Consequences of undetected olfactory loss for human chemosensory communication and well-being

Abstract

Olfactory perception has implications for human chemosensory communication and in a broader context, it affects well-being. However, most of the studies investigating the consequences of olfactory loss have recruited patients who have already been categorized as having a dysfunctional sense of smell and sought help in an ENT clinic. We revisit these findings by distinguishing subjects with olfactory impairment from a group of subjects who all declared a normal sense of smell when enrolling for this study. In the initial sample of 203 individuals, we found 59 to have impaired olfaction and four with marginal olfactory performance, not useful in daily life. Interestingly, we found a significant between-group difference in cognitive functioning, further supporting the notion of the relationship between cognition and olfactory performance. However, their chemosensory communication and well-being appeared not to be different from subjects with normosmia. Impaired olfactory function certainly has a severe impact on daily life but more so in individuals who are bothered with it and decide to seek treatment. The limited-to-no olfactory perception in the fraction of subjects who neither complain about it nor seek help in ENT clinics does not seem to have a major effect on their social, cognitive, emotional and health functioning.

This article is part of the Theo Murphy meeting issue ‘Olfactory communication in humans’.

1. Introduction

The importance of olfaction in daily routine has been commonly highlighted by researchers. The ability to memorize and decode emotions in humans is derived from structures primarily devoted to processing smells [1]. Brain structures engaged in emotion recognition overlap with those used to process olfactory information (e.g. amygdala, orbitofrontal cortex) [2]. Without the presence of odours, positive emotions are recognized faster than negative emotions, but this difference becomes insignificant in the presence of unpleasant odour [3], suggesting an interaction between the two inputs in the limbic system. The olfactory-visual compensatory effect on emotion recognition has recently been reported. Anosmic individuals are able to more quickly and accurately recognize emotions of fear and disgust in other people and this effect is related to the duration of olfactory impairment. This supra-skill is thought to result from the learning process, wherein anosmic individuals, unable to acquire alerts about environmental threats (e.g. rotten food, fire smoke) via the olfactory system, become reliant on the potentially alerting facial expressions of other people [4].

Olfactory loss undermines romantic and sexual activity. Anosmia enhances social insecurity but the consequences depend on sex. Men declare fewer sexual relationships while women feel less secure about their partner [5]. Furthermore, the inability to perceive body odour precludes anosmic patients a wide variety of social-relevant information, such as emotional conditions of others, e.g. fear [6–9] and happiness [10]. Body odour constitutes an individual and sex-specific fingerprint [11–13] that enables women to recognize a partner or a close friend [14]. Body odour is informative of psychological traits, including dominance and neuroticism [15,16]. Furthermore, human sweat is also indicative of a partner's genetic compatibility [17–22] and has been found to be an important factor in family bonding [23–25]. In summary, olfaction plays an important role in regulating human social interactions and has been found to be related to individual health, cognitive and emotional functioning that we further operationalize as ‘well-being’.

The evidence for the importance of olfaction in human social communication and well-being suggests the critical role of a fully functional sense of smell. As a consequence, research including subjects with an olfactory impairment has focused largely on the negative consequences of diminished olfactory perception [26–34]. However, in our olfactory screening practice within studies recruiting samples of normosmic subjects we usually tend to observe a few volunteers presenting with impaired olfactory performance. These subjects are convinced that their sense of smell is normal, yet in the screening test they score in the range of hyposmia or even functional anosmia [35]. This means that subjects who have severely impaired olfactory perception (hyposmia) or even have no functional sense of smell (anosmia) are not able to acquire odour-based social cues or even to detect environmental threats such as gas, fire smoke or rotten food. Despite this, they consider themselves healthy and volunteer for studies with clearly set inclusion criteria of a normal sense of smell. Thus, at least part of the population with severely impaired olfaction is unaware of the lack of sensory input, or is aware of but not bothered with it. In the broader sense, this observation suggests that smell loss might not be as disabling as we tend to assume, at least not in everybody with olfactory loss.

