Abstract

The purpose of this study was to: (1) determine the diagnostic prediction of the global score of the Kaplan–Baycrest Neurocognitive Assessment and (2) determine which subtests discriminate individuals with mild dementia from individuals without dementia. A case–control study of 33 individuals, diagnosed with mild dementia, was matched with 33 individuals with no cognitive or memory complaints, on age, education, and gender. A global measure score correctly classified 94% of the total sample with a sensitivity of 0.91 and specificity of 0.97. A measure composed of three subtest scores, representing verbal episodic memory, visual episodic memory, and semantic fluency, correctly classified 98% of the sample with a sensitivity of 1.00 and specificity of 0.97. The results indicate that the Kaplan–Baycrest Neurocognitive Assessment is a valid tool for prediction of cognitive impairment associated with mild dementia.

Dementia is a mental condition characterized by a decline in cognitive status from a previous level that has an impact on daily function. The most commonly used criteria for diagnosing dementia is the DSM-IV criteria (American Psychiatric Association, 1994). The DSM-IV criteria require the loss of intellectual abilities, memory impairment, and impairment in at least one of the cognitive domains of abstract reasoning, aphasia, apraxia, agnosia, constructional ability, and personality change. The cognitive deficits interfere with daily functioning, and cannot be due to a clouded consciousness because of conditions such as delirium or intoxication. When determining whether a patient meets the criteria for dementia, the initial task for the clinician is to determine whether there is psychometric evidence for impairment, and to define the type(s) of impairment present. Because the diagnostic criteria require impairment in at least two cognitive domains, one of which should be memory, then a test battery should include sampling several cognitive domains including memory.

To determine whether there is evidence of dementia, a neuropsychologist must use tests that evaluate a range of cognitive functions, including memory. Ideally, these tests should be standardized, and should have norms that encompass the age of people who are most probable to develop dementia—those over 60 and at least up to 90 years old. Also, the tests should have equivalent scales to facilitate comparison of performance across different cognitive domains. Test batteries have been published to identify cognitive changes associated with dementia. For example, the Mattis Dementia Rating Scale (DRS) (Mattis, 1988) was originally developed to evaluate the severity of dementia but it has also been shown to have utility in diagnosing dementia. Stuss, Meiran, Guzman, Laflech, and Willmer (1996) reported that the DRS was 83% accurate at discriminating clients with dementia from a clinical sample with no cognitive impairments. The DRS provides a global score as well as scores in five different domains of function. Lucas and colleagues (1998) published age-corrected scores for nine age bands ranging from 56 to 105 years of age for the total score as well as for each of the five subtests. They provided tables for converting raw scores to scaled-scores with an M of 10 and SD of 3, which is equivalent to the scaled-scores used in the Wechsler Adult Intelligence Scale and Wechsler Memory Scale. The Mini-Mental Status Examination (MMSE) was developed as a rapid screen for cognitive status. Education-corrected normative data, expressed in T-scores (M = 50, SD = 10), are available in 14 age bands from 18 to 85 years of age (Folstein, Folstein, & Fanjiang, 2001). Stuss and colleagues (1996) reported finding an optimal diagnostic accuracy of 84% using the MMSE. Although widely used as a tool for screening for mental status, the MMSE yields only a single score, thereby limiting its usefulness for diagnosing dementia although it is beneficial for rating severity of dementia.

Both the DRS and MMSE share a significant shortcoming in the evaluation of dementia in that both lack sufficiently sensitive tests for episodic memory. Both the MMSE and DRS evaluate only episodic recall over a short delay interval of approximately 3–5 min. Moreover, the MMSE evaluates only verbal episodic recall and the DRS Memory scale consists of items that evaluate semantic memory and episodic recall. Assessment of episodic memory is important because tests of episodic recall consistently have been shown to be among the most predictive tests of cognitive decline and the development of dementia (Artero, Tierney, Touchon, & Ritchie, 2003; Saxton et al., 2004; Tierney, Yao, Kiss, & McDowell, 2005).

The Kaplan–Baycrest Neurocognitive Assessment (KBNA) was developed to fulfill the need for a comprehensive set of tests based on a common normative database to assist in the identification of cognitive disorders due to brain dysfunction (Leach, Kaplan, Rewilak, Richards, & Proulx, 2000). The KBNA was standardized on a sample of 700 adults. All normative sampling was conducted in the USA according to a plan developed to insure a representative sample relative to the 1999 U.S. Census according to gender, race/ethnicity, educational level, and geographic region. The standardization sample was divided into seven age bands of 20–29, 30–39, 40–49, 50–59, 60–69, 70–79, and 80–89 years old, with 100 participants in each age band. The KBNA consists of 25 subtests of different cognitive abilities. Of the 25 subtests, 12 subtests yielded normal distributions of raw scores that could be converted to standardized scores. (The remaining 13 subtests yielded significantly skewed or irregular distributions. These subtests facilitate a process-oriented approach to performance interpretation, but these will not be discussed further in this paper.) The raw scores of the 12 subtests were converted to normalized, age-corrected scaled-scores with an M of 10 and SD of 3. The following is a description of each of the 12 subtests.

The Sequences subtest assesses focused attention and working memory. Word Lists 1 subtest evaluates verbal learning and acquisition of verbal information. Word Lists 2-Recall subtest is a delayed recall of the Word Lists 1 words and evaluates retrieval of verbal information from episodic memory. Word Lists 2-Recognition is a delayed, forced-choice recognition of Word Lists 1 items that evaluates consolidation of verbal information in episodic memory. The combined scores of the Complex Figure 1-Copy and Clocks subtests are used as measures of visuospatial and constructive capacity. Complex Figure 1-Recall subtest assesses acquisition and encoding of visual information. Complex Figure 2-Recall subtest is a 20-min delayed recall of the figure from Complex Figure 1-Copy and assesses retrieval of visuospatial information from episodic memory. Complex Figure 2-Recognition assesses retention of visuospatial information in episodic memory. The Verbal Fluency-Phonemic subtest is a measure of executive control of expressive language function. The Verbal Fluency-Semantic subtest is a measure of word retrieval from semantic, lexical memory. The Spatial Location subtest evaluates primary memory (short-term memory) for spatial location. The Practical Problem Solving/Conceptual Shifting subtest consists of two separate components. The Practical Problem Solving component evaluates the examinee's awareness of the need and knowledge of appropriate actions to take in hypothetical emergencies. The Conceptual Shifting component evaluates formation as well as cognitive flexibility.

The sums of the age-corrected scaled-scores from the above 12 subtests are combined into a set of seven indices, covering the cognitive domains of attention/working memory, episodic memory, language, judgment, reasoning, visuospatial, and constructional skills. For each of the seven indices, the age-corrected scaled-scores of the constituent subtests are summed and the sum is converted to a T-score with an M of 50 and SD of 10. The seven indices and their constituent subtests are as follows: Attention/Concentration Index, consists of the Sequences and Spatial Memory subtests; Memory-Immediate Recall Index, consists of the Word Lists 1 and Complex Figure 1-Recall subtests; Memory-Delayed Recall Index, consists of the Word Lists 2-Recall and the Complex Figure 2-Recall subtests; Memory-Delayed Recognition Index, consists of the Word Lists 2-Recognition and the Complex Figure 2-Recognition subtests; Spatial Processing Index, consists of the Complex Figure 1-Copy/Clocks subtests; Verbal Fluency Index, consists of the Verbal Fluency-Phonemic and Semantic subtests; Reasoning/Conceptual Shifting Index, consists of the Practical Problem Solving and Conceptual Shifting subtests. In addition to the seven primary indices, a composite index, Total Index, is obtained. The Total Index is based on the sum of all seven primary indices' T-scores.

