The geriatric California Verbal Learning Test (CVLT-9) is a 9-word list learning task administered over 5 trials with a short and long delay (Libon et al., 1996). Informed by the process approach, we sought to further expand the clinical utility of the CVLT-9 by defining new variables that reflect the nature of perseverations. Performance was compared among 52 individuals with probable Alzheimer's Disease (pAD) and 45 individuals with Ischaemic Vascular Dementia (IVD). While the pAD produced more unique intrusions and perseverated on those same errors across trials, the IVD individuals refrained from producing such perseverations. Additionally, on recognition testing, the IVD group endorsed items from the interference condition, while pAD made errors on both interference and semantically similar foils. These findings underscore the utility of using the process approach for discerning two of the most prevalent dementing disorders.

A detailed analysis of how an individual performs on a task, in contrast to obtaining composite scores, yields more information about a patient's cognitive style and residual strengths. Known as the Boston Process Approach, this form of assessment continues to enhance our understanding of neurological disorders and underlying neural systems (e.g., WAIS-R NI, Kaplan et al., 1991). This is especially true for any task that involves multidimensional cognitive skills, such as verbal declarative memory tasks, which involve competence in attention, concentration, and semantic knowledge. Additionally, this approach can provide innovative ways for differential diagnosis of syndromes that may otherwise appear similar on raw and scaled score measures (e.g., WAIS-R Similarities, Giovannetti et al., 1998; Goldberg Graphical Sequences, Lamar et al., 1997). We believe employing such an approach is essential for describing the nuances of, and ultimately improving the specificity of treatment in, the dementia syndromes.

Using a process-oriented approach to assessment, the current investigation was devised to enhance our understanding of free recall and recognition memory performance among two common forms of dementia; ischaemic vascular dementia marked by subcortical white matter alterations (IVD) and probable Alzheimer's disease (pAD). The geriatric version of the California Verbal Learning Test (CVLT-9), a serial list learning test, drawn from three semantic categories, over five learning trials with an interference condition, long delay free recall and recognition trials, was used as the primary investigative tool. Preliminary findings have demonstrated construct validity, as well as diagnostic specificity, among patients in the early stages of pAD and IVD (Libon et al., 1996). Thus, the CVLT-9 has emerged as a promising assessment of verbal declarative memory in dementia.

Typically, performance on the CVLT-9 is examined by recording the number of intrusions and perseverations during free and cued recall, and the discrimination of target words from foils after a long delay (recognition discriminability; Libon, et al., 1996). However, it is hypothesized that this traditional scoring method does not identify the various manifestations of perseverative behavior that may implicate neural systems compromised by IVD and pAD. For example, using the traditional scoring method, intrusions that are repeated across trials are scored as intrusion errors, despite the notion that they reflect perseverative behavior.

Additionally, there has been little attention attributed to when commission errors occur during list learning trials. As shown by Lamar and colleagues (1999) with the Boston Revision of the Weschler Mental Control subtest and by Price and colleagues (1999) with phonemic fluency on FAS, performance over time is believed to distinguish pAD from IVD and Parkinson's disease. In short, these investigations found that while relatively good performance was elicited from patients with subcortical dementia at the beginning of a task trial, performance deteriorated over time. In light of these results, we suggest that another important aspect of the process approach should include a quantification of when errors occur during tests of free recall.

Thus, the purpose of this investigation was to modify the scoring of errors on the CVLT to more accurately reflect the nature of the error types, especially the nature of perseverative responses. Additionally, we sought to analyze when errors occurred during the free recall acquisition trials. Finally, we sought to demonstrate the utility of such an approach by examining the errors of patients in the early stages of pAD and IVD characterized by subcortical white matter alterations.

Participants and Procedure
Ninety-seven participants (52 pAD; 45 IVD) were drawn from the Geriatric Assessment Program (GAP) of the Alexander Silberman Center of the Crozer-Chester Medical Center, Upland, Pennsylvania, USA. Participants were assigned to groups by meeting the NINCDS-ADRDA criteria (McKhann et al., 1984) and the CADDTC criteria (Chui et al., 1992) for pAD and IVD, respectively. Some of these data were drawn from the cohort from which the CVLT-9 was originally normed (Libon et al., 1996). Groups were comparable on demographic and assessment screening tools as summarized in Table 1. The CVLT-9 was administered according to standard procedures (Delis et al., 1987).

Table 1: Summary of results from demographic and screening measures for partcipants with probable Alzheimer's disease (pAD) and ischaemic vascular dementia (IVD). MMSE: Folstein et al., 1975; GDS: Yesavage, 1986.