Diminished olfactory perception has implications for human social communication and well-being. However, most of the studies looking into consequences of olfactory loss recruited patients who have already been categorized as having a dysfunctional sense of smell and sought help in an ENT clinic [33,34,36]. We revisited these findings by distinguishing subjects with olfactory impairment from a group of subjects who all declared a normal sense of smell and enrolled in a study looking into indicators of human social communication and well-being (measured as health, cognitive and emotional functioning). We post hoc categorized subjects to a normosmic and an impaired olfaction group (including subjects with hyposmia and functional anosmia) on the basis of their Sniffin’ Sticks scores and compared results of those who scored low (in the range of hyposmia and functional anosmia) with those who scored in the normosmic range. Our main objective was to assess a range of socially relevant factors, which have previously been shown to be affected among patients who present themselves to an ENT department, in this interesting sample of subjects who appear to be unaware of their olfactory deficit.

2. Material and methods

(a). Ethics statement

The study was performed in accordance with the Declaration of Helsinki on Biomedical Studies Involving Human Subjects. Informed written consent was obtained from all subjects who were additionally assured they could withdraw their consent at any time without giving reasons. The study design and consent approach were approved by the Ethics Review Board at the TU Dresden (EK167052018).

(b) Participants

Power analysis was performed by the means of G * Power software [37] to estimate a required sample size. In order to compare two groups and control for age with the statistical power of 1 − β = 0.80, we were able to detect a small-to-medium effect size of f = 0.20 with a sample size of 199 subjects. Two hundred and three subjects were enrolled in the study. The convenience sampling was mostly performed in a rural area in the region of Brandenburg, Germany, in a city with approximately 1000 inhabitants. Among the subjects, 134 were women aged from 18 to 92 (M_age = 47 ± 1.7 years) and 69 were men aged from 18 to 86 (M_age = 45.1 ± 2.1 years). In the initial sample, we found four subjects whose Sniffin’ Sticks TDI score (TDI = threshold + discrimination + identification; table 1) indicated functional anosmia (less than or equal to 16 points) and 59 subjects whose score indicated hyposmia (score in the range of 16.25–30.75 points). Therefore, for the following analyses, we combined the two groups, resulting in 63 subjects with an impaired sense of smell. Their descriptive information is presented in table 1.

Table 1.

Descriptive statistics for normosmic (n = 140), hyposmic (n = 59) and functionally anosmic (n = 4) subjects in the total sample. Note: BMI, body mass index; TDI, threshold, discrimination, identification—the total score of the three subtests of the Sniffin’ Sticks; PNIF, peak nasal inflammatory flow (l min⁻¹).

		M	s.e.	95% confidence interval		min	max
				lower	upper
age (years)	normosmia	40.32	1.38	37.59	43.06	18	82
	hyposmia	59.59	2.41	54.76	64.43	21	92
	functional anosmia	62.25	12.22	23.35	101.15	28	86
threshold	normosmia	10.07	0.20	9.67	10.46	4.5	15.5
	hyposmia	5.84	0.32	5.19	6.49	1.5	12.5
	functional anosmia	1.13	0.13	0.73	1.52	1	1.5
discrimination	normosmia	12.49	0.15	12.21	12.78	8	16
	hyposmia	9.80	0.29	9.21	10.38	5	15
	functional anosmia	4.25	1.44	−0.32	8.82	0	6
identification	normosmia	13.42	0.11	13.21	13.64	9	16
	hyposmia	10.47	0.31	9.85	11.10	5	15
	functional anosmia	5.25	1.38	0.87	9.63	2	8
TDI	normosmia	35.98	0.25	35.49	36.47	30.75	43.5
	hyposmia	26.11	0.46	25.18	27.04	18.5	30.5
	functional anosmia	10.63	2.64	2.22	19.03	3	15
PNIF	normosmia	86.9	2.828	81.29	92.5	40	180
	hyposmia	88.07	5.187	77.68	98.46	40	250
	functional anosmia	60	8.165	34.02	85.98	40	80

Olfactory impairment was independent from sex, ${\chi }_{(1)}^2=0.68,$ p = 0.41. Subjects with impaired olfaction (M_age = 59.8 ± s.e._age = 2.4 years) were significantly older than subjects with normosmia (M_age = 40.3 ± 1.4 years), t_106.3 = −7.1, p < 0.001 (corrected for unequal variance). Therefore, for the final analyses, we included age as a covariate.

