- Split View
-
Views
-
Cite
Cite
Kristoffer Romero, Astrid Coleman, Arjan Heir, Larry Leach, Guy B Proulx, Multivariate Base Rates of Low Neuropsychological Test Scores in Cognitively Intact Older Adults with Subjective Cognitive Decline from a Specialist Memory Clinic, Archives of Clinical Neuropsychology, Volume 37, Issue 7, October 2022, Pages 1467–1479, https://doi.org/10.1093/arclin/acac050
- Share Icon Share
Abstract
To avoid misdiagnosing mild cognitive impairment (MCI), knowledge of the multivariate base rates (MVBRs) of low scores on neuropsychological tests is crucial. Base rates have typically been determined from normative population samples, which may differ from clinically referred samples. The current study addresses this limitation by calculating the MVBR of low or high cognitive scores in older adults who presented to a memory clinic experiencing subjective cognitive decline but were not diagnosed with MCI.
We determined the MVBRs on the Kaplan–Baycrest Neurocognitive Assessment for 107 cognitively healthy older adults (M age = 75.81), by calculating the frequency of patients producing n scores below or above different cut-off values (i.e., 1, 1.5, 2.0, 2.5 SD from the mean), stratifying by education and gender.
Performing below or above cut-off was common, with more stringent cut-offs leading to lower base rates (≥1 low scores occurred in 84.1% of older adults at −1 SD, 55.1% at −1.5 SD, and 39.3% at −2 SD below the mean; ≥1 high scores occurred in 80.4% of older adults at +1 SD, 35.5% at +1.5 SD, and 16.8% at +2 SD above the mean). Higher education was associated with varying base rates. Overall, the MVBR of obtaining a low cognitive test score was higher in this clinic sample, compared with prior studies of normative samples.
MVBRs for clinically referred older adults experiencing memory complaints provide a diagnostic benefit, helping to prevent attributing normal variability to cognitive impairment and limiting false positive diagnoses.
Mild cognitive impairment (MCI; Petersen et al., 2009; Albert et al., 2011) is characterized by the presence of subjective cognitive complaints, and evidence of cognitive deficits without significant impairment in performance on functional activities of daily life (Petersen et al., 1999; Winblad et al., 2004). Progression rates from MCI to dementia are |$\sim$|10% per year (varying from 5% to 40%), a significantly higher proportion than in cognitively healthy individuals (Bruscoli & Lovestone, 2004; Boyle, Wilson, Aggarwal, Tang, & Bennett, 2006; Mitchell & Shiri-Feshki, 2009), making MCI a generally useful diagnostic tool for identifying individuals at risk of developing dementia (Bennett et al., 2002; Fisk, Merry, & Rockwood, 2003; Summers & Saunders, 2012). Nevertheless, the diagnosis of MCI lacks precision, with a proportion of individuals meeting criteria for MCI (ranging from 4% to 59%) reverting to normal cognitive functioning within 1 to 4 years of diagnosis (Larrieu et al., 2002; Busse, Hensel, Gühne, Angermeyer, & Riedel-Heller, 2006; Loewenstein et al., 2009; Ganguli et al., 2011; Malek-Ahmadi, 2016). These heterogeneous findings for both MCI conversion rates to dementia and reversion rates to normal cognition suggest more accurate assessment of cognitive impairment is needed to minimize misdiagnosing healthy individuals as cognitively impaired.
When determining psychometric criteria for diagnosing cognitive impairment, it is critical to recognize that variability across tests in a neuropsychological battery is common in the general population. In particular, healthy individuals regularly obtain one or more abnormally low scores when any given battery of cognitive tests is administered (Ingraham & Aiken, 1996; Palmer, Boone, Lesser, & Wohl, 1998; Crawford, Garthwaite, & Gault, 2007; Axelrod & Wall, 2007; Brooks, Iverson, & White, 2007; Brooks, Iverson, Holdnack, & Feldman, 2008; Schretlen, Testa, Winicki, Pearlson, & Gordon, 2008; Brooks et al., 2009a; Brooks, Iverson, & White, 2009b; Crawford, Garthwaite, Sutherland, & Borland, 2011; Gunner, Miele, Lynch, & McCaffrey, 2012; Klekociuk, Summers, Vickers, & Summers, 2014; Mistridis et al., 2015; Holdnack et al., 2017; Karr, Garcia-Barrera, Holdnack, & Iverson, 2018; Kiselica, Webber, & Benge, 2020; Wallace, Schatz, Covassin, & Iverson, 2020). For example, Brooks, Holdnack, and Iverson (2011) determined the base rates of cognitively healthy adults on the Wechsler Adult Intelligence Scale–Fourth Edition and Wechsler Memory Scale–Fourth Edition (WMS-IV). On the 10 composite/index scores, 47.2% of participants obtained one or more low scores when the cut-off criterion was set at the 16th percentile (|$\sim$|≤1.0 SD below the mean), 20.8% when the cut-off criterion was set to the 5th percentile (|$\sim$|≤1.5 SD below the mean), and 10.7% when the cut-off criterion was set to the ≤2nd percentile (|$\sim$| ≤2 SD below the mean). Similarly, Karr and his coworkers (2018) noted 55% of adults scored ≤9th percentile on measures of executive functioning, indicating that low scores in the general population are not limited to the domains of memory and intellectual functioning. Regarding older adults specifically, Brooks and his coworkers (2009b) utilized the Neuropsychological Assessment Battery and found the frequency of low scores in cognitively normal older adults was also relatively common (1+ low scores = 35.6% at ≤16th percentile, 12.2% at ≤5th percentile, and 5.2% at ≤2nd percentile). Similar patterns were observed by Drozdick, Holdnack, Salthouse, and Cullum (2013), who reported that healthy older adults scored below 1.5 SD from the normative mean on 0.4 subtests of the WMS-IV, whereas those with MCI scored below the 5th percentile on 1.8 subtests on average, indicating that although healthy older adults do produce low scores, the multivariate base rate (MVBR) is higher in those with a diagnosis of MCI. Consequently, one could posit that those older adults with subjective cognitive impairment may show MVBRs of low scores in between normative older adult and MCI samples.
Although the occurrence of low scores among healthy individuals is common, the base rate of low scores can vary as a function of demographic characteristics, such as ethnocultural background (Mejia, Gutiérrez, Villa, & Ostrosky-Solis, 2004; Rivera et al., 2019; Rivera et al., 2021) and level of education (Brooks et al., 2007; 2008; 2012; Holdnack et al., 2017). Collectively, these results indicate that to properly interpret assessment results and avoid misdiagnosing healthy individuals as impaired, the informed clinician must consider information regarding the frequency (i.e., base rate) with which low scores occur in the relevant comparison population (Crawford, 2004; Crawford et al., 2007).
Another consideration regarding base rates is the variability in criteria for diagnosing MCI, both in terms of cut-off scores and the number of low test scores required (Jak et al., 2009; Albert et al., 2011; Iverson & Brooks, 2011; APA, 2014; Kiselica et al., 2020). The choice of cut-off score is non-trivial as the specific cut-off criterion and the number of tests in a battery may impact the frequency of obtaining low scores (Iverson & Brooks, 2011): that is, administering a greater number of tests and utilizing a less stringent cut-off threshold may result in a higher number of low scores (Crawford et al., 2007; Schretlen et al., 2008; Mistridis et al., 2015). Consequently, these criteria must be chosen thoughtfully to balance sensitivity and specificity to correctly differentiate those older adults with MCI from those older adults who may have cognitive concerns but no objective cognitive deficits.
Further complicating the diagnosis of MCI is the association between objective cognitive deficits and subjective cognitive decline (SCD). As SCD is based solely on self- or informant-reports of decline in any cognitive domain rather than test performance, SCD may or may not track with objective ability (Jessen et al., 2014). It has been suggested that SCD may present in a preclinical phase of dementia, prior to MCI and any observable cognitive deficits (Ávila-Villanueva & Fernández-Blázquez, 2017; Cheng, Chen, & Chiu, 2017). Indeed, SCD has been associated with brain-based biomarkers of dementia (mainly Alzheimer’s disease), including carriers of the apolipoprotein E ε4 gene variant, neurodegeneration, and amyloid pathology (Jessen et al., 2014; Rabin, Smart, & Amariglio, 2017). Moreover, a meta-analysis performed by Mitchell, Beaumont, Ferguson, Yadegarfar, and Stubbs (2014) found that among older adults with no objective cognitive deficits, those with subjective memory complaints were twice as likely to develop dementia compared with individuals without complaints (annual conversion rate = 2.3% and 1.0%, respectively). In addition, in some cases, worse objective cognitive performance among cognitively normal individuals with SCD has been associated with greater cognitive complaints (Slot et al., 2018).
However, the relationship between SCD and objective cognitive ability is not straightforward, with reporting of SCD influenced by other medical, psychological, and demographic factors (Cheng et al., 2017; Rabin et al., 2017). In addition, Edmonds et al. (2018) found that during a 2-year period, cognitively normal individuals consistently overreported cognitive difficulties, whereas those with MCI underreported cognitive complaints, which also tracked with AD biomarker positivity and informant complaints. Thus, subjective cognitive complaints do not always predict cognitive decline, with self-report becoming increasingly unreliable as objective cognitive impairment progresses. Taken together, these mixed findings demonstrate that SCD is not always a reliable indicator of objective cognitive impairment.
