Assessing olfaction in the neuropsychological exam: The relationship between odor identification and cognition in older adults

Archives of Clinical Neuropsychology 20 (2005) 761–769

Assessing olfaction in the neuropsychological exam: The relationship between odor identification and cognition in older adults Holly James Westervelt a,b,∗ , Jessica Somerville Ruffolo b , Geoffrey Tremont a,b a

Rhode Island Hospital, Physician’s Office Building, Suite 430, 593 Eddy Street, Providence, RI 02903, USA b Brown Medical School, RI, USA Accepted 9 April 2005

Abstract The relationship between odor identification and cognition has not been previously well characterized. The neuroanatomy of the olfactory system and the frequent finding of olfactory dysfunction in neurodegenerative diseases suggest a likely relationship between odor identification and memory, language, and executive functioning, though previous studies have often failed to demonstrate the expected relationship. The current study examined this relationship in across a continuum of ability levels (N = 100). Strongest correlations were found between odor identification and language, most aspects of memory, and a measure of general cognitive functioning. Significant but more modest correlations were seen between odor identification and attention, motor, visuospatial, and executive functions. A regression analysis revealed language as the only significant predictor of olfactory performance. These findings suggest that odor identification is most closely associated with other measures of temporolimbic functioning. The implications of these findings, particularly in consideration of the assessment of older adults, are discussed. © 2005 National Academy of Neuropsychology. Published by Elsevier Ltd. All rights reserved. Keywords: Olfaction; Odor identification; Memory; Cognitive assessment

∗

Corresponding author. Tel.: +1 401 444 4500; fax: +1 401 444 6643. E-mail address: [email protected] (H.J. Westervelt).

0887-6177/$ – see front matter © 2005 National Academy of Neuropsychology. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.acn.2005.04.010

762

H.J. Westervelt et al. / Archives of Clinical Neuropsychology 20 (2005) 761–769

The nature of the relationship between odor identification performance and cognition has not been well established, though an association between olfaction and at least certain aspects of cognition would be expected based on the neuroanatomy of the olfactory system. In particular, a relationship with memory is likely given the anatomic proximity of areas critical for memory processing (i.e., hippocampus) and processing of olfactory information (i.e., piriform cortex with projections to the amygdala, periamygdaloid cortex, and entorhinal cortex). In addition, the prefrontal cortex is a secondary area for olfactory processing, suggesting the likelihood of a relationship between odor identification performance and executive functioning. Functional imaging and resection studies support the importance of these regions in olfactory functioning (Dade, Zatorre, & Jones-Gotman, 2002). Furthermore, deficits in odor identification are commonly observed in neurodegenerative and other dementing diseases which often present with prominent memory disturbance such as Alzheimer’s disease (Serby, Larson, & Kalkstein, 1991), diffuse Lewy body disease (Westervelt, Stern, & Tremont, 2003), and vascular dementia (Knupfer & Spiegel, 1986). Although a relationship between these processes seems intuitive, the few prior studies which have examined the relationship have often failed to find an association. The two studies assessing memory failed to find statistically significant associations between memory functioning and olfaction (Carone et al., 1999; Weber, 2004), and the two assessing the relationship between olfaction and other temporal processes such as confrontation naming had mixed results (c.f., Carone et al., 1999; Larsson, Semb, Winblad, Wahlund, & Backman, 1999). The relationship with executive functioning has also been inconsistent, with one study finding a modest but statistically significant association (Weber, 2004), but another failing to find a relationship (Larsson et al., 2004) In some of these studies, the surprising null results may have reflected the likely limited range of performance in the samples studied (e.g., patients with Alzheimer’s disease) or limited sample sizes. The current study explores the relationship between odor identification and a variety of cognitive domains (i.e., attention, executive functioning, visuospatial skills, language, motor skills, and memory) in a large sample of older adults representing a continuum of performance ability. Based on an anatomic rationale, we anticipated the strongest relationships with tasks involving significant temporal and anterior frontal processing (i.e., memory, executive functioning, and language), and more modest, if any, relationships with other cognitive tasks.

