Professional Documents
Culture Documents
Letter and Category Fluency in Community - Dwelling Canadian Seniors
Letter and Category Fluency in Community - Dwelling Canadian Seniors
To cite this article: Margaret Crossley , Carl D'arcy & Nigel S.B. Rawson (1997) Letter and category fluency
in community-dwelling canadian seniors: A comparison of normal participants to those with dementia of
the Alzheimer or vascular type, Journal of Clinical and Experimental Neuropsychology, 19:1, 52-62, DOI:
10.1080/01688639708403836
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our
licensors make no representations or warranties whatsoever as to the accuracy, completeness, or
suitability for any purpose of the Content. Any opinions and views expressed in this publication
are the opinions and views of the authors, and are not the views of or endorsed by Taylor &
Francis. The accuracy of the Content should not be relied upon and should be independently
verified with primary sources of information. Taylor and Francis shall not be liable for any
losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities
whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or
arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial
or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or
distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use
can be found at http://www.tandfonline.com/page/terms-and-conditions
Journal of Clinical and Experimental Neuropsychology 1380-3395/97/190 1-52$12.00
1997, Vol. 19, NO. 1, pp. 52-62 0 Swets & Zeitlinger
ABSTRACT
Measures of letter (FAS) and category (Animal Naming) fluency were completed by community-dwelling,
cognitively normal seniors (n = 635), and by individuals diagnosed with dementia of the Alzheimer type
(DAT) (n = 155), or with vascular dementia (n = 39). Among normal seniors, category, but not letter flu-
ency, declined with age, females performed better than males on letter but not on category fluency, and
performance increased with education on both tasks. Among participants with DAT, letter fluency was
influenced by level of education, whereas category fluency was education-, age-, and gender-invariant.
Both measures differentiated normal seniors from those with dementia. Category fluency was more im-
paired than letter fluency at both mild and moderate stages of dementia, but neither task differentiated DAT
from vascular dementia.
In the study of normal aging some components ton, 1977) and requiring the time-limited gener-
of language are commonly described as age-in- ation of words that begin with a given letter, is
variant, while other components typically un- typically found to be age-stable in cognitively
dergo age-associated decline (see Bayles & normal individuals (Benton, Eslinger, & Dama-
Kaszniak, 1987 or Kemper, 1992 for compre- sio, 1981; Bolla, Lindgren, Bonaccorsky, &
hensive summaries of this area). For example, Bleecher, 1990; Tomer & Levin, 1993). In con-
comprehension of word meaning and basic read- trast, category or semantic fluency, commonly
ing skills are generally reported as stable in nor- assessed by Animal Naming (Goodglass &
mal aging. In contrast, confrontational naming Kaplan, 1972; Rosen, 1980) and requiring the
appears to be sensitive to the effects of increas- time-limited production of exemplars of a given
ing age (Nicholas, Obler, Albert, & Goodglass, category, is generally found to be age-sensitive
1985). Similarly, verbal fluency or generative (Ska, 1993; Tomer & Levin, 1993), although
naming tasks are differentially affected by nor- Perlmuter and Tun (1987) found no differences
mal aging (Tomer & Levin, 1993). Letter-flu- between middle-aged (55-64 years) and older
ency, commonly assessed by the Controlled Oral participants (65-74 years).
Word Association or FAS test (Spreen & Ben-
* The data reported in this article were collected as part of the Canadian Study of Health and Aging. This was
funded by the Seniors Independence Research Program of Health and Welfare Canada. This multi-centred na-
tional study was coordinated through the University of Ottawa and Health Canada’s Laboratory Centre for Dis-
ease Control. The authors would like to thank all of the participants and staff from across Canada who were
involved in the CSHA.
Address correspondence to: Margaret Crossley, Department of Psychology, Arts Bldg, 9 Campus Drive, Univer-
sity of Saskatchewan, Saskatoon, Saskatchewan, Canada, S7N 5A5.
Accepted for publication: July 7, 1996.
