Adapting the WMS-III Logical Memory Rarely Missing Index (RMI) to the WMS-IV: a study across archived clinical and...

Use of embedded performance validity tests (PVTs) helps efficiently monitor performance within neuropsychological batteries, particularly when embedded within instruments such as the Wechsler Memory Scale (WMS) Logical Memory (LM) subtest. Killgore and DellaPietra's 2000 commonly referenced WMS-III Rarely Missed Index (RMI) was developed through simulated design and not updated using WMS-IV items. This study investigated the utility of the WMS-III RMI in our clinical sample, while also seeking to validate a WMS-IV RMI update utilizing an archival Pearson non-stimulus sample (NSS) and our clinical archive.

Fifty cases from the Pearson NSS archive and clinic archive PVT Pass (N = 195) and Fail (N = 95) cases were included. Determination of PVT Pass-Fail was based on passing ≥2 stand-alone and/or embedded PVTs. The original RMI was updated using WMS-IV questions adapted from WMS-III. The novel WMS-IV RMI was developed by identifying LM recognition items answered with ≥ 70% accuracy in the Pearson NSS or ≥ 90% accuracy in the Pass-PVT group. Items entered into exploratory discriminant function analysis revealed a structure matrix with group correlation of ≥ 0.30, and standardized canonical discriminant function coefficients of 0.59, 0.46, 0.41, 0.33, and 0.23 for WMS-IV LM recognition items 14, 16, 22, 28, and 29, respectively. Items were weighted based on these coefficients, aggregating to an index ranging from 0 to 202.

Unexpectedly, no original RMI items overlapped with the novel RMI. The novel RMI cutoff of ≤140 resulted in 90.3% specificity with 25.3% sensitivity. Novel RMI AUC was 0.66, with an optimal cutoff of ≤190 to maximize sensitivity (64.2%) and specificity (65.4%).

Results caution clinicians and researchers against using dated PVTs, while PVTs determined by simulated designs should not be assumed to stand up to clinical samples.