1. Method 1.1. Participants The study sample consisted of 100 participants (46 men) with a mean age of 70.27 (S.D. = 9.21) and mean educational level of 13.8 (S.D. = 3.20). The study included only older adults (age 50+) because: (1) olfactory measures appear especially well suited for identifying and distinguishing neurodegenerative disease; and (2) the range of performance on the olfaction measure is extremely limited in younger adults (Doty, Marcus, & Lee, 1996). Seventy-four of the participants were patients presenting for assessment of memory concerns to the outpatient service of a large, urban, academic general medical center. Of the 74 patients, 63

H.J. Westervelt et al. / Archives of Clinical Neuropsychology 20 (2005) 761–769

763

presented with questions of possible dementia (of those, 32 met DSM-IV criteria for dementia), 5 with stroke, 2 with multiple sclerosis, 1 with encephalitis, 1 with anoxic injury, 1 with leukoencephalopathy, and 1 with a primary psychiatric condition. The remainder (n = 26) were community dwelling healthy older adults. All participants were fluent in English. Potential participants were excluded if they had histories of significant head injury given the possibility of peripheral injury to cranial nerve I. Other exclusion criteria included the following: (1) self-report of long-standing anosmia; (2) history of brain or nasal surgery; (3) significant nasal congestion at the time of the exam; or (4) self-report of hypersensitivity to odors. To maximize the likelihood that the sample would contain performances within the normal range, healthy community dwelling participants were additionally excluded if they had histories of diseases known to affect olfactory functioning, concerns about their memory functioning, or if they reported any decline in daily functioning. All community participants had current MMSE total scores of 26/30 or better. The study was approved by the institution’s Institutional Review Board, and all participants provided written consent to participate (consent was additionally obtained from a significant other/caregiver when appropriate). Healthy participants were paid $20 for participation; patients were not paid given that their participation involved only consent to allow their clinical data to be used for research purposes. 1.2. Materials and procedures Patients presenting for clinical exam completed a comprehensive neuropsychological evaluation, as well as a measure of odor identification. Because patient participants were being evaluated as part of a clinical examination, some patients did not complete all measures if not deemed clinically appropriate for the patient. In such instances, those data were coded as missing variables, but the remainder of the patient’s scores were included in the analyses to reduce sample bias. Cognitive domain scores were created by collapsing age-corrected (and sometimes education-corrected) T-scores generated from published normative data within each domain into a mean domain T-score. If a participant were missing data for one or more measures, the domain T-score included the mean of all available scores. For each cognitive domain, the total sample size ranged from 98 to 100, with the exception of the Motor domain (N = 86). The domains included the following measures for most patients and all healthy participants: (1) odor identification (Brief Smell Identification Test [BSIT; Doty et al., 1996]); (2) global cognitive functioning (Modified Mini Mental State Examination [3MS; Teng & Chui, 1987]); (3) attention (Digit Span and Mental Control from the Wechsler Memory Scale-III [WMS-III; Wechsler, 1997] and Part A of the Trail Making Test [TMT; Reitan & Wolfson, 1985]); (4) executive functioning (Part B of the TMT [Reitan & Wolfson, 1985], Controlled Oral Word Association using F, A, and S [Benton & Hamsher, 1989], and the Rey-Osterrieth Complex Figure [ROCF] copy Organization score from the Boston Qualitative Scoring System [BQSS; Stern et al., 1999]); (5) language (Boston Naming Test [BNT; Kaplan, Goodglass, & Weintraub, 1983] and animal fluency [Goodglass & Kaplan, 1983]; (6) visuospatial functioning (ROCF BQSS Copy Presence and Accuracy score); (7) motor skills (Grooved Pegboard [GP; Reitan & Halstead, 1985]); and (8) memory (Total Learning trials 1–3 and Delayed Recall from the Hopkins Verbal Learning Test-Revised [HVLT; Benedict, Schretlen, Groninger, & Brandt, 1998],

764

H.J. Westervelt et al. / Archives of Clinical Neuropsychology 20 (2005) 761–769

Immediate and Delayed Recall from the WMS-III Logical Memory [LM; Wechsler, 1997], and Immediate and Delayed Presence and Accuracy scores from the ROCF BQSS [Stern et al., 1999]). We considered analyzing immediate and delayed recall scores separately, but the correlation between these two memory indices was so strong (r = .93) that combining these scores into a single indicator appeared most appropriate.

