Cognition-Mortality Associations Are More Pronounced When Estimated Jointly in Longitudinal and Time-to-Event Models.

With aging populations worldwide, there is growing interest in links between cognitive decline and elevated mortality risk-and, by extension, analytic approaches to further clarify these associations. Toward this end, some researchers have compared cognitive trajectories of survivors vs. decedents while others have examined longitudinal changes in cognition as predictive of mortality risk. A two-stage modeling framework is typically used in this latter approach; however, several recent studies have used joint longitudinal-survival modeling (i.e., estimating longitudinal change in cognition conditionally on mortality risk, and vice versa). Methodological differences inherent to these approaches may influence estimates of cognitive decline and cognition-mortality associations. These effects may vary across cognitive domains insofar as changes in broad fluid and crystallized abilities are differentially sensitive to aging and mortality risk. We compared these analytic approaches as applied to data from a large-sample, repeated-measures study of older adults (N = 5,954; ages 50-87 years at assessment; 4,453 deceased at last census). Cognitive trajectories indicated worse performance in decedents and when estimated jointly with mortality risk, but this was attenuated after adjustment for health-related covariates. Better cognitive performance predicted lower mortality risk, and, importantly, cognition-mortality associations were more pronounced when estimated in joint models. Associations between mortality risk and crystallized abilities only emerged under joint estimation. This may have important implications for cognitive reserve, which posits that knowledge and skills considered well-preserved in later life (i.e., crystallized abilities) may compensate for declines in abilities more prone to neurodegeneration, such as recall memory and problem solving. Joint longitudinal-survival models thus appear to be important (and currently underutilized) for research in cognitive epidemiology.

Illness and intelligence are comparatively strong predictors of individual differences in depressive symptoms following middle age.

OBJECTIVE: We compared the importance of socio-demographic, lifestyle, health, and multiple cognitive measures for predicting individual differences in depressive symptoms in later adulthood. METHOD: Data came from 6203 community-dwelling older adults (age 41-93 years at study entry) from the United Kingdom. Predictors (36 in total) were assessed up to four times across a period of approximately 12 years. Depressive symptoms were measured with the Geriatric Depression Scale. Statistical methods included multiple imputation (for missing data), random forest analysis (a machine learning approach), and multivariate regression. RESULTS: On average, depressive symptoms increased gradually following middle age and appeared to accelerate in later life. Individual differences in depressive symptoms were most strongly associated with differences in combined symptoms of physical illness (positive relation) and fluid intelligence (negative relation). The strength of association between depressive symptoms and fluid intelligence was unaffected by differences in health status within a subsample of chronically depressed individuals. CONCLUSION: Joint consideration of general health status and fluid intelligence may facilitate prediction of depressive symptoms severity during later life and may also serve to identify sub-populations of community-dwelling elders at risk for chronic depression.

Cardiovascular symptoms and longitudinal declines in processing speed differentially predict cerebral white matter lesions in older adults.

It is well established that cerebral white matter lesions (WML), present in the majority of older adults, are associated with cardiovascular and cerebrovascular diseases and also with cognitive decline. However, much less is known about how WML are related to other important individual characteristics and about the generality vs. brain region-specificity of WML. In a longitudinal study of 112 community-dwelling adults (age 50-71 years at study entry), we used a machine learning approach to evaluate the relative strength of 52 variables in association with WML burden. Variables included socio-demographic, lifestyle, and health indices-as well as multiple cognitive abilities (modeled as latent constructs using factor analysis)-repeatedly measured at three- to six-year intervals. Greater chronological age, symptoms of cardiovascular disease, and processing speed declines were most strongly linked to elevated WML burden (accounting for ∼49% of variability in WML). Whereas frontal lobe WML burden was associated both with elevated cardiovascular symptoms and declines in processing speed, temporal lobe WML burden was only significantly associated with declines in processing speed. These latter outcomes suggest that age-related WML-cognition associations may be etiologically heterogeneous across fronto-temporal cerebral regions.

Speed of Visual Search in Old Age: 1950 to 2016.

OBJECTIVES: To review work on the effects of old age on speed of visual search and of discriminations between signals and choices of responses between 1950 and 2016. METHODS: Literature review and Brain Google. RESULTS: Mild existential despondency. DISCUSSION: Models for age changes in discrimination between signals and choices of responses have been based on comparisons of speed. The concept of speed has been used in four distinct ways: as a directly measured task performance index, as a hypothetical functional system performance characteristic, as a factor in psychometric models computing mutual variance in calendar age and as a neurophysiological performance characteristic. Since 1950, these measures have, in turn, determined models for speed of perceptual discriminations. Since the 1990s, advances in neuroimaging have not only transformed the data available, but also the nature of the questions that we ask.

