Correlation of scores for retronasal olfactory testing obtained for test and retest. Data from healthy subjects only (r27 = 0.76). The width of a bubble relates linearly to the number of data points included in the bubble.
Correlation between scores for retronasal testing and tests of orthonasal olfactory function (A, orthonasal odor identification; B, orthonasal odor discrimination; and C, orthonasal butanol odor thresholds; higher scores indicate higher olfactory sensitivity). When considering data from both normosmic and hyposmic patients, the coefficient of correlation was largest for retronasal and orthonasal odor identification (r86 = 0.78; P<.001). In contrast, coefficients of correlation between scores for retronasal olfactory function and orthonasal odor discrimination (r86 = 0.61; P<.001) and butanol thresholds (r86 = 0.57; P<.001) were lower.
Heilmann S, Strehle G, Rosenheim K, Damm M, Hummel T. Clinical Assessment of Retronasal Olfactory Function. Arch Otolaryngol Head Neck Surg. 2002;128(4):414-418. doi:10.1001/archotol.128.4.414
To develop a test kit for the simple assessment of retronasal olfactory function and to compare orthonasal and retronasal olfactory function in healthy subjects and patients with olfactory disorders.
Design and Patients
We tested 230 individuals with normosmia, hyposmia, and anosmia using grocery-available powders. Initially, 30 different substances were investigated. Subjects identified each substance using a list with 4 verbal items (forced choice). After preliminary experiments, 20 items were selected according to the degree to which they were identified by normosmic and anosmic subjects. Orthonasal olfactory function was assessed psychophysically using "sniffin' sticks," which includes tests for odor identification, discrimination, and butanol odor thresholds. In addition, anosmia was confirmed electrophysiologically by means of olfactory-evoked potentials.
In healthy subjects, there was a test-retest reliability correlation of r27 = 0.76 for retronasal olfactory function, which is similar to other odor identification tests. Retronasal testing in normosmic subjects allowed for the discrimination of sex-related differences, with women scoring higher than men (P = .007), and the identification of a slight decrease with age (r120 = −0.20; P = .03). Orthonasal and retronasal identification of odors was found to correlate (r86 = 0.78; P<.001). Retronasal testing allowed for the discrimination between normosmia, hyposmia, and anosmia (P<.001). In addition, retronasal performance of anosmic patients appeared to improve with duration of anosmia (P = .03). No difference was found between patients with anosmia of different origin.
Results of the present investigation indicate that the assessment of retronasal olfactory function is possible using oral stimulus presentation.
IN EVERYDAY LIFE, olfactory-mediated sensations are often confused with gustatory-mediated sensations. Retronasally perceived odors are referred to the oral cavity rather than to the olfactory epithelium. In fact, in some languages the word taste is used to refer to both the olfactory and gustatory components of food. Rozin1 described the sense of smell in terms of a duality, with orthonasal smelling referring to the outside world and retronasal smelling referring to the inside world. Differences between these perceptually different worlds are only partly understood. In fact, hedonic evaluation may be completely different for the same olfactory component when applied either orthonasally or retronasally, which is observed, for example, with cheese or fish.1
Patients with a loss of smell often report gustatory deficits as the first and most prominent symptom, indicating a reduced capacity to identify and appreciate foods.2 Hence, patients experience their olfactory deficit by means of a procedure similar to a retronasal smell test. Despite this fact, retronasal smelling so far has received far less attention than its orthonasal counterpart. Thus, the aims of this study were (1) to develop a test kit for the simple assessment of retronasal olfactory testing that resembles everyday challenges to retronasal olfactory identification abilities and (2) to compare orthonasal and retronasal olfactory function in healthy subjects and patients with olfactory disorders.
To assess retronasal olfactory function, we tested 230 individuals with normosmia (n = 120), hyposmia (n = 37), and anosmia (n = 73) (Table 1). Investigations were performed according to the Declaration of Helsinki on biomedical studies involving human subjects (Summerset West Amendment). Most normosmic subjects were students at the medical schools of the University of Dresden, Dresden, Germany, and the University of Cologne, Köln, Germany (see Table 1 for subject characteristics). In addition, we investigated patients reporting to the Smell Dysfunction Clinic at the Department of Otorhinolaryngology of the University of Dresden. None of the participants received any form of financial compensation.
Orthonasal olfactory function was assessed psychophysically using "sniffin' sticks,"3 which involves tests for butanol odor threshold, odor discrimination, and identification. Results of the 3 subtests were also presented as a composite score derived from the sum of the results obtained for threshold, discrimination, and identification measures (TDI score). This score allows a diagnosis in terms of anosmia, hyposmia, or normosmia.4 Anosmia was additionally confirmed electrophysiologically by means of olfactory-evoked potentials.5 Controls were included on the basis of reports of normal olfactory function and normal test results in either an olfactory screening test (n = 71)6 or the whole sniffin' sticks test battery (n = 49).3 As sniffin' sticks are a well-investigated means to assess orthonasal olfactory function, the comparison with the new retronasal test battery was thought to be useful, especially because the present investigation was primarily concerned with the clinical validity of a new test for retronasal olfactory function.
