Decline in learning ability best predicts future dementia type: the Freedom House Study

Evaluating verbal learning and memory in patients with an at-risk mental state or first episode psychosis using structural equation modelling

BACKGROUND

Verbal learning and memory are impaired not only in patients with a first episode of psychosis (FEP) but also-to a lower extent-in those with an at-risk mental state for psychosis (ARMS). However, little is known about the specific nature of these impairments. Hence, we aimed to study learning and memory processes in ARMS and FEP patients by making use of structural equation modelling.

METHODS

Verbal learning was assessed with the California Verbal Learning Test (CVLT) in 98 FEP patients, 126 ARMS patients and 68 healthy controls (HC) as part of the Basel early detection of psychosis (FePsy) study. The four-factorial CFA model of Donders was used to estimate test performance on latent variables of the CVLT and growth curve analysis was used to model the learning curve. The latter allows disentangling initial recall, which is strongly determined by attentional processes, from the learning rate.

RESULTS

The CFA model revealed that ARMS and FEP patients were impaired in Attention Span, Learning Efficiency and Delayed Memory and that FEP patients were additionally impaired in Inaccurate Memory. Additionally, ARMS-NT, but not ARMS-T, performed significantly worse than HC on Learning Efficiency. The growth curve model indicated that FEP patients were impaired in both initial recall and learning rate and that ARMS patients were only impaired in the learning rate.

CONCLUSIONS

Since impairments were more pronounced in the learning rate than the initial recall, our results suggest that the lower scores in the CVLT reported in previous studies are more strongly driven by impairments in the rate of learning than by attentional processes.

Does ethnicity moderate dementia's biomarkers?

We have used structural equation models to explicitly distinguish functional status, and therefore "dementia-relevant" variance in cognitive task performance (i.e., "δ") from that which is unrelated to a dementing process. Our approach results in a relatively "error-free" continuous variable that can serve as a dementia-specific phenotype. In this study, we associate δ with cytokines and other serum biomarkers in a well characterized Alzheimer's disease cohort, the Texas Alzheimer's Research and Care Consortium. Independent of covariates, δ is associated significantly with 12 serum biomarkers. These associations appear to be specific to non-Hispanic white participants.

Modeling learning and memory using verbal learning tests: results from ACTIVE

OBJECTIVE

To investigate the influence of memory training on initial recall and learning.

METHOD

The Advanced Cognitive Training for Independent and Vital Elderly study of community-dwelling adults older than age 65 (n = 1,401). We decomposed trial-level recall in the Auditory Verbal Learning Test (AVLT) and Hopkins Verbal Learning Test (HVLT) into initial recall and learning across trials using latent growth models.

RESULTS

Trial-level increases in words recalled in the AVLT and HVLT at each follow-up visit followed an approximately logarithmic shape. Over the 5-year study period, memory training was associated with slower decline in Trial 1 AVLT recall (Cohen's d = 0.35, p = .03) and steep pre- and posttraining acceleration in learning (d = 1.56, p < .001). Findings were replicated using the HVLT (decline in initial recall, d = 0.60, p = .01; pre- and posttraining acceleration in learning, d = 3.10, p < .001). Because of the immediate training boost, the memory-trained group had a higher level of recall than the control group through the end of the 5-year study period despite faster decline in learning.

DISCUSSION

This study contributes to the understanding of the mechanisms by which training benefits memory and expands current knowledge by reporting long-term changes in initial recall and learning, as measured from growth models and by characterization of the impact of memory training on these components. Results reveal that memory training delays the worsening of memory span and boosts learning.

JointMMCC: joint maximum-margin classification and clustering of imaging data

A number of conditions are characterized by pathologies that form continuous or nearly-continuous spectra spanning from the absence of pathology to very pronounced pathological changes (e.g., normal aging, mild cognitive impairment, Alzheimer's). Moreover, diseases are often highly heterogeneous with a number of diagnostic subcategories or subconditions lying within the spectra (e.g., autism spectrum disorder, schizophrenia). Discovering coherent subpopulations of subjects within the spectrum of pathological changes may further our understanding of diseases, and potentially identify subconditions that require alternative or modified treatment options. In this paper, we propose an approach that aims at identifying coherent subpopulations with respect to the underlying MRI in the scenario where the condition is heterogeneous and pathological changes form a continuous spectrum. We describe a joint maximum-margin classification and clustering (JointMMCC) approach that jointly detects the pathologic population via semi-supervised classification, as well as disentangles heterogeneity of the pathological cohort by solving a clustering subproblem. We propose an efficient solution to the nonconvex optimization problem associated with JointMMCC. We apply our proposed approach to an medical resonance imaging study of aging, and identify coherent subpopulations (i.e., clusters) of cognitively less stable adults.