The validation study described in the KBNA manual (Leach et al., 2000) included a mixed clinical sample of individuals suffering from dementia, head injury, or stroke. The between-group comparisons revealed that the clinical group performed significantly worse than a matched group of controls on all subtests except for one, thereby providing convergent evidence for the validity of the KBNA as a test of cognitive impairment. That validation study did not specifically address the ability of the KBNA to differentiate individuals suffering from dementia from individuals without dementia, however.

This paper describes the ability of the KBNA to discriminate neurologically normal individuals from those suffering from mild dementia. The study utilized a case–control design using two groups; a group of individuals that had been diagnosed as having dementia and a group of neurologically normal volunteers matched to the group with dementia with respect to age, education, and gender. The diagnosis of dementia was performed according to standard clinical guidelines and preceded evaluation by the KBNA, thereby eliminating any influence of the KBNA on the diagnosis. There were two goals established for this study: The first goal of the study was to establish the sensitivity, specificity, and diagnostic prediction of the KBNA global scores. The second goal was to determine which KBNA subtests provide the best diagnostic prediction of dementia.

Materials and Methods

Participants

Group Dementia

Participants in this group were patients recruited from an outpatient memory clinic at Baycrest Centre for Geriatric Care. The patients underwent a series of investigations as part of a standard workup for dementia that included a history, physical, neurological examination (performed by board-certified neurologists specializing in Behavioral Neurology), laboratory tests, neuroimaging examinations, and neuropsychological evaluation. Some patients may have undergone additional evaluations by a nurse practitioner, psychiatrist, occupational therapist, or speech-language pathologist. On the basis of the outcome of the various evaluations, a consensus diagnosis was reached by the neurologist and neuropsychologist. Diagnosis of dementia was based on the DSM-IV criteria (American Psychiatric Association, 1994). In addition to the diagnosis of dementia, a provisional diagnosis of the type of dementia was made and this diagnosis was communicated to all patients by the attending neurologist. Participants were consecutively enrolled in the study after the clinical diagnosis was made. Patients were contacted by a research assistant and asked to participate in this study. Only patients capable of consenting to the study were considered, and informed consent was obtained prior to participation in this study. Patients were excluded from the study if they had any of the following conditions: Color-blindness, uncorrected hearing loss, uncorrected visual impairment, active treatment for alcohol or drug abuse, or an upper extremity disability that would affect manipulation of a pencil or objects. The KBNA was administered after a diagnosis was made and therefore did not influence the diagnosis. A total of 33 patients met the inclusion criteria for Group Dementia. The composition by diagnosis and number respectively was as follows: Alzheimer dementia, n = 10; combined Alzheimer disease (AD) and vascular dementia, n = 5; Lewy-body disease, n = 7; fronto-temporal lobe dementia, n = 3; primary progressive aphasia, n = 2; vascular dementia, n = 4; progressive supranuclear palsy, n = 1; alcoholic dementia, n = 1.

Group Control

This group consisted of 33 participants; 18 control volunteers were from the KBNA standardization subject pool and 15 were volunteers from the metropolitan Toronto area. All control participants were given a health questionnaire prior to admission to the study. Exclusion criteria were color-blindness, uncorrected hearing loss, uncorrected visual impairment, active treatment for alcohol or drug abuse, current assessment by a physician or other health professional for memory or cognitive problems, an upper extremity disability that would affect manipulation of a pencil or objects, history of unconsciousness lasting longer than 5 min, a head injury that resulted in hospitalization for more than 24 hr, a history of medical or psychiatric condition that could potentially affect cognitive functioning, including, but not limited to, stroke, electroconvulsive therapy, epilepsy, brain surgery, encephalitis, meningitis, multiple sclerosis, AD, Parkinson's disease, Huntington's disease, schizophrenia, depression, and bipolar affective disorder.

Control participants were matched to Dementia clients for age, education, and gender.

Procedures

KBNA Administration and Scoring

The KBNA was administered and scored according to the standardized instructions in the test manual (Leach et al., 2000).

Sequences

The examinee is required to generate sets or sequences of stimuli based on specified rules. The stimuli are based on well-learned sequences, such as the alphabet, letters, or numbers. There are five individual sequences with a total raw score of 57.

Word Lists 1

A list of 12 words is read aloud to the examinee and then she/he is requested to recall as many of the words as possible. This is repeated four times and the scored measure is the total words recalled over the four trials. The maximum total raw score is 48.

Complex Figure 1-Copy/Clocks

For the Complex Figure 1-Copy part, the examinee is presented a sheet with a novel complex figure and is required to make a direct copy of the figure. There is no time limit and the maximum raw score is 20. In the Clocks part, the examinee is required to draw clocks under three different conditions. Under the Free-drawn clock condition (13 points possible) she/he must reproduce the contour of a clock face, all digits and hands placed at 11:10. Under the Pre-drawn clock condition (11 points possible), the examinee is provided with the clock contour but she/he must produce all digits and hands at 8:20. Under the Copy clock condition (13 points possible), the examinee is provided with a diagram of a clock with the time 11:10 represented and she/he must make a copy of the clock. The maximum total raw score for the Clocks is 37. The score of the Copy Figure 1-Copy and the Clocks conditions are combined to form a single subtest score with a maximum total score of 57.

Complex Figure 1-Recall

Immediately following the completion of Complex Figure 1-Copy, the examinee's copy is removed and then she/he is asked to reproduce the figure from memory. The maximum raw score is 20.

Word Lists 2-Recall

Following a delay of at least 20 min after completion of Word Lists 1, the examinee is asked to recall the words, first under both free and the under cued recall conditions. The total number of words recalled under both conditions is used as the performance measure for Word Lists 2 with a maximum total raw score of 24.

Word Lists 2-Recognition

The examinee is required to indicate, by a yes–no response, which words from a total list of 36 words were presented during the Word Lists 1 presentations. The 36 words consist of the original 12 Word Lists 1 words plus 12 new words that are semantically related and 12 new words not semantically related to the Word Lists 1 words. The maximum total raw score is 36.

Complex Figure 2-Recall

After at least a 20 min delay, the examinee is asked to reproduce, from memory, the figure drawn in the Complex Figure 1-Copy subtest. The maximum raw score is 20.

Complex Figure 2-Recognition

After completion of the Complex Figure 2-Reproduction, the examinee is presented with a sheet that contains five details present in the complex figure. Each detail is presented with a set of three foils; a set of four choices is presented on a separate row with only one correct choice in each row. For each row, the examinee is asked to pick out the one detail present in the original figure. After the examinee has picked out the five details, she/he is asked to place the details in a representation of the figure contour that is present on the same sheet. One point is awarded for correct choice and one point for correct location which yields a maximum raw score of 10.