2. Results All analyses were carried out using SPSS software version 10.0. All analyses were twotailed. Given the number of analyses conducted, a conservative level of statistical significance was selected a priori (P < .01). 2.1. Description of the sample The means and standard deviations of the domain T-scores and raw scores of the 3MS and BSIT are presented in Table 1. In all cognitive domains, the level of performance in the group ranged from impaired to average or better. 2.2. Correlational analyses Pearson’s product moment correlations between BSIT and each cognitive domain were analyzed and are presented in Table 2. Moderate to strong correlations were found between the BSIT and the 3MS and the Language and Memory domains. Statistically significant but more modest correlations were found between the BSIT and the Attention, Executive, Motor, and Visuospatial domains. Semi-partial correlations partialing out the effect of age on BSIT performance were also conducted given that raw scores for the BSIT were analyzed; there was no appreciable change in the size of the correlations. When the correlations between domain scores and BSIT were analyzed in the community and patient groups separately, the size and direction of the correlations were similar with the exception of somewhat more robust relationships between BSIT and the Executive and Spatial domains in the patient group (in the patient group r = .32 and .39, respectively, and for community group r = −.01 and .13, respectively), and a more robust relationship between BSIT and the Language domain in the community group (for patient group, r = .44, and for community group, r = .71). Table 1 Means and standard deviations of BSIT, 3MS, and domain scores Domain

M S.D.

BSIT

3MS

Attention

Executive

Motor

Visuospatial

Language

Memory

8.22 2.75

86.60 12.23

46.63 9.23

41.87 9.52

36.74 10.62

43.86 13.19

39.64 15.47

38.89 12.00

Note. BSIT = Brief Smell Identification Test; 3MS = Modified Mini Mental State Examination.

H.J. Westervelt et al. / Archives of Clinical Neuropsychology 20 (2005) 761–769

765

Table 2 Correlation matrix with all domain scores 3MS 3MS (n = 100) Attention (n = 100) Executive (n = 99) Motor (n = 86) Spatial (n = 98) Language (n = 98) Memory (n = 100)

*

Attention

Executive

*

*

.510

.475 .717*

*

*

Motor *

.300 .432* .409* *

Spatial *

.534 .421* .627* .336* *

Language

Memory

BSIT

*

*

.548* .286* .383* .370* .392* .540* .519*

.801 .539* .474* .368* .450* *

.776 .554* .575 .348* .508* .689* *

Note. 3MS = Modified Mini Mental State Examination; BSIT = Brief Smell Identification Test. ∗ P < .01.

2.3. Regression analysis For the regression analysis, the Motor domain was eliminated from analysis to include the maximum number of participants possible (n = 97). A hierarchical regression analysis was conducted with two levels of predictors. The first block of predictors included demographic variables known to be associated with olfactory performance (age and education); the second block included the cognitive domains. The outcome variable was total score on the BSIT. Results of the regression analysis are presented in Table 3. The first set of predictors (age and education) accounted for a statistically significant amount of the variance in BSIT performance (13%, Ps < .002). The addition of the cognitive domain scores significantly incremented the prediction of BSIT (R2 = 36%, Ps < .0001). Examination of the individual predictors in this model indicate that among the cognitive variables, only Language emerged as being significantly predictive of BSIT performance (t = 2.68, P < .009, β = .35). Despite significant zero-order correlations, Memory, Attention, Executive, and Spatial domains were not uniquely predictive of BSIT performance. Table 3 Summary of hierarchical regression analysis for variables predicting BSIT performance (N = 97) Variable

B

S.E. B

β

Step 1 Age Education

−7.06 −.19

.29 .08

−.24* .23*

Step 2 Language Memory Visuospatial Attention Executive

6.12 4.11 2.59 −3.74 3.22

.02 .03 .02 .04 .04

.35* .18 .13 −.13 .11

Note. BSIT = Brief Smell Identification Test; R2 = .13 for Step 1; R2 = .235 for Step 2 (Ps < .0001). ∗ P < .05.