Subjective cognitive complaints and longitudinal changes in memory and brain function

OBJECTIVE

Subjective cognitive complaints are often used in the diagnosis of memory and other cognitive impairment. This study examined whether cognitive complaints are associated with longitudinal changes in cognition and cross-sectional differences in regional brain function during memory performance in 98 participants with a mean age of 75.

METHOD

The Cognitive Failures Questionnaire (CFQ) assessed cognitive complaints and mixed effects regression models were used to determine whether mean CFQ scores predicted rates of change in cognitive function over a period of 11.5 years.

RESULTS

Higher CFQ scores, reflecting increased subjective complaints, were associated with steeper rates of decline in immediate and delayed recall on the California Verbal Learning Test. Voxel-based regression analysis was used to determine the cross-sectional relationship between CFQ scores and regional cerebral blood flow measured by PET during a resting condition and during verbal and figural memory tasks. Higher levels of cognitive complaints were associated with increased activity in insular, lingual and cerebellar areas during memory tasks.

CONCLUSIONS

These findings offer some support for the validity of subjective cognitive complaints as markers of age related changes in memory and brain activity.

Verbal learning changes in older adults across 18 months

The major aim of this study was to investigate individual changes in verbal learning across a period of 18 months. Individual differences in verbal learning have largely been neglected in the last years and, even more so, individual differences in change in verbal learning. The sample for this study comes from the Zurich Longitudinal Study on Cognitive Aging (ZULU; Zimprich et al., 2008a) and comprised 336 older adults in the age range of 65-80 years at first measurement occasion. In order to address change in verbal learning we used a latent change model of structured latent growth curves to account for the non-linearity of the verbal learning data. The individual learning trajectories were captured by a hyperbolic function which yielded three psychologically distinct parameters: initial performance, learning rate, and asymptotic performance. We found that average performance increased with respect to initial performance, but not in learning rate or in asymptotic performance. Further, variances and covariances remained stable across both measurement occasions, indicating that the amount of individual differences in the three parameters remained stable, as did the relationships among them. Moreover, older adults differed reliably in their amount of change in initial performance and asymptotic performance. Eventually, changes in asymptotic performance and learning rate were strongly negatively correlated. It thus appears as if change in verbal learning in old age is a constrained process: an increase in total learning capacity implies that it takes longer to learn. Together, these results point to the significance of individual differences in change of verbal learning in the elderly.

Correlates of individual, and age-related, differences in short-term learning

Latent growth models were applied to data on multitrial verbal and spatial learning tasks from two independent studies. Although significant individual differences in both initial level of performance and subsequent learning were found in both tasks, age differences were found only in mean initial level, and not in mean learning. In neither task was fluid or crystallized intelligence associated with learning. Although there were moderate correlations among the level parameters across the verbal and spatial tasks, the learning parameters were not significantly correlated with one another across task modalities. These results are inconsistent with the existence of a general (e.g., material-independent) learning ability.

Cognitive predictors of mortality in elderly retirees: results from the Freedom House study

OBJECTIVE

The objective of this longitudinal cohort study was to study the cognitive domains associated with five-year longitudinal survival among healthy, well-educated, noninstitutionalized elderly.

METHODS

Survival curves were generated as a function of cross-sectional baseline cognitive test performance.

RESULTS

Nonverbal tests were significantly associated with survival. This finding was markedly consistent. Several nonverbal tasks were each significantly associated with survival independently of age, gender, baseline level of care, and healthcare utilization. In a multivariate model, copying a clock made the strongest, independent contribution to survival.

CONCLUSIONS

Right hemisphere integrity in general and nonverbal drawing tasks in particular have been associated with survival in conditions as diverse as Alzheimer disease, stroke, and epilepsy. This study extends this association to "normal" aging. The mechanism by which nonverbal cognitive function is related to mortality remains unclear but may be mediated by changes in right hemisphere cortical control of autonomic function. Nondemented older persons may be at risk. Clock drawing may provide a simple means of identifying them.