LETTER AND CATEGORY FLJJENCY IN THE CSHA 53
Both letter-generation and category fluency temporal lobe structures, whereas letter-gener-
tasks are frequently included in neuropsycholog- ated fluency may be primarily mediated by the
ical test batteries to assess language and retrie- frontal lobes. In a logical extension of Mosco-
val skills (Benton & Hamsher, 1989) and are vitch’s model, age-associated differences in per-
considered sensitive measures for early detec- formance on the two fluency tasks may reflect
tion of dementia of the Alzheimer type (Bayles, differences in the way that normal aging affects
Tomoeda, & Trosset, 1993; Hart, Smith, & the biological substrate underlying these tasks.
Swash, 1988; Huff, Corkin, & Growdon, 1986; Similarly, differences among subgroups of indi-
Martin & Fedio, 1983; Monsch et al., 1992; viduals with dementia may also reflect differ-
Ober, Dronkers, Koss, Delis, & Friedland, 1986; ences in the way pathological processes affect
Rosen, 1980). Consistent with the superiority of the brain. For example, both normal aging and
Downloaded by [University of California Santa Cruz] at 08:07 16 November 2014
category fluency in detecting normal age-related early-stage dementia of the Alzheimer type are
cognitive change, category fluency is generally known to affect the mesial temporal lobe struc-
found to be superior to letter fluency in discrimi- tures, with relative sparing in the frontal brain
nating early stage dementia from normal aging areas (see Kolb & Whishaw, 1996 for a brief
(Butters, Granholm, Salmon, Grant, & Wolf, review); dementias described as subcortical or
1987; Monsch et al., 1992), although others have vascular may interfere with both temporal and
not demonstrated this difference (Bayles et al., frontal functions.
1989; Ober et al., 1986; Rosen, 1980). Recently category and letter fluency tasks
Category and letter fluency differences are have been used to differentiate dementias that
usually attributed to the relative preservation, in are considered primarily cortical (e.g., Pasquier,
both normal aging and in early stage dementia, Lebert, Grymonprez, & Petit, 1995), and to dis-
of cognitive processes thought to underlie letter tinguish cortical from subcortical dementias
fluency as compared to those associated with (e.g., Monsch et al., 1994; Randolph, Braun,
category fluency. For example, Jorm (1986) ar- Goldberg, & Chase, 1993; Stern & Richards,
gued that the letter-sound associations or phono- 1993; Troster, Salmon, McCullough, & Butters,
logical knowledge necessary for letter fluency 1989). For example, Monsch et al. (1994) com-
may be well preserved compared to semantic- pared category and letter fluency in groups of
exemplar associations necessary for category individuals with dementia of the Alzheimer type
fluency. Similarly, Butters et al. (1987) and (DAT) or Huntington’s disease (HD) and found
Monsch et al. (1992) concluded that the superior that the participants with HD were equally im-
sensitjvity of category fluency is due to its de- paired on both tasks, whereas those with DAT
pendence on the structure of semantic knowl- were proportionally more impaired on the cate-
edge. This store of semantic information is gory task. Differences were attributed to a
thought to deteriorate in early stages of demen- breakdown in the structure of semantic knowl-
tia, and may be increasingly difficult to access edge in DAT in contrast to difficulties in initia-
or retrieve in normal aging. Differences between tion and retrieval in HD.
letter and category fluency may also reflect re- Although current research suggests that corti-
duced general processing capacity or attentional cal and subcortical dementias may be distin-
resources in normal aging and in early-stage guishable using letter and category fluency
dementia, such that accessing semantically cued tasks, these tasks do not consistently distinguish
versus phonemically cued exemplars may be among patients who suffer different types of
relatively difficult or resource-demanding (Hart- cortical dementia. For example, Pasquier et al.
man, 1991; Huff et al., 1986). (1995) compared category and letter fluency
Moscovitch (1995) proposed a neuropsycho- performance in groups of individuals with fron-
logical explanation for differences in perfor- tal lobe dementia or DAT. They found that pa-
mance on letter and category fluency tasks. Ac- tients had lower word fluency than age- and gen-
cording to his model, category or semantic flu- der-matched controls on both tasks, and that cat-
ency tasks may be carried out by the mesial- egory fluency was more impaired than letter flu-
54 MARGARET CROSSLEY ET AL
ency for both dementia types. Neither category groups (although see Benton, Eslinger, & Dama-
nor letter fluency differentiated frontal lobe de- sio, 1981). The current study included a large
mentia and DAT. Finally, Margolin, Pate, Frie- sample of community-dwelling older adults di-
drich, and Elia (1990) compared naming ability vided into young-old (65-74 years), moderately-
in individuals with DAT and anomic stroke pa- old (75-84 years) and old-old (85+ years) age
tients using letter fluency and a semantic naming groups.