(c) Procedure

Subjects were tested during individual sessions with the same experimenter (FK). Sessions were performed in an empty office space with two rooms—room 1 was meant to provide a quiet and private atmosphere for the individual work with the questionnaires and room 2 was used for olfactory testing (odourless space), MOCA (Montreal Cognitive Assessment) and digit span. Each session began with the standard medical interview [38]. The interview includes information on general health, past medical history and factors that could potentially affect olfactory performance (such as head injuries, asthma or chronic diseases). Within the standard medical interview subjects were asked to self-rate their smell abilities on a Likert-type scale ranging from 1—very bad to 7—very good [38]. They answered a series of two-alternative-forced choice (yes versus no) questions about their ability to detect environmental smells (odorous and trigeminal) likely to be encountered in daily life: cigarette smoke, meals, rotten foods, perfume, alcohol, coffee, spices, vinegar, gasoline, flowers and smoke. Subjects were asked to complete the following questionnaires referring to olfactory, cognitive and well-being domains (respectively):

• (1)Individual significance of olfaction [39] consisting of 18 items divided into three subscales: ‘Association’ refers to emotions, memories and evaluations that are triggered by olfactory experiences, ‘Application’ refers to the extent to which a person uses his or her sense of smell in everyday life and ‘Consequence’ that refers to the role of olfaction in daily decisions. We also separately analysed three specific items related to chemosensory communication (‘The smell of a person plays a role in the decision whether I like him/her’ (Association), ‘Sometimes I smell a person (e.g. my partner or my child) to judge, if he/she has drunken alcohol or smoked’) (Application) and ‘If my partner has a nasty smell, I avoid kissing him’ (Consequence)). Reliability reported for this questionnaire is α = 0.77.
• (2)Cognitive Reserve Index questionnaire (CRIq) [40], which evaluates cognitive reserve an individual has in their adult life in education, work and leisure domains. The total score is computed as the average score of the three subscales—a result below 70 points suggests low cognitive reserve; above 130 points suggests high cognitive reserve. Validity reported for the Italian population is α = 0.62. Note that we present separate scores for each subscale.
• (3)Montreal Cognitive Assessment (MOCA) test [41] for screening mild cognitive impairment with a maximum of 30 points and 26 points being used as a cut-off for mild cognitive impairment supplemented with a digit span memory task wherein the score represents the number of digits a subject was able to successfully repeat. Internal consistency is α = 0.83.
• (4)Short Form-36 Health Survey SF-36 [42,43], which is a 36-item inventory consisting of eight subscales scores from 0 to 100, such that the total score varies from 0 to 800 points; the higher the score the better health is reported. Cronbach's α exceeds 0.85.
• (5)Beck Depression Inventory [44] comprising 21 items scored from 0 to 3 points; the higher the result the more intense the depressive symptoms. Reliability of this method is estimated to Cronbach's α = 0.81.

The olfactory performance was measured with the Sniffin’ Sticks test, comprising three subtests: olfactory threshold, discrimination and identification ([45]; for detailed procedure description and normative data see: [35]). We further assessed respiratory performance with peak nasal respiratory flow (PNIF) using an inspiratory peak flow metre (Clement Clarke International Ltd., Harlow, UK). Subjects were asked to inhale as hard and fast as they could through the mask with the mouth closed starting from the end of a full expiration [46].

Questionnaires and olfactory testing were performed within a single run to reduce the risk of dropouts, which were likely to occur if the study was divided into two (or more) sessions. A single session lasted approximately 70–80 min, although completion time varied across the subjects. They were instructed that they were free to take a break at any time. None of the participants reported major fatigue.

