RESEARCH SUMMARY

I. INITIAL MARKETING STUDY
II. ASSESSMENT OF VDR ITEM DIFFICULTY
III. CONSTRUCT VALIDATION OF DAT
IV. RETEST RELIABILITY
V. ALTERNATE FORM RELIABILITY
VI. INITIAL VALIDATION STUDY
VII. SENSITIVITY AND SPECIFICITY
VIII. COMPARISON TO NON-CPNPUTERIZED SCREENING TECHNIQUES
IX. CONSTRUCT VALIDITY ANALYSIS

I. INITIAL MARKETING STUDY
Prior to developing the specific procedures that became the GrayMatters® system, a marketing study was commissioned to evaluate community needs and perceptions with respect to dementia and screening for dementia. The survey was completed in 2002 by an academic organization which had no financial or other ties to Dementia Screening, Inc. There were 125 respondents, community-dwelling individuals from 50 to 90 years of age. Key findings were as follows:

•Concern about being affected by Alzheimer’s disease or other form of dementia: Very Concerned 45%, Somewhat Concerned 30%.

•Willingness to complete a screening test for dementia: Very Willing 60%, Somewhat Willing 29%.

•Preferred place for taking screening test: Doctor’s Office 53%, Home 22%.

•Who would benefit from screening: Me 69%, My Spouse 29%.

These results were interpreted to indicate the following:
There exists in the community significant concern about developing some form of dementia. Most respondents are willing to complete standardized testing to screen for dementia, preferably in a doctor’s office, and there is perceived benefit to one’s self and to one’s spouse.

II. ASSESSMENT OF VDR ITEM DIFFICULTY
One hundred prospective items were developed as possible items to be used in the VDR. These items consisted of a set of 2 or 4 pictures followed by correct or incorrect challenge items which were presented after a 30-second delay period. No distracter task was employed in this phase of test development. Subjects (N=100) were recruited from undergraduate classes at a local university, and the difficulty of individual items was determined by the probability of subject’s success on each individual item. Information from this study was used to select a progression of difficulty in VDR items and to establish alternate forms for the VDR.

III. CONSTRUCT VALIDATION OF DAT
Benge, JF. Validation of a computerized alternation task as a measure of executive functioning in an elderly sample. Unpublished Master’s Thesis, Abilene Christian University, 2003.
The construct validity of the DAT was evaluated by comparing DAT performance with performance on the Wisconsin Card Sort, Category Test, the Rey-Osteritth Complex Figure Test (Recall and Copy), the Paced Auditory Serial Addition Test, and the Trail Making Test. Subjects were 29 elderly individuals living in the community. Several significant relationships were found between DAT variables and the executive measures listed. Strength of the relationships was moderate (r varied from 46 to .58 in the expected direction).

III. RETEST RELIABILITY
Brinkman et al. Validation of a self-administered computerized test to detect cognitive impairment. Manuscript in review.
Ninety subjects who were included in the validation analysis completed the GrayMatters® screening at baseline and again after 30 days. Statistically significant correlations were found for the DAT measures and the VDR measures. The strength of the relationships were comparable to retest reliability scores on subtests of the Wechsler Memory Scale -3 (WMS-3), the most widely used standardized assessment of memory functioning. Retest correlation coefficients for derived scores on the GrayMatters® screening were comparable to derived scores on the WMS-3. For both the DAT scores and VDR scores, there was no statistically significant difference in mean scores from baseline to retesting.


IV. ALTERNATE FORM RELIABILITY
Alternate forms of the VDR were developed based on the item analysis described above (II. Assessment of VDR Item Difficulty). The equivalence of the two forms was evaluated with 170 community-dwelling elderly individuals who were found on the basis of about 2.5 hours of neuropsychological testing to have normal cognitive abilities. Order of test administration was randomized. Alternate form reliability coefficients were very high both for raw scores and derived scores and no mean differences were found. Equivalency of forms was accepted based on the study.

VI. INITIAL VALIDATION STUDY
Brinkman et al. Validation of a self-administered computerized test to detect cognitive impairment. Manuscript in review.
As this manuscript is in review with a refereed journal, only limited information is presented here. In a sample of several hundred subjects, identified as normal or impaired on the basis of more detailed neuropsychological evaluation, statistically significant differences were found between groups for all DAT variables and all VDR variables.


VII. SENSITIVITY AND SPECIFICITY
Brinkman et al. Sensitivity and specificity of the GrayMatters® Dementia Screening System. Manuscript in preparation.
As this manuscript will be submitted to a refereed journal for consideration, only limited information is provided here. In a sample of over 500 subjects meeting the same criteria as described in item VI above, overall accuracy of classification was found to be about 90%, utilizing classification by the neuropsychological test performance as the criterion variable. Sensitivity and specificity can be adjusted to achieve the optimum Positive and Negative Predictive Power for any given setting.

VIII. COMPARISON TO NON-COMPUTERIZED SCREENING TECHNIQUES
Manuscript in Preparation.
Data analysis is in progress to determine how the GrayMatters® system compares to more widely used systems such as the MMSE, the Clock Test, the Verbal Fluency Test, and others. On the basis of comparisons with over 500 cases, the GrayMatters® system is found to be markedly more sensitive than the MMSE, whose specificity is very high but whose sensitivity is acknowledged to be rather low (80 percent or lower). Similarly the GrayMatters® system demonstrated greater sensitivity than a wide range of other more labor-intensive screening procedures. Overall accuracy of classification of the GrayMatters® system is superior to the other screening procedures as well, when classification by neuropsychological testing is used as the criterion variable.

XI. CONSTRUCT VALIDITY ANALYSIS

Construct validity is being analyzed through factor analysis and through comparison of GrayMatters® performance to performance on a number of widely used neuropsychological tests. Results of this analysis will be presented at an upcoming professional meeting in November. Results indicate that the GrayMatters® system measures memory and executive functioning.

.......................
Dated 10-16-07