HIV Articles  
Aging with HIV; Early Cognitive Decline
  from Jules: Aging with HIV is the major theme that emerged this year in the western world. The CDC reported at CROI 08 that in the last few years there has been an increase of 30% in the number of HIV+ individuals over 50 yrs old. This trend will continue. Since HIV+ individuals are at greater risk for diseases associated with aging they are perhaps at greater risk for steeper declines in mental functioning. The authors of this study refer to cardiovascular disease and frailty and lack of exercise that may accelerate declines in mental skills associated with aging. I add to this list bone disease, diabetes, and neurological complications that may be associated with HIV. The study below reports cognitive decline can begin as much as 15 years before death for persons in the study. The study of cognitive decline in the context of neurological complications and other aging-associated diseases for HIV+ individuals would be an important subject to begin thinking about and incorporating into studies. The concerns regarding Aging with HIV need to be addressed by our research community.
Cognitive Decline May Begin 15 Years Before Death

Posted by Jane Akre
Thursday, August 28, 2008
A decline in verbal skills may indicate that an age-related cognitive decline is underway.
A new study reports that key mental skills drop by as much as 15 years before death, even without dementia. The skills in decline include verbal ability, spatial reasoning and perceptual speed.
This is not part of the aging process.
So why are we losing abilities? Researchers from Goteborg University in Sweden, credit undetected dementia, insufficient physical and mental exercise, and perhaps early heart disease that interrupt blood flow to the brain.
Valgeir Thorvaldsson, from the department of psychology at Goteborg University, in Sweden said, "[And] our findings clearly showed there to be a pattern of terminal decline, even among relatively healthy individuals, that the brain changes that influence our cognitive abilities in old age occur over a relatively long period of time, even among individuals who remain non-demented until they die."
The researchers said they were surprised at how far in advance of death the decline begins.
Verbal ability started a sharply accelerated decline more than six years prior to death, while it's 7.8 years for spatial ability.
However, perceptual speed, the ability to quickly compare figures, begins to decline as much as 15 years before death.
In the study, 288 people without dementia were followed from age 70 to their death, which occurred at an average age of 84. Their mental skills were measured up to a dozen times over 30 years.
On a more positive note, the study finds that the elderly remain stable in their verbal abilities until they are burdened by diseases.
"A change in verbal ability might therefore be considered a critical marker for degeneration in health in older people," Thorvaldsson said.
The peak of mental functioning is reached for most people between the ages of 35 to 40 after which a steady decline begins, then speeds up before death. The Swedish team was trying to determine when the acceleration starts.
The research may help those who work with the elderly derive markers when trying to determine a degeneration of mental health.
Age-related cognitive decline has long been evaluated along with dementia.
The value of the study may be the distinction it makes between the normal aging process and the dying process.
The work is presented to today's issue of Neurology.
The changes may not be inevitable.
Evidence suggests among those who participate in regular strenuous aerobic activity, that there is four times less shrinkage in the brain when compared to those who do not exercise.
The average U.S. life expectancy is now 78 years up 30 years since 1900 and 10 years since 1950. And while aging fears are often used in the hiring of corporate America some believe those fears are misplaced.
A 45-year-old and a 75-year-old "absolutely" have the same mental capacity, and energy is a function of health rather than aging, said Neil Resnick, chief of geriatric medicine at the University of Pittsburgh tells the Wall Street Journal.
But much of our genealogy influences how fact we decline. John Rowe, a physician and former Aetna Inc. chair tells the Wall Street Journal that our genes influence how much and how fast we decline. He says genes account for about 30 percent of longevity and perhaps half of age-related changes in the brain.
The implications for U.S. society are huge. The over-65 population will double to 80 million in 30 years when one in five will be older than 65, up from one in eight now.