Scoring
Variables of interest included the traditional measures of performance on the original 16-word CVLT version (Delis et al., 1987). To assess when free recall errors occurred, the frequency of errors within the first and last half of trials were calculated. The following supplemental variables were also of interest:

Free Recall Virgin Intrusions: The number of unique (i.e., non-target) words stated are scored as intrusions. For example, although the word "apple" may appear on trials 3, 4, and 5, only one virgin intrusion is scored (trial 3).

Free Recall Trans-trial Perseverations: The frequency of virgin intrusions repeated in subsequent trials. Although "apple" appears on trials 3, 4, and 5, two trans-trial perseverations would be scored (trial 4 and 5).

Free Recall Imported Perseverations: The number of perseverative responses drawn from the context of the testing environment either through testing room stimuli or previously administered tests (e.g., coffee pot from an examiner's desk, asparagus from the Boston Naming Test).

Free Recall Within-trial Perseverations: The number of perseverations that occur within each trial. These include perseverative responses of non-target words (e.g., apple) and correct target words.

Semantic False Positive (FP) Recognition Errors: The number of FPs within the same three semantic categories as the original target words from list A.

Interference Condition False Positive (FP) Recognition Errors: The number of FPs taken from to the interference condition (List B).

Unrelated False Positive (FP) Recognition Errors: FPs which are neither semantically similar nor included in the interference condition.

Previous research showed that although both pAD and IVD individuals produced the same number of words (i.e., total output including errors) during the free recall trials, AD individuals made more intrusion errors (Libon et al, 1996). The current investigation retrospectively examined these same data and revealed the nature of these intrusion errors. We were able to discern between group differences in the number of unique (i.e., virgin) intrusions and the degree to which these virgin intrusions were repeated across learning trials. We believe these findings provide insight into the integrity of verbal recognition as mediated by intact semantic knowledge in IVD.

Analysis of five free recall acquisition trials revealed that the pAD individuals (M=1.29, SD=1.7) produced more virgin intrusions, overall, than IVD (M=0.58, SD=0.97) [t(82.56)=2.57, p=.012]. These free recall intrusions occurred within the latter half of each trial for both dementia groups [F(1, 95)=25.86, p<.001]. Additionally, those with pAD produced more trans-trial perseverations (M=0.75, SD=1.47) than the IVD (M=0.22, SD=0.70) [t(75.53)=2.31, p=.024]. No between group differences were found among within-trial perseverations. Together, these findings indicate that during the free recall trials of the CVLT-9, both the pAD and IVD suffer intrusions during the latter half of each trial. However, only the pAD repeat these intrusions throughout all five trials.

Source memory may play a key role in the recognition of word lists. Individuals with pAD are known to suffer from source memory deficits (Multaup & Balota, 1997). This may explain the continued perseveration of virgin intrusions (i.e., trans-trial perseverations) found in the current investigation. Indeed, as Luria (1980) postulated, perseverations occur in domains in which patients are experiencing a deficit. Preservation of both source memory and semantic knowledge appear to aid individuals with IVD and may explain their relative success on verbal declarative memory tasks, both in the production of fewer errors during acquisition and in the increased recognition discriminability after a long delay. Independent samples t-tests identified that after a long delay, the IVD individuals produced fewer errors than the pAD on the semantic [t(85.94)=3.93, p<.001], interference [t(95)=2.02, p=.047], and unrelated recognition condition [t(95)=2.38, p=.02]. However, pairwise t-tests revealed that individuals with IVD endorsed more recognition items from the interference condition than unrelated words [t(44)=3.89, p<.001]. There was also a trend for IVD to endorse more items from the interference condition than semantically similar foils [pairwise t(44)=1.73, p=.09]. In constrast, the pAD were just as likely to endorse semantically similar foils as that from the interference condition [t(51)=.82, p=.41]. Similar to the IVD, the pAD endorsed fewer unrelated words [t(51)=2.22, p<.03]. These data are commissurate with the relatively preserved recognition discriminability index scores of the IVD when compared to the pAD (Libon, et al., 1996).

These results can be interpreted within the context of the underlying neuropathology of IVD and pAD. Deficits in both semantic knowledge and declarative memory in AD are consistent with infiltration of neuropathology and atrophy in parahippocampal areas of the temporal lobe. Additionally, pAD is usually associated with intrusion errors. In contrast, there is relative preservation of both temporal lobe structure and function with subsequent integrity of semantic processes in IVD (Giovannetti et al., 1997). As one of the hallmarks of IVD is perseveration of behavior, and lesions occur within subcortical white matter tracts projecting to the frontal lobes, these findings may appear to be counterintuitive. However, the perseverative behavior in pAD may be explained by deficits in semantic processes and impaired source memory rather than executive functions mediated by frontal systems.

LE FastCounter