For retronasal stimulation, grocery store condiments and food items available in powder form were used (eg, spices, instant drinks, and instant soups; Table 2). The substances were applied using squeezable plastic vials with a 6-cm long spout. Substances were selected according to their texture (ie, nonsticky, small grains) and the degree to which they are known in the general population. Subjects were free to sample as much stimulant as needed for identification. This approach also minimized the problem of standardizing the area of stimulation, differences in tongue, or oral cavity size. In a typical trial, through the wide-opened mouth, the experimenter placed approximately 0.05 g on the middle of the tongue inside the oral cavity. Before application of the first stimulant and after each trial, subjects rinsed with tap water; this helped to minimize interindividual differences in salivation, which might interfere with the release of the odorants. The procedure was self-timed, although test intervals were usually 1 minute due to rinsing. Each substance was identified by means of a closed set with 4 verbal items using a forced-choice procedure (subjects also indicated the taste of the substances, the results of which will be reported elsewhere). Responses were not analyzed with regard to "near" or "far" misses. All descriptors used indicated odors that had previously been shown to be familiar to more than 75% of a test population.3 Scores were obtained by adding the number of correct identifications of retronasal stimuli. For investigations of the test-retest reliability, subjects without nasal pathologic features were tested twice at an interval of 1 to 7 days.
Results were analyzed with SPSS 9.0 for Windows (SPSS Inc, Chicago, Ill). Because olfactory sensitivity varies as a function of age,7 subjects were separated into 4 age groups (groups A-D). Ages of subjects in the groups were as follows: group A, 6 to 15 years; group B, 16 to 35 years; group C, 36 to 55 years; group D, older than 55 years. To explore olfactory sensitivity in relation to age and sex, data were submitted to analyses of variance using the general linear model and multifactorial design analysis of variance with "between-subject factors," "group," and "sex" with Bonferroni post hoc tests. Correlations were performed using Pearson statistics. The α level was .05.
From 30 items used for retronasal testing, 20 were selected according to the degree to which they could be identified (Table 3). Specifically, substances that were poorly recognized by normosmic subjects (ie, caraway, anise, blueberry, almond, sour cherry, bacon, mustard, and coconut; identification rate in normosmic subjects <70%) and substances that were identified to a similar degree by hyposmic, anosmic, and normosmic subjects (ie, lemon and pepper) were left out. All results reported in this section relate to the test based on 20 items only (compare with Table 2).
In healthy subjects, there was a test-retest reliability correlation of r27 = 0.76 for retronasal olfactory function (Figure 1). Results for retronasal testing in normosmic subjects (55 men and 65 women; mean age, 33.5 years) allowed for the discrimination of sex-related differences, with women scoring higher than men (t120 = 2.77; P = .007), and the identification of a slight age-related decrease (r120 = −0.20; P = .03).
The correlation between retronasal olfactory function and scores for the 3 orthonasal tests for normosmic and hyposmic subjects varied between r86 = 0.78 for odor identification (P<.001), r86 = .61 for odor discrimination (P<.001), and r86 = 0.57 for butanol thresholds (P<.001) (Figure 2).
Retronasal testing allowed for the discrimination between normosmic, hyposmic, and anosmic subjects (F2,227 = 279; P<.001). Specifically, there was almost no overlap between scores of anosmic (score of 12.3 at the 95th percentile) and normosmic (score of 13 at the 5th percentile) subjects, whereas scores of hyposmic subjects exhibited overlap with both anosmic and normosmic subjects (Table 4).
Retronasal identification scores in anosmic patients improved slightly with duration of anosmia (in this analysis all patients with congenital olfactory loss were excluded; in addition, only those patients in whom the duration of olfactory loss was known were taken into consideration). Patients were divided in 2 groups with duration of olfactory loss of 2 years or less (n = 26; mean [range] age, 58 [28-89] years) and more than 2 years (n = 34; mean [range] age, 56 [22-84] years). When duration of anosmia was 2 years or less, patients scored lower compared with patients with longer duration of anosmia (duration ≤2 years: mean ± SEM score of 7.9 ± 0.4; duration >2 years: mean ± SEM score of 9.3 ± 0.4; t58 = 2.18; P = .03).