(Boston Naming Test) task. Both patients with Both above- and below-average educational
DAT and anomic stroke were impaired on the levels can confound the assessment of normal
semantic naming task as compared to normals, elderly and the diagnosis of those with early-
but those with DAT were significantly more im- stage dementia (see Lezak, 1995 for a discussion
paired than those with anomic stroke. A very of this issue). In addition, gender differences are
Downloaded by [University of California Santa Cruz] at 08:07 16 November 2014
different pattern occurred for the letter fluency frequently observed in measures of verbal flu-
task. The patients with anomic stroke were im- ency with normal individuals (Bolla et al., 1990;
paired as compared to both normal participants Lezak, 1995), but few have investigated these
and patients with DAT, whereas the DAT group differences in individuals with dementia
did not differ from normal controls. These find- (although see Monsch et al., 1992). In the cur-
ings at least tentatively suggest that fluency rent study, we systematically examined educa-
measures may successfully differentiate individ- tional and gender effects on verbal fluency per-
uals with DAT from those with vascular demen- formance for both normal participants and indi-
tia; however, to date, no studies have compared viduals diagnosed with DAT.
the performance of these two groups on letter As suggested by Butters et al. (1987) and
and category fluency. Margolin et al. (1990), lack of consistency
The current study was conducted in order to across studies designed to investigate letter and
replicate previous findings with normal older category fluency performance in dementia may
adults and those with DAT using a large com- be partially explained by differences in level of
munity-dwelling sample identified as part of the severity in the patient groups. For example, let-
Canadian Study on Health and Aging (CSHA; ter fluency may diminish later in the course of
Canadian Study of Health and Aging Work DAT than category fluency (Margolin et al.,
Group, 1994a). In addition, we aimed to extend 1990). In an attempt to control for the effect of
the use of category and letter fluency tasks in the illness severity, we subgrouped participants with
investigation of dementia subtypes by including vascular dementia and DAT into mild and mod-
the sample of participants from the CSHA with erate impairment categories.
vascular dementia.
The CSHA was a nationwide epidemiological
study of the cognitive status of adults aged 65 METHOD
years and older that provided estimates of the
prevalence of dementia in Canada (CSHA Participants
Working Group, 1994a), identified risk factors As part of the CSHA, 9,008 community-dwelling
for dementia (CSHA Working Group, 1994b), seniors from 18 centres across Canada were
and described the impact on caregivers of per- screened for cognitive impairment using the Modi-
fied Mini-Mental State Examination (3MS; Teng
sons with dementia (CSHA Working Group, & Chui, 1987). Individuals ( n = 1,165) who fell
1 9 9 4 ~ ) .The epidemiological nature of the below a predetermined cut-off score for normal
CSHA allowed us to address a number of meth- performance (i.e., 78/100 on the 3MS), and a sub-
odological issues associated with the study of sample of individuals (n = 494) who performed
verbal fluency in normal and pathological aging. above the cut-off score, were invited to participate
For example, normal aging studies have typi- in the clinical component of the study. The clinical
component was designed to confirm the presence
cally compared young, middle-aged, and older or absence of cognitive impairment, and, in the
individuals, and have seldom differentiated former case, to identify the type of impairment.
young-old, moderately-old, and old-old age For individuals scoring above 50 on the 3MS, the
LETTER AND CATEGORY FLUENCY IN THE CSHA 55
clinical component comprised a nurse’s evaluation, by gender, age, education, and consensus diagno-
a standardized medical examination, and a detailed sis. Almost half the sample had less than 10 years
neuropsychological assessment. Those scoring be- of formal education.