(d) Statistical analyses

Data were analysed with IBM SPSS Software v. 25 with the level of significance set to α = 0.05. We examine daily life odour perception in the two groups by comparing their level of self-assessed olfactory performance with the independent sample t-test and the distribution of the claims to be able to smell particular environmental smells between subjects with normosmia and impaired olfaction with the χ²-test. To examine the effect of anosmia unawareness on chemosensory communication, we compared individual chemosensory communication scores with the average scores on the remaining items from Association, Application and Consequence subscales of the Individual significance of olfaction questionnaire. Three repeated measures general linear models (GLM) were ran separately for Association, Application and Consequence subscales with domain (human chemosensory communication versus other domain) as the within-subjects factor, group (normosmia versus impaired olfaction) as a between-subject factor and age as a covariate. Furthermore, we compared well-being of the two groups in terms of cognitive reserve index, self-reported health, depressive symptoms and cognitive abilities (MOCA) using univariate ANCOVAs with age included as the covariate.

3. Results

(a) Daily life odour perception

Self-rated ability to smell was significantly higher in subjects presenting with normosmia (M = 4.81 ± .09) than in those with impaired olfaction (M = 4.10 ± .19), t₂₀₁ = 3.48, p = 0.001 [CI: 0.31; 1.13]. Despite this, we found no significant difference in the distribution of declarations to be able to smell particular environmental smells (${\chi }_{(1)}^2>1.59,$ p > 0.26) except for coffee, which was significantly more frequently declared to be perceived by normosmic subjects than those with impaired olfaction. Exact coefficients can be found in table 2.

Table 2.

Declared ability to detect certain environmental odours across subjects with impaired olfaction and normosmia.

odour	impaired olfaction	normosmia	χ2	p
cigarette smoke	30 (47.6%)	77 (55%)	0.95	0.33
meals	13 (20.6%)	35 (25%)	0.46	0.50
rotten foods	27 (42.9%)	60 (42.9%)	0	1
perfume	21 (33.3%)	51 (36.4%)	0.18	0.67
alcohol	10 (15.9%)	25 (17.9%)	0.12	0.73
coffee	6 (9.5%)	30 (21.4%)	4.22	0.04
spices	9 (14.3%)	24 (17.1%)	0.26	0.61
vinegar	12 (19%)	37 (26.4%)	1.29	0.26
gasoline	20 (31.7%)	48 (34.3%)	0.13	0.72
flowers	7 (11.1%)	23 (16.4%)	0.98	0.32
smoke	18 (28.6%)	51 (36.4%)	1.20	0.27

(b) Chemosensory communication

The first model comparing chemosensory communication in the context of memories and emotions evoked by odours (Association) revealed a significant main effect of group (F_1,200 = 15.1, p < 0.001), suggesting significantly stronger odour-based associations in normosmic (M = 2.14 ± 0.05) than in subjects with impaired olfaction (M = 1.78 ± 0.08). This implies that normosmic subjects self-declared significantly more odour-based associations, regardless of whether it concerned chemosensory communication with others or other daily-life situations. There were no other main or interaction effects (all F < 1.3, p > 0.26). The second model examining the use of odours in everyday life (Application) yielded a significant main effect of domain (F_1,200 = 15.9, p < 0.001), suggesting that regardless of the group, the individual significance of odours was stronger in non-human interaction situations (M = 1.87 ± 0.05) than in situations that involved human chemosensory communication (M = 1.17 ± 0.08). There were no other main or interaction effects (all F < 1.7, p > 0.07). Finally, the third model examining the behavioural consequences of olfactory experience (Consequence) showed no significant main or interaction effects (all F < 2, p > 0.10).

(c) Well-being

No significant differences were found in CRI scores between subjects with normosmia and impaired olfaction, except for CRI leisure (F_1,200 = 11.5, p = 0.001), where subjects with a normal sense of smell (M = 221 ± 8.5) declared higher cognitive reserve than those whose smell was impaired (M = 165 ± 13.2). MOCA scores where subjects with a normal sense of smell (M = 26.9 ± .24) outperformed those whose smell was impaired (M = 25.4 ±0 .37; F_1,200 = 10.7, p = 0.001). With regard to SF-36, we only found a significant between-group difference in physical functioning (F_1,200 =4.71, p = 0.03) where subjects with olfactory impairment rated their physical health lower (M = 81.5 ± 1.5) than subjects with normosmia (M = 87.9 ± 1.5). There were no other significant between-group differences with regard to BDI or digit span scores. Results of both groups are graphically presented in figure 1.

