Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Jan 1.
Published in final edited form as: Neuropsychology. 2019 Jun 13;34(1):31–42. doi: 10.1037/neu0000576

Intra-Individual Variability in Verbal Fluency Performance is Moderated by and Predictive of Mild Cognitive Impairments

Roee Holtzer 1, Sydney Jacobs 2, Eleni Demetriou 3
PMCID: PMC6908771  NIHMSID: NIHMS1037401  PMID: 31192656

Abstract

Objective:

The current study was designed to achieve two primary objectives: 1) determine the moderating effect of Mild Cognitive Impairments (MCI) on intra-individual variability in semantic and letter fluency performance taking into account longitudinal annual assessments; 2) establish predictive utility for intra-individual variability in semantic and letter fluency performance as a risk factor of incident MCI.

Methods:

Participants were community-residing older adults (n=514; mean age =75.89±6.45; percent female=55.1). Sixty participants were diagnosed with MCI at baseline and 50 developed incident MCI during the follow-up. We operationalized intra-individual variability as word generation slopes derived from three consecutive time intervals during the standard one-minute administration of both letter and semantic fluency tasks (i.e. 0–20sec, 21–40sec, 41–60sec).

Results:

Linear Mixed Effects Models revealed significant within task slope effects for semantic (estimate=−8.350; p<0.0001; 95%CI=−8.604 to −8.095) and letter (estimate=−5.068; p<0.0001; 95%CI=−5.268 to −4.869) fluency indicating that word generation declined over the course of both tasks. The two-way interactions of MCI x slope were significant for semantic (estimate=1.34; p=0.001; 95%CI=0.551 to 2.126) and letter (estimate=0.733; p=0.020; 95%CI=0.116 to 1.350) fluency indicating attenuated slopes among MCI participants compared to controls taking into account repeated annual assessments. Cox proportional-hazards models revealed that attenuated word generation slope, at baseline, in semantic (HR=2.063; p=0.015; 95%=1.149 to 3.702) but not letter (HR=0.704; p=0.243; 95%CI=0.391 to 1.269) fluency was associated with increased risk of incident MCI.

Conclusion:

Intra-individual variability in verbal fluency performance has clinical and predictive utility; it can be easily incorporated into testing batteries in clinical and research settings.

Keywords: Intra-Individual Variability, Verbal Fluency, Aging, Mild Cognitive Impairments

Semantic and letter fluency tests require the individual to generate, under timed conditions, words that belong to a specific category or begin with a specified letter, respectively (Lezak, Howieson, Loring, Hannay, & Fischer, 2004). These tests are often included in neuropsychological testing batteries employed in clinical and research settings because of their brevity and established norms (Holtzer, Goldin, et al., 2008), as well as utility at elucidating the diagnosis of neurodegenerative diseases based on distinct performance patterns (Marczinski & Kertesz, 2006; Monsch et al., 1992). For example, disproportionate impairment in semantic relative to letter fluency is frequently observed in early Alzheimer’s disease (Henry, Crawford, & Phillips, 2004; Monsch et al., 1992), which is followed by letter fluency decline that worsens as a function of disease progression (Keilp, Gorlyn, Alexander, Stern, & Prohovnik, 1999). Although limited, longitudinal patterns of performance among older adults suggest a more prominent downward trajectory in semantic fluency in cognitively healthy individuals at risk of Alzheimer’s disease (i.e. Ab positive) (Papp et al., 2016; Snitz et al., 2013) as well as preclinical Alzheimer’s disease (Clark et al., 2009) compared to older adults who remain cognitively intact or have no genetic risk. Findings concerning letter fluency are less consistent with reports of intact performance in individuals who remained cognitively healthy over time and in those who converted to preclinical Alzheimer’s disease (Clark et al., 2009). Improved letter fluency performance over repeated evaluations in both controls and individuals at risk, however, has also been reported (Snitz et al., 2013).

Verbal Fluency and Mild Cognitive Impairments

Mild Cognitive Impairments (MCI) is a transition state in aging that is associated with increased risk of incident dementia. The diagnosis of MCI requires evidence for objective cognitive impairment (Petersen et al., 2014), based on performance cutoff scores typically at 1 or 1.5 standard deviations below the mean on neuropsychological test scores (Albert et al., 2011). The presence of subjective cognitive complaints is required for the diagnosis of MCI; while activities of daily living are preserved, relatively minor declines are observed in instrumental activities of daily living (Petersen et al., 2014; Winblad et al., 2004). Subtle cognitive impairments in individuals with MCI may manifest across different cognitive domains that include episodic memory (Summers & Saunders, 2012), semantic memory (Wilson, Leurgans, Boyle, & Bennett, 2011), processing speed, attention, working memory (Summers & Saunders, 2012), and executive functions (Brandt et al., 2009; Summers & Saunders, 2012; Traykov et al., 2007). Different MCI subtypes have been proposed to capture this variability. The Amnestic MCI (aMCI) subtype is characterized by impairments in memory whereas the non-amnestic (naMCI) subtype requires impairments in cognitive domains other than memory (Winblad et al., 2004). MCI can be further classified as single or multiple domains (Petersen, 2004).

Review of the literature revealed mixed findings with respect to the effect of MCI on verbal fluency performance. While some studies reported that participants with MCI performed worse than controls (Brandt & Manning, 2009; Malek-Ahmadi, Small, & Raj, 2011; Murphy, Rich, & Troyer, 2006; Price et al., 2012), other studies found that performance on these measures was not associated with MCI status (Traykov et al., 2007). Older adults with multiple cognitive impairments, similarly to those diagnosed with Alzheimer’s Dementia (AD), typically have worse semantic than letter fluency performance (Brandt & Manning, 2009; Nutter-Upham et al., 2008). It is noteworthy, however, that differences in verbal fluency performance between individuals with MCI and healthy controls are relatively small (Malek-Ahmadi et al., 2011; Murphy et al., 2006). Furthermore, because standard fluency scores have limited utility at discriminating MCI and normal aging (Radanovic et al., 2009), it is of interest to examine whether novel methods that quantify performance on these tests, such as within task intra-individual variability, are associated with prevalent MCI and predict its incidence.

Intra-individual variability in verbal fluency and MCI

Intra-individual variability in cognitive performance has been conceptualized and operationalized in multiple ways (Holtzer, Verghese, Wang, Hall, & Lipton, 2008). Common across different conceptual and operational definitions is the notion that intra-individual variability measures a change in cognitive function within a person (MacDonald, Nyberg & Backman 2006). Evidence suggests intra-individual variability is a sensitive marker of nervous system integrity (Hultsch, MacDonald, Hunter, Levy-Bencheton, & Strauss, 2000) that provides incremental information beyond central tendency measures (Dixon et al., 2007). In verbal fluency, intra-individual variability can be operationalized using the slopes of word generation during the standard one-minute administration of the task. Specifically, word generation performance declines during the course of the task (Butters, Granholm, Salmon, Grant, & Wolfe, 1987; Crowe, 1998; Demetriou & Holtzer, 2017; Fernaeus & Almkvist, 1998; Ober, Dronkers, Koss, Delis, & Friedland, 1986; Raboutet et al., 2010); initially the individual rapidly generates words that are readily available, but as time passes by word retrieval becomes more effortful and slower (Crowe, 1998). Previous work provided initial evidence that examination of within task performance variability may be a more sensitive indicator of age related transition states. Separate cross-sectional studies revealed that within task trajectories in semantic fluency varied as function of dementia subtype (Lamar, Price, Davis, Kaplan, & Libon, 2002) and letter fluency trajectories varied across MCI subtypes (Eppig et al., 2012). Individuals with MCI demonstrated attenuated declines in performance over the course of the standard one-minute administration of both semantic and letter fluency in comparison to healthy older adults (Demetriou & Holtzer, 2017). These findings were also based on cross-sectional data. Longitudinal effects on within task intra-individual variability in verbal fluency measures or the predictive utility of such measures have yet to be established.