When distinguishing between SCD and true cognitive impairments, knowledge of base rates and the frequency of scores that fall below a cut-off in the normative population can act as informative reference tools. Brooks and his coworkers (2009b) noted that recognizing the frequency of low scores in the population, in conjunction with the sensitivity and specificity of the chosen cut-off scores, helps to minimize both false positive and false negative diagnoses by supplementing interpretation of objective cognitive results and clinical judgement. One key limitation of the existing work on base rates in geriatric assessments is that the base rates for test batteries have all been calculated using normative samples designed to reflect the population at-large. However, patients seen for neuropsychological assessment of suspected MCI are often seen in secondary or even tertiary settings. The base rate of impairment may differ in patient samples compared with the general population as these samples would likely have already received some form of prior screening assessment to flag possible cognitive impairment and are more likely to have comorbid health conditions that can affect cognition. Indeed, progression rates from normal cognition to MCI have been found to be higher in clinical compared with community-based populations: Cheng and his coworkers (2017) showed that during a 7-year longitudinal study, annual conversion rates to MCI were 30% for individuals in the clinical group and only 5% among the community sample. Furthermore, progression from MCI to dementia is also elevated for clinically referred individuals. A systematic review by Bruscoli and Lovestone (2004) suggests that the source of participants (community vs. clinic) is the largest contributing factor to widespread variance among annual conversion rates from MCI to dementia. Indeed, Farias, Mungas, Reed, Harvey, and DeCarli (2009) found annual conversion rates of MCI to dementia were 13% and 3% for clinic and community samples, respectively. Taken together, these findings suggest that individuals referred to clinical care may demonstrate a higher rate of low scores, which may or may not reflect an underlying neurodegenerative disorder. Therefore, base rates computed using normative samples may underestimate the base rate of cognitive impairment in clinical samples of those referred for suspected cognitive impairment.
A related topic is the MVBR of unusually high scores that are well above the normative mean. Recent studies suggest that high scores on neuropsychological tests are common within normative samples of commonly used neuropsychological tests, such as the Delis–Kaplan Executive Functioning System (Karr, Garcia-Barrera, Holdnack, & Iverson, 2020) and the NIH Toolbox Cognition Battery (Karr & Iverson, 2020). This work suggests those with higher levels of education would likely show above average scores on some measures, and that in those with higher levels of education and/or intellectual functioning, the absence of above average scores could be indicative of cognitive impairment. Preliminary findings suggest a similar pattern in older adults who completed the NIH Toolbox Cognition Battery (Iverson & Karr, 2021). Thus, knowledge of the MVBR for scores above average would be useful when conducting a neuropsychological assessment for MCI in those patients with higher levels of education, as such data would provide guidance as to what a clinician could expect regarding the likelihood of n scores falling above or at the normative mean. To our knowledge, there have been no studies of the MVBR of above average test scores in older adults with SCD.
The increased likelihood of scoring below a cut-off score in this clinical referral stream, the gap in knowledge regarding the MVBR of above average scores in geriatric samples, and the utility of including base rates in clinical decision-making suggest a diagnostic benefit to generating base rates for older adults who present to secondary or tertiary care with SCD but do not have cognitive impairment. Knowledge of the rate at which this sample produces impaired scores would provide another reference point to aid in clinicians’ decision-making regarding the appropriateness of an MCI diagnosis. Thus, we sought to report the MVBRs of producing low cognitive scores at different cut-off points (i.e., 1.0 SD, 1.5 SD, 2.0 SD, and 2.5 SD below the mean) among older adults with SCD who were referred to a specialist memory clinic for suspected cognitive impairment but who were cognitively intact. We also stratified participants by level of education or by gender to explore the interaction between these variables and MVBRs. Moreover, we conducted exploratory analyses to report the frequency of above average scores across neuropsychological tests (i.e., scores 1.0 SD, 1.5 SD, 2.0 SD, and 2.5 SD above the mean), and to determine the likelihood of achieving unusually high scores among cognitively healthy older adults with SCD.
Methods
Subjects
We retrospectively reviewed the charts of 687 patients between 2001 and 2017 with subjective cognitive complaints, who were referred from a specialist memory clinic or behavioral neurology clinic at Baycrest Health Sciences for formal neuropsychological assessment because of suspected cognitive impairment. This initial sample included only patients who had completed data for all neuropsychological, demographic, and clinical variables of interest. Referral sources for the memory clinics and behavioral neurology clinics included primary care physicians and other hospitals within the Greater Toronto Area. We included patients who had a subjective cognitive complaint (self or informant), but who did not show evidence of global cognitive impairment or episodic memory deficits. Specifically, we included patients only if they had a score of 27 or higher on the Mini Mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975) and if they scored above the fifth percentile on the Wechsler Memory Scale-Revised (WMS-R; Wechsler, 1987) Logical Memory delayed recall subtest, using age- and education-adjusted norms (Ivnik et al., 1992). We specifically used these inclusion criteria as they are independent of the patient diagnoses and the tests used to calculate MVBR, thereby avoiding logical issues of circularity in our analyses. Previous studies of MCI using larger datasets have taken a similar approach of a cognitive screen + neuropsychological test to classify patients as cognitively intact (Edmonds et al., 2018), and the combined use of an MMSE score of ≥27 with verbal memory delayed recall scores has been shown to be useful in differentiating MCI from dementia (Pozueta et al., 2011). Patients without MMSE or WMS-R scores, or who were referred for neuropsychological testing but did not receive an assessment were excluded from the study. Patients were also excluded from the study if there was a medical history of a comorbid neurodegenerative disease or neurological condition that could affect cognition (i.e., multiple sclerosis; Parkinson’s disease; stroke, complicated mild traumatic brain injury (TBI), moderate or severe TBI; normal pressure hydrocephalus; cancer), or if patients were receiving treatment for a comorbid psychiatric condition at the time of the assessment.
This resulted in a final sample of 107 older adults who had subjective complaints but did not show global cognitive impairment. Note, of these participants, 17 were missing at least one neuropsychological test score used in the calculation of MVBRs (missing 1 score: n = 3, missing 2 scores: n = 14). Given that these missing data constituted 2.41% of the total dataset, we opted to impute data using expectation–maximization to estimate missing values. This form of imputation is appropriate when the missing data constitute a small proportion of the total data, and when no hypothesis testing is conducted on the imputed dataset (Graham, 2009). An imputed dataset was created using the missMethods package in R (function: impute_EM).
Materials
Demographics
We extracted the following information from participants’ clinical files: age, gender, years of education, employment history, number of languages spoken, source of subjective cognitive complaint (self or other), formal diagnoses, comorbidities, and the number of cardiovascular risk factors and/or cardiovascular diseases reported (i.e., hypertension, hypercholesterolemia, coronary artery disease, diabetes, smoking, white matter hyperintensities, transient ischemic attack, sleep apnea, obesity). Patients were classified according to the presence of 0, 1–2, or ≥ 3 cardiovascular risk factors, based on recent studies examining the role of cardiovascular health on cognition (Wooten et al., 2019). All research was approved by the Research Ethics Boards of Baycrest Health Sciences and the University of Windsor.
Neuropsychological tests
The Kaplan–Baycrest Neurocognitive Assessment (KBNA; Leach, Kaplan, Rewilak, Richards, & Proulx, 2000) is a neuropsychological test battery covering major cognitive functions of interest, including attention/concentration, immediate recall, delayed recall, visuospatial functioning, language, and reasoning/problem-solving. Specific subtests consist of: Sequences, Word List recall (immediate recall, delayed recall, recognition), Complex Figure (immediate recall, delayed recall, recognition), Visuospatial Functioning (clock copy, complex figure copy), Visual Short-Term Memory (object location recall), Phonemic and Semantic Fluency, and Reasoning (practical problem solving, conceptual shifting). For our analyses, we included the following subtests: Sequences; Word List Immediate Recall, Delayed Recall and Yes/No Recognition; Complex Figure Immediate Recall, Delayed Recall, and Recognition; Phonemic Fluency; Semantic Fluency; and Visuospatial Functioning (Clock Drawing, Complex Figure Copy); Short-Term Memory (Spatial Location); Reasoning (Practical Problem Solving); and Conceptual Shifting. Note that the latter two measures are combined into an overall measure of problem solving/reasoning (Preas).
The KBNA standardization sample consisted of 700 adults in the U.S., collected to be representative of the U.S. population with respect to gender (Male, Female), race/ethnicity (White, African American, Hispanic, Other), age, education level (≤8, 9–11, 12, 13–15, ≥16 years), and geographical location (Northeast, North Central, South, West) based on categories from the updated 1999 U.S. Bureau of the Census data. Specifically, 100 adults were recruited across seven age groups (20–29, 30–39, 40–49, 50–59, 60–69, 70–79, 80–89). Participants were screened for medical and psychiatric disorders via self-report. The resulting normative tables were thus corrected for age, but not gender, race/ethnicity, or education level (Leach et al., 2000, p. 79).
Analyses
We calculated the MVBR as the cumulative percentage of subjects producing n number of scores falling below different cut-offs. In line with prior work, we calculated the MVBR of obtaining a low score on a cognitive test according to the following cut-off scores: ≤1, ≤1.5, ≤2, and ≤ 2.5 SD below the normative mean. We also calculated the rate at which subjects scored above the normative mean using the same numerical cut-off values, i.e., ≥1, ≥1.5, ≥2, and ≥ 2.5 SD. For both high and low cut-off scores, the sample was stratified by years of education (≤12, 13–15, or ≥ 16). In addition to reporting the MVBR of low scores according to the number of subtests, we also calculated the percentage of subjects falling above or below different cut-offs according to specific subtests of the KBNA. This would provide complimentary data to compare on which tests these individuals would be more likely to fall below the cut-off score.
Results
The demographic characteristics of the 107 participants are presented in Table 1. On average, participants were 75.81 (SD = 7.54) years old with 13.45 (SD = 3.13) years of education. Unfortunately, complete information regarding the number of languages spoken or the source of SCD (i.e., self or informant) was missing for a subset of participants (n = 11). For the 96 older adults who did provide responses to these variables, all spoke English. For half the participants (50%), English was their only language, slightly less than one-third (31.3%) spoke two languages, less than one-tenth (9.4%) spoke three languages. Descriptive statistics for group-level scores on the 12 KBNA subtests can be found in Table 2.