Figure 1.

Results obtained by subjects with normosmia and subjects with impaired olfaction. Error bars represent standard error of the mean. (a) CRI, cognitive reserve index; (b) BDI, Beck depression inventory; (c) cognitive performance—includes MOCA, Montreal cognitive assessment and digit span scores; (d) SF-36, short form-36 health survey; Phys. func., physical functioning; Phys. role. func., physical role functioning; Emo. role. func., emotional role functioning; Vital., vitality; Soc. role. func., social role functioning; Gen. health, general health. *p < 0.05, **p < 0.01. (Online version in colour.)

4. Discussion

It is commonly understood that normal olfactory function is required for good health and a fully satisfying life. However, we have found that a notable proportion of subjects who believe they have a normal sense of smell actually fall into the range of hyposmia or anosmia. By contrast to patients diagnosed by an ENT specialist, this group had never sought help in clinics specializing in smell rehabilitation. In the current study, we were able to investigate this interesting group and explore the effects of their unsuspected impaired olfaction on aspects of their well-being and use of odour in social interaction. In the initial sample of declared healthy subjects, we found 2% individuals with anosmia and 29% with hyposmia. The conclusion that one out of three subjects in our sample misjudged their smell abilities is in agreement with previous findings that people inaccurately assess their olfactory abilities [47–51]. The fraction of functionally anosmic subjects found in our sample corresponds with former reports but the number of hyposmic individuals slightly exceeds epidemiological [52–54] and normative data [55]. This discrepancy is likely to stem from the recruitment procedure. In the case of studies on human olfaction, subjects are usually recruited in a university context—either students, patients or their family and friends. Here, we avoided this bias by conducting this study largely in a small non-academic town.

The comparison with healthy subjects yielded a major pattern of non-significant between-group differences, with the exceptions of the self-assessed olfactory ability, odour association and cognitive performance (CRI leisure, MOCA) and physical health. This finding is in line with previously reported results [56]. The lack of apparent differences in emotional functioning, chemosensory communication and well-being provides boundaries to the assumption that olfactory impairment is associated with cognitive performance, the effectiveness of chemosensory communication, well-being and general health (see [57,58] for review). This assumption has largely been built on the conclusions from studies including patients who decided to seek help because they associated their various health problems with the lack of the sense of smell [4,27,29], and was not confirmed in studies including subjects who consider themselves healthy independent of potential olfactory deficits. Interestingly, we found that subjects with impaired olfaction rated their ability to smell lower than the subjects with normosmia. This undermines our notion that the subjects with undetected olfactory loss were unaware of the significant smell deficits, but apparently, their emotional functioning, chemosensory communication and health performance were not robustly decreased because of the impaired olfaction. However, it has to be maintained that an inclusion criterion was that subjects considered their olfactory function as normal. Hence, the ability to smell was rated within the context of ‘normal olfaction’, which was one of the study preconditions, and the differences between participants should be interpreted only in this range.

(a) Consequences of undetected olfactory loss for human chemosensory communication