Verbal Fluency-Phonemic

The examinee is required to generate words beginning with the letter C within a 1 min time limit. The examinee is instructed not to use proper nouns and not to repeat words, even with different endings (i.e., can but not cans). The score is the total number of words correctly generated in 1 min.

Verbal Fluency-Semantic

This subtest consists of two parts. The first part requires the examinee to generate the names of animals within a 1 min time limit. The second part requires the examinee to generate words first names within a 1 min time limit. The score is the total number of words generated in both animal and first-name conditions.

Spatial Location

The subject is asked to view a card that contains a box containing a number of black dots for 10 s. At the end of the 10 s viewing period, the card is removed and the examinee is required to place tokens on a matrix in the same locations as the black dots appeared on the card. There are 10 cards in total and the number of dots varies from 3 to 7 on each card. The maximum total raw score is 48.

Practical Problem Solving/Conceptual Shifting

This subtest consists of two distinct components. In the Practical Problem Solving component, the examinee is presented with a set of five hypothetical situations (e.g., “What would you do if you smelled smoke in your house?”) and asked to give two distinct and effective responses to resolve the situation. There are two points possible for each situation for a maximum total score of 10. In the Conceptual Shifting component, the examinee is presented a card with four designs and asked to indicate which three of the four designs are similar and to explain the similarity. Following his/her first choice, the examinee is then asked to identify which three of the four designs are similar in a different manner. The examiner must indicate the correct designs and give the correct reason for credit. There are two points possible for each card and there are 10 cards in all for a total of 20 points. The measure is the sum of the Practical Problem Solving and Conceptual Shifting components that yields a maximum total of 30 points.

For each subtest, the raw score is converted to an age-corrected scaled-score using tables provided in the KBNA manual. To calculate the seven individual index scores, the subtest scaled-scores comprising each index are summed and these sums are converted to T-scores using tables provided in the KBNA manual (Leach et al., 2000). The Total Index is calculated based on the sum of the T-scores of the seven indices, and this Index Sum is converted to a T-score using a table in the KBNA manual.

Time to complete the KBNA ranged from 90 to 120 min.

Mattis Dementia Rating Scale

In addition to the KBNA, participants in the Dementia group were administered the Mattis DRS (Mattis, 1988). The DRS provided an independent measure of severity of cognitive impairment (Meiran, Stuss, Guzman, Lafleche, & Willmer, 1996; Stuss et al., 1996). The DRS was administered and scored according to the standard instructions described in the administration manual (Mattis, 1988). Both the KBNA and DRS were administered in a single session lasting between 120 and 150 min.

Global Deterioration Scale

The level of functional impact of cognitive decline was rated according to the Global Deterioration Scale (GDS). The GDS is a 7-point rating scale, ranging from 1 (no cognitive decline) to 7 (very severe cognitive decline). The information used to make the rating was based on the clinical notes from the clinical neuropsychological evaluation, including interviews with the client and a family member or friend who knew the client well. Two independent raters performed the GDS rating and the final GDS score was taken as the average of the two ratings. The GDS has been shown to correlate significantly with psychometric measures of cognitive functioning (Reisberg, Ferris, de Leon, & Crook, 1982) and has excellent inter-rater reliability (Foster, Sclan, Welkowitz, Boksay, & Seeland, 1988; Gottlieb, Gur, & Gur, 1988).

Data Analysis

Data for the 12 subtests and 8 composite indices of the KBNA were analyzed. The data for the 12 subtests were age-corrected scaled-scores (M = 10, SD = 3) and the data for the eight indices were age-corrected T-scores (M = 50, SD = 10). A one-way ANOVA for the between-group difference was conducted on the Total Index score. Between-group differences on the remaining KBNA measures were evaluated by means of two separate MANOVAs. One MANOVA evaluated the between-group differences on the following seven KBNA indices: Attention/Concentration, Memory-Immediate Recall, Memory-Delayed Recall, Memory-Delayed Recognition, Spatial Processing, Verbal Fluency, and Reasoning/Conceptual Shifting. The second MANOVA assessed the between-group differences on the following 12 subtests: Sequences, Word Lists 1, Word Lists 2-Recall, Word Lists 2-Recognition, Complex Figure 1-Recall, Complex Figure 2-Recall, Complex Figure 2-Recognition, Complex Figure 1-Copy/Clocks, Spatial Location, Verbal Fluency-Phonemic, Verbal Fluency-Semantic, Practical Problem Solving/Conceptual Shifting. Planned comparisons were performed to determine the between-group effects on the 7 KBNA indices and 12 subtest scores using the Bonferonni correction for multiple comparisons and setting the experiment-wise α at 0.05.

In addition to the standard index and subtest scores, the following variables were calculated and analyzed: The total raw scores of the 12 subtests (Raw Score) and the sum of the T-scores of the seven indices (Index Sum) used to obtain the Total Index score. The clinical discriminability of the KBNA was determined by individual discriminant function analyses (DFAs) of the Raw Score, Total Index, the Index Sum, and a step-wise DFA of the 12 subtest scores. Individual receiver-operating curves (ROC) were generated for the Raw Score, Total Index, and Index Sum measures using SPSS Version 13. Based on the ROCs, optimal cut-off scores were obtained at the point of maximum correct classification. For each cut-off, the sensitivity, specificity, positive predictive ability (PPA), negative predictive ability (NPA), likelihood ratio of a positive response (LRPR), likelihood ratio of a negative response (LRNR), and overall classification accuracy were calculated.

Results

The mean age of Group Dementia was 70.8 (SD = 8.5, range 50–86) and the mean age of Group Control was 71.8 (SD = 8.5, range 57–87). The mean education of Group Dementia was 12.7 (SD = 3.1, range 7–17) and the mean education of Group Control was 13.1 (SD = 2.6, range 9–16). The gender composition of Group Dementia was 12F;21M and that of Group Control was 13F;20M. The two groups did not differ with respect to age, U = 534.00, p = .89, education, U = 233.50, p = .75, or gender composition, χ2 (1, N = 66) = 0.06, p = .80. The mean group Dementia GDS rating was 3.3 (SD = .91); this level represents mild cognitive decline. The mean group Dementia DRS score was 115.3 (SD = 3.3).

The group by KBNA Index analysis is presented in Fig. 1. The mean T-score is presented with the 95% confidence intervals plotted. The MANOVA revealed a significant multivariate group effect, Wilks' Λ = 0.219, F(7, 58) = 25.37, p < .001, η2 = 0.781. All between-group comparisons for the KBNA Indices were significant (p < .05) with the Dementia group obtaining lower scores than the Control group. Relative to the Control group, the Dementia group also obtained a significantly lower Total Index score, F(1, 64) = 157.30, p < .001. The between-group univariate statistics and effect sizes (Cohen's d) for the KBNA Indices are presented in Table 1. All effect sizes were large, ranging from 0.97 (Spatial Processing) to 3.05 (Total Index).

Table 1.