low 50 on the 3MS were assessed by the physician The consensus diagnosis, by definition, re-
and the nurse, but were considered to be too im- flected a consideration of the neuropsychological
paired to complete neuropsychological testing. test results, including the results of the category
A consensus diagnosis by the clinical research and letter fluency measures. To address any con-
team in each centre was derived using the criteria cern about circularity, all analyses in the current
for primary degenerative dementia from the Diag- study were repeated on groups formed by the phy-
nostic and Statistical Manual of Mental Disorders, sicians’ preliminary diagnosis. There were no sig-
third edition, revised (DSM-111-R; American Psy- nificant differences between the two sets of analy-
chiatric Association, 1987) and the criteria for ses. Since the groupings generated by consensus
Downloaded by [University of California Santa Cruz] at 08:07 16 November 2014
DAT developed by the National Institute of Neu- diagnosis better reflect the criteria for diagnosis as
rological and Communicative Disorders and set out by DSM-111-R and by NINCDS-ADRDA,
Stroke and the Alzheimer’s Disease and Related that specifically require a consideration of neuro-
Disorders Association (NINCDS-ADRDA; psychological test performance, the results report-
McKhann et al., 1984). Participants were classi- ed in this study are based on the consensus group-
fied as having either no cognitive loss, cognitive ings.
loss but no dementia, or some form of dementia The Canadian Study of Health and Aging
(i.e., Possible or Probable Alzheimer type, Vascu- (1994a) provides a detailed overview of the study
lar, Mixed, Other, or Unclassifiable). Table 1 sum- participants, methods, and findings, and Tuokko,
marizes the demographic characteristics of the Kristjansson, and Miller (1995) describe the neuro-
community-dwelling participants of CSHA who psychological component of the CSHA. The large-
completed the neuropsychological component of scale survey and sampling procedures of the
the clinical assessment. Participants are grouped CSHA provided an unusually broad and represen-
Age Education
M SD n M SD n
Gender
Males 78.8 6.7 545 8.7 4.4 544
Females 80.5 6.9 764 9.0 3.9 763
Education (years)
0-6 79.2 7.1 37 1 4.1 1.7 37 1
7-9 79.8 6.9 378 8.0 0.7 378
10--12 80.1 6.7 368 11.0 0.8 368
13+ 80.0 6.4 190 15.7 2.6 190
Consensus Diagnosis
Normal 78.8 6.8 645 9.5 4.1 644
Cognitive loss 80.1 7.1 420 8.1 4.0 420
Alzheirner 82.5 5.9 164 8.4 3.9 163
Vascular 80.8 6.0 40 9.6 4.3 40
OtherlLinident. 80.0 6.4 40 8.2 4.3 40
All participants 79.8 6.8 1309 8.8 4.1 1307
56 MARGARET CROSSLEY ET AL.
Table 17. Letter and Category Fluency Performance for Cognitively Normal Participants according to Age,
Gender, and Educational Level
Fluency Task
FASa (letter fluency) Animal Namingb (category fluency)
Group M SD n M SD n
Age Group (years)
65-74 24.0 12.4 139 14.2 4.3 144
75-84 25.8 11.5 343 14.2 3.8 343
85 + 24.0 10.8 146 12.5** 3.8 148
Downloaded by [University of California Santa Cruz] at 08:07 16 November 2014
Gender
Male 23.2 12.1 258 14.2 4.2 258
Female 26.2* 11.0 370 13.6 3.9 317
Education (years)
0-6 16.2 6.9 140 12.1 3.1 149
7-9 23.7 9.9 170 13.4 3.8 169
10-12 27.0 10.2 202 14.1 3.9 203
13+ 34.2** 12.6 115 16.3** 4.1 113
Sample 25.0 11.6 628 13.8 4.3 635
Note. Values for fluency tasks are compared vertically, in table columns, according to age, gender, and educa-
tional level.
a The values are the total number of words beginning with specified letters reported orally in three 1 min. trials.
The values are the number of animal names reported orally in I-min.
