Eur J Nutr DOI 10.1007/s00394-015-0921-2

ORIGINAL CONTRIBUTION

Olfactory impairment in older adults is associated with poorer diet quality over 5 years Bamini Gopinath1 · Joanna Russell2 · Carolyn M. Sue3 · Victoria M. Flood4 · George Burlutsky1 · Paul Mitchell1 

Received: 12 January 2015 / Accepted: 4 May 2015 © Springer-Verlag Berlin Heidelberg 2015

Abstract  Purpose  Decreased smell could cause appetite suppression and malnutrition. However, there is a paucity of longitudinal data between olfaction and nutritional status in older adults. We aimed to prospectively examine the relationship between olfactory impairment and overall diet quality (reflecting adherence to dietary guidelines) in a population-based cohort of older adults. Methods  We used 5-year follow-up data from 557 adults (aged 60+ years at baseline) whose olfaction was measured using the San Diego Odor Identification Test (SDOIT). Dietary data were collected using a validated semiquantitative food frequency questionnaire. A total diet score (TDS) was calculated for intake of selected food groups and nutrients for each participant as described in the national dietary guidelines. Final scores ranged from 0 to 20; higher scores indicated closer adherence to dietary guidelines. Electronic supplementary material  The online version of this article (doi:10.1007/s00394-015-0921-2) contains supplementary material, which is available to authorized users. * Bamini Gopinath [email protected] 1

Centre for Vision Research, Department of Ophthalmology and Westmead Millennium Institute, University of Sydney, Sydney, Australia

2

School of Health and Society, University of Wollongong, Wollongong, Australia

3

Departments of Neurology and Neurogenetics, Kolling Institute, University of Sydney, Sydney, Australia

4

Faculty of Health Sciences, University of Sydney and St Vincent’s Hospital, Sydney, Australia

5

Centre for Vision Research, Westmead Hospital, Westmead, NSW 2145, Australia





Results  After adjusting for all potential confounders, older adults with moderate/severe olfactory impairment (SDOIT score ≤ 3; lower scores indicate impairment) compared with those with no olfactory impairment had significantly lower adjusted mean (±SE) TDS, 9.09 (0.40) versus 9.94 (0.10), p  = 0.04. Women with moderate/severe impaired olfaction (i.e., scored poorly on the odor identification test) compared with those with normal olfaction had significantly lower adjusted mean TDS, 8.87 (0.69) versus 10.31 (0.13), p  = 0.04. No associations were observed between olfaction and TDS in men. Conclusions  Olfactory impairment in older women could signal an increased risk of poorer diet quality, defined as adherence to national dietary guidelines. Additional longitudinal studies are needed to confirm or refute the observed link between olfactory loss and overall patterns of food intake in older adults. Keywords  Olfactory impairment · Diet quality · Older adults · Blue Mountains Eye Study

Introduction A decrease in olfactory function with increasing age is well established in the literature [1, 2]. We previously showed that among older adults aged 60 and over, 27.0 % had olfactory impairment [3]. While the prevalence of impaired olfaction in the ‘healthy’ aging population could be greater than previously thought [3], many older people may not raise the issue of their loss of the sense of smell because they are not always aware of it [4, 5]. Moreover, the potential importance of olfaction is not well documented, compared with that of other senses such as vision and hearing. Although it is easy to appreciate the role of hearing and

13



seeing in humans, the sense of smell encompasses a wide range of functions [5, 6]. Olfaction plays an important role in eating habits and nutritional intake [7]. Decreased smell has been shown to result in appetite suppression leading to weight loss and malnutrition [5, 8]. However, there is little research on the impacts of olfactory dysfunction on dietary intakes in older adults, and results are contradictory [9] and largely cross-sectional. A Swedish clinic-based study showed that patients who scored lower on the Brief Smell Identification Test had lower intakes of protein, folate, magnesium, and phosphorous [10]. In contrast, Ferris et al. [11] showed in a cross-sectional sample of hyposmic and anosmic patients that although enjoyment of food was lower and eating habits changed, nutrient intakes were found to be adequate. Recently, a German cross-sectional study of 191 geriatric patients showed that neither subjective nor objective olfactory dysfunction was associated with impaired nutritional status in older adults [9]. To our best knowledge, there have been no longitudinal population-based studies that have assessed the association between olfactory impairment (including severity of the impairment) and overall patterns of food intake in older adults. Hence, to address this gap in knowledge, we used a cohort of adults aged 60+ years to determine whether olfaction influences diet quality (an assessment of how well an individual’s diet adhered to national dietary guidelines) over 5 years. Additionally, we assessed whether the severity of olfactory dysfunction is related to diet quality independent of potential confounders in the longer term.