Onset of terminal decline in cognitive abilities in individuals without dementia
Neurology 2008 Aug 27 2008
V. Thorvaldsson MSc*, S. M. Hofer PhD, S. Berg PhD, I. Skoog MD, PhD, S. Sacuiu MD, and B. Johansson PhD
From the Department of Psychology (V.T., B.J.), and Institute of Neuroscience and Physiology, Sahlgrenska Academy (I.S., S.S.), Goteborg University, Sweden; Department of Human Development and Family Sciences (S.M.H.), Oregon State University, Corvallis; and Institute of Gerontology (S.B.), School of Health Sciences, Jonkoping University, Sweden.
"Our purpose was to examine terminal decline in three cognitive abilities in an age- homogeneous, population-based sample without dementia, followed from age 70 to the end of life. The main finding was that accelerated cognitive change, indicating a terminal decline phase, occurred on average 6.5 years prior to death for verbal ability, about 8 years for spatial ability, and almost 15 years before death for perceptual speed. In all cognitive abilities, we found a substantial acceleration in decline in the terminal decline phase relative to earlier age-related within-person change....
..... There are a number of interacting factors that may explain these findings of terminal decline. Cardiovascular conditions (e.g., heart diseases) and undetected preclinical dementia (e.g., Alzheimer disease) are two such candidates. Increased medical burden and frailty in old age often further leads to inactivity (i.e., lack of physical exercise and cognitive stimulation) that can accelerate cognitive decline."

Objective: To identify time of onset and rate of mortality-related change (terminal decline) in cognitive abilities in later life.
Method: The sample consisted of 288 individuals without dementia (born 1901-1902) drawn from the population of Goteborg, Sweden. Participants were followed from age 70 until death, with up to 12 measurement occasions on three cognitive abilities. Change-point analysis was performed using an automated piecewise linear mixed modeling approach to identify the inflection point indicating accelerated within-person change related to mortality. A profile likelihood method was used to identify the change point that best fit the data for each of three cognitive abilities.
Results: Onset of terminal decline was identified 6.6 years prior to death for verbal ability, 7.8 years for spatial ability, and 14.8 years for perceptual speed.
Conclusions: There is substantial acceleration in cognitive decline many years prior to death among individuals without dementia. Time of onset and rate of terminal decline vary considerably across cognitive abilities.
Terminal decline in cognition refers to acceleration in within-person change prior to death and is distinct from, but possibly moderated by, aging-related changes.1,2 Several studies have demonstrated terminal decline by comparing survivals and nonsurvivals but relatively few longitudinal studies have identified terminal decline at the level of the individual with complete information about age at death in a population-based representative sample.3,4 Little is therefore known about time of onset of terminal decline, rate of change, and whether terminal decline varies across cognitive abilities.
We are aware of two studies that have identified terminal decline as acceleration in within-person change before death and relative to prior within-person age-related change. One study5 reported a terminal decline period ranging from 2.75 years to 6 years before death on various cognitive abilities based on data from a sample of Roman Catholic nuns, priests, and brothers. Another study4 reported a longer terminal decline period ranging over 8.4 years in a measure of episodic memory based on data from a sample of community-based North Americans. Explanations for the discrepancy between these findings in terms of the length, and magnitude, of the terminal decline phase is somewhat unclear but is likely related to different sample compositions, testing instruments, and design (i.e., differences in length of the follow-ups).
Dementia, including its preclinical phase, has been shown in longitudinal studies to be a major contributor for age-related heterogeneity in cognitive change. Most studies of terminal decline, however, exclude individuals with dementia only at the initial assessment,6,7 leaving estimates of cognitive change affected by a subsequent incidence of dementia. To our knowledge, little is known about terminal decline in populations that remain without dementia until death. The aim of this study is to examine time of onset (i.e., change point) and rate of change in terminal decline across three distinct cognitive abilities (i.e., verbal ability, spatial ability, and perceptual speed). The change point for terminal decline is identified within the context of two time-dependent change processes-age and proximity to death-in a population-based age-homogeneous sample of individuals without dementia followed from age 70 to death.