Current Study

The current study was designed to examine longitudinal associations of MCI with intra-individual variability in verbal fluency performance. We operationalized intra-individual variability as word generation slopes derived from three consecutive time intervals during the standard one-minute administration of both letter and semantic fluency tasks (i.e. 0–20sec, 21–40sec, 41–60sec). First, we evaluated the moderating effect of MCI on word generation slopes over repeated annual assessments while adjusting for total fluency scores. We specifically assessed whether annual follow-up influenced the moderating effect of MCI on word generation slopes. Second, we determined via correlational analysis whether individual word generation slopes measured a different construct than total verbal fluency scores. We then determined the predictive utility of individual word generation slopes (our proxy of intra-individual performance variability) as a risk factor of incident MCI. Because total fluency scores but not individual word generation slopes were used to determine MCI status, using the former scores in predictive models of incident MCI introduced diagnostic circularity. Hence, the primary predictive models of incident MCI excluded total verbal fluency scores. We hypothesized that the presence of MCI would be associated with attenuated word generation slopes during the standard one-minute administration of letter and semantic fluency tasks. We further hypothesized that attenuated word generation slopes in letter and semantic fluency at baseline would predict increased risk of incident MCI.

METHODS

Participants:

Participants were community-residing older adults (age≥65yrs) enrolled in “Central Control of Mobility in Aging” (CCMA), a cohort study designed to determine cognitive and brain predictors of mobility. CCMA procedures were described in previous publications (Holtzer, Mahoney, & Verghese, 2014). Participants whose baseline verbal fluency data were previously reported (Demetriou & Holtzer, 2017) were included in the current longitudial study. Potential participants were identified from population lists of lower Westchester County, NY. Structured telephone interviews were administered to obtain verbal assent and determine initial eligibility including screens for dementia (Galvin, Roe, Xiong, & Morris, 2006; Lipton et al., 2003) and assessments of medical history and current physical function levels. Individuals who passed the telephone interview were invited to two annual in-person study visits during which trained research assistants administered comprehensive neuropsychological, psychological, and mobility assessments. The neuropsychological testing battery assessed cognitive domains that included premorbid ability, current level of cognitive status, processing speed, attention, visual spatial processing, visual and verbal memory and executive functions. The study clinician conducted structured neurological examinations. Exclusion criteria were: current or history of age-related neuro-degenerative diseases, dementia, severe neurological or psychiatric disorders, inability to ambulate independently, significant loss of vision and/or hearing that threatened the validity of the testing procedures, and recent or anticipated medical procedures that may affect ambulation and functional capacity. Dementia diagnoses were determined via consensus diagnostic case conference procedures (Holtzer, Verghese, et al., 2008). Participants who developed a neurodegenerative disease during the follow-up (e.g., dementia, PD) were excluded from the current study.

MCI Status:

The presence of MCI was determined during consensus diagnostic case conference procedures based on published guidelines (Albert et al., 2011; Winblad et al., 2004). The following criteria were used: performance at or below 1.5 standard deviation below the mean for age and education adjusted norms in at least two tests in one or more cognitive domains, relatively preserved instrumental activities of daily living, and absence of dementia. Cognitive complaints were assessed through structured clinical interviews and questionnaires (Galvin et al., 2005; Katz, 1983).

The study was in compliance with the Code of Ethics of the World Medical Association (Declaration of Helsinki) and with the standards established by the institutional review board of Albert Einstein College of Medicine. Written informed consents were obtained in-person and approved by the institutional Review Board.

Outcome Measures

The Control Word Oral Associated Test (COWAT; Spreen & Benton, 1977) was administered according to standard procedures. Letter Fluency: Using the letters F, A and S, each administered in a separate 60-sec trial, participants were instructed to provide as many words that begin with the specified letter. Participants were instructed to avoid providing responses that consisted of proper nouns or responses with different suffixes. Proper nouns, words with different endings, repetitions, and perseverations were considered incorrect responses. Semantic Fluency: Using the categories of fruits, animals, and vegetables, each administered in a separate 60-sec trial, participants were required to name as many words as possible that belonged to each of the specified categories. Repetitions and perseverations were classified as incorrect responses. For both letter and semantic fluency only correct responses were used to determine performance and standing relative to normative data. Only correct responses were used for the statistical analyses as well. Time Intervals: Different time intervals have been used in the literature to examine intra-individual performance patterns within the standard 60s administration of verbal fluency ranging from 10 to 30 seconds (Fernaeus & Almkvist, 1998; Hurks et al., 2010; Raboutet et al., 2010; Weakley, Schmitter-Edgecombe, & Anderson, 2013). Consistent with previous research (Demetriou & Holtzer, 2017), responses were recorded at 0–20, 21–40 and 41–60sec time intervals without altering the standard administration of the tests. The number of words produced at 0–20 s [T1], 21–40 s [T2] and 41–60 s [T3] across the three trials of each fluency task was summed and used to derive individual word production slopes. Letter fluency was administered first followed by semantic fluency.

Covariates:

The Wide Range Achievement Test third edition (WRAT-3) provided estimates of premorbid intellectual function (Wilkinson, 1993) The Geriatric Depression Scale (GDS) was used to assess depressive symptoms (Yesavage et al., 1982). Demographic and health information were assessed via structured clinical interviews including a structured medical evaluation conducted by the study clinician. Using these data, a general health status (GHS) summary score was computed based on dichotomous rating indicating the presence or absence of 10 chronic conditions for each participant with scores ranging from 0–10 (Holtzer, Verghese, et al., 2008). The GHS included the following conditions: diabetes, chronic heart failure, arthritis, hypertension, depression, stroke, Parkinson’ s disease, chronic obstructive lung disease, angina, and myocardial infarction (Holtzer, Verghese, et al., 2008).

Statistical Analysis

Separate Linear Mixed Effects Models (LMEMs) were used to determine the effects of group, time (annual assessments), within task word generation slopes and their interactions on letter and category fluency performance. Specifically, in each model, cognitive status (MCI vs. controls) served as a two-level covariate. Annual assessments served as the repeated measures time variable. Within task time intervals (T1, T2 and T3) also served as a repeated measures variable that was used to derive individual word generation slopes for each fluency task. The total number of correct words in each of the three time intervals served as the dependent measures. The moderating effects of MCI on within task word generation slopes and on annual changes in performance were tested via two-way interactions. Three-way interactions (group status x time x within task word generation slopes) were used to determine whether the moderating effects of MCI on word generation slopes changed over annual assessments. Age, gender, education, WRAT-3 score, letter and semantic fluency scores (in their respective models), GDS score, and global health status score were used as covariates. Cross-sectional correlations between slopes and verbal fluency performance levels were examined among participants who did not have MCI at baseline, had a least one year of follow-up and complete diagnostic workup, based on case conference procedures, during their follow-up. Separate Cox proportional-hazards models were used to examine whether baseline within task word generation slopes of letter and semantic fluency (dichotomized at the median) predicted the risk of incident MCI. We used age as the time scale given that previous literature suggested age is a more appropriate time scale than follow-up time in cohort studies (Thiebaut & Benichou, 2004). Age at event was defined as the age when incident MCI occurred or age at final study contact. Prevalent MCI cases were excluded from this analysis. Models were checked for proportional hazard assumptions and were adequately met. Because letter and semantic fluency word generation slopes have not been used in prior research to predict incident MCI, adjustments for covariates was sequential. First, unadjusted analyses (other than age which served as the time scale) were examined. Second, analyses adjusted for gender, education, WRAT-3 score, GDS score, and global health status score. In additional sensitivity analyses letter and semantic fluency slopes were used simultaneously as predictors in one Cox model. Letter and semantic fluency normative scores were added to separate cox models, though it should be recognized that their addition to the models introduces circularity because scores on both tests were used to determine diagnostic status. Finally, to address the decline in sample size over repeated annual assessments additional sensitivity analyses (both LMEMs and cox) excluded the last two years of follow-up. SPSS statistical software package, (version 25; SPSS, Inc., Chicago, IL) was used for statistical analysis.