KBNA Index scores—between-group univariate statistics

Index Statistics
 
 F(1,64) p Cohen's d [95% CI] 
Attention/Concentration 32.29 <.001 1.38 [0.84, 1.92] 
Memory-Immediate Recall 150.02 <.001 2.99 [2.28, 3.67] 
Memory-Delayed Recall 146.00 <.001 2.94 [2.24, 3.64] 
Memory-Delayed Recognition 140.47 <.001 2.88 [2.19, 3.57] 
Spatial Processing 15.73 <.001 0.97 [0.46, 1.48] 
Verbal Fluency 63.33 <.001 1.94 [1.35, 2.52] 
Reasoning/Conceptual Shifting 33.27 <.001 1.40 [0.86, 1.94] 
Total Index 157.30 <.001 3.05 [2.34, 3.76] 
Index Statistics
 
 F(1,64) p Cohen's d [95% CI] 
Attention/Concentration 32.29 <.001 1.38 [0.84, 1.92] 
Memory-Immediate Recall 150.02 <.001 2.99 [2.28, 3.67] 
Memory-Delayed Recall 146.00 <.001 2.94 [2.24, 3.64] 
Memory-Delayed Recognition 140.47 <.001 2.88 [2.19, 3.57] 
Spatial Processing 15.73 <.001 0.97 [0.46, 1.48] 
Verbal Fluency 63.33 <.001 1.94 [1.35, 2.52] 
Reasoning/Conceptual Shifting 33.27 <.001 1.40 [0.86, 1.94] 
Total Index 157.30 <.001 3.05 [2.34, 3.76] 
Fig. 1.

Mean KBNA index scores of Group Dementia and Control expressed as (M = 50, SD = 10). Error bars represent 95% confidence intervals. Abbreviations: AC = Attention/Concentration; MIR = Memory-Immediate Recall; MD = Memory-Delayed Recall; MDRec = Memory-Delayed Recognition; SP = Spatial Processing; VF = Verbal Fluency; RCS = Reasoning/Conceptual shifting; TI = Total Index.

Fig. 1.

Mean KBNA index scores of Group Dementia and Control expressed as (M = 50, SD = 10). Error bars represent 95% confidence intervals. Abbreviations: AC = Attention/Concentration; MIR = Memory-Immediate Recall; MD = Memory-Delayed Recall; MDRec = Memory-Delayed Recognition; SP = Spatial Processing; VF = Verbal Fluency; RCS = Reasoning/Conceptual shifting; TI = Total Index.

The group by KBNA subtest analysis is presented in Fig. 2. The mean age-corrected scaled-score and the 95% confidence intervals are plotted for each subtest. The MANOVA revealed a significant multivariate group effect, Wilks' Λ = 0.196, F(12, 53) = 18.15, p < .001, η2 = 0.804. All 12 between-group subtest comparisons were significant (p < .05) with the Dementia group obtaining lower scores than the Control group. The univariate statistics and effect sizes (Cohen's d) are presented in Table 2. All effect sizes for subtest scores were large, ranging from 0.94 (Complex Figure 1-Copy/Clock Drawing) to 2.73 (Word List 1).

Table 2.

KBNA subtest scores—between-group univariate statistics

Subtest Statistics
 
 F(1,64) p Cohen's d [95% CI] 
Sequences 16.51 <.001 0.99 [0.48, 1.50] 
Spatial location 30.27 <.001 1.34 [0.80, 1.87] 
Word List 1 125.44 <.001 2.73 [2.06, 3.40] 
Word List 2-Recall 103.50 <.001 2.48 [1.83, 3.12] 
Word List 2-Recognition 108.49 <.001 2.53 [1.89, 3.18] 
Complex Figure 1-Recall 102.31 <.001 2.46 [1.82, 3.10] 
Complex Figure 2-Recall 89.72 <.001 2.30 [1.68, 2.93] 
Complex Figure 2-Recognition 71.93 <.001 2.06 [1.47, 2.66] 
Complex Figure 1-Copy/Clocks 14.98 <.001 0.94 [0.43, 1.45] 
Verbal Fluency-Phonemic 31.30 <.001 1.36 [0.82, 1.90] 
Verbal Fluency-Semantic 73.61 <.001 2.09 [1.49, 2.69] 
Practical Problem Solving/Conceptual Shifting 33.45 <.001 1.41 [0.87, 1.95] 
Subtest Statistics
 
 F(1,64) p Cohen's d [95% CI] 
Sequences 16.51 <.001 0.99 [0.48, 1.50] 
Spatial location 30.27 <.001 1.34 [0.80, 1.87] 
Word List 1 125.44 <.001 2.73 [2.06, 3.40] 
Word List 2-Recall 103.50 <.001 2.48 [1.83, 3.12] 
Word List 2-Recognition 108.49 <.001 2.53 [1.89, 3.18] 
Complex Figure 1-Recall 102.31 <.001 2.46 [1.82, 3.10] 
Complex Figure 2-Recall 89.72 <.001 2.30 [1.68, 2.93] 
Complex Figure 2-Recognition 71.93 <.001 2.06 [1.47, 2.66] 
Complex Figure 1-Copy/Clocks 14.98 <.001 0.94 [0.43, 1.45] 
Verbal Fluency-Phonemic 31.30 <.001 1.36 [0.82, 1.90] 
Verbal Fluency-Semantic 73.61 <.001 2.09 [1.49, 2.69] 
Practical Problem Solving/Conceptual Shifting 33.45 <.001 1.41 [0.87, 1.95] 
Fig. 2.

Mean KBNA subtest scores of Group Dementia and Control expressed as age-corrected scaled-scores (M = 10, SD = 3). Error bars represent 95% confidence intervals. Abbreviations: Seq = Sequences; SpLoc = Spatial location; WL1 = Word List 1; CF1 = Complex Figure 1-Recall; WL2 = Word List 2-Recall; CF2 = Complex Figure 2-Recall; WLRec = Word List Recognition; CFRec = Complex Figure 2-Recognition; VisSP = Complex Figure 1-Copy/Clocks; PhF = Verbal Fluency-Phonemic; SemF = Verbal Fluency-Semantic; Preas = Practical Problem Solving/Conceptual Shifting.

Fig. 2.

Mean KBNA subtest scores of Group Dementia and Control expressed as age-corrected scaled-scores (M = 10, SD = 3). Error bars represent 95% confidence intervals. Abbreviations: Seq = Sequences; SpLoc = Spatial location; WL1 = Word List 1; CF1 = Complex Figure 1-Recall; WL2 = Word List 2-Recall; CF2 = Complex Figure 2-Recall; WLRec = Word List Recognition; CFRec = Complex Figure 2-Recognition; VisSP = Complex Figure 1-Copy/Clocks; PhF = Verbal Fluency-Phonemic; SemF = Verbal Fluency-Semantic; Preas = Practical Problem Solving/Conceptual Shifting.