. | . | . | Total . | Women . | Men . | |
---|---|---|---|---|---|---|
M (SD)/N | M (SD)/N | M (SD)/N | ||||
Age | 75.81 (7.54)/107 | 76.92 (7.49)/60 | 74.40 (7.44)/47 | |||
Education | 13.45 (3.13)/106 | 13.20 (2.94)/60 | 13.78 (3.36)/46 | |||
Mini Mental State Exam | 28.84 (0.98)/107 | 28.87 (0.98)/60 | 28.81 (0.99)/47 | |||
Logical Memory II (scaled score) | 10.69 (2.83)/107 | 10.97 (2.66)/60 | 10.34 (3.02)/47 | |||
# | % | % | % | |||
Number of cardiovascular risk factors | 0 | 24.3 | 25.0 | 23.4 | ||
1–2 | 43.0 | 43.3 | 42.6 | |||
≥ 3 | 32.7 | 31.7 | 34.0 | |||
Number of languages spoken | 1 2 | 50.0 31.3 | 56.4 27.3 | 41.5 36.6 | ||
3 | 9.4 | 9.1 | 9.8 | |||
≥ 4 | 9.4 | 7.3 | 12.2 | |||
SCD source (self:other) | 80.2:19.8 | 81.8:18.2 | 78.1:22.0 |
. | . | . | Total . | Women . | Men . | |
---|---|---|---|---|---|---|
M (SD)/N | M (SD)/N | M (SD)/N | ||||
Age | 75.81 (7.54)/107 | 76.92 (7.49)/60 | 74.40 (7.44)/47 | |||
Education | 13.45 (3.13)/106 | 13.20 (2.94)/60 | 13.78 (3.36)/46 | |||
Mini Mental State Exam | 28.84 (0.98)/107 | 28.87 (0.98)/60 | 28.81 (0.99)/47 | |||
Logical Memory II (scaled score) | 10.69 (2.83)/107 | 10.97 (2.66)/60 | 10.34 (3.02)/47 | |||
# | % | % | % | |||
Number of cardiovascular risk factors | 0 | 24.3 | 25.0 | 23.4 | ||
1–2 | 43.0 | 43.3 | 42.6 | |||
≥ 3 | 32.7 | 31.7 | 34.0 | |||
Number of languages spoken | 1 2 | 50.0 31.3 | 56.4 27.3 | 41.5 36.6 | ||
3 | 9.4 | 9.1 | 9.8 | |||
≥ 4 | 9.4 | 7.3 | 12.2 | |||
SCD source (self:other) | 80.2:19.8 | 81.8:18.2 | 78.1:22.0 |
Note. Based on an n = 107. Descriptive statistics are reported for age and education in years. Cumulative percentages are reported for both the number of cardiovascular risk factors and languages spoken, as is the percent of self versus other reported subjective cognitive decline (SCD). Due to missing data during the collection process, percentages reported for languages spoken and SCD source are based on n = 96 participants (female n = 55; male n = 41).
. | . | . | Total . | Women . | Men . | |
---|---|---|---|---|---|---|
M (SD)/N | M (SD)/N | M (SD)/N | ||||
Age | 75.81 (7.54)/107 | 76.92 (7.49)/60 | 74.40 (7.44)/47 | |||
Education | 13.45 (3.13)/106 | 13.20 (2.94)/60 | 13.78 (3.36)/46 | |||
Mini Mental State Exam | 28.84 (0.98)/107 | 28.87 (0.98)/60 | 28.81 (0.99)/47 | |||
Logical Memory II (scaled score) | 10.69 (2.83)/107 | 10.97 (2.66)/60 | 10.34 (3.02)/47 | |||
# | % | % | % | |||
Number of cardiovascular risk factors | 0 | 24.3 | 25.0 | 23.4 | ||
1–2 | 43.0 | 43.3 | 42.6 | |||
≥ 3 | 32.7 | 31.7 | 34.0 | |||
Number of languages spoken | 1 2 | 50.0 31.3 | 56.4 27.3 | 41.5 36.6 | ||
3 | 9.4 | 9.1 | 9.8 | |||
≥ 4 | 9.4 | 7.3 | 12.2 | |||
SCD source (self:other) | 80.2:19.8 | 81.8:18.2 | 78.1:22.0 |
. | . | . | Total . | Women . | Men . | |
---|---|---|---|---|---|---|
M (SD)/N | M (SD)/N | M (SD)/N | ||||
Age | 75.81 (7.54)/107 | 76.92 (7.49)/60 | 74.40 (7.44)/47 | |||
Education | 13.45 (3.13)/106 | 13.20 (2.94)/60 | 13.78 (3.36)/46 | |||
Mini Mental State Exam | 28.84 (0.98)/107 | 28.87 (0.98)/60 | 28.81 (0.99)/47 | |||
Logical Memory II (scaled score) | 10.69 (2.83)/107 | 10.97 (2.66)/60 | 10.34 (3.02)/47 | |||
# | % | % | % | |||
Number of cardiovascular risk factors | 0 | 24.3 | 25.0 | 23.4 | ||
1–2 | 43.0 | 43.3 | 42.6 | |||
≥ 3 | 32.7 | 31.7 | 34.0 | |||
Number of languages spoken | 1 2 | 50.0 31.3 | 56.4 27.3 | 41.5 36.6 | ||
3 | 9.4 | 9.1 | 9.8 | |||
≥ 4 | 9.4 | 7.3 | 12.2 | |||
SCD source (self:other) | 80.2:19.8 | 81.8:18.2 | 78.1:22.0 |
Note. Based on an n = 107. Descriptive statistics are reported for age and education in years. Cumulative percentages are reported for both the number of cardiovascular risk factors and languages spoken, as is the percent of self versus other reported subjective cognitive decline (SCD). Due to missing data during the collection process, percentages reported for languages spoken and SCD source are based on n = 96 participants (female n = 55; male n = 41).
KBNA subtest . | M (SD) . |
---|---|
SEQ | 11.08 (2.36) |
WL1 | 8.82 (3.77) |
CF1 | 9.77 (2.90) |
VisSP | 9.25 (2.72) |
WL2 | 11.24 (2.59) |
WLRec | 9.93 (3.04) |
CF2 | 10.57 (2.95) |
CFRec | 9.27 (3.08) |
SpLoc | 8.26 (3.95) |
PhF | 9.73 (3.22) |
SemF | 9.69 (2.99) |
Preas | 8.09 (4.72) |
KBNA subtest . | M (SD) . |
---|---|
SEQ | 11.08 (2.36) |
WL1 | 8.82 (3.77) |
CF1 | 9.77 (2.90) |
VisSP | 9.25 (2.72) |
WL2 | 11.24 (2.59) |
WLRec | 9.93 (3.04) |
CF2 | 10.57 (2.95) |
CFRec | 9.27 (3.08) |
SpLoc | 8.26 (3.95) |
PhF | 9.73 (3.22) |
SemF | 9.69 (2.99) |
Preas | 8.09 (4.72) |
Note. N = 107. Scores listed are scaled scores with an M = 10 and SD = 3. Scores for the n = 17 participants with missing values were imputed using the impute_EM function from the missMethods package in R. Abbreviations: KBNA = Kaplan–Baycrest Neurocognitive Assessment; SEQ = sequences; WL1 = word list immediate recall; CF1 = complex figure immediate recall; VisSP = visuospatial functioning (complex figure copy/clocks); WL2 = word list delayed recall; WLRec = word list yes/no recognition; CF2 = complex figure delayed recall; CFRec = complex figure recognition; SpLoc = spatial location; PhF = phonemic fluency; SemF = semantic fluency; Preas = reasoning (practical problem solving/conceptual shifting).
KBNA subtest . | M (SD) . |
---|---|
SEQ | 11.08 (2.36) |
WL1 | 8.82 (3.77) |
CF1 | 9.77 (2.90) |
VisSP | 9.25 (2.72) |
WL2 | 11.24 (2.59) |
WLRec | 9.93 (3.04) |
CF2 | 10.57 (2.95) |
CFRec | 9.27 (3.08) |
SpLoc | 8.26 (3.95) |
PhF | 9.73 (3.22) |
SemF | 9.69 (2.99) |
Preas | 8.09 (4.72) |
KBNA subtest . | M (SD) . |
---|---|
SEQ | 11.08 (2.36) |
WL1 | 8.82 (3.77) |
CF1 | 9.77 (2.90) |
VisSP | 9.25 (2.72) |
WL2 | 11.24 (2.59) |
WLRec | 9.93 (3.04) |
CF2 | 10.57 (2.95) |
CFRec | 9.27 (3.08) |
SpLoc | 8.26 (3.95) |
PhF | 9.73 (3.22) |
SemF | 9.69 (2.99) |
Preas | 8.09 (4.72) |
Note. N = 107. Scores listed are scaled scores with an M = 10 and SD = 3. Scores for the n = 17 participants with missing values were imputed using the impute_EM function from the missMethods package in R. Abbreviations: KBNA = Kaplan–Baycrest Neurocognitive Assessment; SEQ = sequences; WL1 = word list immediate recall; CF1 = complex figure immediate recall; VisSP = visuospatial functioning (complex figure copy/clocks); WL2 = word list delayed recall; WLRec = word list yes/no recognition; CF2 = complex figure delayed recall; CFRec = complex figure recognition; SpLoc = spatial location; PhF = phonemic fluency; SemF = semantic fluency; Preas = reasoning (practical problem solving/conceptual shifting).