With regard to chemosensory communication, subjects whose Sniffin’ Sticks TDI score was in the range of hyposmia or anosmia scored lower in the Association subscale of the Individual Significance of Olfaction scale [39]. Items in this subscale evaluate emotions, memories and evaluations triggered by a perceived odour. People with reduced sense of smell form lower associations between odours and feelings and memories in comparison to normosmic people, and that includes also associations between body odour and liking of a partner. We hypothesize that the association between emotions, memories and odours might not be developed in subjects with impaired olfaction or might have weakened along the period of limited-to-no odour perception, but to verify this hypothesis one would have to control for the onset of olfactory impairment. This, however, is difficult outside the clinic and even more so in subjects who maintain that they have a normal sense of smell. A detailed analysis of the ‘Application’ subscale, which compared the use of odour in detecting whether people had consumed alcohol or smoked, against other non-social uses of odour revealed that all subjects applied their sense of smell significantly less often to detect consumption of alcohol or cigarettes in other people than in non-social applications. This result can be explained by the relatively unlikely context of this item (people might know smoking and drinking habits of their relatives or friends). Alternatively, this particular item could be susceptible to social desirability bias (e.g. by admitting such sniffing checks our subjects might have pictured them to appear controlling or suspicious). Altogether, our results indicate that despite the diminished olfactory performance, people with undetected hyposmia or anosmia are able to successfully navigate in their social environment, and that they recognize socially relevant traits of other people with comparable accuracy to normosmics. Beyond our study, examples of good coping mechanisms in the absence of one modality include famous chefs whose smell is significantly deteriorated, or normally behaving infants whose hearing deficits may not be recognized until language acquisition problems become apparent. The mechanism explaining this null-effect could relate to the increased meaning of input from other modalities (e.g. vision, audition) that in the absence of the sense of smell can carry more weight in social cognition [4,59].

(b) Consequences of undetected olfactory loss for well-being

Subjects with hyposmia or anosmia scored lower in the MOCA test. Although the absolute mean difference between the two groups was 1.5 points and the average score was close to the 26 points considered as a cut-off point for mild cognitive impairment, some participants scored below this threshold. We believe that the lower MOCA scores in subjects with impaired olfaction may indicate an early sign of cognitive deterioration [60,61]. To fully address this speculation, it would be necessary to follow individuals with lower scores and monitor their olfactory performance in the coming years. In this context it is interesting to note that in a longitudinal study [62], an improvement in olfactory function was accompanied by an improvement in semantic–categorical verbal fluency, highlighting the association between the functioning of the sense of smell and cognitive abilities.

A limitation of the current study refers to a significant age difference between subjects with normosmia and subjects with impaired olfaction. Although we controlled for subjects' age during the analyses, one cannot rule out possible artefacts. There is the possibility that the significant difference in physical functioning of the two groups might be attributed more to the age difference than to the olfactory abilities, but to determine this a replication with more age-balanced groups would be appropriate. We have tested this in a reduced sample where age was perfectly balanced between the groups and this difference was not significant, further suggesting the effect of age rather than olfactory performance on physical functioning.

The universe of odours, including the complex mixtures of compounds that constitute body odours, may have more sophisticated impact on human social behaviour than arbitrary odours used in the standard olfactory assessment. We used a standardized Sniffin’ Sticks battery to categorize subjects as normosmic or with impaired olfaction, but further studies undertaking this line of reasoning should use natural body odour samples to deepen our understanding of olfactory drivers of human social behaviour. More evidence is needed to fully understand how olfactory test performance relates with performance in social olfaction.

To conclude, olfactory loss certainly has a severe impact on daily life but more so in individuals who are bothered with it and decide to seek treatment. The limited-to-no olfactory perception in the fraction of subjects who do not complain about it or do not seek help in ENT clinics does not seem to have a major impact on their emotional and health functioning. However, people with low olfactory scores exhibit small but significant weaknesses in terms of cognitive function, which may be considered an early sign of future cognitive loss. This highlights the association between the sense of smell and cognition.

Ethics

The study was performed in accordance with the Declaration of Helsinki on Biomedical Studies Involving Human Subjects. Informed written consent was obtained from all subjects, who were additionally assured they could withdraw their consent at any time without giving reasons. The study design and consent approach were approved by the Ethics Review Board at the TU Dresden (EK167052018).

Data accessibility

The datasets supporting this article have been uploaded as the supplementary material.

Authors' contributions

All authors (A.O., F.K., M.L. and T.H.) contributed to the planning of the study, the final interpretation of the results and the writing of the paper. F.K. collected all the data.

Competing interests

We declare we have no competing interests.

Funding

A.O. was supported by the Ministry of Science and Higher Education (grant no. 626/STYP/12/2017); T.H. received support from the Deutsche Forschungsgemeinschaft (grant no. DFG HU411/18-1).