In addition to the Total Index score, the mean Raw Score (the sum of the 12 individual subtest raw scores) and the mean Index Sum (the sum of the T-scores of the seven indices comprising the Total Index) were analyzed for between-group differences by individual ANOVAs. A significant difference, F(1,64) = 126.84, p < .001; d = 2.74, 95% CI (2.07, 2.74), obtained between the mean Raw Score of the Dementia group, M = 215, SD = 50, 95% CI (197, 233), and the Control group, M = 334, SD = 50, 95% CI (322, 346). Similarly, the difference between the mean Index Sum of the Dementia group, M = 240, SD = 48, 95% CI (223, 257), and that of the Control group, M = 378, SD = 42, 95% CI (363, 393), was significant, F(1,64) = 155.02, p < .001, d = 3.03, 95% CI (2.32, 3.74). All three global measures, considered individually, were good predictors of the functional status of the Dementia group. The GDS was correlated with the Raw Score, Total Index, and Index Sum, r(31) = .50, .44 and .46, respectively, all p < .05. In addition, the DRS was correlated, p < .05, with the Raw Score, r(31) = .83, the Total Index, r(31) = .70, and the Index Sum, r(31) = .72.

The finding that three KBNA global measures of functioning, Total Index, Raw Score and Index Sum, yield large effect sizes and correlate with functional decline, suggests that the measures could be useful clinical indicators of dementia. The DFA for the three global measures confirmed that they were capable of discriminating the Dementia and Control group members. The Raw Score correctly classified 92.4% of the participants, χ2(1, N = 66) = 69.35, and both the Index Total and Index Sum correctly classified 93.9% of the participants, χ2 (1, N = 66) = 78.77 and 77.73, respectively. A step-wise DFA was performed using the age-corrected scaled-scores of the 12 KBNA subtests. Three subtests, Complex Figure 1-Recall, Word List 2-Recognition and Verbal Fluency-Semantic, correctly classified 97% of the cases, χ2 (1, N = 66) = 93.01.

To further evaluate the predictive clinical utility of the KBNA for dementia, clinical cut-offs were established for the Index Sum. The Index Sum was chosen over the Total Index and Raw Score for the following reasons: (i) in contrast to the Raw Score, the Index Sum is age corrected by virtue of being a composite of age-corrected scaled-scores and (ii) although equivalent to the Total Index in discriminating the Clinical and Dementia groups in the DFA, the Index Sum was a better predictor of functional status than the Total Index. The sensitivity and specificity of the entire range of scores was calculated using the ROC analysis function of SPSS–13. Using methods described by Sackett, Haynes, Guyatt, and Tugwell (1991) and Altman (2000), the following parameters of class prediction, along with the corresponding 95% confidence intervals, were calculated; sensitivity, specificity, PPA, NPA, LRPR, LRNR, and correct classification. Table 3 provides the parameters of prediction for the Index Sum, in five-point intervals, encompassing the lowest score obtained by a group Control participant (295) to the highest score obtained by a group Dementia participant (327). The Index Sum of 305 yielded an optimal cut-off between the two groups; 94% correct classification, sensitivity = 0.91, specificity = 0.97, χ2 (1, N = 66) = 51.2, although all scores in the range of 295–330 yielded significant classification rates, χ2 (1, N = 66) = 48.6–55.0.

Table 3.

Diagnostic characteristics of KBNA Index Sum score

Index Sum score Sensitivity Specificity PPA NPA LRPR LRNR Classification 
295 0.85 [0.69, 0.93] 1.00 [0.90, 1.00] 1.00 [0.88, 1.00] 0.87 [0.73, 0.94] a 0.15a 0.92 [0.86, 0.99] 
300 0.88 [0.73, 0.95] 0.97 [0.85, 0.99] 0.97 [0.83, 0.99] 0.89 [0.75, 0.96] 29.00 [4.2, 200.6] 0.12 [0.05, 0.31] 0.92 [0.86, 0.99] 
305 0.91 [0.76, 0.97] 0.97 [0.85, 0.99] 0.97 [0.84, 0.99] 0.91 [0.78, 0.97] 30.00 [4.3, 207.3] 0.09 [0.03, 0.28] 0.94 [0.88, 1.00] 
310 0.91 [0.76, 0.97] 0.97 [0.85, 0.99] 0.97 [0.84, 0.99] 0.91 [0.78, 0.97] 30.00 [4.3, 207.3] 0.09 [0.03, 0.28] 0.94 [0.88, 1.00] 
315 0.91 [0.76, 0.97] 0.94 [0.80, 0.98] 0.94 [0.80, 0.90] 0.91 [0.78, 0.97] 15.00 [3.9, 57.7] 0.10 [0.03, 0.29] 0.92 [0.86, 0.99] 
320 0.91 [0.76, 0.97] 0.94 [0.80, 0.98] 0.94 [0.80, 0.90] 0.91 [0.78, 0.97] 15.00 [3.9, 57.7] 0.10 [0.03, 0.29] 0.92 [0.86, 0.99] 
325 0.91 [0.76, 0.97] 0.91 [0.76, 0.97] 0.91 [0.76, 0.97] 0.91 [0.76, 0.97] 10.00 [3.4, 29.6] 0.10 [0.03, 0.29] 0.91 [0.84, 0.98] 
330 1.00 [0.89, 1.00] 0.91 [0.76, 0.97] 0.92 [0.76, 0.97] 1.00 [0.89, 1.00] 11.00a 0.00a 0.95 [0.90, 1.00] 
Index Sum score Sensitivity Specificity PPA NPA LRPR LRNR Classification 
295 0.85 [0.69, 0.93] 1.00 [0.90, 1.00] 1.00 [0.88, 1.00] 0.87 [0.73, 0.94] a 0.15a 0.92 [0.86, 0.99] 
300 0.88 [0.73, 0.95] 0.97 [0.85, 0.99] 0.97 [0.83, 0.99] 0.89 [0.75, 0.96] 29.00 [4.2, 200.6] 0.12 [0.05, 0.31] 0.92 [0.86, 0.99] 
305 0.91 [0.76, 0.97] 0.97 [0.85, 0.99] 0.97 [0.84, 0.99] 0.91 [0.78, 0.97] 30.00 [4.3, 207.3] 0.09 [0.03, 0.28] 0.94 [0.88, 1.00] 
310 0.91 [0.76, 0.97] 0.97 [0.85, 0.99] 0.97 [0.84, 0.99] 0.91 [0.78, 0.97] 30.00 [4.3, 207.3] 0.09 [0.03, 0.28] 0.94 [0.88, 1.00] 
315 0.91 [0.76, 0.97] 0.94 [0.80, 0.98] 0.94 [0.80, 0.90] 0.91 [0.78, 0.97] 15.00 [3.9, 57.7] 0.10 [0.03, 0.29] 0.92 [0.86, 0.99] 
320 0.91 [0.76, 0.97] 0.94 [0.80, 0.98] 0.94 [0.80, 0.90] 0.91 [0.78, 0.97] 15.00 [3.9, 57.7] 0.10 [0.03, 0.29] 0.92 [0.86, 0.99] 
325 0.91 [0.76, 0.97] 0.91 [0.76, 0.97] 0.91 [0.76, 0.97] 0.91 [0.76, 0.97] 10.00 [3.4, 29.6] 0.10 [0.03, 0.29] 0.91 [0.84, 0.98] 
330 1.00 [0.89, 1.00] 0.91 [0.76, 0.97] 0.92 [0.76, 0.97] 1.00 [0.89, 1.00] 11.00a 0.00a 0.95 [0.90, 1.00] 

Note: Values in brackets represent 95% confidence intervals.