Verbal Learning and Everyday Functioning in Dementia: An Application of Latent Variable Growth Curve Modeling

This study used latent variable growth curve modeling to identify predictors and correlates of verbal learning over trials on a list-learning task in patients with dementia. Data from 116 patients evaluated at the Detroit satellite of the Michigan Alzheimer's Disease Research Center were incorporated in the present analyses. Patients were administered the Fuld Object Memory Evaluation, examined independently by a geriatrician, and, if appropriate, given a diagnosis of probable Alzheimer's disease according to criteria from the National Institute of Neurological and Communicative Disorders and Stroke/Alzheimer's Disease and Related Disorders Association. The presence of dementia significantly predicted both the intercept (i.e., level of performance) and the slope (i.e., learning over trials), with dementia patients demonstrating lower overall levels of performance and less verbal learning over trials. Rate of verbal learning over trials was a significant predictor of everyday functioning (instrumental activities of daily living) above and beyond general cognitive impairment and demographics.

A growth curve model of learning acquisition among cognitively normal older adults

The objective of this study was to model recall and learning on the Auditory Verbal Learning Test using latent growth curve techniques. Participants were older adults recruited for the ACTIVE cognitive intervention pilot. A series of nested models revealed that an approximately logarithmic growth curve model provided optimal fit to the data. Although recall and learning factors were statistically uncorrelated, a fitted multivariate model suggested that initial recall was significantly associated with demographic characteristics but unrelated to health factors and cognitive abilities. Individual differences in learning were related to race/ethnicity, speed of processing, verbal knowledge, and global cognitive function level. These results suggest that failing to recognize initial recall and learning as distinct constructs clouds the interpretation of supraspan memory tasks.

Executive control mediates memory's association with change in instrumental activities of daily living: the Freedom House Study

OBJECTIVES

To assess the relative independent contribution of changes in executive control function (ECF) and memory to changes in functional status.

DESIGN

Three-year longitudinal cohort study.

SETTING

A comprehensive care retirement community.

PARTICIPANTS

Five hundred forty-seven noninstitutionalized people aged 70 and older.

MEASUREMENTS

The California Verbal Learning Test (CVLT) and Executive Interview (EXIT25). Functional status was assessed using instrumental activities of daily living (IADLs). Latent growth curves of CVLT, EXIT25, and IADLs were modeled. The rate of change in IADLs (adjusted for baseline IADLs and cognition) was regressed on the rate of change in each cognitive measure. Models were also adjusted for baseline age, level of care, and comorbid illnesses.

RESULTS

There was significant variability around the baseline means and slopes for all variables. The rate of change in EXIT25 was independently correlated with the rate of change in IADLs (correlation coefficient (r)=-0.52, P<.001). This remained significant after adjusting for baseline EXIT25 scores, IADLs, age, comorbid disease, and level of care. The EXIT25's effect on the rate of change in IADLs was stronger than those of age, baseline IADLs, comorbid disease, or level of care. The rate of change in CVLT scores was not significantly associated with the rate of change in IADLs.

CONCLUSION

ECF is a strong, significant, and independent correlate of functional status in normal aging. In contrast, decline in memory, as measured using the CVLT, has no independent association with the rate of change in functional status. This suggests that amnestic mild cognitive impairment can be associated with dementia only though the subsequent or comorbid development of ECF impairment.

Using generalized additive models to reduce residual confounding

Traditionally, confounding by continuous variables is controlled by including a linear or categorical term in a regression model. Residual confounding occurs when the effect of the confounder on the outcome is mis-modelled. A continuous representation of a covariate was previously shown to result in a less biased estimate of the adjusted exposure effect than categorization provided the functional form of the covariate-outcome relationship is correctly specified. However, this is rarely known. In contrast to parametric regression, generalized additive models (GAM) fit a smooth dose-response curve to the data, without requiring a priori knowledge of the functional form. We used simulations to compare parametric multiple logistic regression vs its non-parametric GAM extension in their ability to control for a continuous confounder. We also investigated several issues related to the implementation of GAM in this context, including: (i) selecting the degrees of freedom; and (ii) alternative criteria for inclusion/exclusion of the potential confounder and for choosing between parametric and non-parametric representation of its effect. The impact of the shape and strength of the confounder-disease association, sample size, and the correlation between the confounder and exposure were investigated. Simulations showed that when the confounder has a non-linear association with the outcome, compared to a parametric representation, GAM modelling (i) reduced the mean squared error for the adjusted exposure effect; (ii) avoided inflation of the type I error for testing the exposure effect. When the true confounder-outcome relationship was linear, GAM performed as well as the parametric logistic regression. When modelling a continuous exposure non-parametrically, in the presence of a continuous confounder, our results suggest that assuming a linear effect of the confounder and focussing on the non-linearity of the exposure-outcome relationship leads to spurious findings of non-linearity: joint non-linear modelling is necessary. Overall, our results suggest that the use of GAM to reduce residual confounding offers several improvements over conventional parametric modelling.