RESULTS

Older adults (n=514; mean age=75.89±6.45ys; mean education=14.6±2.92ys; %female=55.1) who were recruited and tested between June 2011 and May 2018 participated in the current study. The mean WRAT-3 standard score (106.92±10.11) was indicative of average pre-morbid function. Mean disease comorbidity index (1.64±1.09) and GDS (4.62±3.90) scores were low indicating the sample was relatively healthy and with minimal depressive symptoms, respectively. The mean follow-up time for the sample was 2.95±1.64ys. The mean time from baseline to MCI diagnosis was 1.80±1.01ys. Individuals diagnosed with dementia over the course of the study (n=34) were excluded from analyses. Additionally, individuals who had not been evaluated in diagnostic case conferences at the time of data analysis (n=44) were excluded from analyses. At baseline, cognitively normal and MCI participants did not differ in terms of GHS (p=0.341), GDS (p=0.124), and gender (p=0.402). Participants with MCI at baseline were older (p=0.004), more ethnically diverse (p < 0.001), reported fewer years of education (p < 0.001), and had lower standardized scores on the WRAT and verbal fluency measures (p < 0.001) compared to cognitively normal individuals. Participants who were included in the current study were not different from those excluded in terms of gender (p=0.514), education (p=0.534), and GDS (p=0.126). However, excluded participants were significantly older (p=0.004) and reported lower GHS scores (p<0.001). There were 60 participants diagnosed with MCI at baseline and 50 participants who developed incident MCI during the follow-up period. Demographic information and descriptive statistics were summarized and stratified by MCI status in Table 1. Summary of the total number of correct words generated in letter and semantic fluency stratified by within task time interval, annual assessments and MCI status is presented in Table 2.

Table 1:

Demographic characteristics at baseline for the total sample stratified by MCI group status

All (n=514) Normal (n=454) Incident MCI (n=50) Prevalent MCI (n=60) Normal vs. Prevalent MCI
Variable M SD M SD M SD M SD p
Age (years) 75.89 6.45 75.60 6.35 77.42 6.24 78.13 6.83 .004
Education (years) 14.6 2.92 14.8 2.84 15.30 2.69 13.10 3.12 < .001
WRAT-3 Standard Score 106.92 10.11 107.89 9.52 107.22 9.85 99.60 11.44 < .001
GHS 1.64 1.09 1.62 1.09 1.70 1.12 1.77 1.16 .341
GDS 4.62 3.90 4.53 3.94 5.54 4.52 5.35 3.53 .124
Semantic Fluency Z-score 0.15 1.14 0.23 1.12 −0.24 1.29 −0.50 1.09 < .001
Letter Fluency Z-score 0.27 1.23 0.42 1.16 −0.09 1.06 −0.91 1.09 < .001
Sex (%female) 55.1 55.7 54.0 50.0 .402
Ethnic (%) < .001
   Caucasian 80.5 83.0 82.0 61.7
   Other 19.5 17.0 18.0 38.3
Open in a new tab

Note: Dataset includes participants who completed all baseline measures from June 2011 – May 2018; T tests for independent samples and chi-square were used to assess group differences on continuous and categorical variables, respectively; MCI = mild cognitive impairment; WRAT = Wide Range Achievement Test (reading standard score) third edition; GHS = general health status; GDS = geriatric depression scale. Z scores for semantic and letter fluency are based on published normative data (Holtzer et al., 2008).

Table 2:

Verbal fluency performance stratified by year and MCI group status

Semantic Fluency Letter Fluency
Variable M SD M SD M SD M SD M SD M SD
Year 1 All (n=514) Normal (n=454) MCI (n=60) All (n=514) Normal (n=454) MCI (n=60)
0–20 s 23.87 5.18 24.50 4.96 19.12 4.30 19.14 5.82 19.74 5.61 14.67 5.42
21–40 s 10.81 4.03 11.26 3.95 7.43 3.02 11.45 4.93 11.76 4.88 8.88 4.51
41–60 s 7.22 3.73 7.50 3.71 4.90 2.89 9.20 4.56 9.54 4.56 6.50 3.53
Total 41.89 9.89 43.26 9.34 31.45 7.46 39.76 13.23 41.04 12.89 30.05 11.75
Year 2 All (n=395) Normal (n=353) MCI (n=42) All (n=395) Normal (n=353) MCI (n=42)
0–20 s 23.30 5.19 23.87 4.82 18.50 5.78 19.60 5.56 20.13 5.35 15.12 5.32
21–40 s 11.27 4.43 11.75 4.29 7.21 3.48 11.60 4.81 12.00 4.75 8.21 3.91
41–60 s 7.71 3.80 8.02 3.80 5.14 2.71 9.12 4.19 9.43 4.11 6.52 4.01
Total 42.28 10.64 43.64 10.03 30.86 8.66 40.32 12.79 41.56 12.40 29.86 11.22
Year 3 All (n=299) Normal (n=278) MCI (n=21) All (n=300) Normal (n=279) MCI (n=21)
0–20 s 24.23 5.05 24.64 4.87 18.81 4.29 19.93 5.60 20.33 5.39 14.62 5.84
21–40 s 10.97 4.45 11.24 4.41 7.48 3.34 11.94 4.75 12.22 4.67 8.24 4.44
41–60 s 7.73 3.95 7.94 3.91 4.86 3.32 9.40 4.41 9.59 4.40 6.86 3.86
Total 42.93 10.39 43.82 9.96 31.14 8.83 41.27 12.79 42.14 12.45 29.71 11.88
Year 4 All (n=254) Normal (n=233) MCI (n=21) All (n=254) Normal (n=233) MCI (n=21)
0–20 s 23.41 5.48 23.78 5.40 19.24 4.62 20.10 5.57 20.47 5.38 15.95 6.06
21–40 s 10.85 4.44 11.13 4.43 7.71 3.24 11.96 4.77 12.27 4.67 8.57 4.71
41–60 s 7.82 4.02 8.07 3.95 5.00 3.73 9.50 4.48 9.68 4.47 7.38 4.09
Total 42.07 11.30 42.98 11.04 31.95 9.17 41.55 13.13 42.42 12.76 31.90 13.65
Year 5 All (n=185) Normal (n=173) MCI (n=12) All (n=184) Normal (n=172) MCI (n=12)
0–20 s 23.79 5.23 24.01 5.15 20.75 5.75 20.43 5.64 20.84 5.40 14.58 6.10
21–40 s 10.90 4.27 11.16 4.23 7.17 2.95 12.28 4.75 12.58 4.66 7.92 3.94
41–60 s 7.49 3.89 7.68 3.85 4.67 3.31 10.06 4.42 10.28 4.40 7.00 3.57
Total 42.18 10.59 42.85 10.31 32.58 9.66 42.76 13.13 43.68 12.66 29.50 12.31
Year 6 All (n=120) Normal (n=112) MCI (n=8) All (n=120) Normal (n=112) MCI (n=8)
0–20 s 23.36 5.42 23.71 5.32 18.50 4.75 20.36 5.61 20.85 5.36 13.50 4.75
21–40 s 10.95 4.65 11.21 4.66 7.25 2.31 12.26 4.54 12.57 4.47 7.88 3.09
41–60 s 7.04 3.88 7.21 3.94 4.63 1.77 9.53 4.66 9.77 4.60 6.13 4.36
Total 41.35 11.26 42.13 11.07 30.38 6.89 42.14 12.89 43.19 12.36 27.50 10.65
Year 7 All (n=24) Normal (n=22) MCI (n=2) All (n=24) Normal (n=22) MCI (n=2)
0–20 s 23.63 6.95 24.82 5.81 10.50 4.95 21.88 6.80 23.14 5.39 8.00 5.66
21–40 s 10.79 5.44 11.41 5.22 4.00 2.83 12.88 5.74 13.50 5.26 6.00 8.49
41–60 s 7.42 4.54 7.77 4.51 3.50 3.54 9.63 5.11 10.05 4.89 5.00 7.07
Total 41.83 13.34 44.00 11.68 18.00 3.29 44.38 15.65 46.68 13.49 19.00 16.43
Open in a new tab

Note: Dataset includes participants who completed all baseline measures from June 2011 – May 2018; MCI = mild cognitive impairment

Linear Mixed Effects Models: MCI Effects on Word Generation Slopes

Semantic Fluency:

LMEM revealed a significant slope effect (estimate=−8.350; p<0.0001) indicating that correct word generation declined during the course of the task among controls. The two-way interaction of MCI x slope was significant indicating an attenuated slope among MCI participants compared to control (estimate=1.34; p=0.001). The slope among MCI participants (deduced from the LMEM based on the above two estimates) was −7.01. The three-way interaction of MCI x slope x time was not significant indicating that the effect of MCI on word generation slopes during the course of semantic fluency did not change across annual assessments (estimate=−0.044; p=0.741). Summary of the complete LMEM is presented in Table 3.