PPA = positive predictive ability; NPA = negative predictive ability; LRPR = likelihood ratio of a positive response; LRNR = likelihood ratio for a negative response.

aValue cannot be calculated because it involves division by 0.

Discriminant function analysis revealed that the Complex Figure 1-Recall, Word List 2-Recognition, and Verbal Fluency-Semantic subtests were the best predictors of class membership. In order to evaluate the predictive utility of these three subtests, the sum of the age-corrected scaled-scores (3-Discriminant) was calculated. There was little overlap in the distribution of the scores of the Control and Dementia groups. Group Dementia obtained a significantly lower mean 3-Discriminant, M = 14.2; SD = 4.7, than Group Control, M = 33.1; SD = 5.8, t(64) = 14.6, p < .001, d = 3.55, 95% CI (2.76, 4.32). Group Dementia 3-Discriminant scores were significantly correlated with the GDS rating, r(31) = .41, p < .05, and with the DRS scores, r(29) = .43, p < .05. Using the methods described earlier for the Index Sum, the sensitivity, specificity, PPA, NPA, LRPR, LRNR, and correct classification for the 3-Discriminant score were calculated. The parameters of classification for each 3-Discriminant score, in two-point intervals, between the lowest score (14) obtained by a group Control subject to the highest score (24) obtained by a Dementia group subject are given in Table 4. A score of 24 provided the optimal cut-off correctly classifying 98%, χ2 (1, N = 66) = 62.1, of the groups although all scores in the range of 14–24 yielded significant discrimination, χ2 (1, N = 66), range 21.1–62.1, p < .001.

Table 4.

Diagnostic characteristic of KBNA 3-Discriminant score

3- Discriminant Score Sensitivity Specificity PPA NPA LRPR LRNR Correct classification 
14 0.48 [0.32, 0.65] 1.00 [0.90, 1.00] 1.00 [0.81, 1.00] 0.66 [0.55, 0.78] a 0.52 [0.60, ∞] 0.74 [0.64, 0.85] 
16 0.58 [0.41, 0.73] 0.97 [0.85, 1.00] 0.95 [0.76, 0.99] 0.70 [0.55, 0.81] 19.00 [2.70, 133.84] 0.48 [0.29, 0.65] 0.77 [0.67, 0.87] 
18 0.67 [0.50, 0.80] 0.97 [0.85, 1.00] 0.96 [0.79, 0.99] 0.74 [0.60, 0.85] 22.00 [3.14, 153.87] 0.34 [0.21, 0.56] 0.82 [0.73, 0.91] 
20 0.82 [0.66, 0.91] 0.97 [0.85, 1.00] 0.96 [0.82, 0.99] 0.84 [0.70, 0.92] 27.00 [3.90, 187.30] 0.19 [0.09, 0.39] 0.89 [0.82, 0.97] 
22 0.97 [0.85, 1.00] 0.97, 0.85, 1.00] 0.97 [0.85, 1.00] 0.97 [0.85, 1.00] 32.00 [0.64, 220.70] 0.03 [0.005, 0.22] 0.97 [0.93, 1.00] 
24 1.00 [0.90, 1.00] 0.97 [0.85, 1.00] 0.97 [0.85, 1.00] 1.00 [0.89, 1.00] 33.00a 0.00a 0.98 [0.96, 1.00] 
3- Discriminant Score Sensitivity Specificity PPA NPA LRPR LRNR Correct classification 
14 0.48 [0.32, 0.65] 1.00 [0.90, 1.00] 1.00 [0.81, 1.00] 0.66 [0.55, 0.78] a 0.52 [0.60, ∞] 0.74 [0.64, 0.85] 
16 0.58 [0.41, 0.73] 0.97 [0.85, 1.00] 0.95 [0.76, 0.99] 0.70 [0.55, 0.81] 19.00 [2.70, 133.84] 0.48 [0.29, 0.65] 0.77 [0.67, 0.87] 
18 0.67 [0.50, 0.80] 0.97 [0.85, 1.00] 0.96 [0.79, 0.99] 0.74 [0.60, 0.85] 22.00 [3.14, 153.87] 0.34 [0.21, 0.56] 0.82 [0.73, 0.91] 
20 0.82 [0.66, 0.91] 0.97 [0.85, 1.00] 0.96 [0.82, 0.99] 0.84 [0.70, 0.92] 27.00 [3.90, 187.30] 0.19 [0.09, 0.39] 0.89 [0.82, 0.97] 
22 0.97 [0.85, 1.00] 0.97, 0.85, 1.00] 0.97 [0.85, 1.00] 0.97 [0.85, 1.00] 32.00 [0.64, 220.70] 0.03 [0.005, 0.22] 0.97 [0.93, 1.00] 
24 1.00 [0.90, 1.00] 0.97 [0.85, 1.00] 0.97 [0.85, 1.00] 1.00 [0.89, 1.00] 33.00a 0.00a 0.98 [0.96, 1.00] 

Note: Values in brackets represent 95% confidence intervals.

PPA = positive predictive ability; NPA = negative predictive ability; LRPR = likelihood ratio for a positive response; LRNR = likelihood ratio for a negative response.

aValue cannot be calculated because it involves division by 0.

In order to appreciate the clinical utility of the KBNA in identifying cases of dementia, it is also necessary to consider the effects of prevalence, or base rate, of dementia on the test's diagnostic ability. If sensitivity and specificity are held constant, then as prevalence of disease increases the PPA increases but the NPA decreases; conversely, as prevalence decreases, NPA increases and PPA decreases (Sackett et al., 1991). This effect of prevalence on test prediction can best be appreciated by plotting positive and negative abilities of a test against prevalence. Fig. 3 presents the positive and NPA of three KBNA Index Sums, 295, 305, and 330, as a function of prevalence. (Fig. 3 was generated by a modified version of the Excel spreadsheet baysgrph.xls by Robert Hamm available at http://www.fammed.ouhsc.edu/robhamm/cdmcalc.htm.). These three KBNA Index Scores were chosen because a score of 295 represents the “rule-in” score (i.e., the lowest score obtained by all Normal group subjects in the validation study), a score of 305 represents the optimal cut-off score, and a score of 330 represents the “rule-out” score (i.e., the highest score obtained by all Dementia group subjects). Fig. 3 demonstrates that, at low levels of prevalence, a score <295 identified all cases of dementia (i.e., rule-in dementia) whereas a score of 330 identified all normal controls (i.e., rule-out dementia). For an Index Sum cut-off score of 305, the PPA drops below 0.90 when the prevalence of dementia is <0.23, but at that 0.23 prevalence level the NPA would be 0.97. In contrast, the NPA of a cut-off of 305 would drop below 0.90 when the prevalence of dementia was above 0.54 but at that prevalence level the PPA would be 0.97.

Fig. 3.