The prevalence rates of older adults obtaining low scores on the KBNA, when the 12 subtests are considered simultaneously, are provided in Table 3. In the total sample, at a cut-off threshold of 1 SD below the mean, obtaining low scores was very common. Having one or more low scores occurred in most participants (MVBR = 84.1%), and two or more scores below cut-off occurred in more than two-thirds of participants (MVBR = 68.2%), with MVBRs decreasing to about half of participants (MVBR = 51.4%) having three or more low scores. At the slightly stricter criterion of 1.5 SD below the mean, one or more low scores occurred in 55.1% of participants, dropping to 37.4% at two or more and 18.7% at three or more low scores, respectively. When the threshold was increased to 2 SD below the mean, we observed further decreases in the prevalence of low scores. One or more low scores occurred in 39.3% of participants, two or more in 29.9% of participants, with three scores below cut-off occurring infrequently (7.5% of participants). These results are consistent with previous findings that suggest as cut-off criteria became more extreme, fewer low scores were obtained (Iverson & Brooks, 2011).
Low score . | % of n . | Years of education . | ||||
---|---|---|---|---|---|---|
. | Total . | Women . | Men . | ≤12 . | 13–15 . | ≥16 . |
N | 107 | 60 | 47 | 42 | 32 | 32 |
≤ −1 SD | ||||||
0 scores | 15.9 | 21.7 | 8.5 | 11.9 | 15.6 | 21.9 |
1+ scores | 84.1 | 78.3 | 91.5 | 88.1 | 84.4 | 78.1 |
2+ scores | 68.2 | 61.7 | 76.6 | 71.4 | 68.8 | 62.5 |
3+ scores | 51.4 | 48.3 | 55.3 | 52.4 | 46.9 | 56.3 |
4+ scores | 34.6 | 31.7 | 38.3 | 33.3 | 40.6 | 31.3 |
5+ scores | 21.5 | 13.3 | 31.9 | 28.6 | 12.5 | 21.9 |
6+ scores | 13.1 | 8.3 | 19.2 | 16.7 | 9.4 | 12.5 |
≤ −1.5 SD | ||||||
0 scores | 44.9 | 55 | 31.9 | 42.9 | 43.8 | 50 |
1+ scores | 55.1 | 45 | 68.1 | 57.1 | 56.3 | 50 |
2+ scores | 37.4 | 25 | 53.2 | 42.9 | 37.5 | 31.3 |
3+ scores | 18.7 | 13.3 | 25.5 | 21.4 | 15.7 | 18.8 |
4+ scores | 7.5 | 5 | 10.6 | 14.3 | 3.1 | 3.1 |
5+ scores | 3.7 | 1.7 | 6.4 | 4.8 | 3.1 | 3.1 |
6+ scores | 1.9 | 0 | 4.3 | 2.4 | 0 | 0 |
≤ −2 SD | ||||||
0 scores | 60.8 | 68.3 | 51.1 | 50 | 62.5 | 68.8 |
1+ scores | 39.3 | 31.7 | 48.9 | 50 | 37.5 | 31.3 |
2+ scores | 29.9 | 21.7 | 40.4 | 31 | 28.1 | 31.3 |
3+ scores | 7.5 | 5 | 10.6 | 11.9 | 3.1 | 6.3 |
4+ scores | 1.9 | 0 | 4.3 | 4.8 | 0 | 0 |
5+ scores | 0.9 | 0 | 2.1 | 2.4 | 0 | 0 |
≤ −2.5 SD | ||||||
0 scores | 73.8 | 81.7 | 63.8 | 73.8 | 71.9 | 75 |
1+ scores | 26.2 | 18.3 | 36.2 | 26.2 | 28.1 | 25 |
2+ scores | 23.4 | 16.7 | 31.9 | 26.2 | 25 | 18.8 |
3 scores | 2.8 | 1.7 | 4.3 | 2.4 | 3.1 | 3.1 |
Low score . | % of n . | Years of education . | ||||
---|---|---|---|---|---|---|
. | Total . | Women . | Men . | ≤12 . | 13–15 . | ≥16 . |
N | 107 | 60 | 47 | 42 | 32 | 32 |
≤ −1 SD | ||||||
0 scores | 15.9 | 21.7 | 8.5 | 11.9 | 15.6 | 21.9 |
1+ scores | 84.1 | 78.3 | 91.5 | 88.1 | 84.4 | 78.1 |
2+ scores | 68.2 | 61.7 | 76.6 | 71.4 | 68.8 | 62.5 |
3+ scores | 51.4 | 48.3 | 55.3 | 52.4 | 46.9 | 56.3 |
4+ scores | 34.6 | 31.7 | 38.3 | 33.3 | 40.6 | 31.3 |
5+ scores | 21.5 | 13.3 | 31.9 | 28.6 | 12.5 | 21.9 |
6+ scores | 13.1 | 8.3 | 19.2 | 16.7 | 9.4 | 12.5 |
≤ −1.5 SD | ||||||
0 scores | 44.9 | 55 | 31.9 | 42.9 | 43.8 | 50 |
1+ scores | 55.1 | 45 | 68.1 | 57.1 | 56.3 | 50 |
2+ scores | 37.4 | 25 | 53.2 | 42.9 | 37.5 | 31.3 |
3+ scores | 18.7 | 13.3 | 25.5 | 21.4 | 15.7 | 18.8 |
4+ scores | 7.5 | 5 | 10.6 | 14.3 | 3.1 | 3.1 |
5+ scores | 3.7 | 1.7 | 6.4 | 4.8 | 3.1 | 3.1 |
6+ scores | 1.9 | 0 | 4.3 | 2.4 | 0 | 0 |
≤ −2 SD | ||||||
0 scores | 60.8 | 68.3 | 51.1 | 50 | 62.5 | 68.8 |
1+ scores | 39.3 | 31.7 | 48.9 | 50 | 37.5 | 31.3 |
2+ scores | 29.9 | 21.7 | 40.4 | 31 | 28.1 | 31.3 |
3+ scores | 7.5 | 5 | 10.6 | 11.9 | 3.1 | 6.3 |
4+ scores | 1.9 | 0 | 4.3 | 4.8 | 0 | 0 |
5+ scores | 0.9 | 0 | 2.1 | 2.4 | 0 | 0 |
≤ −2.5 SD | ||||||
0 scores | 73.8 | 81.7 | 63.8 | 73.8 | 71.9 | 75 |
1+ scores | 26.2 | 18.3 | 36.2 | 26.2 | 28.1 | 25 |
2+ scores | 23.4 | 16.7 | 31.9 | 26.2 | 25 | 18.8 |
3 scores | 2.8 | 1.7 | 4.3 | 2.4 | 3.1 | 3.1 |
Note. Values represent cumulative percentages of low scores ranging from 1 SD to 2.5 SD below the mean on the 12 subtests of the Kaplan–Baycrest Neurocognitive Assessment (KBNA). Scores for the n = 17 participants with missing values were imputed using the impute_EM function from the missMethods package in R.
Low score . | % of n . | Years of education . | ||||
---|---|---|---|---|---|---|
. | Total . | Women . | Men . | ≤12 . | 13–15 . | ≥16 . |
N | 107 | 60 | 47 | 42 | 32 | 32 |
≤ −1 SD | ||||||
0 scores | 15.9 | 21.7 | 8.5 | 11.9 | 15.6 | 21.9 |
1+ scores | 84.1 | 78.3 | 91.5 | 88.1 | 84.4 | 78.1 |
2+ scores | 68.2 | 61.7 | 76.6 | 71.4 | 68.8 | 62.5 |
3+ scores | 51.4 | 48.3 | 55.3 | 52.4 | 46.9 | 56.3 |
4+ scores | 34.6 | 31.7 | 38.3 | 33.3 | 40.6 | 31.3 |
5+ scores | 21.5 | 13.3 | 31.9 | 28.6 | 12.5 | 21.9 |
6+ scores | 13.1 | 8.3 | 19.2 | 16.7 | 9.4 | 12.5 |
≤ −1.5 SD | ||||||
0 scores | 44.9 | 55 | 31.9 | 42.9 | 43.8 | 50 |
1+ scores | 55.1 | 45 | 68.1 | 57.1 | 56.3 | 50 |
2+ scores | 37.4 | 25 | 53.2 | 42.9 | 37.5 | 31.3 |
3+ scores | 18.7 | 13.3 | 25.5 | 21.4 | 15.7 | 18.8 |
4+ scores | 7.5 | 5 | 10.6 | 14.3 | 3.1 | 3.1 |
5+ scores | 3.7 | 1.7 | 6.4 | 4.8 | 3.1 | 3.1 |
6+ scores | 1.9 | 0 | 4.3 | 2.4 | 0 | 0 |
≤ −2 SD | ||||||
0 scores | 60.8 | 68.3 | 51.1 | 50 | 62.5 | 68.8 |
1+ scores | 39.3 | 31.7 | 48.9 | 50 | 37.5 | 31.3 |
2+ scores | 29.9 | 21.7 | 40.4 | 31 | 28.1 | 31.3 |
3+ scores | 7.5 | 5 | 10.6 | 11.9 | 3.1 | 6.3 |
4+ scores | 1.9 | 0 | 4.3 | 4.8 | 0 | 0 |
5+ scores | 0.9 | 0 | 2.1 | 2.4 | 0 | 0 |
≤ −2.5 SD | ||||||
0 scores | 73.8 | 81.7 | 63.8 | 73.8 | 71.9 | 75 |
1+ scores | 26.2 | 18.3 | 36.2 | 26.2 | 28.1 | 25 |
2+ scores | 23.4 | 16.7 | 31.9 | 26.2 | 25 | 18.8 |
3 scores | 2.8 | 1.7 | 4.3 | 2.4 | 3.1 | 3.1 |
Low score . | % of n . | Years of education . | ||||
---|---|---|---|---|---|---|
. | Total . | Women . | Men . | ≤12 . | 13–15 . | ≥16 . |
N | 107 | 60 | 47 | 42 | 32 | 32 |
≤ −1 SD | ||||||
0 scores | 15.9 | 21.7 | 8.5 | 11.9 | 15.6 | 21.9 |
1+ scores | 84.1 | 78.3 | 91.5 | 88.1 | 84.4 | 78.1 |
2+ scores | 68.2 | 61.7 | 76.6 | 71.4 | 68.8 | 62.5 |
3+ scores | 51.4 | 48.3 | 55.3 | 52.4 | 46.9 | 56.3 |
4+ scores | 34.6 | 31.7 | 38.3 | 33.3 | 40.6 | 31.3 |
5+ scores | 21.5 | 13.3 | 31.9 | 28.6 | 12.5 | 21.9 |
6+ scores | 13.1 | 8.3 | 19.2 | 16.7 | 9.4 | 12.5 |
≤ −1.5 SD | ||||||
0 scores | 44.9 | 55 | 31.9 | 42.9 | 43.8 | 50 |
1+ scores | 55.1 | 45 | 68.1 | 57.1 | 56.3 | 50 |
2+ scores | 37.4 | 25 | 53.2 | 42.9 | 37.5 | 31.3 |
3+ scores | 18.7 | 13.3 | 25.5 | 21.4 | 15.7 | 18.8 |
4+ scores | 7.5 | 5 | 10.6 | 14.3 | 3.1 | 3.1 |
5+ scores | 3.7 | 1.7 | 6.4 | 4.8 | 3.1 | 3.1 |
6+ scores | 1.9 | 0 | 4.3 | 2.4 | 0 | 0 |
≤ −2 SD | ||||||
0 scores | 60.8 | 68.3 | 51.1 | 50 | 62.5 | 68.8 |
1+ scores | 39.3 | 31.7 | 48.9 | 50 | 37.5 | 31.3 |
2+ scores | 29.9 | 21.7 | 40.4 | 31 | 28.1 | 31.3 |
3+ scores | 7.5 | 5 | 10.6 | 11.9 | 3.1 | 6.3 |
4+ scores | 1.9 | 0 | 4.3 | 4.8 | 0 | 0 |
5+ scores | 0.9 | 0 | 2.1 | 2.4 | 0 | 0 |
≤ −2.5 SD | ||||||
0 scores | 73.8 | 81.7 | 63.8 | 73.8 | 71.9 | 75 |
1+ scores | 26.2 | 18.3 | 36.2 | 26.2 | 28.1 | 25 |
2+ scores | 23.4 | 16.7 | 31.9 | 26.2 | 25 | 18.8 |
3 scores | 2.8 | 1.7 | 4.3 | 2.4 | 3.1 | 3.1 |
Note. Values represent cumulative percentages of low scores ranging from 1 SD to 2.5 SD below the mean on the 12 subtests of the Kaplan–Baycrest Neurocognitive Assessment (KBNA). Scores for the n = 17 participants with missing values were imputed using the impute_EM function from the missMethods package in R.