Table 3:

Linear mixed effects model examining the effects of time, MCI status, year and their interactions on semantic fluency performance

MCI
Variable Estimate t 95% CI p
Slope −8.35 −64.31 [−8.60, −8.10] < .001
MCI −2.77 −3.22 [−4.45, −1.08] .001
Year −0.11 −1.30 [−0.27, −0.06] .195
Slope x MCI 1.34 3.33 [0.55, 2.13] .001
Year x MCI 0.15 0.51 [−0.42, 0.71] .610
Year x Slope 0.05 1.27 [−0.03, 0.13] .203
Year x Slope x MCI −0.04 −0.33 [−0.31, 0.22] .741
Age −0.03 −5.99 [−0.04, −0.02] < .001
GHS 0.001 0.03 [−0.07, 0.07] .981
Gender 0.01 0.10 [−0.13, 0.15] .917
Education 0.26 17.29 [0.23, 0.29] < .001
GDS −0.003 −0.30 [−0.02, 0.02] .766
WRAT-3 Standard Score 0.01 2.14 [0.001, 0.02] .033
Semantic fluency z-score 2.51 81.66 [2.44, 2.57] < .001
Open in a new tab

Note: MCI = mild cognitive impairment; GHS = general health status; GDS = geriatric depression scale; WRAT = Wide Range Achievement Test (reading standard score) third edition; The slope of correct word generation is derived from three consecutive time intervals in semantic fluency

Letter Fluency:

LMEM revealed a significant slope effect (estimate=−5.068; p<0.0001) indicating that word generation declined during the course of the task among controls. The two-way interaction of MCI x slope was significant indicating an attenuated slope among MCI participants compared to controls (estimate=0.733; p=0.020). The slope among MCI participants (deduced from the LMEM based on the above two estimates) was −4.335. The longitudinal effect of annual assessments was significant indicating a slight improvement in mean word generation over time (0.155; p<0.020). The interaction of time x slope was significant indicating that the slope of correct word generation increased over time (−0.084; p=0.006). The three-way interaction of MCI x slope x time, although statistically insignificant, was suggestive of a trend whereby that the effect of MCI on word generation slopes during the course of letter fluency was attenuated over repeated annual assessments (estimate=0.197; p=0.062). Summary of the complete LMEM is presented in Table 4.

Table 4:

Linear mixed effects model examining the effects of time, MCI status, year and their interactions on letter fluency performance

MCI
Variable Estimate t 95% CI p
Slope −5.07 −49.84 [−5.27, −4.87] < .001
Year 0.16 2.33 [0.02, 0.29] .020
MCI −1.45 −2.13 [−2.78, −0.11] .034
Slope x MCI 0.73 2.33 [0.12, 1.35] .020
Year x MCI −0.40 −1.74 [−0.85, 0.05] .083
Year x Slope −0.08 −2.73 [−0.15, −0.02] .006
Year x Slope x MCI 0.20 1.86 [−0.01, 0.41] .062
Age −0.02 −4.11 [−0.04, −0.01] < .001
GHS 0.01 0.15 [−0.07, 0.08] .879
Gender 0.06 0.80 [−0.09, 0.22] .426
Education 0.50 29.15 [0.47, 0.54] < .001
GDS 0.01 1.08 [−0.01, 0.03] .280
WRAT-3 Standard Score 0.02 3.60 [0.01, 0.03] < .001
Letter fluency z-score 3.56 95.29 [3.49, 3.63] < .001
Open in a new tab

Note: MCI = mild cognitive impairment; GHS = general health status; GDS = geriatric depression scale; WRAT = Wide Range Achievement Test (reading standard score) third edition; the slope of correct word generation is derived from three consecutive time intervals in letter fluency

Correlations of slopes with performance levels in letter and semantic fluency.

The correlations of slopes with performance levels were examined in 344 participants who did not have MCI at baseline and were included in the cox analysis described below. Pearson correlations revealed that in semantic fluency the slope and total number of words generated were significantly correlated (r=−.193, p<0.0001); in letter fluency the slope and total number of words generated were also signficantly correlated (r=−.168, p=0.0002).

Cox Proportional Hazard Models: Word Generation Slopes and Incident MCI

Word generation slopes of semantic and letter fluency (dichotomized at the median) were used as predictors of incident MCI in Cox proportional-hazards models (n=344). Results revealed that attenuated slopes in semantic fluency, at baseline, predicted a two-fold increase in the risk of developing MCI in unadjusted (HR=2.110, p=0.010) and fully adjusted (HR=2.063, p=0.015) analyses (see Table 5 and Figure 1). In contrast, attenuated slopes in letter fluency, at baseline, were not significant predictors of incident MCI in unadjusted (HR=0.747, p=0.317) or fully adjusted (HR=0.704, p=0.243) analyses (see Table 5 and Figure 2).

Table 5:

Cox analysis: Hazard ratios with 95% CI for risk of incident MCI as a function of baseline verbal fluency slopes

Semantic Fluency Letter Fluency
Variable HR (95% CI), p value HR (95% CI), p value
Model 1
Slope 2.119 (1.201–3.739), 0.010 0.747 (0.422–1.323), 0.317
Model 2
Slope 2.063 (1.149–3.702), 0.015 0.704 (0.391–1.269), 0.243
GHS 1.059 (0.789–1.420), 0.702 1.014 (0.757–1.358), 0.926
Gender 0.897 (0.489–1.644), 0.725 0.936 (0.513–1.707), 0.830
Education 1.205 (1.053–1.379), 0.007 1.200 (1.050–1.371), 0.007
GDS 1.070 (0.997–1.148), 0.059 1.074 (1.002–1.150), 0.044
WRAT-3 Standard Score 0.954 (0.917–0.993), 0.021 0.945 (0.909–0.983), 0.005
Open in a new tab

Note: MCI = mild cognitive impairment; GHS = general health status; GDS = geriatric depression scale; WRAT = Wide Range Achievement Test (reading standard score) third edition; the slopes of correct word generation are derived from three consecutive time intervals in letter and semantic fluency

Figure 1.

Figure 1

Open in a new tab

Survival curve for the risk of incident MCI as a function of baseline semantic fluency slope dichotomized at the median. Below the median indicates an attenuated within task slope.

Figure 2.

Figure 2

Open in a new tab

Survival curve for the risk of incident MCI as a function of baseline letter fluency slope dichotomized at the median. Below the median indicates an attenuated within task slope.

