-reported smell and taste altera-
insulin system in the olfactory epithelium: first approaches to its role and tions: results from the 2011-2012 US National Health and Nutrition
regulation. J Neuroendocrinol. 20:1176–1190. Examination Survey (NHANES). Chem Senses. 41:69–76.
Le Floch JP, Le Lièvre G, Labroue M, Paul M, Peynegre R, Perlemuter L. 1993. Rawal S, Hoffman HJ, Honda M, Huedo-Medin TB, Duffy VB. 2015. the taste
Smell dysfunction and related factors in diabetic patients. Diabetes Care. and smell protocol in the 2011-2014 US National Health and Nutrition
16:934–937. Examination Survey (NHANES): Test-Retest Reliability and Validity
Manschot SM, Brands AM, van der Grond J, Kessels RP, Algra A, Kappelle LJ, Testing. Chemosens Percept. 8:138–148.
Biessels GJ; Utrecht Diabetic Encephalopathy Study Group. 2006. Brain Roy T, Lloyd CE. 2012. Epidemiology of depression and diabetes: a systematic
magnetic resonance imaging correlates of impaired cognition in patients review. J Affect Disord. 142 Suppl:S8–21.
with type 2 diabetes. Diabetes. 55:1106–1113. Sanke H, Mita T, Yoshii H, Yokota A, Yamashiro K, Ingaki N, Onuma T,
McCrimmon RJ, Ryan CM, Frier BM. 2012. Diabetes and cognitive dysfunc- Someya Y, Komiya K, Tamura Y, et al. 2014. Relationship between olfac-
tion. Lancet. 379:2291–2299. tory dysfunction and cognitive impairment in elderly patients with type 2
Mehdizadeh Seraj J, Mehdizadeh Seraj S, Zakeri H, Bidar Z, Hashemi S, diabetes mellitus. Diabetes Res Clin Pract. 106:465–473.
Mahdavi Parsa F, Yazdani N. 2015. Olfactory dysfunction in Iranian dia- Schöpf V, Kollndorfer K, Pollak M, Mueller CA, Freiherr J. 2015. Intranasal
betic patients. Acta Med Iran. 53:204–206. insulin influences the olfactory performance of patients with smell loss,
Menke A, Casagrande S, Geiss L, Cowie CC. 2015. Prevalence of and dependent on the body mass index: a pilot study. Rhinology. 53:371–378.
trends in diabetes among adults in the United States, 1988-2012. JAMA. Serby M, Larson P, Kalkstein D. 1991. The nature and course of olfactory
314:1021–1029. deficits in Alzheimer’s disease. Am J Psychiatry. 148:357–360.
Mesholam RI, Moberg PJ, Mahr RN, Doty RL. 1998. Olfaction in neurode- Tiehuis A, van der Meer F, Mali W, Pleizier M, Biessels GJ, Kappelle J,
generative disease: a meta-analysis of olfactory functioning in Alzheimer’s Luijten P. 2010. MR spectroscopy of cerebral white matter in type 2 dia-
and Parkinson’s diseases. Arch Neurol. 55:84–90. betes; no association with clinical variables and cognitive performance.
Musen G, Jacobson AM, Bolo NR, Simonson DC, Shenton ME, McCartney Neuroradiology. 52:155–161.
RL, Flores VL, Hoogenboom WS. 2012. Resting-state brain functional Várkonyi T, Körei A, Putz Z, Kempler P. 2014. Olfactory dysfunction in dia-
connectivity is altered in type 2 diabetes. Diabetes. 61:2375–2379. betes: a further step in exploring central manifestations of neuropathy?
Palouzier-Paulignan B, Lacroix MC, Aimé P, Baly C, Caillol M, Congar P, Angiology. 65:857–860.
Julliard AK, Tucker K, Fadool DA. 2012. Olfaction under metabolic influ- Weinstock RS, Wright HN, Smith DU. 1993. Olfactory dysfunction in diabetes
ences. Chem Senses. 37:769–797. mellitus. Physiol Behav. 53:17–21.