Think Fast, Feel Fine, Live Long: A 29-Year Study of Cognition, Health, and Survival in Middle-Aged and Older Adults.

In a 29-year study of 6,203 individuals ranging in age from 41 to 96 years at initial assessment, we evaluated the relative and combined influence of 65 mortality risk factors, which included sociodemographic variables, lifestyle attributes, medical indices, and multiple cognitive abilities. Reductions in mortality risk were most associated with higher self-rated health, female gender, fewer years as a smoker, and smaller decrements in processing speed with age. Thus, two psychological variables-subjective health status and processing speed-were among the top predictors of survival. We suggest that these psychological attributes, unlike risk factors that are more narrowly defined, reflect (and are influenced by) a broad range of health-related behaviors and characteristics. Information about these attributes can be obtained with relatively little effort or cost and-given the tractability of these measures in different cultural contexts-may prove expedient for prevention, diagnosis, and treatment of conditions related to increased mortality risk in diverse human populations.

Life span decrements in fluid intelligence and processing speed predict mortality risk.

We examined life span changes in 5 domains of cognitive performance as predictive of mortality risk. Data came from the Manchester Longitudinal Study of Cognition, a 20-plus-year investigation of 6,203 individuals ages 42-97 years. Cognitive domains were general crystallized intelligence, general fluid intelligence, verbal memory, visuospatial memory, and processing speed. Life span decrements were evident across these domains, controlling for baseline performance at age 70 and adjusting for retest effects. Survival analyses stratified by sex and conducted independently by cognitive domain showed that lower baseline performance levels in all domains-and larger life span decrements in general fluid intelligence and processing speed-were predictive of increased mortality risk for both women and men. Critically, analyses of the combined predictive power of cognitive performance variables showed that baseline levels of processing speed (in women) and general fluid intelligence (in men), and decrements in processing speed (in women and in men) and general fluid intelligence (in women), accounted for most of the explained variation in mortality risk. In light of recent evidence from brain-imaging studies, we speculate that cognitive abilities closely linked to cerebral white matter integrity (such as processing speed and general fluid intelligence) may represent particularly sensitive markers of mortality risk. In addition, we presume that greater complexity in cognition-survival associations observed in women (in analyses incorporating all cognitive predictors) may be a consequence of longer and more variable cognitive declines in women relative to men.

Between-individual variability and interpretation of associations between neurophysiological and behavioral measures in aging populations: comment on Salthouse (2011).

Salthouse (2011) argued that (a) variance between individuals on cognitive test scores remains constant between 20 and 90 years of age and (b) widely recognized problems of deducing functional relationships from patterns of correlations between measurements become especially severe for neuropsychological indices, especially for gross indices of age-related brain changes (e.g., losses of brain volume or increases in white matter lesions). I argue that between-individual variability on cognitive tests does increase with age and provides useful information on causes of age-related cognitive decline. I suggest that problems of inference from correlations are just as difficult for behavioral as for neurophysiological indices and that inclusion, in analyses, of even gross measures of brain status such as loss of volume and white matter lesions can correct misinterpretations that occur when only behavioral data are examined.

Terminal pathologies affect rates of decline to different extents and age accelerates the effects of terminal pathology on cognitive decline.

OBJECTIVES: To test whether different terminal pathologies are associated with different rates of age-related decline in fluid and crystallized mental abilities and whether pathology-associated declines are accelerated by age. METHODS: During a 20-year longitudinal study, 6203 participants were quadrennially assessed on the Heim's (Heim, A 1970) The AH4 series of intelligence tests Slough, U.K.: NEP) AH4-1 and AH4-2 tests of fluid intelligence and on the Raven's (Raven, J. C. 1965) The Mill Hill Vocabulary Scale London: H.K. Lewis) Mill Hill A and B tests of recognition and production vocabulary. Dates and proximate causes of death were logged for 2499 participants. Multilevel modelling compared rates of decline after effects of sex, demographics, and practice were taken into consideration. RESULTS: Rates of cognitive decline markedly differed across pathologies, being most rapid for dementias and infections, slower for malignancies, and most prolonged for cardiovascular conditions. Pathologies were associated with faster declines in older individuals. DISCUSSION: After sex, age, and demographics have also been considered, different terminal pathologies are associated with markedly different rates of decline. Age accelerates pathology-related decline. This raises the further question as to whether any, or how much of, age-related cognitive decline is brought about by other causes than an increasing burden of pathologies.

Developing a self-reported comorbidity index to predict mortality of community-dwelling older adults.