Sensitivity Analysis:

A fully adjusted Cox proportional-hazards model that simultaneously included the slopes of both fluency tasks as predictors revealed that attenuated slope of semantic fluency remained a significant predictor of incident MCI (HR=2.154, p=0.011, 95%=1.197 to 3.879) even when adjusting for the slope of letter fluency (HR=0.652, p=0.156, 95%CI=0.361 to 1.177, see supplementary table 1). A fully adjusted Cox analysis that also included the mean baseline normative semantic fluency score as a predictor revealed that the effect of the slope in semantic fluency was diminished (HR=1.645, p=0.106, 95%=0.900 to 3.006). Higher normative semantic fluency scores at baseline were associated with a reduced risk of incident MCI (HR=0.607, p=0.001, 95%CI=0.454 to 0.811; see supplementary table 2). A fully adjusted Cox analysis that included the mean baseline normative letter fluency score as a predictor revealed that the effect of the slope in letter fluency remained insignificant (HR=0.670, p=0.186, 95%=0.370 to 1.213). Higher normative letter fluency scores at baseline were associated with a reduced risk of incident MCI (HR=0.723, p=0.020, 95%CI=0.551 to 0.949; see supplementary table 3). LMEMs excluding the final two years of follow-up revealed that the moderating effects of MCI on letter fluency slopes (estimate=0.92; 95%CI=0.24 to 1.60; p=0.008; see supplementary table 4) and on semantic fluency slopes (estimate=1.68; 95%CI=0.80 to 2.55; p<0.001; see supplementary table 5) remained statistically significant. Cox analyses excluding the last two years of follow-up revealed that the slope of semantic fluency remained a significant predictor of incident MCI (HR=2.274; p=0.008; 95%CI=1.242 to 4.164) while the slope of letter fluency remained an insignificant predictor of incident MCI (HR=0.673; p=0.199; 95%CI=0.367 to1.232; see supplementary table 6).

At baseline, the control and prevalent MCI groups varied in terms of ethnic composition but including ethnicity as an additional covariate in separate LMEMs for semantic and letter fluency revealed it was not associated with the outcomes nor did it influence the main effects or interactions of interest (data not shown). The incident MCI and control groups, however, were not significantly different in terms of ethnic composition (p=0.835). Adding ethnicity to fully adjusted Cox models did not change the results summarized in table 5; the slope of semantic fluency remained a significant risk factor of incident MCI (HR=1.957, p=0.027) and the slope of letter fluency remained insignificant (HR=0.700, p=0.234).

DISCUSSION

The current study was designed to determine whether intra-individual variability in semantic and letter fluency performance, operationalized as within task word generation slopes, was associated with prevalent MCI and predictive of its incidence in a cohort of community-residing older adults. Consistent with previous research (Crowe, 1998; Fernaeus & Almkvist, 1998; Raboutet et al., 2010), performance slopes were negative indicating a decline in word generation during the course of both fluency tasks. As expected (Brandt & Manning, 2009; Malek-Ahmadi et al., 2011; Murphy et al., 2006; Price et al., 2012), prevalent MCI was associated with worse overall verbal fluency performance. Confirming the first study hypothesis and extending previous work (Demetriou & Holtzer, 2017) we found that, compared to controls, participants with MCI demonstrated attenuated word generation slopes in both semantic and letter fluency and that this effect remained consistent over repeated annual assessments. This evidence of stability is critical but should be interpreted in the context of the characteristics of the study cohort, which consisted of relatively healthy community-residing older adults; participants with dementia were excluded. Furthermore, task specific differences emerged as well. Letter fluency performance improved over time but the interaction of time x slope suggested the improvement was limited to improved word generation in the first time interval only (see Table 2 for descriptives). The effect of repeated assessments and practice on verbal fluency performance is discussed in further details in a subsequent paragraph. The three-way interaction of MCI x time x slope in letter fluency approached significance suggesting a trend whereby differences in word generation slopes between the controls and MCI participants increased over time. This trend, while speculative, could be attributed to further cognitive decline among the MCI participants resulting in less efficient word recall, notably in the first 20-sec time interval.

The correlations of slopes and performance levels in both semantic and letter fluency were significant. A more negative slope value is indicative of greater decline in word production. Hence, as expected, greater slopes were associated with a larger number of words produced in both fluency tasks. The magnitudes of the correlations, however, were small with shared variance of less than four percent. This finding suggests that the slopes of word generation in verbal fluency tasks capture a separate construct than the total number of words generated, which is typically used in clinical and research setting to determine normative performance levels. Indeed, the decline in word generation during the course of semantic and letter fluency has been interpreted as evidence for two distinct processes that include a semi-automatic retrieval process at the initial stages of the tasks, and effortful retrieval in the later stages. Word production is more efficient and thus greater during the early stages of the task (Crowe, 1998; Fernaeus & Almkvist, 1998) during which individuals access their long-term store of easy to retrieve words (Crowe, 1998). When, however, this store is exhausted, individuals switch to a more effortful process that entails retrieving words from a larger pool of word store (Crowe, 1998); which in turn make the search process less efficient and more time-consuming and difficult (Crowe, 1998; Raboutet et al., 2010). Our findings suggest that word retrieval is more difficult and effortful for older adults with MCI compared to healthy controls in the early semi-automatic stages of verbal fluency and that this difference remains consistent over repeated annual assessments. We emphasize that if MCI had comparable negative effects on both early automatic and later effortful retrieval stages of semantic and letter fluency the moderation effect of MCI on word generation slopes would not have been significant. Furthermore, the analysis also controlled for verbal fluency performance strongly suggesting that intra-individual variability in verbal fluency is uniquely associated with prevalent MCI providing incremental information that is not available through standard letter and semantic fluency scores.

To determine the predictive utility of intra-individual variability in verbal fluency performance we examined whether differences in individual word generation slopes, assessed at baseline, predicted incident MCI. We found that individuals whose word generation slopes in semantic fluency were attenuated had a two-fold increase in the risk of developing MCI during the follow-up. It is noteworthy that this association remained significant even after controlling for a number of covariates and when the slopes of both semantic and letter fluency were simultaneously used as predictors in the same model. Additional sensitivity analysis revealed that including semantic fluency scores as a covariate in the cox model reduced the predictive utility of the slope. However, the HR of the slope in this analysis, while statistically insignificant, still indicated a 64.5 percent increase in the risk of developing MCI among individuals whose word generation slope was attenuated. The effect of semantic fluency performance on the predictive utility of the slope is not surprising because semantic fluency scores (among other tests) were used to determine MCI diagnostic status. Hence, in a scenario where two performance indices of the same test are used to predict a diagnostic outcome, MCI in this case, it is expected that the index used to determine diagnostic status (i.e., semantic fluency standard score) will have a stronger predictive power. One advantage of using the slope of semantic fluency in predictive models is that it avoids, at least in part, the issue of diagnostic circularity; that is utilizing as predictors of a clinical outcome the same measures used to define it. The slope of word generation in letter fluency, however, was not a significant predictor of incident MCI. Previous research suggested that semantic fluency was superior to letter fluency at detecting individuals who develop AD (Clark et al., 2009; Fernaeus, Ostberg, Hellstrom, & Wahlund, 2008). A separate study reported that semantic fluency performance also predicted conversion to MCI (Loewenstein et al., 2012). Hence, establishing predictive utility for the slope of semantic but not letter fluency, as reported herein, is consistent with the above referenced studies. Future longitudinal studies should examine the differential predictive utility of semantic and letter fluency slopes across MCI and dementia subtypes. The results of the current study revealed that higher semantic and letter normative scores at baseline were associated with a reduced risk of developing incident MCI. However, it is noteworthy that, on average, baseline semantic and letter fluency performance was within the normal range in this cohort (see table 1) suggesting a substantial overlap in performance between the healthy control and prevalent MCI groups. Hence, our findings concerning the incremental and predictive utility of word generation slopes in semantic fluency has significant clinical implications. Measuring intra-individual variability in verbal fluency, as proposed herein, is accomplished without changing the standard administration and scoring procedures of these tests and can thus be incorporated into routine neuropsychological testing batteries in clinical and research settings with relatively minimal effort.