Positive (left ordinate) and negative (right ordinate) predictive abilities of KBNA Index Sum score as a function of prevalence of dementia. Curves represent predictions for Index Sum of 295 (rule-in dementia), 305 (optimal level for classification), and 330 (rule-out dementia). Diagonal line represents classification if no test given.

Fig. 3.

Positive (left ordinate) and negative (right ordinate) predictive abilities of KBNA Index Sum score as a function of prevalence of dementia. Curves represent predictions for Index Sum of 295 (rule-in dementia), 305 (optimal level for classification), and 330 (rule-out dementia). Diagonal line represents classification if no test given.

Discussion

This study utilized age- and education-matched cohorts of individuals diagnosed as having dementia and volunteers with no cognitive complaints. Dementia diagnosis was determined by the results of neurological, functional, and neuropsychological evaluations that preceded and that were independent of the KBNA results and therefore the KBNA results had no influence on the diagnosis of dementia. The KBNA was found to be a clinically valid test for identifying individuals with mild dementia. The presence of mild dementia was found to have a significant effect on all KBNA global measures of cognitive functioning as well as on all 12 subtests of specific domains of cognitive functioning. Global measures of cognitive functioning were found to be sensitive and specific to the presence of dementia; one measure of global functioning, the Index Sum, correctly classified 94% of demented and non-demented clients. Moreover, it was found that a composite measure of performance on tests of immediate visual memory, delayed verbal recognition, and semantic fluency correctly classified 98% of the clients.

The finding that episodic recall and semantic fluency are the best predictors of dementia is consistent with published findings. Tierney and colleagues (2005) examined the ability of neuropsychological tests to predict incident AD after 5 and 10 years. Tierney and colleagues found that recall of word lists (Rey Auditory Verbal Learning-Short Delay) best predicted which participants would go on to develop AD 10 years after their initial assessment and a combination of scores on tests of short delayed verbal recall, animal fluency, and information predicted AD 5 years after initial assessment. Artero and colleagues (2003) gave a battery of tests to non-demented individuals with memory complaints and followed up with re-assessment after 2 years. Artero and colleagues found that tests of delayed auditory verbal recall, construction, and category fluency predicted development of AD. Saxton and colleagues (2004) performed annual neuropsychological assessments on community-dwelling individuals over an 8 year period. They found that performances on tests of delayed episodic recall, naming, and semantic fluency best predicted which individuals would develop AD.

The ability of the KBNA to identify an individual as demented approached that of a neuropsychological test battery composed of many tests using different normative sets. At the upper limits of the confidence intervals, the positive and negative prediction of the KBNA were equivalent to the outcome of the clinical assessment used to classify the individuals with dementia in this study. At the lower limits of confidence, the positive prediction of the KBNA was still good at 88% while the negative prediction was acceptable at 78% (Table 3). One major advantage of the KBNA is shorter administration time; the KBNA was completed within 90–120 min, but the administration of the battery used to classify individuals with dementia required 180–300 min to complete.

As with all diagnostic tests, however, if sensitivity and specificity are held constant, then positive and negative predictive abilities (PPA and NPA, respectively) vary according to the base rate, or prevalence, of the condition. Fig. 3 demonstrates that as prevalence increases, PPA increases and NPA decreases, but as prevalence decreases, PPA decreases and NPA increases. When interpreting diagnostic test results, it is crucial that the clinician be aware of the prevalence of the condition in question, and to appreciate that the prevalence or base rate can vary depending on characteristics of the person being tested as well as the setting. For example, the prevalence of dementia increases with age; therefore, the prevalence for a 65 year old is lower than that of an 85-year-old individual. Also, the principal reason for referral to a memory clinic is a complaint of memory failure; therefore, the base rate of dementia presenting in a memory disorder clinic will be greater than the base rate in the primary care physician's office.

A KBNA Index Sum cut-off score of 305 was found to give an optimal classification rate, but it should be emphasized that this score may not be the best choice for all uses. For example, if a neuropsychologist was asked to confirm a finding of dementia, then the rule-in score of 295 would be preferable because false positives are minimized. On the other hand, if the question involves excluding dementia, then the rule-out score of 330 would minimize the risk of missing a case (false negative) of dementia.

A global test score should not be the sole criteria for diagnosing, or ruling out, dementia or significant cognitive impairment. The most compelling reason not to use a single test score to diagnose dementia is that the various diagnostic criteria require impairment in at least two domains of cognitive functioning in order to diagnose dementia. In the case of the Index Sum or 3-Discriminant scores, it is feasible that an individual could obtain a score below cut-off but not obtain impaired scores on any of the individual subtest that comprised these composite scores. The composite scores should be used as a critical indicator that there is a high likelihood of dementia. If such a case is identified, then the neuropsychologist needs to determine whether there is impairment in two or more domains and there is functional decline as a consequence of the cognitive impairment before diagnosing dementia.

Another reason for not relying on a global test score, such as the Index Sum, for defining dementia is that it is possible to obtain an Index Sum score below cut-off but still be functionally capable and thereby not meeting one criterion for dementia. This situation could arise if the person's premorbid functioning was below average or the person could meet criteria for mild cognitive impairment (MCI). In the first case, comparison of the individual's KBNA scores to premorbid ability estimates, such as measures of crystallized intelligence, may identify this type of situation. In the latter case, MCI-Amnestic variety could be identified if the individual's scores on the memory tests were the only scores that were in the impaired range.

With respect to identifying clients with MCI, another pattern may emerge. For high functioning individuals, the Index Sum score can fall above cut-off but the 3-Discriminant score could fall below cut-off. In such a case, further comparison of the person's Total Index score and a measure of crystallized intelligence may reveal a statistically significant difference, thereby supporting a conclusion of a decline in cognitive function. One potentially useful crystallized intelligence measure is the Verbal IQ measure provided by the Wechsler Abbreviated Scale of Intelligence (WASI, The Psychological Corporation, 1999). The WASI Verbal IQ is particularly appropriate as it is developed using the same normative sample as the KBNA (Leach et al., 2000).

An illustrative example of the potential of the KBNA for identifying significant cognitive decline in high functioning individuals was found in one of the outliers in the group of individuals with dementia. This individual was a high functioning professional with a 2–3 year history of memory problems. He had retired from a successful professional practice only 6 months prior to the initial clinical neuropsychological workup. His wife reported that he had been forgetting recent events, had difficulty learning new information, and had a marked tendency to repeat the same questions or comments within the course of a day. At work, he had not made any significant errors, but he had support from his secretary-receptionist who kept his appointment calendar and client records, and from a bookkeeper that handled his accounts. His wife had become very concerned about his functioning and the risk of malpractice due to his memory lapses and had finally convinced him to retire. At the time of the first neuropsychological assessment he was independent in his self-care but his wife had assumed all household banking functions. The neuropsychological assessment found that his overall intellectual functioning was in the superior range. Relative deficits were noted on the tests of episodic recall (California Verbal Learning Test, Wechsler Memory Scale-Revised, Rey-Osterrieth Complex Figure Recall), semantic fluency (animal names), and naming (Boston Naming Test). His scores on tests of focused attention and psychomotor speed (Trail Making Test Form A and B), visuoconstructional function (Rey-Osterrieth Complex Figure Copy, WAIS-R Block Design), semantic memory (WAIS-R Vocabulary, Similarities) were all in the average range or higher and his DRS score was 135/144, which is above the published cut-offs (Mattis, 1988; Stuss et al., 1996). His KBNA Index Sum of 325 was above the optimal cut-off for identifying dementia, but his 3-Discriminant score of 20 was below cut-off. Examination of the 3-Discriminant subtest scores indicate that he obtained a low average score (25th percentile) on the Word List 2-Recognition and the Verbal Fluency-Semantic Fluency (16th percentile) and a borderline normal score (5th percentile) on the Complex Figure 1-Recall subtests. Although his Total Index score was rated as low average–average (31st percentile), it was significantly lower than his FSIQ (90th percentile) for his age. By taking advantage of the ability to compare normalized scores of tests, this individual could be identified early in the stage of dementia. Subsequent serial reassessments over a 3-year period revealed a progressive deterioration of his memory and loss of functional independence confirming a diagnosis of probable AD.