Eur J Nutr

and 1149 participants (55.4 % of survivors) with complete data were re-examined, respectively. In the current study, we used 5-year follow-up data from BMES-3 to BMES-4, given that olfactory data were only obtained from BMES-3 onwards. The University of Sydney and the Western Sydney Area Human Ethics Committees approved the study, and written informed consent was obtained from all participants at each examination. Olfaction examination The San Diego Odour Identification Test (SDOIT) [13] and related olfactory and taste questions were a component of the BMES-3 examination, and complete olfaction and taste data were obtained from 1636 of 1952 (83.8 %) BMES-3 participants. Participants were tested individually with the SDOIT, an 8-item odor identification test with a test–retest reliability relatively similar to that for the 40-item University of Pennsylvania Smell Identification Test (UPSIT) (r  = 0.86 SDOIT; r  = 0.94 UPSIT) [14]. Odorants were presented to participants in random order, in an opaque container covered with gauze. An inter-stimulus pause of 45 s was used to prevent adaptation [15]. A picture board illustrating the odorants as well as distracters was used for participants to identify each odorant. Scores were calculated from the number of odorants identified correctly. We defined mild olfactory impairment as less than six but greater than three correct responses and moderate/severe as three or less correct responses out of a total of eight possible responses. Dietary assessment

Methods Study population The Blue Mountains Eye Study (BMES) is a populationbased cohort study of common eye diseases and other health outcomes in a suburban Australian population located west of Sydney. Study methods and procedures have been described elsewhere [12]. Participants were non-institutionalized residents who were invited to attend a detailed baseline eye examination after a door-to-door census of the study area. Selection bias at baseline was minimized after multiple callback visits, including doorknocking, telephone reminders, and letters at recruitment. Baseline examinations of 3654 residents aged >49 years were conducted during 1992–1994 (BMES-1, 82.4 % participation rate). Surviving baseline participants were invited to attend examinations after 5 years (1997–1999, BMES-2), 10 years (2002–2004, BMES-3), and 15 years (2007–2009, BMES-4). At BMES-2, BMES-3, and BMES4, 2334 (75.1 % of survivors), 1952 (75.6 % of survivors)

13

Dietary data were collected using a 145-item self-administered food frequency questionnaire (FFQ), modified for Australian diet and vernacular from an early Willett FFQ [16], and included reference portion sizes. Participants used a nine-category frequency scale to indicate the usual frequency of consuming individual food items during the past year. The FFQ was validated by comparing nutrients from the FFQ to 4-day-weighed food records collected over 1 year (n  = 79). Most nutrient correlations were between 0.50 and 0.60 for energy-adjusted intakes, similar to other validated FFQ studies [17, 18]. A dietitian coded data from the FFQ into a customized database that incorporated the Australian Tables of Food Composition 1990 (NUTTAB 90) [19]. A modified version of the Australian diet quality index [20], based on the Dietary Guidelines for Australian Adults [21] and the Australian Guide to Healthy Eating [22], was used to establish the total diet score (TDS), assessing adherence to the Australian dietary guidelines. In the current study, TDS data obtained both at BMES-3 and BMES-4