The overall trajectory of semantic fluency over repeated annual assessments was negative but statistically insignificant, and it was not modified by the presence of MCI at baseline. The overall trajectory of letter fluency over repeated annual assessments, however, was positive and significant indicating that performance improved over time irrespective of MCI status, though a trend for a moderating effect of MCI on the time trajectory was observed. Practice effects were reported in various cognitive domains, including verbal fluency (Cooper, Lacritz, Weiner, Rosenberg, & Cullum, 2004; Snitz et al., 2013), with individuals with MCI showing different trajectories both prior to diagnosis relative to older adults who remain cognitively healthy (Machulda et al., 2013), as well as following diagnosis (Howieson et al., 2008) and in the progression to Alzheimer’s disease (Cloutier, Chertkow, Kergoat, Gauthier, & Belleville, 2015). However, it has been noted that serial assessments and subsequent practice effects may obscure early identification of cognitive decline or its extent (Elman et al., 2018; Goldberg, Harvey, Wesnes, Snyder, & Schneider, 2015; Mathews et al., 2014). This is particularly important considering the long presymptomatic course of neurodegenerative diseases and the advent of secondary prevention initiatives. The use of age-based norms may distort cognitive trajectories by failing to account for practice effects, which in cognitive aging may present not only as performance gains over time (Elman et al., 2018) but also as stability of performance (Duff et al., 2011) over repeated assessments or even attenuated declines (Elman et al., 2018).

Study strengths, limitations and future directions:

MCI is a heterogeneous transition state with different underlying brain pathologies but the current study was underpowered to examine whether or not the predictive utility of word generation slopes varied as a function of MCI subtypes. Longitudinal studies should further determine whether differences in the slopes of word generation predict incidence dementia and its subtypes. Additionally, informant information was not available to support decisions regarding diagnostic status. The participants in the current study were relatively healthy community-residing older adults. Future research should examine the generalizability of these findings to populations that are more diverse in terms of demographic and physical characteristics. Consistent with standard scoring procedures verbal fluency performance and slopes were measured using the number of correct responses. Research, however, demonstrated that the number of errors increased with old age (McDowd et al., 2011); and that individuals with AD produced more errors than healthy older adults (Haugrud, Crossley, & Vrbancic, 2011). Using errors to measure differences in intra-individual variability in word production, though, might prove difficult if not impossible to implement due to the relatively small number of errors produced, notably when stratified by 20-sec time intervals. The notable decline in sample size over repeated annual assessments was largely attributed to the ongoing enrollment of participants during the course of the study, which resulted in a shorter follow-up for individuals who were recruited in later years. Nonetheless, additional sensitivity analyses revealed that the moderating effects of MCI on semantic and letter fluency slopes remained significant even after excluding the last two years of follow-up. Similarly, the slope of word generation in semantic fluency remained a significant predictor of incident MCI in analysis that excluded the last two years of follow-up. It would also be of interest to consider the effect of established biological markers such as amyloid burden as well as the effect of pharmacological agents such as acetylcholinesterase inhibitors on the slopes of word generation within these tasks at cross-section and longitudinally. The literature identified shared (Henry et al., 2004; McDowd et al., 2011; Monsch et al., 1994) and distinct (Martin, Wiggs, Lalonde, & Mack, 1994; Murphy et al., 2006) brain regions that have been implicated in letter and semantic fluency performance. Less is known about the structural and functional brain systems that underlie intra-individual variability in verbal fluency performance. One study reported that the presence of white matter hypersensitivities was associated with impaired initiation of word retrieval during the initial 30 seconds of letter fluency (Fernaeus et al., 2001). Future research should examine whether measures of structural brain integrity and functional networks are differentially related to intra-individual variability in verbal fluency performance as compared to standard scores in normal, transition states and disease populations.

In conclusion, the present study provided evidence that prevalent MCI was associated with attenuated word generation slopes in both semantic and letter fluency compared to healthy controls and that this effect remained consistent over annual assessments. Furthermore, attenuated word generation slopes in semantic fluency, at baseline, predicted increased risk of incident MCI. These findings suggest that intra-individual variability in verbal fluency performance provides incremental information that has clinical and predictive utility. Measures of intra-individual variability in verbal fluency can be incorporated into testing batteries in both clinical and research settings with little effort and without changing standard administration and scoring procedures.

Supplementary Material

Supplemental Material

Significance Statement:

Identifying older adults at risk of cognitive decline has important individual and public health implications. Here we demonstrated that a change in performance within a person during the course of a commonly used neuropsychological test predicted the risk of future cognitive decline in a cohort of community-residing older adults. The method used to assess the change in performance during the course of the task could be easily incorporated into standard testing procedures in both clinical and research settings.

Acknowledgements:

This research was supported by the National Institutes on Aging grants (R01AG036921, R01AG044007).

Contributor Information

Roee Holtzer, Department of Neurology, Albert Einstein College of Medicine, and Ferkauf Graduate School of Psychology, Yeshiva University.

Sydney Jacobs, Ferkauf Graduate School of Psychology, Yeshiva University.

Eleni Demetriou, Ferkauf Graduate School of Psychology, Yeshiva University.

