Multivariable analysis of the relative utility and additive value of eight embedded performance validity tests for classifying invalid neuropsychological test performance.

Introduction: This study investigated a combination of eight embedded performance validity tests (PVTs) derived from commonly administered neuropsychological tests to optimize sensitivity/specificity for detecting invalid neuropsychological test performance. The goal of this study was to evaluate what combination of these common embedded PVTs that have the most robust predictive power for detecting invalid neuropsychological test performance in a single diverse clinical sample.

Method: Eight previously validated memory- and nonmemory-based embedded PVTs were examined among 231 patients undergoing neuropsychological evaluation. Patients were classified into valid/invalid groups based on four independent criterion PVTs. Embedded PVT accuracy was assessed using standard and stepwise multiple logistic regression models.

Results: Three PVTs, the Brief Visuospatial Memory Test-Revised Recognition Discrimination (BVMT-R-RD), Rey Auditory Verbal Learning Test Forced Choice, and WAIS-IV Digit Span Age Corrected Scaled Score, predicted 45.5% of the variance in validity group membership. BVMT-RD independently accounted for 32% of the variance in prediction of independent, criterion-defined validity group membership.

Conclusions: This study demonstrated the incremental predictive power of multiple embedded PVTs derived from common neuropsychological measures in detecting invalid test performance and those measures accounting for the greatest portion of the variance. These results provide guidance for evaluating the most fruitful embedded PVTs and proof of concept to better guide selection of embedded validity indices. Further, this offers clinicians an efficient, empirically derived approach to assessing performance validity when time restraints potentially limit the use of freestanding PVTs.