Eur J Nutr

(i.e., 5-year follow-up data) were analyzed. The methodology used to develop TDS has been previously reported [23]. Briefly, intakes of selected food groups and nutrients for each participant as described in the Dietary Guidelines for Australian Adults comprised the TDS (details for FFQ food groupings available in Supplementary Table 1). The TDS is divided into ten components, and each component has a possible score ranging from 0 to 2. A maximum score of 2 was given to subjects who met the recommendations with prorated scores for lower intakes. These were then summed providing a final score ranging between 0 and 20 with higher scores indicating closer adherence to the dietary guidelines. The TDS accounts for both food intake and optimal choice with scores allocated to reflect intake characteristics from both sources. Food intake scores were based on total intakes of the following components: (1) vegetables and fruit; (2) cereals and breads; (3) meat, fish, poultry and/or alternatives; (4) dairy (5) sodium; (6) alcohol; (7) sugar, and (8) extra or discretionary foods intakes. Optimal choices scores determined intakes of foods with greater dietary benefits including servings of whole grain cereals, lean red meat, low- or reduced-fat milk versus whole milk, low saturated fat intake, and fish consumption. Alcohol cutpoints refer to the Australian recommendations for alcohol consumption per day (e.g., two standard drinks for men and one standard drink for women) [21]. For men, up to two standard drinks is the recommend daily intake of alcohol, and therefore if participants consumed two or less drinks per day, they would be given a score of 2 and a 0 score if they drank more than two standard alcoholic drinks per day. For women, one standard drink per day was recommended, and therefore female participants received a maximum of two points if they drank one or less standard drinks per day; >1–2 drinks per day was scored as one point, and alcoholic drinks greater than two standard drinks per day scored as 0. To follow dietary guideline recommendations as closely as possible, the non-dietary component of the Australian Guide to Healthy Eating, preventing weight gain, was included in the total dietary score as the tenth component. Half the score was assigned to energy balance, calculated as the ratio of energy intake to energy expenditure with a maximum score of 1 given for ratios falling between 0.76 and 1.24 (and 0 was given for scores 1.24), defined as the 95 % confidence levels of agreement between energy intake and expenditure [24]. The other half of the score was assigned to leisure time physical activity. Details of walking exercise and the performance of moderate or vigorous activities were used to calculate metabolic equivalents (METs) [25]. Subjects in the highest METs tertile scored one point reducing to a zero point score for subjects in the lowest METs tertile (Supplementary Table 1).

Collection of other information Information on covariates collected at BMES-3 (or baseline in the context of this study) was included in all analyses. At face-to-face interviews with trained interviewers, a comprehensive medical history that included information about socio-demographic factors such as level of education, living status, and receipt of pension payment was obtained from all participants. Body mass index (BMI) was calculated as weight divided by height squared (kg/m2). Statistical analysis SAS statistical software (SAS Institute, Cary, NC, USA) version 9.2 was used for analyses including t tests, Chisquare tests, and linear regression. Mean TDS (assessed as a continuous variable) was the dependent variable, and presence and severity of olfactory impairment (assessed as categorical variable) was the independent variable. Analysis of covariance (general linear model, GLM) was used to assess associations between olfactory impairment at baseline with adjusted mean TDS 5 years later. That is, adjusted mean TDS was compared across severity levels of olfactory impairment and the presence/absence of olfactory loss at baseline. Analyses were first adjusted for age and sex and then further adjusted for: education level, living status, receipt of pension, and BMI. Standardized beta coefficients and corresponding SE for all covariates included in the multivariable model is presented in Supplementary Table 2. After multivariable adjustment, ANCOVA analyses did not indicate significant interactions between sex and the association of severity of olfactory impairment with mean TDS, P interaction = 0.32. However, we still stratified the analyses by gender as there is biological plausibility that sex-specific associations regarding olfactory impairment were previously observed in our cohort [3].

Results Of the 1636 participants at BMES-3 (baseline) who had SDOIT data, 557 participants had complete dietary data at both baseline (BMES-3) and 5 years later (BMES-4), and thus, were included for longitudinal analyses. Table 1 compares study characteristics between participants versus nonparticipants, i.e., those who had died over the 5 years or who were lost to follow-up or had missing data (either dietary data or SDOIT information). Participants versus nonparticipants were more likely to be younger, have tertiary qualifications and higher mean TDS, but less likely to live alone and receive a pension. Also, participants compared with those who were not followed up or had missing data were more likely to be male, and participants compared

13



Eur J Nutr

Table 1  Comparison of study characteristics between participants and non-participants in the Blue Mountains Eye Study Participants (n = 557)

Died (n = 320)

Not followed up or had missing data (n = 1075)

p value

Age mean (SD), years

70.7 (6.5)

79.7 (7.9)

73.8 (7.5)