*p<.o1.**p<.oo1.
Table 3. Letter and Category Fluency Performance for Community-Dwelling Elderly Diagnosed with Dementia
of the Alzheimer Type.
Fluency Task
Group M SD n M SD n
Age Group (years)
65-74 17.0 8.4 13 8.2 2.1 13
75-84 16.4 10.6 73 8.7 3.9 75
85+ 16.0 9.1 66 8.3 3.4 67
Downloaded by [University of California Santa Cruz] at 08:07 16 November 2014
Gender
Male 15.5 10.1 56 9 .O 4.6 58
Female 16.8 9.6 96 8.2 2.7 97
Education (years)
0-6 11.9 7.4 48 9.1 3.7 49
7-9 14.6 8.9 45 7.9 3.5 46
10-12 20.1 10.0 39 8.3 2.9 40
13+ 22.6** 10.3 19 8.7 4.2 19
Impairment
Mild 19.5 10.1 46 9.5 3.5 46
Moderate 14.9 9.4 105 8.1* 3.5 108
Total N 16.3 9.8 152 8.5 3.5 155
Note. Values for fluency tasks are compared vertically, in table columns, according to age, gender, and education,
and impairment.
a The values are the total number of words beginning with specified letters reported orally in three I-min trials.
The values are the number of animal names reported orally in 1-min.
p < .05. ** p c .001.
Table 4. Letter and Category Fluency Performance for Community-Dwelling Cognitively-Normal Elderly and
for Elderly with Dementia of the Alzheimer or Vascular Type at Mild and Moderate Levels of Impair-
ment.
Fluency Task
Group M SD n M SD n
Note. Values for fluency tasks are compared vertically, in table columns, according to cognitive group and level
of impairment.
a The values are the total number of words beginning with specified letters reported orally in three 1-min trials.
The values are the number of animal names reported orally in 1 min.
Mild impairment refers to 3MS scores > 70.
Moderat:*impairment refers to 3MS scores I 70.
p < .01. p < .001.
LETTER AND CATEGORY FLUENCY IN THE CSHA 59
than the normal participants while performing both measures to standard scores (i.e., z scores),
letter fluency, F(2,714) = 38.57, p < .001, and expressed in standard deviations from the mean
category fluency, F(2,714) = 97.12, p < .001, of the normal seniors’ performance (see Figure
tasks. In summary, both fluency measures suc- 1). A 2 (DAT vs. vascular dementia) x 2 (mild
cessfully distinguished between the cognitively vs. moderate severity) x 2 (letter vs. category
normal group and the two dementia groups at fluency) ANOVA was performed on these stan-
both mild and moderate levels of impairment, dard scores. A significant main effect for level
but failed to differentiate DAT from vascular of severity, F(1, 374) = 1 3 . 8 5 , <
~ .001, revealed
dementia at either impairment level. Level of that moderately impaired participants performed
performance on the two tasks was significantly more poorly than mildly impaired participants
correlated for normal participants (Spearman’s for both fluency tasks, regardless of diagnosis.
Downloaded by [University of California Santa Cruz] at 08:07 16 November 2014
rho = .44, p < .001), for those with DAT (Spear- A significant main effect for fluency task, F(1,
man’s rho = .29, p < .OOl), and for those with 374) = 44.17, p < .001, confirmed that category
vascular dementia (Spearman’s rho = .36, p = fluency was significantly more affected than
.014). letter fluency at both mild and moderate levels
of impairment, regardless of diagnosis. There
Letter Versus Category Fluency in DAT and were no significant effects for dementia type
Vascular Dementia (DAT vs. vascular dementia), nor were there any
Letter and category scores were directly com- significant interactions among the three factors.
pared by converting each participant’s score on
0 LETTER
1 CATEGORY
-1
-2 I I I I
MILD DAT MOD DAT MILD VASC MOD VASC
Standard scores of participants with mild dementia of the Alzheimer type (Mild DAT), moderate de-
mentia of the Alzheimer type (Mod DAT), mild vascular dementia (Mild Vasc), and moderate vascular
dementia (Mod Vasc) on letter and category fluency tasks.