References

  1. Albert MS, DeKosky ST, Dickson D, Dubois B, Feldman HH, Fox NC, … Phelps CH (2011). The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimer’s & Dementia: the Journal of the Alzheimer’s Association, 7(3), 270–279. doi: 10.1016/j.jalz.2011.03.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Brandt J, Aretouli E, Neijstrom E, Samek J, Manning K, Albert MS, & Bandeen-Roche K (2009). Selectivity of executive function deficits in mild cognitive impairment. Neuropsychology, 23(5), 607–618. doi: 10.1037/a0015851 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Brandt J, & Manning KJ (2009). Patterns of word-list generation in mild cognitive impairment and Alzheimer’s disease. Clinical Neuropsychologist, 23(5), 870–879. doi: 10.1080/13854040802585063 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Butters N, Granholm E, Salmon DP, Grant I, & Wolfe J (1987). Episodic and semantic memory: a comparison of amnesic and demented patients. Journal of Clinical and Experimental Neuropsychology, 9(5), 479–497. doi: 10.1080/01688638708410764 [DOI] [PubMed] [Google Scholar]
  5. Clark LJ, Gatz M, Zheng L, Chen YL, McCleary C, & Mack WJ (2009). Longitudinal verbal fluency in normal aging, preclinical, and prevalent Alzheimer’s disease. American Journal of Alzheimer’s Disease and Other Dementias, 24(6), 461–468. doi: 10.1177/1533317509345154 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cloutier S, Chertkow H, Kergoat MJ, Gauthier S, & Belleville S (2015). Patterns of Cognitive Decline Prior to Dementia in Persons with Mild Cognitive Impairment. Journal of Alzheimer’s Disease, 47(4), 901–913. doi: 10.3233/jad-142910 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cooper DB, Lacritz LH, Weiner MF, Rosenberg RN, & Cullum CM (2004). Category fluency in mild cognitive impairment: reduced effect of practice in test-retest conditions. Alzheimer Disease and Associated Disorders, 18(3), 120–122. [DOI] [PubMed] [Google Scholar]
  8. Crowe SF (1998). Decrease in performance on the verbal fluency test as a function of time: evaluation in a young healthy sample. Journal of Clinical and Experimental Neuropsychology, 20(3), 391–401. doi: 10.1076/jcen.20.3.391.810 [DOI] [PubMed] [Google Scholar]
  9. Demetriou E, & Holtzer R (2017). Mild Cognitive Impairments Moderate the Effect of Time on Verbal Fluency Performance. Journal of the International Neuropsychological Society, 23(1), 44–55. doi: 10.1017/s1355617716000825 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Dixon RA, Garrett DD, Lentz TL, MacDonald SW, Strauss E, & Hultsch DF (2007). Neurocognitive markers of cognitive impairment: exploring the roles of speed and inconsistency. Neuropsychology, 21(3), 381–399. doi: 10.1037/0894-4105.21.3.381 [DOI] [PubMed] [Google Scholar]
  11. Duff K, Lyketsos CG, Beglinger LJ, Chelune G, Moser DJ, Arndt S, … McCaffrey RJ (2011). Practice effects predict cognitive outcome in amnestic mild cognitive impairment. The American Journal of Geriatric Psychiatry: Official Journal of the American Association for Geriatric Psychiatry, 19(11), 932–939. doi: 10.1097/JGP.0b013e318209dd3a [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Elman JA, Jak AJ, Panizzon MS, Tu XM, Chen T, Reynolds CA, … Kremen WS (2018). Underdiagnosis of mild cognitive impairment: A consequence of ignoring practice effects. Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring, 10, 372–381. doi: 10.1016/j.dadm.2018.04.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Eppig J, Wambach D, Nieves C, Price CC, Lamar M, Delano-Wood L, … & Lippa C (2012). Dysexecutive functioning in mild cognitive impairment: Derailment in temporal gradients. Journal of the International Neuropsychological Society, 18(1), 20–28. doi: 10.1017/S1355617711001238 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Fernaeus SE, & Almkvist O (1998). Word production: dissociation of two retrieval modes of semantic memory across time. Journal of Clinical and Experimental Neuropsychology, 20(2), 137–143. doi: 10.1076/jcen.20.2.137.1170 [DOI] [PubMed] [Google Scholar]
  15. Fernaeus SE, Almkvist O, Bronge L, Ostberg P, Hellstrom A, Winblad B, & Wahlund LO (2001). White matter lesions impair initiation of FAS flow. Dementia and Geriatric Cognitive Disorders, 12(1), 52–56. doi: 10.1159/000051235 [DOI] [PubMed] [Google Scholar]
  16. Fernaeus SE, Ostberg P, Hellstrom A, & Wahlund LO (2008). Cut the coda: early fluency intervals predict diagnoses. Cortex, 44(2), 161–169. doi: 10.1016/j.cortex.2006.04.002 [DOI] [PubMed] [Google Scholar]
  17. Galvin JE, Roe CM, Powlishta KK, Coats MA, Muich SJ, Grant E, … Morris JC (2005). The AD8: a brief informant interview to detect dementia. Neurology, 65(4), 559–564. doi: 10.1212/01.wnl.0000172958.95282.2a [DOI] [PubMed] [Google Scholar]
  18. Galvin JE, Roe CM, Xiong C, & Morris JC (2006). Validity and reliability of the AD8 informant interview in dementia. Neurology, 67(11), 1942–1948. doi:67/11/1942 [pii] 10.1212/01.wnl.0000247042.15547.eb [doi] [DOI] [PubMed] [Google Scholar]
  19. Goldberg TE, Harvey PD, Wesnes KA, Snyder PJ, & Schneider LS (2015). Practice effects due to serial cognitive assessment: Implications for preclinical Alzheimer’s disease randomized controlled trials. Alzheimer’s dementia (Amsterdam, Netherlands), 1(1), 103–111. doi: 10.1016/j.dadm.2014.11.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Haugrud N, Crossley M, & Vrbancic M (2011). Clustering and switching strategies during verbal fluency performance differentiate Alzheimer’s disease and healthy aging. Journal of the International Neuropsychological Society, 17(6), 1153–1157. doi: 10.1017/s1355617711001196 [DOI] [PubMed] [Google Scholar]
  21. Henry JD, Crawford JR, & Phillips LH (2004). Verbal fluency performance in dementia of the Alzheimer’s type: a meta-analysis. Neuropsychologia, 42(9), 1212–1222. doi: 10.1016/j.neuropsychologia.2004.02.001 [DOI] [PubMed] [Google Scholar]
  22. Holtzer R, Goldin Y, Zimmerman M, Katz M, Buschke H, & Lipton RB (2008). Robust norms for selected neuropsychological tests in older adults. Archives of Clinical Neuropsychology, 23(5), 531–541. doi: 10.1016/j.acn.2008.05.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Holtzer R, Mahoney J, & Verghese J (2014). Intraindividual variability in executive functions but not speed of processing or conflict resolution predicts performance differences in gait speed in older adults. Journals of Gerontology. Series A: Biological Sciences and Medical Sciences, 69(8), 980–986. doi: 10.1093/gerona/glt180 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Holtzer R, Verghese J, Wang C, Hall CB, & Lipton RB (2008). Within-person across-neuropsychological test variability and incident dementia. JAMA: Journal of the American Medical Association, 300(7), 823–830. doi:300/7/823 [pii] 10.1001/jama.300.7.823 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Howieson DB, Carlson NE, Moore MM, Wasserman D, Abendroth CD, Payne-Murphy J, & Kaye JA (2008). Trajectory of mild cognitive impairment onset. Journal of the International Neuropsychological Society, 14(2), 192–198. doi: 10.1017/s1355617708080375 [DOI] [PubMed] [Google Scholar]
  26. Hultsch DF, MacDonald SW, Hunter MA, Levy-Bencheton J, & Strauss E (2000). In traindividual variability in cognitive performance in older adults: comparison of adults with mild dementia, adults with arthritis, and healthy adults. Neuropsychology, 14(4), 588–598. [DOI] [PubMed] [Google Scholar]
  27. Hurks PP, Schrans D, Meijs C, Wassenberg R, Feron FJ, & Jolles J (2010). Developmental changes in semantic verbal fluency: analyses of word productivity as a function of time, clustering, and switching. Child Neuropsychology, 16(4), 366–387. doi: 10.1080/09297041003671184 [DOI] [PubMed] [Google Scholar]
  28. Katz S (1983). Assessing self-maintenance: activities of daily living, mobility, and instrumental activities of daily living. Journal of the American Geriatrics Society, 31(12), 721–727. [DOI] [PubMed] [Google Scholar]
  29. Keilp JG, Gorlyn M, Alexander GE, Stern Y, & Prohovnik I (1999). Cerebral blood flow patterns underlying the differential impairment in category vs letter fluency in Alzheimer’s disease. Neuropsychologia, 37(11), 1251–1261. [DOI] [PubMed] [Google Scholar]
  30. Lamar M, Price CC, Davis KL, Kaplan E, & Libon DJ (2002). Capacity to maintain mental set in dementia. Neuropsychologia, 40(4), 435–445. [DOI] [PubMed] [Google Scholar]