A number of factors are known to be involved in the retronasal perception of olfactory stimuli. It has been shown that adequate orthonasal olfactory function is a prerequisite for good retronasal olfactory performance.8 However, in some cases good orthonasal olfactory function goes along with impaired retronasal smelling. This may be due to oral conditions influencing retronasal perception.8 In addition, mouth movements, mastication, and swallowing are known to influence the perception of retronasal stimuli.9,10 Pierce and Halpern11 reported that odor identification is better when stimuli were presented orthonasally compared with retronasal odor presentation. This was partly improved by training. That is, subjects using a certain breathing technique to enhance retronasal stimulation scored higher in retronasal identification tasks. However, retronasal odor identification abilities still did not reach the level of orthonasal odor identification. It was hypothesized that this difference was due to the differential efficacy in odor presentation to the olfactory epithelium.
Apart from these factors, information-processing patterns and interactions with other oral senses may influence the evaluation of olfactory information obtained via the retronasal route (eg, as demonstrated in the poster by Fast et al12). Different hypotheses have been presented on interactions between taste and smell.13- 18 Gillan15 compared both odor-odor and taste-taste mixtures with odor-taste mixtures. He was able to show that suppression between senses was smaller than suppression induced by mixtures stimulating the same sensory system. Burdach et al16 reported that measured sensitivity to stimuli decreased when odor-taste mixtures were administered. This means that thresholds might increase as a consequence of suppression caused by a combination of different stimuli. On the other hand, when magnitude estimation procedures were used for the classification of stimuli, a number of authors reported an independent contribution of smell and taste to overall intensity.13,14,17,18
The test used in the present study addressed retronasal stimulation similar to everyday challenges. Specifically, subjects were allowed to distribute the powders intraorally, perform masticationlike movements, and to swallow (compare with the articles by Burdach and Doty9 and Harrison et al 10). However, the "real-life stimuli" used in the present study not only activated the olfactory system, but additionally provided sensations mediated by the gustatory and trigeminal systems. Based on this sensory information, some of the stimulants (eg, lemon) could be identified by more than 80% of the anosmic patients. After excluding these substances from the test, it appeared that olfactory-mediated information was a major factor in the identification of the 20 substances that remained in the test. This hypothesis is supported by the separation of anosmic, hyposmic, and normosmic subjects, which was possible by means of the 20-item retronasal test kit. The use of stimuli that do not specifically activate the olfactory system is also applied in most tests of orthonasal odor identification (eg, the University of Pennsylvania Smell Identification Test19). In these tests, it is fundamentally assumed that the identification of most odors is possible through some degree of olfactory function but to a much smaller degree through, for example, trigeminal sensitivity.
The discrimination between different degrees of olfactory loss is also the prerequisite for clinical applications of this test. In addition, its test-retest reliability correlation of r27 = 0.76 is similar to other olfactory tests using a similar number of items. For example, a 12-item odor identification reached a coefficient of correlation of 0.71,20 and a 16-item odor identification test reached a coefficient of correlation of 0.73.3 Also, it has to be considered that the coefficient of correlation for test and retest was established in a relatively homogeneous population with little interindividual variance. Thus, it can be assumed that the coefficient of correlation would be higher if different populations had been examined (eg, normosmic and hyposmic subjects), which will be investigated in future studies.
About the test's validity, it also allowed for the identification of effects related to both age and sex. This indicates that retronasal function follows the course of well-known, age-related changes in orthonasal olfactory function.7 In addition, the better performance of women in orthonasal olfactory testing has been described frequently21- 23; this appears to be equally present in retronasal olfactory testing. Previous studies already described the decline of retronasal olfactory function with age, using solely the oral presentation of ethyl butyrate with unpinched or pinched nose (thus switching retronasal smelling on and off)24 or using a single food ingredient (marjoram).25 Our study extended these findings with regard to retronasal odor identification.
Most interestingly, retronasal identification abilities of anosmic patients appeared to improve slightly with duration of anosmia. In an adaptive sense, patients seem to become more efficient in terms of the use of gustatory and/or trigeminal information. It may be speculated that this indicates cross-modal plasticity as it has been hypothesized to occur, for example, for visual processing in congenitally deaf subjects.26
In conclusion, results of the present investigation indicate that the assessment of retronasal olfactory function is possible using oral stimulus presentation. Future developments will focus on an extension of the number of test items used to improve its sensitivity.
Accepted for publication September 10, 2001.
We would like to thank John Prescott, PhD, University of Otago, Dunedin, New Zealand, for his thoughtful comments on earlier versions of the manuscript. We also would like to thank our referees for their most helpful suggestions.
Corresponding author: Thomas Hummel, MD, Department of Otorhinolaryngology, University of Dresden Medical School, Fetscherstr 74, 01307 Dresden, Germany (e-mail: firstname.lastname@example.org).