Our purpose was to examine terminal decline in three cognitive abilities in an age- homogeneous, population-based sample without dementia, followed from age 70 to the end of life. The main finding was that accelerated cognitive change, indicating a terminal decline phase, occurred on average 6.5 years prior to death for verbal ability, about 8 years for spatial ability, and almost 15 years before death for perceptual speed. In all cognitive abilities, we found a substantial acceleration in decline in the terminal decline phase relative to earlier age-related within-person change
The terminal decline phases in our study were considerably longer than the estimates from the two other studies reporting onset of terminal decline. One study5 found evidence for terminal decline 3.33 years before death in semantic memory, 6 years in visuospatial ability, and only 2.75 years in perceptual speed. The other study4 found a terminal decline phase on episodic memory ranging over 8.4 years before death. Our findings for both verbal ability and spatial ability are similar to this estimate, but our estimate for perceptual speed is much longer. Measures of processing speed are typically one of the most sensitive markers of age-related change and between-person age differences. Our finding that speed was the earliest marker of mortality-related decline is in line with these findings. Otherwise, we can only speculate about the various reasons that might contribute to the discrepancy between the present and previous findings but most likely the explanation is related to differences in health-related characteristics of the samples and length of follow-ups.
The steep average estimate of verbal ability change in the terminal decline period, along with the nonsignificant age slope, and consequently relatively large terminal decline effects, indicates that individuals remain stable in this ability unless affected by mortality-related pathology. Change in verbal ability might therefore be considered a relatively robust indicator of mortality-related processes in old age and might consequently be considered as a critical marker for compromised health in older individuals. The terminal decline effect in spatial ability and perceptual speed was also relatively large. For example, in spatial ability this estimate was 90% of the average age-based change and in perceptual speed 100% of the average age-based change. Note, however, the substantial individual differences in the effects of terminal decline in verbal and spatial ability as indicated by the significant estimates of death slope random effects.
One contribution of this study is that it allows evaluation of terminal decline in a population-based sample examined over 30 years on three cognitive abilities. Information about age at death is available for all deceased individuals (i.e., 99% of the sample). However, as in all studies, there are several shortcomings to be addressed. One relates to a possible lack of generalizability of findings in an age-homogenous sample.15 That is, other birth cohorts may differ in onset and rate of terminal decline. Age-homogeneity can, however, also be considered as a major strength as it reduces potential bias due to cohort differences in estimates of between-person differences in both aging and mortality-related change. Another potential shortcoming relates to the relatively large measurement occasion intervals in the study, ranging from 1 to 5 years. This might reduce sensitivity in identification of the change points. Retest effects might present another shortcoming.16 Analyses of this issue suggest small retest effects on verbal ability and spatial ability but not on perceptual speed.17 Initial and subsequent attrition (not due to mortality) from the H70 study is relatively small.8,10
The left censoring of data is an additional issue that should be addressed. In verbal ability and spatial ability this is less of a problem because the majority (i.e., 59% and 61%) of the sample is contributing to both sides of the terminal change point. In perceptual speed, however, this might present problems because only 41% of the sample is contributing to both sides of the change point, a fact that encourages a careful interpretation. There are few studies18 that would allow a within-person identification of terminal decline change points over a period ranging over 14 years. Although there might be bias in the estimate of the change point in perceptual speed we are inclined to interpret our findings in terms of a substantially longer terminal decline period for speed compared with that for verbal and spatial ability.
There are a number of interacting factors that may explain these findings of terminal decline. Cardiovascular conditions (e.g., heart diseases) and undetected preclinical dementia (e.g., Alzheimer disease) are two such candidates. Increased medical burden and frailty in old age often further leads to inactivity (i.e., lack of physical exercise and cognitive stimulation) that can accelerate cognitive decline. Subsequent studies need to illuminate these issues and identify the neurobiological markers for terminal decline. The analytical framework that we used in the present analyses provides a general framework to test hypotheses underlying terminal decline as it separates between-person differences in age-related change and terminal decline at the level of the individual. Time-constant or time-varying covariates can also be included on either slope and thereby obtain separate estimates of the determinants for age-related and terminal decline trajectories.