60 MARGARET CROSSLEY ET AL.
(FAS) task was age-stable. Females performed sensitivity to the effects of education, age, and
better than males on the letter fluency task, but gender among individuals with DAT. In con-
there was no significant gender difference on the trast, interpretation of performance on the letter
category task. Taken together, these differential fluency task must take into account the effects
findings provide some additional support for the of educational level that is highly significant for
thesis that letter and category fluency tasks are both normal older individuals and those with
measuring unique aspects of language, and may DAT.
have different brain-based substrates that deter- One goal of this study was to examine the
mine their vulnerability to the effects of normal ability of the letter and category fluency tasks to
aging. differentiate those with DAT from vascular de-
Educational level had a significant effect on mentia at mild and moderate stages of impair-
the performance of normal seniors for both the ment. Although the relatively small sample of
letter and category naming tasks. Since 50% of individuals with vascular dementia in the CSHA
the elderly participants reported less than nine may have limited our investigation, these results
years of formal education, and 21% had five or suggest that category and letter fluency tasks are
fewer years of schooling, the importance of con- sensitive to dementia and level of impairment,
sidering this demographic variable is clear. In regardless of type, but are unable to distinguish
contrast to these data, studies of normal elderly DAT from vascular dementia.
are typically based on convenience samples that In summary, these data provide some addi-
are frequently biased in the direction of relati- tional support for the thesis that category and
vely high education and socioeconomic levels. letter fluency tasks measure unique aspects of
The relatively large sample of individuals cognition. In particular, syntactical or phonolog-
with DAT allowed an investigation of the effects ical knowledge (required for letter fluency) may
of age, gender, education, and level of impair- be less affected than semantic knowledge (re-
ment (i.e., mild or moderate) on fluency perfor- quired for category fluency), both by normal
mance in the DAT group. Age and gender did aging and by early-stage dementia. Consistent
not influence performance of DAT participants with this thesis is the proposition that syntactical
on either fluency measure. Educational level and semantic knowledge systems may have dif-
significantly influenced performance on the let- ferent underlying biological substrates. Alter-
ter fluency task, but not on the category fluency nately, fluency tasks may differentially draw
task. Interestingly, highly educated participants resources from a limited capacity store that is
with DAT performed similarly on the letter flu- affected both by normal aging and by dementia.
ency task to cognitively normal participants with In any case, letter and category fluency did not
low levels of formal education. In contrast, on differentiate DAT and vascular dementia, sug-
the category task, even the most highly educated gesting that both forms of dementia may simi-
participants with DAT performed well below the larly affect the brain-based processes underlying
level of the least well educated normal partici- these tasks.
LETTER AND CATEGORY FLUENCY IN THE CSHA 61
Parkinson’s,and Huntington’s disease: Dissocia- Teng, E. L., & Chui, H. C. (1987). The modified
tion of storage and retrieval failures. Neuropsycho- Mini-Mental State (3MS) Examination. Journal of
logy, 7, 82-88. Clinical Psychiatry, 48, 314-318.
Rosen, W. G . (1980). Verbal fluency in aging and de- Tomer, R., & Levin, B. E. (1993). Differential effects
mentia. Journal of Clinical Neuropsychology, 2, of aging on two verbal fluency tasks. Perceptual
135-146. and Motor Skills, 76, 465-466.
Ska, B. (1993). Quelques observations a propos de Troster, A.I., Salmon, D. P., McCullough, D., & But-
modifications lexico-semantiques lors du vieilliss- ters, N. (1989). A comparison of the category flu-
ement normal. (1993). Revue de Neuropsychologie, ency deficits associated with Alzheimer’s and
3, 157-180. Huntington’s disease. Brain and Language, 37,
Spreen, O., & Benton, A. L. (1977). Neurosensory 500-513.
centre comprehensive examination for aphasia: Tuokko, H., Kristjansson, B., & Miller, J. (1995).
Manual of instructions. Victoria, B.C.: University Neuropsychological detection of dementia: An
Downloaded by [University of California Santa Cruz] at 08:07 16 November 2014