  31. Lezak MD, Howieson DB, Loring DW, Hannay HJ, & Fischer JS (2004). Neuropsychological assessment Oxford, NY: Oxford University Press. [Google Scholar]
  32. Lipton RB, Katz MJ, Kuslansky G, Sliwinski MJ, Stewart WF, Verghese J, … Buschke H (2003). Screening for dementia by telephone using the memory impairment screen. Journal of the American Geriatrics Society, 51(10), 1382–1390. [DOI] [PubMed] [Google Scholar]
  33. Loewenstein DA, Greig MT, Schinka JA, Barker W, Shen Q, Potter E, … Duara R (2012). An investigation of PreMCI: subtypes and longitudinal outcomes. Alzheimer’s & Dementia: the Journal of the Alzheimer’s Association, 8(3), 172–179. doi: 10.1016/j.jalz.2011.03.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Machulda MM, Pankratz VS, Christianson TJ, Ivnik RJ, Mielke MM, Roberts RO, … Petersen RC (2013). Practice effects and longitudinal cognitive change in normal aging vs. incident mild cognitive impairment and dementia in the Mayo Clinic Study of Aging. Clinical Neuropsychologist, 27(8), 1247–1264. doi: 10.1080/13854046.2013.836567 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Malek-Ahmadi M, Small BJ, & Raj A (2011). The diagnostic value of controlled oral word association test-FAS and category fluency in single-domain amnestic mild cognitive impairment. Dementia and Geriatric Cognitive Disorders, 32(4), 235–240. doi: 10.1159/000334525 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Marczinski CA, & Kertesz A (2006). Category and letter fluency in semantic dementia, primary progressive aphasia, and Alzheimer’s disease. Brain and Language, 97(3), 258–265. doi: 10.1016/j.bandl.2005.11.001 [DOI] [PubMed] [Google Scholar]
  37. MacDonald SW, Nyberg L, Backman L. (2006). Intra-individual variability in behavior: links to brain structure, neurotransmission and neuronal activity. Trends in Neuroscience, 29(8), 474–80. doi: 10.1002/sim.2098 [DOI] [PubMed] [Google Scholar]
  38. Martin A, Wiggs CL, Lalonde F, & Mack C (1994). Word retrieval to letter and semantic cues: a double dissociation in normal subjects using interference tasks. Neuropsychologia, 32(12), 1487–1494. [DOI] [PubMed] [Google Scholar]
  39. Mathews M, Abner E, Kryscio R, Jicha G, Cooper G, Smith C, … Schmitt FA (2014). Diagnostic accuracy and practice effects in the National Alzheimer’s Coordinating Center Uniform Data Set neuropsychological battery. Alzheimer’s & Dementia: the Journal of the Alzheimer’s Association, 10(6), 675–683. doi: 10.1016/j.jalz.2013.11.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. McDowd J, Hoffman L, Rozek E, Lyons KE, Pahwa R, Burns J, & Kemper S (2011). Understanding verbal fluency in healthy aging, Alzheimer’s disease, and Parkinson’s disease. Neuropsychology, 25(2), 210–225. doi: 10.1037/a0021531 [DOI] [PubMed] [Google Scholar]
  41. Monsch AU, Bondi MW, Butters N, Paulsen JS, Salmon DP, Brugger P, & Swenson MR (1994). A comparison of category and letter fluency in Alzheimer’s disease and Huntington’s disease. Neuropsychology, 8(1), 25–30. doi:http://psycnet.apa.org/doi/10.1037/0894-4105.8.1.25 [Google Scholar]
  42. Monsch AU, Bondi MW, Butters N, Salmon DP, Katzman R, & Thal LJ (1992). Comparisons of verbal fluency tasks in the detection of dementia of the Alzheimer type. Archives of Neurology, 49(12), 1253–1258. [DOI] [PubMed] [Google Scholar]
  43. Murphy KJ, Rich JB, & Troyer AK (2006). Verbal fluency patterns in amnestic mild cognitive impairment are characteristic of Alzheimer’s type dementia. Journal of the International Neuropsychological Society, 12(4), 570–574. [DOI] [PubMed] [Google Scholar]
  44. Nutter-Upham KE, Saykin AJ, Rabin LA, Roth RM, Wishart HA, Pare N, & Flashman LA (2008). Verbal fluency performance in amnestic MCI and older adults with cognitive complaints. Archives of Clinical Neuropsychology, 23(3), 229–241. doi: 10.1016/j.acn.2008.01.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Ober BA, Dronkers NF, Koss E, Delis DC, & Friedland RP (1986). Retrieval from semantic memory in Alzheimer-type dementia. Journal of Clinical and Experimental Neuropsychology, 8(1), 75–92. doi: 10.1080/01688638608401298 [DOI] [PubMed] [Google Scholar]
  46. Papp KV, Mormino EC, Amariglio RE, Munro C, Dagley A, Schultz AP, … Rentz DM (2016). Biomarker validation of a decline in semantic processing in preclinical Alzheimer’s disease. Neuropsychology, 30(5), 624–630. doi: 10.1037/neu0000246 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Petersen RC (2004). Mild cognitive impairment as a diagnostic entity. Journal of Internal Medicine, 256(3), 183–194. doi: 10.1111/j.1365-2796.2004.01388.x [DOI] [PubMed] [Google Scholar]
  48. Petersen RC, Caracciolo B, Brayne C, Gauthier S, Jelic V, & Fratiglioni L (2014). Mild cognitive impairment: a concept in evolution. Journal of Internal Medicine, 275(3), 214–228. doi: 10.1111/joim.12190 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Price SE, Kinsella GJ, Ong B, Storey E, Mullaly E, Phillips M, … Perre D (2012). Semantic verbal fluency strategies in amnestic mild cognitive impairment. Neuropsychology, 26(4), 490–497. doi: 10.1037/a0028567 [DOI] [PubMed] [Google Scholar]
  50. Raboutet C, Sauzeon H, Corsini MM, Rodrigues J, Langevin S, & N’Kaoua B (2010). Performance on a semantic verbal fluency task across time: dissociation between clustering, switching, and categorical exploitation processes. Journal of Clinical and Experimental Neuropsychology, 32(3), 268–280. doi: 10.1080/13803390902984464 [DOI] [PubMed] [Google Scholar]
  51. Radanovic M, Diniz BS, Mirandez RM, Novaretti TM, Flacks MK, Yassuda MS, & Forlenza OV (2009). Verbal fluency in the detection of mild cognitive impairment and Alzheimer’s disease among Brazilian Portuguese speakers: the influence of education. International Psychogeriatrics, 21(6), 1081–1087. doi: 10.1017/s1041610209990639 [DOI] [PubMed] [Google Scholar]
  52. Snitz BE, Weissfeld LA, Lopez OL, Kuller LH, Saxton J, Singhabahu DM, … Dekosky ST (2013). Cognitive trajectories associated with β-amyloid deposition in the oldest-old without dementia. Neurology, 80(15), 1378–1384. doi: 10.1212/WNL.0b013e31828c2fc8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Spreen O, & Benton AL (1977). Neurosensory center comprehensive examination for aphasia: Manual of directions Victoria, British Columbia: University of Victoria. [Google Scholar]
  54. Summers MJ, & Saunders NL (2012). Neuropsychological measures predict decline to Alzheimer’s dementia from mild cognitive impairment. Neuropsychology, 26(4), 498–508. doi: 10.1037/a0028576 [DOI] [PubMed] [Google Scholar]
  55. Thiebaut AC, & Benichou J (2004). Choice of time-scale in Cox’s model analysis of epidemiologic cohort data: a simulation study. Statistics in Medicine, 23(24), 3803–3820. doi: 10.1002/sim.2098 [DOI] [PubMed] [Google Scholar]
  56. Traykov L, Raoux N, Latour F, Gallo L, Hanon O, Baudic S, … Rigaud AS (2007). Executive functions deficit in mild cognitive impairment. Cognitive and Behavioral Neurology, 20(4), 219–224. doi: 10.1097/WNN.0b013e31815e6254 [DOI] [PubMed] [Google Scholar]
  57. Weakley A, Schmitter-Edgecombe M, & Anderson J (2013). Analysis of verbal fluency ability in amnestic and non-amnestic mild cognitive impairment. Archives of Clinical Neuropsychology: the Official Journal of the National Academy of Neuropsychologists, 28(7), 721–731. doi: 10.1093/arclin/act058 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Wilkinson GS (1993). Wide range achievement test: WRAT3: Revised 3, 1993 edition. Wilmington, DE: Wide Range, Inc. ; Los Angeles, Calif.: Western Psychological Services; [distributor], 1993. [Google Scholar]
  59. Wilson RS, Leurgans SE, Boyle PA, & Bennett DA (2011). Cognitive decline in prodromal Alzheimer disease and mild cognitive impairment. Archives of Neurology, 68(3), 351–356. doi: 10.1001/archneurol.2011.31 [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Winblad B, Palmer K, Kivipelto M, Jelic V, Fratiglioni L, Wahlund LO, … Petersen RC (2004). Mild cognitive impairment--beyond controversies, towards a consensus: report of the International Working Group on Mild Cognitive Impairment. Journal of Internal Medicine, 256(3), 240–246. doi: 10.1111/j.1365-2796.2004.01380.x [DOI] [PubMed] [Google Scholar]
  61. Yesavage JA, Brink TL, Rose TL, Lum O, Huang V, Adey M, & Leirer VO (1982). Development and validation of a geriatric depression screening scale: a preliminary report. Journal of Psychiatric Research, 17(1), 37–49. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material
NIHMS1037401-supplement-Supplemental_Material.pdf (42.5KB, pdf)

RESOURCES