Cognition-Mortality Associations Are More Pronounced When Estimated Jointly in Longitudinal and Time-to-Event Models (original) (raw)
Related papers
Joint modelling of survival and cognitive decline in the Australian Longitudinal Study of Ageing
Journal of the Royal Statistical Society Series C, 2011
The paper describes the use of a longitudinal tobit model to characterize cognitive decline over a 13-year period in a cohort of 2087 elderly Australians. Use of a tobit formulation allows for the so-called 'ceiling effect' wherein many subjects achieve perfect test scores. A Bayesian hierarchical joint model is presented that allows for random subject-specific intercepts and slopes, as well as for informative dropout. Results suggest several potential areas of intervention. For example, there is a clear dose-response effect of exercise whereby increasing levels of exercise are associated with higher cognitive scores.
Intelligence, 2009
The association between cognitive function and mortality is of increasing interest. We followed 1870 men aged 55-69 years at cognitive assessment for 16 years to establish associations with all case and cause specific mortality. Cognitive assessment included AH4, 4 choice reaction time (used as estimates of mid-life cognition) and the National Adult Reading Test (used as an estimate of early-life cognition). Causal models were tested for the effects of a) early-life cognition, b) confounding through mid-life disease, and c) the effects of sociodemographic and lifestyle factors. A fully adjusted model was also tested. Age adjusted associations with mid-life cognitive function were found with mortality from circulatory, coronary, respiratory and digestive disease but not from cancer mortality. Age adjusted associations were attenuated and in some cases nullified by further adjustment for each of early-life cognition, mid-life disease risk and sociodemographic and lifestyle factors. These associations cannot be assumed to be unbiased estimates of effect due to the complex confounding structures that exist in these data. Future studies should explore natural experiments, use different populations where the confounding structures may be different and evaluate more complex methods that may be able to deal with the inherent complexities of a life course perspective.
Cognitive lifestyle jointly predicts longitudinal cognitive decline and mortality risk
European Journal of Epidemiology, 2014
Cognitive lifestyle measures such as education, occupation, and social engagement are commonly associated with late-life cognitive ability although their associations with cognitive decline tend to be mixed. However, longitudinal analyses of cognition rarely account for death and dropout, measurement error of the cognitive phenotype, and differing trajectories for different population subgroups. This paper applies a joint latent class mixed model (and a multi-state model in a sensitivity analysis) that accounts for these issues to a large (n = 3,653), population-based cohort, Paquid, to model the relationship between cognitive lifestyle and cognitive decline. Cognition was assessed over a 20-year period using the Mini-Mental State Examination. Three cognitive lifestyle variables were assessed: education, mid-life occupation, and late-life social engagement. The analysis identified four latent sub-populations with class-specific longitudinal cognitive decline and mortality risk. Irrespective of the cognitive trajectory, increased social engagement was associated with a decreased mortality risk. High education was associated with the most favourable cognitive trajectory, and after adjusting for cognitive decline, with an increased mortality risk. Mid-life occupational complexity was also associated with more favourable trajectories but not with mortality risk. To realistically examine the link between cognitive lifestyle and cognitive decline, complex statistical models are required. This paper applies and compares in a sensitivity analysis two such models, and shows education to be linked to a compression of cognitive morbidity irrespective of cognitive trajectory. Furthermore, a potentially modifiable variable, late-life social engagement is associated with a decreased mortality risk in all of the population sub-groups.
Cognitive Functioning and Health as Determinants of Mortality in an Older Population
American Journal of Epidemiology, 1999
The authors studied whether the ability of cognitive functioning to predict mortality is pervasive or specific, and they considered the role of health in the cognition-mortality association. Data were taken from a sample of 2,380 persons aged 55-85 years who took part in the Netherlands' Longitudinal Aging Study Amsterdam in 1992-1993. Five cognitive measures were distinguished: general cognitive functioning, information processing speed, fluid intelligence, learning, and proportion retained. Mortality data were obtained during an average follow-up period of 1,215 days. Cox proportional hazards regression models revealed that all cognitive functions predicted mortality independent of age, sex, education, and depressive symptoms. When health (self-rated health, medication use, physical performance, functional limitations, lung function, specific chronic diseases) was also taken into account, information processing speed, fluid intelligence, and proportion retained remained independent predictors of mortality, whereas the ability of general cognitive functioning and learning to determine mortality was lost. The authors concluded that the ability of cognitive functioning to predict mortality is pervasive to all cognitive functions that were included in the study when age, sex, education, and depressive symptoms are considered and is more specific to some functions when also controlling for health. Am J Epidemiol 1999; 150:978-86.
Joint longitudinal-survival models are useful when repeated measures and event time data are available and possibly associated. The application of this joint model in aging research is relatively rare, albeit particularly useful, when there is the potential for nonrandom dropout. In this article we illustrate the method and discuss some issues that may arise when fitting joint models of this type. Using prose recall scores from the Swedish OCTO-Twin Longitudinal Study of Aging, we fitted a joint longitudinal-survival model to investigate the association between risk of mortality and individual differences in rates of change in memory. A model describing change in memory scores as following an accelerating decline trajectory and a Weibull survival model was identified as the best fitting. This model adjusted for random effects representing individual variation in initial memory performance and change in rate of decline as linking terms between the longitudinal and survival models. Me...