766

H.J. Westervelt et al. / Archives of Clinical Neuropsychology 20 (2005) 761–769

3. Discussion Results of the current study demonstrate that odor identification ability is moderately correlated with general cognitive functioning, language skills, and memory. The relationships with other areas of cognition such as attention, executive functioning, motor skills, and visuospatial skills were statistically significant but more modest. In the regression analysis, the Language domain emerged as the only significant predictor of olfaction, though understanding of the relationship amongst the cognitive and olfaction measures is likely more complex than the regression model indicates due to the substantial overlap in the variables (particularly language and memory, and, in this study, the Visuospatial and Memory domains given shared test variance). These findings should, therefore, be considered to be exploratory. Overall, the findings were mostly consistent with our expectations given the neuroanatomy of the olfactory system, though the relationship between olfaction and executive functioning was weaker than expected. The strongest correlations with the Language and Memory domains suggest that, like language and memory functioning, odor identification ability may be highly reliant on temporo-limbic processing. The relatively weak correlation with executive functioning was more surprising. These findings may suggest that frontal lobe processing is less critical than previously thought in odor identification performance. Alternatively, the relatively modest correlations may be a function of the measures used to assess this region, as traditional executive measures tend to be less reflective of orbito-frontal processing (i.e., the area thought to be important in olfactory processing) than processing in other frontal regions (i.e., dorsolateral prefrontal). Future studies which assess orbito-frontal processing more directly could better address this question. The substantial correlations between odor identification and memory performance in this study are discrepant with previous findings of modest and, typically, statistically nonsignificant correlations. This discrepancy may reflect the use of a larger sample size and fuller range of performance in the current study. Regression findings indicate that cognition as measured in this study accounts for approximately 36% of the variance in BSIT performance. Although this suggests a meaningful relationship between these variables, the amount of variance unaccounted for is notable. Other factors which likely contribute to BSIT performance may include factors such as olfactory sensitivity. This variable was not measured in the current study given that there was no theoretical rationale to expect that olfactory sensitivity would have any systematic relationship with cognition, and therefore would not contribute to the purpose of this study in understanding the relationship between olfaction and cognition. The nature of correlational research inevitably leads to speculation about the possibility of a causal relationship between variables. In the current study, the question arises as to whether the skills most highly related to odor identification performance subserve this ability. That is, because of the substantial relationship between odor identification and language and memory, is successful performance on odor identification tasks dependent on the integrity of these cognitive functions? Several researchers have suggested that odor identification ability is reliant on central processing, whereas some other olfactory abilities such as threshold may be more reflective of peripheral functioning (Koss, Weiffenbach, Haxby, & Friedland, 1988). Fewer