Current common comorbidity measures have poor to moderate predictive validity of mortality of community-dwelling older adults. Hence, our aim is to develop a simpler resource-efficient self-reported comorbidity index in the prediction of survival. 113 older adults in Greater Manchester, United Kingdom attended a routine medical examination whereby information gathered was used to construct Charlson Comorbidity Index (CCI). They completed the Cornell Medical Index (CMI) questionnaire and reported the number of medication prescribed to them. We compared the ability of CCI, CMI, number of medication, age and sex to predict mortality of the sample over 7-year period using Cox-regression and Kaplan-Meier plot and rank test. None of the variables individually was significant when tested using either Cox-regression via ENTER method or Kaplan-Meier test. Remarkably, by means of forward step-wise Cox-regression, two variables emerged significant: (i) number of medicine (beta coefficient=0.229, SE=0.090 and p=0.011) and (ii) age (beta coefficient=0.106, SE=0.051 and p=0.037). We demonstrated that simple count of medication predicted mortality of community-dwelling older adults over the next 7 years more accurately than CMI or CCI. Further works involving a larger scale of subjects is needed for use in epidemiological study of survival where cost and resources are concerned.

Investigation of a functional quinine oxidoreductase (NQO2) polymorphism and cognitive decline.

NAD(P)H dehydrogenase quinone 2 (NQO2) is a quinone reductase whose functions include the reduction of both oxidative stress during the redox cycle and neurotoxicity caused by the metabolism of catecholamines. We have investigated a functional non-synonymous exon 3 single nucleotide polymorphism (rs1143684) within the NQO2 gene for association with cognitive decline using a cohort of 722 community-dwelling older individuals aged 50 years and over. The volunteers had completed tests that measured fluid intelligence, processing speed, immediate/delayed verbal recall and semantic memory. We observed a nominal significant association between this polymorphism and the trajectory of delayed memory recall over time (p=0.029). No other associations were seen with the decline of other cognitive abilities.

No association between Cholinergic Muscarinic Receptor 2 (CHRM2) genetic variation and cognitive abilities in three independent samples.

Cognitive ability has a substantial genetic component and more than 15 candidate genes have been identified over the past 8 years. One gene that has been associated with general cognitive ability is the cholinergic muscarinic 2 receptor (CHRM2). In an attempt to replicate this finding we typed marker rs8191992 (the originally reported CHRM2 SNP) in two population based cohorts-one Scottish aged over 50 years (N = 2,091) and the other English comprising non-demented elderly participants (N = 758)-and a family-based Australian adolescent sample (N = 1,537). An additional 29 SNPs in CHRM2 were typed in the Australian sample and a further seven in the English cohort. No significant association was found between CHRM2 and diverse measures of cognitive ability in any of the samples. In conclusion, this study does not support a role for CHRM2 in cognitive ability.

Further analyses of the effects of practice, dropout, sex, socio-economic advantage, and recruitment cohort differences during the University of Manchester longitudinal study of cognitive change in old age.

A sample of 4,314 volunteers who, when first recruited, were aged from 41 to 93 years were quadrennially tested from 2 to 4 occasions during the next 4 to 20 years on the Cattell Culture Fair intelligence test, 2 tests of information-processing speed, the Wechsler Adult Intelligence Scale (WAIS) vocabulary test, and 3 memory tests. After significant effects of practice, sex, demographics, socio-economic advantage, and recruitment cohort had been identified and considered, performance on all tests declined with age. These age-related declines accelerated for the Cattell and WAIS, 2 tests of information speed, and 2 of the memory tests. For all tests individuals' trajectories of age-related change diverged with increasing age but, unexpectedly, were not affected by demographic factors. Practice gains from an initial experience of the cognitive tests remained undiminished as the interval before the second experience increased from 4 to 8 + years.

Death, dropout, and longitudinal measurements of cognitive change in old age.

During a 20-year longitudinal study of cognitive change in old age 2,342 of 5,842 participants died and 3,204 dropped out. To study cognitive change as death approaches, we grouped participants by survival, death, dropout, or dropout followed by death. Linear mixed-effects pattern-mixture models compared rates of cognitive change before death and dropout from four quadrennial administrations of tests of fluid intelligence, vocabulary, and verbal learning. After we took into account the significant effects of age, gender, demographics, and recruitment cohorts, we found that approach to death and dropout caused strikingly similar reductions in mean test scores and amounts of practice gains between successive quadrennial testing sessions. Participants who neither dropped out nor died showed significant but slight cognitive declines. These analyses illustrate how neglect of dropout miscalculates effects of death, of worsening health, and of all other factors affecting rates of cognitive change.

Sudden declines in intelligence in old age predict death and dropout from longitudinal studies.

