A brief clinical neuropsychologic screening battery: Statistical classification trials

Abstract

Classified independently 150 neuropsychiatric referrals into Definite (DBD), Suspected (SDB) or NO Brain Damage (NDB) groups on the basis of non‐neuropsychological evaluative data. Each S was examined with a brief, 2‐hour neuropsychologic screening battery, and the test data were analyzed by a stepwise linear discriminant function program. Statistical classifications were compared with criteria for NBD vs. SBD + DBD and for SBD vs. DBD groupings. Hit rates ranged from 95% for the former to 81.7% for the latter, with low false positives and negatives. Hit and miss rates were not as strong in a validation step, but exceeded chance levels.

A brief clinical neuropsychologic battery: Clinical classification trials

“…This hit rate exceeds the overall rate of 71.4% noted in the validation step of the statistical classification trial (Barrett et al, 1982). The false positive and false negative rates in the clinical classification trials were 16.3% and 16.896, respectively.…”

“…In comparing the ratings of the two neuropsychologists with the neurological criteria, 84% of the NBD cases and 84% of the combined SBD + DBD cases were classified correctly, which yielded an overall hit rate of 84% (K = .667,p <.001). This hit rate exceeds the overall rate of 71.4% noted in the validation step of the statistical classification trial (Barrett et al, 1982). The false positive and false negative rates in the clinical classification trials were 16.3% and 16.896, respectively.…”

“…In the clinical classifications study, it was suggested that the interrater variability might have been reduced if the neuropsychologists had been permitted to review the statistical data and had been advised of the discriminant variables that optimally separated the groups. As noted in the statistical classification trials (Barrett et al, 1982). the combinations and weightings of variables changed as a function of the specific discrimination: For the NBD vs. SBD + DBD discriminations, level of performance variables optimally separated the groups, while for the SBD vs. DBD discriminations, pathognomic and left -right difference variables optimally separated the groups.…”

“…The clinical classification hit rates compared favorably with the statistical classification hit rates of the validity study (Barrett et al, 1982). This comparison is felt to be fair because the validity classifications did not have the statistical advantage of the original linear discriminant analyses, nd the clinical classifications did not have the advantage of knowing the population demographics.…”

“…No studies that used brief batteries have compared statistical vs. clinical classification trials. Barrett, Wheatley, and La Plant (1982) introduced a brief, 2-hour neuropsychological battery and used a step-wise linear discriminant function program to evaluate 150 neuropsychiatric referrals, who were classified independently into definite (DBD), suspected (SBD), or no brain damage (NBD) groups. Statistical classifications were compared with criteria for NBD vs. SBD + DBD and for SBD vs. DBD groups.…”