European journal of epidemiology, 2018
Despite several studies demonstrating an independent and inverse association between cognition and mortality, the nature of this association still remains unclear. To examine the association of cognition and mortality after accounting for sociodemographic, health and lifestyle factors and to explore both test and population characteristics influencing this relationship. In a population based cohort of 8585 men and women aged 48-92 years, who had cognitive assessments in 2006-2011 and were followed up till 2016 for mortality, we examined the relationship between individual cognitive tests as well as a global cognition score to compare their ability in predicting mortality and whether these differed by population characteristics. Risk of death was estimated using Cox proportional hazard regression models including sociodemographic, lifestyle and health variables, and self-reported comorbidities, as covariates in the models. Poor cognitive performance (bottom quartile of combined cogni...
Alzheimer's & dementia : the journal of the Alzheimer's Association, 2018
This study examines the role of educational attainment, an indicator of cognitive reserve, on transitions in later life between cognitive states (normal Mini-Mental State Examination (MMSE), mild MMSE impairment, and severe MMSE impairment) and death. Analysis of six international longitudinal studies was performed using a coordinated approach. Multistate survival models were used to estimate the transition patterns via different cognitive states. Life expectancies were estimated. Across most studies, a higher level of education was associated with a lower risk of transitioning from normal MMSE to mild MMSE impairment but was not associated with other transitions. Those with higher levels of education and socioeconomic status had longer nonimpaired life expectancies. This study highlights the importance of education in later life and that early life experiences can delay later compromised cognitive health. This study also demonstrates the feasibility and benefit in conducting coordi...
Methodological Issues in the Study of Cognitive Decline
American Journal of Epidemiology, 1999
In recent years, epidemiologists have given increased attention to cognition, especially to the dementing illnesses that occur in old age. Central to this study is the measurement of change in cognition as opposed to cognition measured at a single point in time. This article addresses conceptual and methodological issues in the study of changes in cognitive function, including: 1) difficulties encountered with the use of single measurements of cognition and the importance of measuring changes in cognition; 2) sources of measurement variation and its potential effects; 3) the importance of careful modeling of age and education; 4) considerations in categorizing outcomes or combining the results of cognitive tests; and 5) the benefits of using multiple-outcome statistical models. Am J Epidemiol 1999;149:789-93.
Alzheimer's Research & Therapy, 2015
Introduction: Many factors influence late-life cognitive changes, and evaluating their joint impact is challenging. Typical approaches focus on average decline and a small number of factors. We used multistate transition models and index variables to look at changes in cognition in relation to frailty (accumulation of health deficits), social vulnerability, and protective factors in the Honolulu-Asia Aging Study (HAAS). Methods: The HAAS is a prospective cohort study of 3,845 men of Japanese descent, aged 71 to 93 years at baseline. Cognitive function was measured using the Cognitive Abilities Screening Instrument (CASI). Baseline index variables were constructed of health deficits (frailty), social vulnerabilities, and protective factors. The chances of improvement/ stability/decline in cognitive function and death were simultaneously estimated using multistate transition modeling for 3-and 6-year transitions from baseline. Results: On average, CASI scores declined by 5.3 points (standard deviation (SD) = 10.0) over 3 years and 9.5 points (SD = 13.9) over 6 years. After adjusting for education and age, baseline frailty was associated with an increased risk of cognitive decline at 3 years (β = 0.18, 95% confidence interval (CI), 0.08 to 0.29) and 6 years (β = 0.40, 95% CI, 0.27 to 0.54). The social vulnerability index was associated with 3-year changes (β = 0.16, 95% CI, 0.09 to 0.23) and 6-year changes (β = 0.14, 95% CI, 0.05 to 0.24) in CASI scores. The protective index was associated with reductions in cognitive decline over the two intervals (3-year: β = −0.16, 95% CI, −0.24 to −0.09; 6-year: β = −0.21, 95% CI, −0.31 to -0.11,). Conclusions: Research on cognition in late life needs to consider overall health, the accumulation of protective factors, and the dynamics of cognitive change. Index variables and multistate transition models can enhance understanding of the multifactorial nature of late-life changes in cognition.
The Journals of Gerontology: Series B, 2022
Objectives: To better understand the temporal dynamics of progression from cognitive decline to onset of dementia in the dementia-free older population in the United States. Methods: We used longitudinal data from a diverse national population-based sample of older adults (N = 531) in the Aging, Demographics, and Memory Study from the Health and Retirement Study with repeated measures of cognitive function and dementia diagnosis during 12 years of follow-up from 1996 to 2009. We employed joint latent class mixed models to estimate the association between cognitive change and competing risks of dementia and nondementia death and identify heterogeneity in the age profiles of such association adjusting for baseline characteristics. Results: Our analyses found 3 latent classes with distinct age profiles of cognitive decline and associated risk of dementia and mortality: "Rapid Cognitive Decline" (19.6%), "Moderate Progression" (44.6%), and "Optimal Cognitive Aging" (35.8%). When simultaneously accounting for cognitive trajectories and time-to-dementia/death, we also found associations of baseline covariates with slope of cognitive decline (e.g., steeper decline among non-Hispanic Blacks and more educated) and risk of dementia (e.g., greater risk for females and apolipoprotein E-4 carriers, but no difference by education level) that differ substantially from those in separate longitudinal mixed models or survival models. Discussion: The differential age patterns of cognitive decline predicting dementia incidences identified in this study suggest variation in the course of cognitive aging in older adults that may inform future etiological and intervention studies.