There is increasing awareness of the need to account for population mortality selection in analysis of aging-related change in later life. Changes in cognition are related to changes in health status and mortality with a pronounced decline typically observed in individuals who are more likely to die, producing a selection effect in the population. This can contribute to a bias in an age-based description of the population trend and obscure identification of between-person differences in within-person change if analyses are not conducted as conditional on subsequent mortality. The present findings strongly support this notion by demonstrating that cognitive change in a sample that remains without dementia until death is related to survival.

The study is part of the Gerontological and Geriatric Population Studies in GoNteborg, Sweden (H70).8,9 The sample consisted of 482 participants, systematically selected at age 70 in 1971 from the total population in GoNteborg born June 1, 1901- June 1, 1902. A total of 70 individuals declined participation (leaving 86% response rate), 10 individuals had diagnosed dementia at baseline, and 114 were diagnosed with dementia over the course of the study. The remaining sample of patients without dementia consisted of 288 participants (women=162, men =126) measured at ages 70, 75, 79, 81, 85, 88, 90, 92, 95, 97, 99, and 100 with occasional dropout. At each occasion between ages 70 to 85 diagnosis of dementia was based on information about severe disorientation for time and place or a longstanding severe memory impairment as measured by rating scales and information from case records or relatives. After age 85, the diagnoses were based using the DSM-III-R with information obtained from psychiatric examinations and interview with relatives (further information about the diagnosis procedures is reported elsewhere10,11). Confirmed date of death was available for 284 individuals and 4 individuals were still living at the last update on June 1, 2003, at the approximate age of 101. In order to prevent missing data on the age at death variable we imputed the date of last update as the date of death for these four individuals (i.e., 101 years). The average age at death was 83.9 years (SD =8.3).
Cognitive measurements. We used three cognitive tests from the Dureman and Salde12 psychometric test battery which was widely used in Sweden at the time of the original start of the H70 study. The Synonyms test was used throughout all measurements occasions, while the Block Design test was omitted from the testing at age 81 and the Figure Identification test was omitted from testing at age 81 and age 100. Otherwise, the same testing procedure was conducted throughout all measurement occasions. Further information about the testing procedures and the statistical properties of the tests can be found elsewhere.13 Raw score trajectories for each individual and cognitive test used in the analyses, plotted as a conditional function of mortality, can be found as supplemental data (figures e-1- e-3 on the Neurology Web site at All test scores were standardized to a pseudo T-score metric with mean of 50 and SD of 10 in order to facilitate comparability across cognitive domains. We used mean and SD at age 70 (i.e., first occasion of measurement) for each test as a base for the standardization procedure.
The Synonyms test is a measure of verbal ability, or more generally, a test of an individual's ability to understand ideas expressed in words. Participants match a target word with one synonym among five choices. The test has a time limit of 7 minutes and a maximum score of 30 points. The words were presented in a magnified form if participants had visual impairments. Block Design test is a measure of spatial ability where participants are presented with two-color blocks and asked to construct replicas of a prototype model design which have been presented to the participant. The test has a time limit of 20 minutes and a maximum score of 42 points. Figure Identification is a test of perceptual speed where participants match, as quickly as possible, a target figure with one identical figure that is placed in line among five figures. The test has a time limit of 4 minutes and a maximum score of 60 points.