H.J. Westervelt et al. / Archives of Clinical Neuropsychology 20 (2005) 761–769

767

investigators have considered the specific requisite cognitive abilities for adequate performance on odor identification tasks. Presumably, successful performance on odor identification tasks require that a patient retrieve an odor label that is associated with memory of a particular scent. As such, both memory and lexical access skills appear necessary. However, the BSIT task is a multiple choice measure, thereby requiring minimal effort in searching the semantic network for the appropriate label. If a deficit on this task were related to a deficit in the semantic network, failure would then suggest some loss of semantic knowledge, rather than simply a retrieval deficit. To briefly explore this consideration, we examined BSIT performance in participants with impaired (T < 30) cued BNT scores (i.e., number correct when provided with phonemic and/or semantic cues) and those with intact (T > 45) cued BNT scores. As expected, BSIT performance was poorer in the impaired cued BNT group (mean BSIT score = 6.63) than in the intact cued BNT group (mean BSIT score = 9.34). However, in reviewing the frequency distributions of the BSIT scores, we found that both intact and impaired BSIT performances occurred in both BNT groups (impaired group range = 1–12; intact group range = 4–12). This suggests that, although there is a relationship between these variables, intact BNT performance is neither necessary nor sufficient for intact BSIT performance. These findings argue against a direct causal relationship. The association amongst these measures is likely complex, but may at least in part reflect the overall integrity of the temporo-limbic region. In sum, the results of the current study illustrate a pattern in the relationship between odor identification and cognition which suggests that odor identification is most strongly associated with other measures of temporo-limbic functioning (i.e., language and memory). These olfaction measures may offer some unique information about brain functioning that is not being fully assessed by other, more typically used, neuropsychological measures, though more complete exploration of this hypothesis is beyond the scope of the current study. Fuller evaluation of temporo-limbic regions may be especially useful in assessing older adults presenting with questions of dementia. It is now believed that AD likely begins in the entorhinal cortex, a critical area for olfactory processing. As such, it is not surprising that odor identification tasks appear to be useful in the early detection of AD, with impairments on these tasks demonstrated in patients with early dementia (Morgan, Nordin, & Murphy, 1995) and suspected preclinical disease (Devanand et al., 2000). Apparent but more mild deficits have even been shown in at-risk non-demented individuals, including those with APOE ␧4 allele genotypes (Murphy, Bacon, Bondi, & Salmon, 1998) and first degree relatives of AD patients (Serby, Mohan, Aryan, Williams, Mohs, & Davis, 1996). Furthermore, even very brief (i.e., three item) odor identification measures can distinguish patients with Alzheimer’s disease from non-demented elderly adults (McCaffrey, Duff, & Solomon, 2000), and odor identification tasks have been shown to also differentiate dementia subtypes (Duff, McCaffrey, & Solomon, 2002; Knupfer & Spiegel, 1986; Westervelt et al., 2003). Although odor identification tasks have been demonstrated to be useful in both the early detection and differential diagnosis of dementia, olfactory measures are underutilized in both neurologic and neuropsychologic exams. From the perspective of the neuropsychologist, this may reflect some unfamiliarity with these instruments and lack of clarity of how these measures fit within a more traditional neuropsychological exam. The results of the current study may, therefore, be useful to practitioners by providing a context in which to interpret performance on these measures.

768

H.J. Westervelt et al. / Archives of Clinical Neuropsychology 20 (2005) 761–769

Acknowledgements This study was conducted with grant support from the National Academy of Neuropsychology awarded to Dr. Westervelt. The authors wish to gratefully acknowledge Deb Shayer for her assistance in recruiting community participants, as well as Drs. Jennifer Davis, Richard Temple, C. Lee Bishop, Mary Beth Spitznagel, and Angela Jefferson for their assistance in recruiting patient participants.