Table 3 also contains the MVBR of low KNBA scores, stratified by gender. The frequency of obtaining low scores across all cut-offs was higher for men than women, demonstrating a potential cognitive advantage for women in older age. For example, at a cut-off threshold of 1 SD below the mean, one or more low scores occurred in 91.5% of male and 78.3% of female participants. As the cut-off criterion increased to 1.5 SD below the mean, this trend continued with one or more low scores occurring in less than half of women (MVBR = 45%) and over two-thirds of men (MVBR = 68.1%). Moreover, at 2 SD below the mean, one or more low scores were obtained by approximately half of male (MVBR = 48.9%) and less than one-third of female participants (MVBR = 31.7%). This pattern of results was maintained as the number of low scores at each threshold increased, reflecting a possible cognitive advantage for women in this sample.
Stratification of low KBNA scores by education can also be found in Table 3. When considering 1 or 2 low scores, individuals with fewer years of education typically produced a greater number of low scores, whereas individuals with higher education typically produced fewer low scores. For example, at a cut-off threshold of 1 SD below the mean, one or more low scores occurred in 88.1% of individuals with ≤12 years of education, 84.4% of individuals with 13–15 years of education, and 78.1% of those with ≥16 years of education. Similarly, at a cut-off threshold of 1.5 SD below the mean, one or more low scores occurred in 57.1% of individuals with ≤12 years of education, 56.3% of individuals with 13–15 years of education, and 50% of those with ≥16 years of education; two or more low scores occurring in 42.9%, 37.5%, and 31.3%, respectively. Taken together, there appears to be an advantage of high education, in line with previous work suggesting lower base rates of producing low test scores in those who are highly educated (Brooks et al., 2007; Brooks et al., 2008; Brooks, Iverson, Lanting, Horton, & Reynolds, 2012; Holdnack et al., 2017).
The prevalence of high scores is presented in Table 4. When considering the 12 KBNA subtests simultaneously, achieving high scores was quite common. At a cut-off criterion of 1 SD above the mean, most participants (MVBR = 80.4%) had one or more high scores, more than half (MVBR = 56.1%) showing two or more high scores, three or more scores occurred in slightly above one-third (MVBR = 36.5%) and four or more scores in approximately one-fourth of participants (MVBR = 26.2%). Consistent with the pattern of low score prevalence rates, as the cut-off criterion became more stringent, fewer high scores were obtained. When a criterion of 1.5 SD above the mean was employed, approximately one-third of participants (MVBR = 35.5%) had at least one high score. This rate decreased to 18.7% for two or more and 10.3% for three or more high scores, with four high scores (MVBR = 3.7%) being the maximum number of high scores achieved at this threshold. Increasing the cut-off criterion to 2 SD above the mean again reduced the prevalence of high scores, with one or more high scores occurring in approximately one-sixth (MVBR = 16.8%) of participants and two or more in 6.5%; no participants obtained more than two scores above 2 SD from the mean.
High score . | % of n . | Years of education . | ||||
---|---|---|---|---|---|---|
. | Total . | Women . | Men . | ≤12 . | 13–15 . | ≥16 . |
N | 107 | 60 | 47 | 42 | 32 | 32 |
≥ 1 SD | ||||||
0 scores | 19.6 | 15 | 25.5 | 26.2 | 9.4 | 21.9 |
1+ scores | 80.4 | 85 | 74.5 | 73.8 | 90.6 | 78.1 |
2+ scores | 56.1 | 61.7 | 48.9 | 50 | 50 | 68.8 |
3+ scores | 36.5 | 43.3 | 27.7 | 28.6 | 28.1 | 53.1 |
4+ scores | 26.2 | 36.7 | 12.8 | 19.1 | 25 | 37.5 |
5+ scores | 15.9 | 21.6 | 8.5 | 11.9 | 18.8 | 18.8 |
6+ scores | 11.2 | 13.3 | 8.5 | 9.5 | 9.4 | 15.6 |
≥ 1.5 SD | ||||||
0 scores | 64.5 | 55 | 76.6 | 73.8 | 65.6 | 50 |
1+ scores | 35.5 | 45 | 23.4 | 26.2 | 34.4 | 50 |
2+ scores | 18.7 | 26.7 | 8.5 | 9.5 | 18.8 | 31.3 |
3+ scores | 10.3 | 13.3 | 6.4 | 7.1 | 6.3 | 18.3 |
4 scores | 3.7 | 6.7 | 0 | 4.8 | 3.1 | 3.1 |
≥ 2 SD | ||||||
0 scores | 83.2 | 78.3 | 89.4 | 92.9 | 78.1 | 75 |
1+ scores | 16.8 | 21.7 | 10.6 | 7.1 | 21.9 | 25 |
2 scores | 6.5 | 8.3 | 4.3 | 0 | 6.3 | 15.6 |
≥ 2.5 SD | ||||||
0 scores | 99.1 | 98.3 | 100 | 100 | 96.9 | 100 |
1 score | 0.9 | 1.7 | 0 | 0 | 3.1 | 0 |
High score . | % of n . | Years of education . | ||||
---|---|---|---|---|---|---|
. | Total . | Women . | Men . | ≤12 . | 13–15 . | ≥16 . |
N | 107 | 60 | 47 | 42 | 32 | 32 |
≥ 1 SD | ||||||
0 scores | 19.6 | 15 | 25.5 | 26.2 | 9.4 | 21.9 |
1+ scores | 80.4 | 85 | 74.5 | 73.8 | 90.6 | 78.1 |
2+ scores | 56.1 | 61.7 | 48.9 | 50 | 50 | 68.8 |
3+ scores | 36.5 | 43.3 | 27.7 | 28.6 | 28.1 | 53.1 |
4+ scores | 26.2 | 36.7 | 12.8 | 19.1 | 25 | 37.5 |
5+ scores | 15.9 | 21.6 | 8.5 | 11.9 | 18.8 | 18.8 |
6+ scores | 11.2 | 13.3 | 8.5 | 9.5 | 9.4 | 15.6 |
≥ 1.5 SD | ||||||
0 scores | 64.5 | 55 | 76.6 | 73.8 | 65.6 | 50 |
1+ scores | 35.5 | 45 | 23.4 | 26.2 | 34.4 | 50 |
2+ scores | 18.7 | 26.7 | 8.5 | 9.5 | 18.8 | 31.3 |
3+ scores | 10.3 | 13.3 | 6.4 | 7.1 | 6.3 | 18.3 |
4 scores | 3.7 | 6.7 | 0 | 4.8 | 3.1 | 3.1 |
≥ 2 SD | ||||||
0 scores | 83.2 | 78.3 | 89.4 | 92.9 | 78.1 | 75 |
1+ scores | 16.8 | 21.7 | 10.6 | 7.1 | 21.9 | 25 |
2 scores | 6.5 | 8.3 | 4.3 | 0 | 6.3 | 15.6 |
≥ 2.5 SD | ||||||
0 scores | 99.1 | 98.3 | 100 | 100 | 96.9 | 100 |
1 score | 0.9 | 1.7 | 0 | 0 | 3.1 | 0 |
Note. Values represent cumulative percentages of high scores ranging from 1 SD to 2.5 SD above the mean on the 12 subtests of the Kaplan–Baycrest Neurocognitive Assessment (KBNA). Scores for the n = 17 participants with missing values were imputed using the impute_EM function from the missMethods package in R.