It is well known that approaching death accelerates cognitive decline. The converse issue, that is, the question of whether rapid declines in cognitive ability are risk factors for imminent death, has not been investigated. Every 4 years between 1983 and 2003, we gave 1,414 healthy community residents who were aged between 49 and 93 years the Heim AH4-1 test of fluid intelligence. A modified Andersen-Gill model evaluated AH4-1 scores at entry to the study and changes in scores between successive quadrennial test sessions as risk factors for death and dropout. Deaths, dropouts, age, gender, occupational categories, and recruitment cohorts were also taken into account. Participants with lower AH4-1 scores on entry were significantly more likely to die or to drop out. At all ages and levels of baseline intelligence, the risks of deaths and dropouts further increased if test scores fell by 10%, and again increased if they fell by 20% during 4-year intervals between successive assessments.

Age and ability affect practice gains in longitudinal studies of cognitive change.

During a 20-year longitudinal study, 5,842 participants aged 49 to 93 years significantly improved over two to four successive experiences of the Heim AH4-1 intelligence test (first published in 1970), even with between-test intervals of 4 years and longer. After we considered significant attrition by death and dropout and the effects of gender, socioeconomic advantage, and recruitment cohort, we found that participants with high intelligence test scores showed greater improvement than did those with lower intelligence test scores. Practice gains also reduced with age, even after we took into consideration the individual differences in intelligence test scores. This emphasizes the methodological point that neglect of individual differences in improvement during longitudinal studies underestimates age-related changes in younger and more able participants and the theoretical point that, like all experiences during everyday life, participation in longitudinal studies alters the ability of aging humans to cope with cognitive demands to different extents according to their baseline abilities.

Age-associated losses of brain volume predict longitudinal cognitive declines over 8 to 20 years.

Absolute differences in global brain volume predict differences in cognitive ability among healthy older adults. However, absolute differences confound lifelong differences in brain size with amounts of age-related shrinkage. Measurements of cerebrospinal fluid (CSF) volume were made to estimate age-related shrinkage in 93 healthy volunteers aged 63 to 86 years. Their current levels of brain shrinkage predicted their amounts of decline over the previous 8 to 20 years on repeated assessments during a longitudinal study on the Cattell "Culture Fair" Intelligence Test, on two tests of information processing speed, and marginally on the Wechsler Adult Intelligence Scale (D. Wechsler, 1981), but not on three memory tests. Loss of brain volume is an effective marker both for current cognitive status and for amounts and rates of previous age-related cognitive losses.

Effects of global atrophy, white matter lesions, and cerebral blood flow on age-related changes in speed, memory, intelligence, vocabulary, and frontal function.

Brain images were obtained from 133 healthy people of ages 61-85 years who completed 20 tests of information processing speed, intelligence, frontal and executive function, memory, and vocabulary. Structural equation models examined relationships between cognitive test scores, ages and measurements of global age-associated atrophy, white matter lesions, and cerebral blood flow. These neurophysiological measures jointly account for all age-related variance in information processing speed. Speed entirely mediated relationships between neurophysiological measures and memory and partly mediated relationships between neurophysiological measures and intelligence and frontal function. Neurophysiological measures, but not calendar age, accounted for vocabulary scores. Cognitive slowing was responsible for some, but not all, age-related declines in mental function. Age-related declines in intelligence, frontal function, and speed were due to changes in different functional systems.

White matter lesions account for all age-related declines in speed but not in intelligence.

MRI scans measured white matter lesion prevalence (WMLP) in 65 people ages 65-84 years who also took 17 cognitive tests: 3 tests of general fluid intelligence, 3 of vocabulary, 2 of episodic and 3 of working memory, 2 of processing speed, and 4 of frontal and executive function. Entry of age with WMLP into regression equations as predictors of test scores showed that inferences about the functional relationships between markers of brain aging and cognitive impairments are seriously misleading if they are based on simple correlations alone. A new finding that WMLP accounts for all of the age-related variance between individuals in tests of speed and executive ability but for none of the age-related variance in intelligence revises current hypotheses that gross brain changes affect general fluid intelligence and other mental abilities solely through their effects on information-processing speed.

Losses in gross brain volume and cerebral blood flow account for age-related differences in speed but not in fluid intelligence.

Age-related gross head size; adjusted age-related change in brain volume and carotid and basilar blood flow; as well as scores on 3 tests of fluid intelligence (gf), 2 tests of information-processing speed, 2 memory tests, and 3 tests of executive function were obtained from 69 volunteers aged from 62 to 84 years. Brain volume negatively predicted scores on all 10 cognitive tasks, accounting for up to 78% of age-related variance in scores on the speed tasks and on 1 executive task. Cerebral blood flow (CBF) negatively predicted scores on 8 cognitive tasks, accounting for up to 36% of age-related variance in speed scores. However, neither brain volume nor CBF accounted for significant age-related variance between individuals on any of 3 gf tests. We conclude that speed, but not gf, is an exceptionally sensitive behavioral index of the progress of gross brain changes that affect cognition in old age and that speed and gf do not reflect integrity of the same functional systems.