References Benedict, R. H. B., Schretlen, D., Groninger, L., & Brandt, J. (1998). The Hopkins Verbal Learning Test-Revised: Normative data and analysis of inter-form and test-retest reliability. The Clinical Neuropsychologist, 12, 43–55. Benton, A. L., & Hamsher, K. (1989). Multilingual Aphasia Examination. Iowa City: AJA Associates. Carone, D. A., Jr., Thompson, M. D., Knee, K., Golden, C. J., Hanson, M. R., Salanga, V. D., et al. (1999). Odor identification in Alzheimer’s disease: Support for a degradation of olfactory knowledge hypothesis. Archives of Clinical Neuropsychology, 14, 618–619. Dade, L., Zatorre, R. J., & Jones-Gotman, M. (2002). Olfactory learning: Convergent findings from lesion and brain imaging studies in humans. Brain, 125, 86–101. Devanand, D. P., Michaels-Marston, K. S., Liu, X., Pelton, G. H., Padilla, M., Marder, K., et al. (2000). Olfactory deficits in patients with mild cognitive impairment predict Alzheimer’s disease at follow-up. American Journal of Psychiatry, 157, 1399–1405. Doty, R. L., Marcus, A., & Lee, W. W. (1996). Development of the 12-Item Cross-Cultural Smell Identification Test (CC-SIT). Laryngoscope, 106, 353–356. Duff, K., McCaffrey, R. J., & Solomon, G. S. (2002). The Pocket Smell Test: Successfully discriminating probable Alzheimer’s Dementia from Vascular Dementia and Major Depression. Journal of Psychiatry and Clinical Neurosciences, 14, 197–201. Goodglass, H., & Kaplan, E. (1983). The assessment of aphasia and related disorders (2nd ed.). Philadelphia: Lea & Febiger. Kaplan, E. F., Goodglass, H., & Weintraub, S. (1983). The Boston Naming Test. Philadelphia: Lea and Febiger. Knupfer, L., & Spiegel, R. (1986). Differences in olfactory test performance between normal aged, Alzheimer, and vascular type dementia individuals. International Journal of Geriatric Psychiatry, 1, 3–14. Koss, E., Weiffenbach, J. M., Haxby, J. V., & Friedland, R. P. (1988). Olfactory detection and identification performance are dissociated in early Alzheimer’s disease. Neurology, 38, 1228–1232. Larsson, M., Nilsson, L., Olofsson, J. K., & Nordin, S. (2004). Demographic and cognitive predictors of cued odor identification: Evidence from a population-based study. Chemical Senses, 29, 547–554. Larsson, M., Semb, H., Winbald, B., Amberla, K., Wahlund, L. O., & Backman, L. (1999). Odor identification in normal aging and early Alzheimer’s disease: Effects of retrieval support. Neuropsychology, 13, 47–53. McCaffrey, R. J., Duff, K., & Solomon, G. S. (2000). Olfactory dysfunction discriminates probable Alzheimer’s dementia from major depression: A cross-validation and extension. Journal of Neuropsychiatry and Clinical Neurosciences, 12, 29–33. Morgan, C. D., Nordin, S., & Murphy, C. (1995). Odor identification as an early marker for Alzheimer’s disease: Impact of lexical functioning and detection sensitivity. Journal of Clinical and Experimental Neuropsychology, 17, 793–803. Murphy, C., Bacon, A. W., Bondi, M. W., & Salmon, D. P. (1998). Apolipoprotein E status is associated with odor identification deficits in nondemented older persons. Annals of the New York Academy of Sciences, 855, 744–750. Reitan, R. M., & Wolfson, D. (1985). The Halstead-Reitan neuropsychological test battery: Theory and interpretation. Tuscon, AZ: Neuropsychology Press.

H.J. Westervelt et al. / Archives of Clinical Neuropsychology 20 (2005) 761–769

769

Serby, M., Larson, P., & Kalkstein, D. (1991). The nature and course of olfactory deficits in Alzheimer’s disease. American Journal of Psychiatry, 148, 357–360. Serby, M., Mohan, C., Aryan, M., Williams, L., Mohs, R. C., & Davis, K. L. (1996). Olfactory identification deficits in relatives of Alzheimer’s disease patients. Biological Psychiatry, 39, 375–377. Stern, R. A., Javorsky, D. J., Singer, E. A., Harris, N. G. S., Somerville, J. A., Duke, L. M., et al. (1999). The Boston qualitative scoring system. Odessa, FL: Psychological Assessment Resources, Inc. Teng, E. L., & Chui, H. C. (1987). The Modified Mini Mental State (3MS) examination. Journal of Clinical Psychiatry, 48, 314–318. Weber, M. T. (2004). Odor memory in older adults with and without cognitive impairments (Doctoral dissertation, State University of New York at Albany, 2004). Dissertation Abstracts International, 64, 4641. Wechsler, D. (1997). Wechsler Memory Scale-III. San Antonio, TX: Psychological Corporation. Westervelt, H. J., Stern, R. A., & Tremont, G. T. (2003). Odor identification deficits in diffuse Lewy body disease. Cognitive and Behavioral Neurology, 16, 93–99.