High score . | % of n . | Years of education . | ||||
---|---|---|---|---|---|---|
. | Total . | Women . | Men . | ≤12 . | 13–15 . | ≥16 . |
N | 107 | 60 | 47 | 42 | 32 | 32 |
≥ 1 SD | ||||||
0 scores | 19.6 | 15 | 25.5 | 26.2 | 9.4 | 21.9 |
1+ scores | 80.4 | 85 | 74.5 | 73.8 | 90.6 | 78.1 |
2+ scores | 56.1 | 61.7 | 48.9 | 50 | 50 | 68.8 |
3+ scores | 36.5 | 43.3 | 27.7 | 28.6 | 28.1 | 53.1 |
4+ scores | 26.2 | 36.7 | 12.8 | 19.1 | 25 | 37.5 |
5+ scores | 15.9 | 21.6 | 8.5 | 11.9 | 18.8 | 18.8 |
6+ scores | 11.2 | 13.3 | 8.5 | 9.5 | 9.4 | 15.6 |
≥ 1.5 SD | ||||||
0 scores | 64.5 | 55 | 76.6 | 73.8 | 65.6 | 50 |
1+ scores | 35.5 | 45 | 23.4 | 26.2 | 34.4 | 50 |
2+ scores | 18.7 | 26.7 | 8.5 | 9.5 | 18.8 | 31.3 |
3+ scores | 10.3 | 13.3 | 6.4 | 7.1 | 6.3 | 18.3 |
4 scores | 3.7 | 6.7 | 0 | 4.8 | 3.1 | 3.1 |
≥ 2 SD | ||||||
0 scores | 83.2 | 78.3 | 89.4 | 92.9 | 78.1 | 75 |
1+ scores | 16.8 | 21.7 | 10.6 | 7.1 | 21.9 | 25 |
2 scores | 6.5 | 8.3 | 4.3 | 0 | 6.3 | 15.6 |
≥ 2.5 SD | ||||||
0 scores | 99.1 | 98.3 | 100 | 100 | 96.9 | 100 |
1 score | 0.9 | 1.7 | 0 | 0 | 3.1 | 0 |
High score . | % of n . | Years of education . | ||||
---|---|---|---|---|---|---|
. | Total . | Women . | Men . | ≤12 . | 13–15 . | ≥16 . |
N | 107 | 60 | 47 | 42 | 32 | 32 |
≥ 1 SD | ||||||
0 scores | 19.6 | 15 | 25.5 | 26.2 | 9.4 | 21.9 |
1+ scores | 80.4 | 85 | 74.5 | 73.8 | 90.6 | 78.1 |
2+ scores | 56.1 | 61.7 | 48.9 | 50 | 50 | 68.8 |
3+ scores | 36.5 | 43.3 | 27.7 | 28.6 | 28.1 | 53.1 |
4+ scores | 26.2 | 36.7 | 12.8 | 19.1 | 25 | 37.5 |
5+ scores | 15.9 | 21.6 | 8.5 | 11.9 | 18.8 | 18.8 |
6+ scores | 11.2 | 13.3 | 8.5 | 9.5 | 9.4 | 15.6 |
≥ 1.5 SD | ||||||
0 scores | 64.5 | 55 | 76.6 | 73.8 | 65.6 | 50 |
1+ scores | 35.5 | 45 | 23.4 | 26.2 | 34.4 | 50 |
2+ scores | 18.7 | 26.7 | 8.5 | 9.5 | 18.8 | 31.3 |
3+ scores | 10.3 | 13.3 | 6.4 | 7.1 | 6.3 | 18.3 |
4 scores | 3.7 | 6.7 | 0 | 4.8 | 3.1 | 3.1 |
≥ 2 SD | ||||||
0 scores | 83.2 | 78.3 | 89.4 | 92.9 | 78.1 | 75 |
1+ scores | 16.8 | 21.7 | 10.6 | 7.1 | 21.9 | 25 |
2 scores | 6.5 | 8.3 | 4.3 | 0 | 6.3 | 15.6 |
≥ 2.5 SD | ||||||
0 scores | 99.1 | 98.3 | 100 | 100 | 96.9 | 100 |
1 score | 0.9 | 1.7 | 0 | 0 | 3.1 | 0 |
Note. Values represent cumulative percentages of high scores ranging from 1 SD to 2.5 SD above the mean on the 12 subtests of the Kaplan–Baycrest Neurocognitive Assessment (KBNA). Scores for the n = 17 participants with missing values were imputed using the impute_EM function from the missMethods package in R.
Table 4 also contains gender stratified high KNBA scores. Similar to the pattern observed for low scores, there appears to be a cognitive advantage for women, with men consistently achieving fewer high scores. At a cut-off of 1 SD above the mean, the likelihood of obtaining both one or two high scores was |$\sim$|10% larger for women (1+ scores: MVBR = 85%; 2+ scores: MVBR = 61.7%) compared with men (1+ scores: MVBR = 74.5%; 2+ scores: MVBR = 48.9). Similarly, achieving one or more high scores at a threshold of 1.5 SD above the mean occurred in 45% of women relative to 23.4% of men. Moreover, two or more high scores occurred in 26.7% of women and 8.5% of men at this cut-off. Importantly, women obtained more high scores across all cut-offs, demonstrating that above average performance was more in prevalent in women compared with men, providing additional support for a cognitive advantage for women in this sample.
Regarding the impact of education on obtaining high scores, the results suggested an advantage of more years of education. For example, at a cut-off of 1 SD above the mean, high scores occurred in roughly half of individuals with either ≤12 or 13–15 years of education (MVBR = 50%) and approximately two-thirds of individuals with ≥16 years of education (MVBR = 68.8). As the cut-off criteria became more stringent, the advantage observed by the highly educated group persisted. For example, one or more high scores occurred in half (MVBR = 50%) and a quarter (MVBR = 25%) of individuals with ≥16 years of education at 1.5 SD and 2 SD above the mean, respectively, whereas the other groups produced above average scores at a lower base rate (13–15 years of education MVBR = 34.4% at 1.5 SD and 21.9% at 2 SD; ≤12 years of education MVBR = 26.2% at 1.5 SD, 7.1% at 2 SD above the mean). These findings suggest that obtaining high scores is common, particularly at liberal thresholds and for individuals with higher education.
In addition to the MVBR of obtaining a low score on n subtests, to provide additional context we also calculated the percentage of individuals who scored below cut-off as a function of each KBNA subtest, which would provide a complementary picture as to which subtests older adults were most likely to provide low scores. Tables 5 and 6 contain the percentages of participants who obtained low and high scores, respectively. There was a large amount of variability on performance across subtests, particularly regarding low scores. On two of the 12 subtests, few participants (<10%) received scores 1 SD below the mean (Sequences MVBR = 7.5%; Word List Delayed Recall MVBR = 8.4%). Frequencies at this cut-off were higher across the remaining 10 subtests, with over 30% of individuals obtaining low scores on four subtests (Word List Immediate Recall MVBR = 33.6%; Complex Figure Recognition MVBR = 34.6%; Spatial location MVBR = 38.3%; Practical Problem Solving MVBR = 35.5%). When increasing the cut-off criterion to 1.5 SD below the mean, low scores were less frequent, occurring in > 10% of individuals only for the above three subtests and Phonemic Fluency (i.e., Word List Immediate Recall MVBR = 15.9%; Complex Figure Recognition MVBR = 15.0%; Spatial location MVBR = 17.8%; Practical Problem Solving MVBR = 29.0%; Phonemic Fluency MVBR = 11.2%). At a cut-off threshold of 2 SD below the mean, frequencies of low scores again decreased, with only scores on three subtests occurring in > 5% of the sample (Word List Immediate Recall MVBR = 13.1%; Spatial location MVBR = 15.0%; Practical Problem Solving MVBR = 27.1%). Regarding variability of unusually high scores on the 12 KBNA subtests, the percent of participants with scores at and exceeding 1 SD above the mean ranged from 14.0% on Visuospatial Functioning to 32.7% on Sequences. Scoring 1.5 SD and above the mean ranged from 0.9% on Complex Figure Recognition up to 14.0% on Word List Delayed Recall. At a cut-off of 2 SD above the mean, scores were not obtained on three subtests (i.e., Sequences, Complex Figure Delayed Recall, and Complex Figure Recognition) and ranged up to 4.7% on Phonemic Fluency. These findings demonstrate the large variability across KBNA subtests for both high and low scores at different cut-off thresholds.