Analyses. We used a series of mixed linear models with a change point in order to determine an average time point when rate of decline in the three examined cognitive abilities starts to accelerate prior to death. We varied the location of the change point by 0.01 year increment up to a maximum of 20 years before death and used a profile-likelihood to select the best fitting model.14 Prior to the estimated change point, we structured individual differences in change as a conditional function of chronological age (centered at baseline) but after that time point we structured changes as a conditional function of time to death. Fixed effects in these models refer to the estimated average performances and changes. The random effects refer to the variability components. We estimated parameters in the models with full maximum likelihood. Detailed description of this modeling procedure can be found elsewhere.4
In figures 1-3, the estimated likelihood functions for each cognitive test are plotted against time before death with 0.1 year increment. The best fitting model determines the maximum likelihood estimates of the terminal decline change point as relative to age-related change. The reference lines in the figures indicate a 95% CI where estimates above the line fall within the interval while estimates below the line fall outside the interval. The estimated change points before death for verbal ability (Synonyms) was 6.58 (4.33, 11.67) years, for spatial ability (Block Design) 7.83 (6.25, 10.58), and for perceptual speed (Figure Identification) 14.83 (10.75, 16.58). These figures indicate the estimated average onset of terminal decline on the cognitive abilities in the population.
Estimates for the best fitting models are shown in the table. The fixed effects intercepts refer to an average standardized performance at age 70. The fixed effects age slopes show the estimated annual average change in the three cognitive abilities before the terminal decline change point. This estimate is nonsignificant for verbal ability indicating no average change until 6.58 years prior to death. For spatial ability the age slope indicates a significant average annual decline by 0.031 standard deviations until 7.83 years prior to death. For the perceptual speed the age slope indicates a significant average annual decline by 0.022 standard deviations until 14.83 years prior to death.
The fixed effects death slopes indicate the average acceleration in change on the cognitive abilities after the estimated terminal decline change point. Estimates for mortality slopes were significant across cognitive abilities. On verbal ability the death slope indicates an average annual decline of 0.056 standard deviations within 6.58 years before death. For spatial ability the death slope estimate was higher with an average annual decline of 0.090 standard deviations within 7.83 years before death. On perceptual speed the death slope estimate indicated an average annual decline of 0.066 standard deviations within 14.83 years before death. The fixed effects based on these models are plotted in figure 4 and demonstrate the onset of the terminal decline period and the average rate of change within each period on the cognitive outcomes. Note that for each of the cognitive abilities the estimated average change within the terminal decline period is indicated by the sum of the age and death slopes. In subsequent analyses, we included a fixed quadratic term for the death slopes. This estimate was occasionally significant but produced only minor differences in the estimate of the change point. In order to facilitate the comparability across the domains on both the fixed and random effects we retained the linear model assumption.
The differences between the age and death slopes provide an estimate of the additional effect of terminal decline. Due to the relatively low estimate in the age slope, this estimate was particularly large in verbal ability or -0.48 (p <0.001) where the post change point slope was 13 times larger than the age slope. In spatial ability the terminal decline effects estimate was -0.28 (p <0.001) and the post change slope was almost twice as large as the age slope. In perceptual speed the terminal decline effects estimate was -0.22 (p <0.001) and the post change slope was twice as large as the age slope. These findings demonstrate that the effect of terminal decline is large relative to age-related change on all cognitive abilities. Furthermore, there is a substantial proportional difference in change across abilities that is partly due to variability in age-related change and partly due to terminal decline.
The lower part of the table shows the random effects (i.e., the variability and co-variability components) for the intercept, age slope, and death slope. All variability components (shown in the diagonals) are significant (p <0.01 determined by the likelihood ratio test) except for the death slope in perceptual speed that had a p value of 0.07. The co-variability among age slope and death slope was nonsignificant for all three cognitive abilities indicating that the rate of change before the terminal decline change point was unrelated to the rate of change after the change point. Co-variability between intercept and death slope were significant for verbal ability (r= -0.42, p <0.05) and spatial ability (r= -0.78, p <0.001) indicating that individuals who declined faster in the terminal phase generally had a higher estimate at baseline. Co-variability between intercept and the age slope was significant for perceptual speed (r = -0.40, p <0.001) indicating that individuals who decline faster before the terminal phase generally show higher baseline estimate. These estimates must, however, be interpreted with caution and may be influenced by left-censoring (unknown information about cognitive performance prior to baseline) of individuals already in a decline phase at baseline and more reliable measurements in the higher range of cognitive functioning. Other co-variability estimates were nonsignificant.
  iconpaperstack View older Articles   Back to Top