Limitations of the Present Study

One limitation regarding the use of the cut-off scores presented in this, or any, study arises from the dependence of the sensitivity and specificity of a test on the severity of the disease (i.e., dementia) in the defining clinical group. In this study, the individuals with dementia were mildly demented overall; therefore the test characteristics described herein applies to individuals with similar dementia severity. Sensitivity and specificity would be higher if a more severely affected group is used and conversely lower if a less-affected group is used.

The study examined the ability of the KBNA to identify individuals with dementia regardless of the etiology. It is expected that different etiologies of dementia would produce different patterns of test results, but our sample size did not allow us to make meaningful comparisons. Studies are underway that will examine the patterns of KBNA subtests scores that result from AD, Lewy-body disease, vascular dementia, MCI, and depression.

The groups used for comparison involved individuals with memory complaints and a diagnosis of dementia versus individuals without memory complaints and without known neurological disease. Had the study involved two groups taken from the memory clinic, one determined to be demented and one group determined not be non-demented, then the size of the effect between the two groups would have been smaller and the classification rate would have been lower. The use of samples drawn from individuals presenting at a memory clinic would have introduced a potential confound, and involved a different clinical question, however. There would have been a likelihood that individuals classified as non-demented could be classified as MCI or as having memory complaints due to normal aging or those associated with depression. Thus the questions would have been how well does the KBNA discriminate among the categories of individuals with dementia from those with MCI, normal aging, or depression. This study sought to determine how well the KBNA identifies individuals with dementia from those without any memory complaint; nevertheless, the ability of the KBNA to distinguish among these conditions needs to be explored.

The two groups used in this study were balanced with respect to education, but education could have an influence on an individual's score. The KBNA manual (p. 112) provides a table indicating a modest, but consistent, effect of education on T-scores; therefore, the user should take education into consideration before rigidly applying the cut-off scores. Notwithstanding the effect of level of education on interpretation of an individual's score relative to the cut-offs provided in this paper, the discriminatory power of the KBNA should be relatively unchanged.

Conflict of Interest

Both the author and Baycrest Geriatric Health Care System are in contractual agreement with PsychCorp to receive royalties from sales of the KBNA.

References

Altman
D. G.
Diagnostic tests
 , 
2000
2nd ed.
Oxford, UK
British Medical Journal Books
American Psychiatric Association
Diagnostic and statistical manual of mental disorders
 , 
1994
4th ed.
Washington, DC
Author
Artero
S.
Tierney
M. C.
Touchon
J.
Ritchie
K.
Prediction of transition from cognitive impairment to senile dementia: a prospective, longitudinal study
Acta Psychiatrica Scandinavica
 , 
2003
, vol. 
107
 (pg. 
390
-
393
)
Folstein
M. F.
Folstein
S. E.
Fanjiang
G.
MMSE Mini-mental state examination: Clinical guide
 , 
2001
Lutz, FL
Psychological Assessment Resources
Foster
J. R.
Sclan
S.
Welkowitz
J.
Boksay
I.
Seeland
I.
Psychiatric assessment in medical long-term care facilities: Reliability of commonly used rating scales
International Journal of Geriatric Psychiatry
 , 
1988
, vol. 
3
 (pg. 
229
-
233
)
Gottlieb
G. L.
Gur
R. E.
Gur
R. C.
Reliability of psychiatric scales in patients with dementia of the Alzheimer type
American Journal of Psychiatry
 , 
1988
, vol. 
45
 (pg. 
857
-
859
)
Leach
L.
Kaplan
E.
Rewilak
D.
Richards
B.
Proulx
G.
The Kaplan–Baycrest neurocognitive assessment (KBNA): Test manual
 , 
2000
San Antonio, TX
Harcourt Assessment
Lucas
J. A.
Ivnik
R. J.
Smith
G. E.
Bohac
D. L.
Tangalos
E. G.
Kokmen
E.
, et al.  . 
Normative data for the Mattis Dementia Rating Scale
Journal of Clinical and Experimental Neuropsychology
 , 
1998
, vol. 
20
 (pg. 
536
-
547
)
Mattis
S.
Dementia rating scale: Professional manual
 , 
1988
Odessa, FL
Psychological Assessment Resources
Meiran
N.
Stuss
D.
Guzman
A.
Lafleche
G.
Willmer
J.
Diagnosis of dementia: Methods for interpretation of scores of 5 neuropsychological tests
Archives of Neurology
 , 
1996
, vol. 
53
 (pg. 
1043
-
1054
)
[PubMed]
Reisberg
B.
Ferris
S. H.
de Leon
M. J.
Crook
T.
The Global Deterioration Scale for assessment of primary degenerative dementia
American Journal of Psychiatry
 , 
1982
, vol. 
139
 (pg. 
1136
-
1139
)
[PubMed]
Sackett
D. L.
Haynes
R. B.
Guyatt
G. H.
Tugwell
P.
Clinical epidemiology: A basic science for clinical medicine
 , 
1991
Boston, MA
Little, Brown
Saxton
J.
Lopez
O. L.
Ratcliff
G.
Dulberg
C.
Fried
L. P.
Carlson
M. C.
, et al.  . 
Preclinical Alzheimer disease: Neuropsychological test performance 1.5 to 8 years prior to onset
Neurology
 , 
2004
, vol. 
63
 (pg. 
2341
-
2347
)
[PubMed]
Stuss
D.
Meiran
N.
Guzman
A.
Laflech
G.
Willmer
J.
Do long tests yield a more accurate diagnosis of dementia than short tests? A comparison of 5 neuropsychological tests
Archives of Neurology
 , 
1996
, vol. 
53
 (pg. 
1033
-
1039
)
[PubMed]
The Psychological Corporation
Wechsler abbreviated scale of intelligence
 , 
1999
San Antonio, TX
Author
Tierney
M. C.
Yao
C.
Kiss
A.
McDowell
I.
Neuropsychological tests accurately predict incident Alzheimer disease after 5 and 10 years
Neurology
 , 
2005
, vol. 
64
 (pg. 
1853
-
1859
)