KBNA subtest . | Score % . | . | . | . |
---|---|---|---|---|
≤ −1 SD . | ≤ −1.5 SD . | ≤ −2 SD . | ≤ −2.5 SD . | |
SEQ | 7.5 | 1.9 | 1.9 | 0.9 |
WL1 | 33.6 | 15.9 | 13.1 | 9.5 |
CF1 | 19.6 | 7.5 | 3.7 | 0 |
VisSP | 24.3 | 9.4 | 1.9 | 0.9 |
WL2 | 8.4 | 0.9 | 0.9 | 0 |
WLRec | 24.3 | 5.6 | 0.9 | 0 |
CF2 | 15.9 | 3.7 | 1.9 | 0 |
CFRec | 34.6 | 15.0 | 4.7 | 0 |
SpLoc | 38.3 | 17.8 | 15.0 | 14.0 |
PhF | 22.4 | 11.2 | 4.7 | 1.9 |
SemF | 25.3 | 6.5 | 3.7 | 0.9 |
Preas | 35.5 | 29.0 | 27.1 | 24.3 |
KBNA subtest . | Score % . | . | . | . |
---|---|---|---|---|
≤ −1 SD . | ≤ −1.5 SD . | ≤ −2 SD . | ≤ −2.5 SD . | |
SEQ | 7.5 | 1.9 | 1.9 | 0.9 |
WL1 | 33.6 | 15.9 | 13.1 | 9.5 |
CF1 | 19.6 | 7.5 | 3.7 | 0 |
VisSP | 24.3 | 9.4 | 1.9 | 0.9 |
WL2 | 8.4 | 0.9 | 0.9 | 0 |
WLRec | 24.3 | 5.6 | 0.9 | 0 |
CF2 | 15.9 | 3.7 | 1.9 | 0 |
CFRec | 34.6 | 15.0 | 4.7 | 0 |
SpLoc | 38.3 | 17.8 | 15.0 | 14.0 |
PhF | 22.4 | 11.2 | 4.7 | 1.9 |
SemF | 25.3 | 6.5 | 3.7 | 0.9 |
Preas | 35.5 | 29.0 | 27.1 | 24.3 |
Note. N = 107. Values represent cumulative percentages of low scores ranging from 1 SD to 2.5 SD below the mean. Scores for the n = 17 participants with missing values were imputed using the impute_EM function from the missMethods package in R. Abbreviations:: KBNA = Kaplan–Baycrest Neurocognitive Assessment; SEQ = sequences; WL1 = word list immediate recall; CF1 = complex figure immediate recall; VisSP = visuospatial functioning (complex figure copy/clocks); WL2 = word list delayed recall; WLRec = word list yes/no recognition; CF2 = complex figure delayed recall; CFRec = complex figure recognition; SpLoc = spatial location; PhF = phonemic fluency; SemF = semantic fluency; Preas = reasoning (Practical Problem Solving/Conceptual Shifting).
KBNA subtest . | Score % . | . | . | . |
---|---|---|---|---|
≤ −1 SD . | ≤ −1.5 SD . | ≤ −2 SD . | ≤ −2.5 SD . | |
SEQ | 7.5 | 1.9 | 1.9 | 0.9 |
WL1 | 33.6 | 15.9 | 13.1 | 9.5 |
CF1 | 19.6 | 7.5 | 3.7 | 0 |
VisSP | 24.3 | 9.4 | 1.9 | 0.9 |
WL2 | 8.4 | 0.9 | 0.9 | 0 |
WLRec | 24.3 | 5.6 | 0.9 | 0 |
CF2 | 15.9 | 3.7 | 1.9 | 0 |
CFRec | 34.6 | 15.0 | 4.7 | 0 |
SpLoc | 38.3 | 17.8 | 15.0 | 14.0 |
PhF | 22.4 | 11.2 | 4.7 | 1.9 |
SemF | 25.3 | 6.5 | 3.7 | 0.9 |
Preas | 35.5 | 29.0 | 27.1 | 24.3 |
KBNA subtest . | Score % . | . | . | . |
---|---|---|---|---|
≤ −1 SD . | ≤ −1.5 SD . | ≤ −2 SD . | ≤ −2.5 SD . | |
SEQ | 7.5 | 1.9 | 1.9 | 0.9 |
WL1 | 33.6 | 15.9 | 13.1 | 9.5 |
CF1 | 19.6 | 7.5 | 3.7 | 0 |
VisSP | 24.3 | 9.4 | 1.9 | 0.9 |
WL2 | 8.4 | 0.9 | 0.9 | 0 |
WLRec | 24.3 | 5.6 | 0.9 | 0 |
CF2 | 15.9 | 3.7 | 1.9 | 0 |
CFRec | 34.6 | 15.0 | 4.7 | 0 |
SpLoc | 38.3 | 17.8 | 15.0 | 14.0 |
PhF | 22.4 | 11.2 | 4.7 | 1.9 |
SemF | 25.3 | 6.5 | 3.7 | 0.9 |
Preas | 35.5 | 29.0 | 27.1 | 24.3 |
Note. N = 107. Values represent cumulative percentages of low scores ranging from 1 SD to 2.5 SD below the mean. Scores for the n = 17 participants with missing values were imputed using the impute_EM function from the missMethods package in R. Abbreviations:: KBNA = Kaplan–Baycrest Neurocognitive Assessment; SEQ = sequences; WL1 = word list immediate recall; CF1 = complex figure immediate recall; VisSP = visuospatial functioning (complex figure copy/clocks); WL2 = word list delayed recall; WLRec = word list yes/no recognition; CF2 = complex figure delayed recall; CFRec = complex figure recognition; SpLoc = spatial location; PhF = phonemic fluency; SemF = semantic fluency; Preas = reasoning (Practical Problem Solving/Conceptual Shifting).
KBNA subtest . | . | Score % . | . | . |
---|---|---|---|---|
≥ 1 SD . | ≥ 1.5 SD . | ≥ 2 SD . | ≥ 2.5 SD . | |
SEQ | 32.7 | 2.8 | 0 | 0 |
WL1 | 17.8 | 2.8 | 1.9 | 0 |
CF1 | 15.0 | 6.5 | 4.7 | 0 |
VisSP | 14.0 | 1.9 | 1.9 | 0 |
WL2 | 28.0 | 14.0 | 2.8 | 0.9 |
WLRec | 23.4 | 8.4 | 0.9 | 0 |
CF2 | 29.0 | 11.2 | 0 | 0 |
CFRec | 16.8 | 0.9 | 0 | 0 |
SpLoc | 15.0 | 5.6 | 1.9 | 0 |
PhF | 18.7 | 5.6 | 4.7 | 0 |
SemF | 18.7 | 4.7 | 1.9 | 0 |
Preas | 18.7 | 3.7 | 2.8 | 0 |
KBNA subtest . | . | Score % . | . | . |
---|---|---|---|---|
≥ 1 SD . | ≥ 1.5 SD . | ≥ 2 SD . | ≥ 2.5 SD . | |
SEQ | 32.7 | 2.8 | 0 | 0 |
WL1 | 17.8 | 2.8 | 1.9 | 0 |
CF1 | 15.0 | 6.5 | 4.7 | 0 |
VisSP | 14.0 | 1.9 | 1.9 | 0 |
WL2 | 28.0 | 14.0 | 2.8 | 0.9 |
WLRec | 23.4 | 8.4 | 0.9 | 0 |
CF2 | 29.0 | 11.2 | 0 | 0 |
CFRec | 16.8 | 0.9 | 0 | 0 |
SpLoc | 15.0 | 5.6 | 1.9 | 0 |
PhF | 18.7 | 5.6 | 4.7 | 0 |
SemF | 18.7 | 4.7 | 1.9 | 0 |
Preas | 18.7 | 3.7 | 2.8 | 0 |
Note. N = 107. Values represent cumulative percentages of high scores ranging from 1 SD to 2.5 SD above the mean. Scores for the n = 17 participants with missing values were imputed using the impute_EM function from the missMethods package in R. Abbreviations: SEQ = sequences; WL1 = word list immediate recall; CF1 = complex figure immediate recall; VisSP = visuospatial functioning (complex figure copy/clocks); WL2 = word list delayed recall; WLRec = word list yes/no recognition; CF2 = complex figure delayed recall; CFRec = complex figure recognition; SpLoc = spatial location; PhF = phonemic fluency; SemF = semantic fluency; Preas = reasoning (practical problem solving/conceptual shifting).
KBNA subtest . | . | Score % . | . | . |
---|---|---|---|---|
≥ 1 SD . | ≥ 1.5 SD . | ≥ 2 SD . | ≥ 2.5 SD . | |
SEQ | 32.7 | 2.8 | 0 | 0 |
WL1 | 17.8 | 2.8 | 1.9 | 0 |
CF1 | 15.0 | 6.5 | 4.7 | 0 |
VisSP | 14.0 | 1.9 | 1.9 | 0 |
WL2 | 28.0 | 14.0 | 2.8 | 0.9 |
WLRec | 23.4 | 8.4 | 0.9 | 0 |
CF2 | 29.0 | 11.2 | 0 | 0 |
CFRec | 16.8 | 0.9 | 0 | 0 |
SpLoc | 15.0 | 5.6 | 1.9 | 0 |
PhF | 18.7 | 5.6 | 4.7 | 0 |
SemF | 18.7 | 4.7 | 1.9 | 0 |
Preas | 18.7 | 3.7 | 2.8 | 0 |
KBNA subtest . | . | Score % . | . | . |
---|---|---|---|---|
≥ 1 SD . | ≥ 1.5 SD . | ≥ 2 SD . | ≥ 2.5 SD . | |
SEQ | 32.7 | 2.8 | 0 | 0 |
WL1 | 17.8 | 2.8 | 1.9 | 0 |
CF1 | 15.0 | 6.5 | 4.7 | 0 |
VisSP | 14.0 | 1.9 | 1.9 | 0 |
WL2 | 28.0 | 14.0 | 2.8 | 0.9 |
WLRec | 23.4 | 8.4 | 0.9 | 0 |
CF2 | 29.0 | 11.2 | 0 | 0 |
CFRec | 16.8 | 0.9 | 0 | 0 |
SpLoc | 15.0 | 5.6 | 1.9 | 0 |
PhF | 18.7 | 5.6 | 4.7 | 0 |
SemF | 18.7 | 4.7 | 1.9 | 0 |
Preas | 18.7 | 3.7 | 2.8 | 0 |
Note. N = 107. Values represent cumulative percentages of high scores ranging from 1 SD to 2.5 SD above the mean. Scores for the n = 17 participants with missing values were imputed using the impute_EM function from the missMethods package in R. Abbreviations: SEQ = sequences; WL1 = word list immediate recall; CF1 = complex figure immediate recall; VisSP = visuospatial functioning (complex figure copy/clocks); WL2 = word list delayed recall; WLRec = word list yes/no recognition; CF2 = complex figure delayed recall; CFRec = complex figure recognition; SpLoc = spatial location; PhF = phonemic fluency; SemF = semantic fluency; Preas = reasoning (practical problem solving/conceptual shifting).
Discussion
In our sample of older adults referred to a memory clinic with SCD but who did not evince cognitive impairment, obtaining both abnormally high and low scores was common. Regarding low scores among SCD individuals, at a cut-off of 1 SD below the mean obtaining one or more (MVBR = 84.1%) or two or more (MVBR = 68.2%) low scores was common. As expected, base rates decreased as cut-offs became more stringent with one or more low scores occurring at rates of 55.1% at 1.5 SD, 39.3% at 2 SD, and 26.2% at 2.5 SD below the mean. Similarly, Brooks et al. (2013b) reported the MVBR of low scores on the WMS-IV, noting that on average cognitively intact older adults scored 1 SD below the norm on 1.3 WMS-IV subtests, whereas those diagnosed with MCI scoring below the same threshold on 3.1 WMS-IV subtests on average. These previous studies presented MVBR derived from normative samples of medically- and psychiatrically-healthy community dwelling older adults, whereas participants in our referred sample had SCD and had comorbid medical conditions. These findings are consistent with the notion that patients with memory complaints from secondary or tertiary care referral streams score below cut-off more often than community-based normative samples who are less likely to have comorbid health conditions. Thus, there is a need to consider the source-specific base rates of obtaining a low score on a cognitive test as a supplemental diagnostic tool when determining the presence or absence of objective cognitive impairment.
Interestingly, within our sample of older adults with SCD, the majority of participants obtained at least one score above 1 SD from the mean (MVBR = 80.4%). The substantial number of participants exceeding 1 SD on at least one subtest may be due in part to the highly educated nature of this sample, with 29.9% having ≥16 years of education. Indeed, we also observed high scores occurring more frequently in highly educated individuals. These findings parallel recent work showing a high base rate of above average scores on neuropsychological tests in those with higher levels of education and/or intellectual functioning. These studies, coupled with our own findings, bolster the notion that when assessing neuropsychological test performance in those with higher levels intellectual functioning, the absence of scores falling above the normative mean could be indicative of cognitive deficits. As our sample size was modest, future research is needed to accurately estimate the impact of high education at these more extreme cut-offs. In addition, there may be factors that affect the likelihood of geriatric patients at different levels of education to report cognitive concerns and/or seek out medical care, which would affect the distribution of levels of education amongst clinic samples as compared with normative samples that used more deliberate sampling rules.
In addition, there were observed differences across genders in the MVBR of obtaining low and high cognitive test scores. The MVBR of scoring below a cut-off was much higher in men versus women: The converse was also observed, with women scoring above the normative mean at a higher frequency. The cause for this discrepancy is not clear: Studies of MVBR on neuropsychological tests do not typically demonstrate large sex/gender differences (Schretlen et al., 2008; Rivera et al., 2019): Within the general population, there are sex differences on some neuropsychological tasks, particularly verbal memory, but whether these differences persist or worsen in post-menopausal women is also unclear, particularly with respect to the effects of hormonal replacement therapy (Nebel et al., 2018). Some longitudinal data in non-clinical older adults suggest that females show a consistent advantage on verbal and processing speed tasks, with males showing a steeper decline in global cognition over time (McCarrey, An, Kitner-Triolo, Ferrucci, & Resnick, 2016). Experimental studies also suggest females outperform males on episodic memory tasks for words, faces, odors, and colors, whereas males have a memory advantage for spatial stimuli, but these effects are small (Asperholm, Högman, Rafi, & Herlitz, 2019). Given the multitude of biological risk factors, social determinants of health, and other protective factors such as educational attainment that may differentially influence cognitive functioning across sex/gender, further research is needed to confirm the reliability of our findings.
In considering the utility of these results, we are also mindful of our selection of participants and the subsequent impact on MVBRs in the present study. Specifically, we defined participants as cognitively intact based on previously used cut-off scores on the MMSE and WMS-R Logical Memory II. Although a valid approach, we still cannot state with full certainty that all participants were cognitively healthy and would remain so over time. Given that our sample was derived from a group of patients presenting to tertiary clinics, the likelihood of objective cognitive impairment is higher than in the general population, and participants may have passed the cut-off criteria but showed low cognitive scores in other domains. Indeed, a minority of participants produced numerous scores below cut-off, which would not be expected in a sample of “cognitively healthy” older adults (see also Gavett, 2015). In sum, these MVBRs are influenced by our criteria for intact cognitive functioning, and subsequent studies using different criteria may produce slightly different results.
There are other caveats to the calculated MVBR from the present investigation. The base rate of low scores on Word List Delayed Recall was much smaller than on other measures, presumably because participants were selected partly based on their performance on a verbal memory delayed recall test (i.e., WMS-R Logical Memory II). Thus, our inclusion/exclusion criteria likely suppressed the base rate of low scores on Word List delayed recall in our sample to some extent. Moreover, our MVBR were calculated as cumulative percentages, and reflect the likelihood of this sample producing n or more scores above/below a given threshold: Consequently, these data cannot determine the likelihood that a participant would produce exactly n scores above/below threshold (Gavett, 2015).
To our knowledge, this is the first study investigating base rates in a clinically referred population with SCD but no global cognitive impairment. Acknowledging the base rate of low scores in this sample likely reflects both normal variation in cognition and some level of cognitive deficit, the frequencies of low scores may nonetheless help to reduce misclassifications of MCI. For example, consider a 75-year-old woman with 14 years of education who scores 27/30 on the MMSE, but has significant cognitive concerns and is thus referred for neuropsychological assessment for possible MCI. Based on our data, the clinician should be wary of over-interpreting a single score falling 1.5 SD below the mean, as over 45% of women with a similar level of education show one or more scores falling below this threshold. Indeed, recent evidence using Uniform Data Set Neuropsychological Battery 3.0 from the National Alzheimer’s Coordinating Center also suggests that the base rate of producing low scores is common in older adults, with 58% of older adults who did not convert to dementia having two or more scores below the 16th percentile (Kiselica et al., 2020). On the other hand, given that this sample was derived from a tertiary clinic, one would expect the base rate of objective cognitive impairment to be higher in this sample, relative to a normative sample, even if a patient passes an initial cognitive screen. Careful consideration of the MVBR of impairment produced in this study, alongside the broader context of the sample characteristics, is important for differentiating patterns of cognitive performance that are indicative of cognitive decline, from typically occurring variability in test scores.
These data may also aid in our understanding of the association between objective and subjective cognitive functioning. That is, perhaps the complex relationships and mixed results surrounding the association between SCD and objective impairment (Rabin et al., 2017) could be partially attributed to a lack of consideration of base rates specific to this population. As low scores in normative samples tended to occur more frequently in our study (Brooks et al., 2007; Brooks et al., 2008), the association between SCD and objective impairment may need to factor in MVBR depending on the source of the population in question (i.e., community or clinically referred).
This study involved several limitations. First, the sample size was modest (n = 107), and included only patient files with complete data, which represented only a fraction of all patients seen in the memory clinic over the period sampled. This limited sample prevented us from stratifying base rates by intellectual ability, which is associated with higher base rates of obtaining low scores on cognitive tests in normative samples (Iverson & Brooks, 2011; Brooks et al., 2013a). Future studies with larger samples should be used to further refine the estimate of the MVBR of true impairment in clinically referred samples with SCD and determine the contribution of these demographic variables on MVBR. An additional limitation involved the lack of biomarkers or neuroimaging measures available to confirm diagnoses, and the lack of follow-up data available to confirm the stability of the participants’ cognitive status over time. Therefore, it is possible that some participants may have been on a neurodegenerative trajectory (i.e., preclinical AD) with their subjective complaints preceding later objective impairments. Furthermore, we were unable to report on the race/ethnicity characteristics of the sample, as these data were unavailable. Given that differences in normative scores can exist between different ethnocultural groups (Wang et al., 2021), and that the improper application and interpretation of normative scores can decrease the accuracy of diagnosing cognitive impairment in racialized communities (Rivera et al., 2019; Werry, Daniel, & Bergstrom, 2020), there is a pressing need for more research in this area. It is germane to note that there is virtually no normative data on cognitive test performance across the most common ethnicities within the Canadian context, which has a different demographic distribution to the U.S.
Despite these limitations, this study provides the first MVBR for an understudied population, individuals with subjective cognitive complaints considered cognitively normal based on intact performance on a cognitive screener and a measure of verbal memory. These results are meant to serve as a preliminary reference point when judging the likelihood of objective cognitive impairment. Recognizing the higher base rates in this clinic sample should help minimize false positives of cognitive impairment and may also help disentangle the complex relationship between objective and subjective cognitive abilities. Future research using larger samples and source-specific (i.e., clinic vs. community) base rates is needed to further refine these findings. Ideally, future studies with longitudinal data that include assessment of biomarkers and/or neuroimaging measures would help validate the non-neurodegenerative nature of the cognitively healthy SCD individuals and provide more definitive estimates of the MVBR of cognitive impairment. Nevertheless, knowledge of the MVBR of low scores in the general population as well as a memory clinic referral population can provide a more comprehensive context for clinician decision-making. Recognizing the higher base rates in clinically referred individuals with SCD versus rates in normative samples should act to minimize false positive diagnoses of cognitive impairment and may also help to tease apart the complex relationship between objective and subjective cognitive abilities.
Author contributions
K. R. and L. L. helped with study conceptualization, interpretation, and manuscript writing. A. C. and A. H. helped with data collection and manuscript writing. G.B. P. helped with data interpretation and manuscript writing.
Funding
None.
Conflicts of Interest
Guy Proulx and Larry Leach are co-authors of the KBNA, and receive royalties from Pearson Inc.