The Relationship Between Age and Cognitive Function in HIV-Infected Men

The data were analyzed using a 4×2 Analysis of Variance (Disease Status by Age group). As mentioned above, this analysis revealed a significant effect of age on education (F=5.28, df=1, 234, p=0.022). It was therefore essential to include education as a covariate in our analyses in order to examine the interaction effect of age and disease level on cognitive performance independent of this confounding variable.

The initial analysis of neuropsychological performance was based on an overall impairment score that was calculated by the total number of measures on which an individual achieved a score that was one SD below the mean of the comparison group. As presented in Table 2 and Table 3, after controlling for education, there were significant effects for age group (F=15.84, df=1, 245, p<0.001) and HIV status (F=6.09, df=3, 245, p=0.001) on the overall measure of impairment, but there was not a significant interaction effect (F=0.273, df=3, 245, p=0.844).

In addition to analyzing overall cognitive impairment, it was of interest to determine whether any specific areas of neuropsychological performance were impacted by the interaction of age and disease status. This was accomplished by a series of ANCOVAs, controlling for the effect of education. As can be seen in Table 3, while results of these analyses revealed several significant main effects for both age group and HIV status, respectively, there were no significant interaction effects on any of the neuropsychological measures. Since the cell size in the older subgroups was considerably smaller than in the younger groups, it was essential to address the potential impact on these analyses. Analyses of homogeneity of variance (Levene’s test) demonstrated no significant differences in variances among the age X HIV status subgroups, which suggests that the lack of differences between group means cannot be attributed to cell size. Furthermore, the data were reanalyzed after combining the two symptomatic subgroups to increase sample size. These reanalyses again demonstrated significant effects for HIV status and age but again failed to demonstrate any significant interaction effects.

In addition, it was recognized that the criterion for defining the older age group was different than that used in other studies (45 years of age versus 50 years of age in other studies).^8–9 Regardless of the criterion used, however, this is a relatively arbitrary distinction. Therefore, to avoid the problems inherent with such an arbitrary criterion, it was decided to directly examine the relationship between aging and cognitive performance in the HIV-positive and HIV-negative subgroups. Correlation coefficients (controlling for the effects of education) were computed separately, and the significance of the difference between correlations was calculated using Fisher’s Z-transformation. This allowed an analysis of all available data. It was hypothesized that an interaction between HIV stage and age would result in significantly different correlations between age and performance in the comparison and HIV-positive subjects. The results of these analyses failed to demonstrate any significant differences in these correlations.

These data fail to provide support for the hypothesis that advancing age is a risk factor for the development of HIV-related cognitive impairment. Age and disease status had independent effects on cognitive function, but there were no significant interactions either on a summary measure of performance or on individual test scores. Both age and disease status were shown to have effects on measures of executive function (VCAT and Trail Making B), memory (Selective Reminding – Delayed Recall), and dexterity (Grooved Pegboard). Disease stage also affected measures of reaction time, information processing speed, and fluency (verbal and nonverbal). The demonstrated effects of HIV are consistent with the characterization of HIV-associated cognitive impairments as a subcortical dementia.¹⁰ These results cannot be attributed to education, which was included as a covariate in all analyses. In addition, the groups did not differ regarding depression or estimates of weekly alcohol consumption. Therefore, the observed results cannot be attributed to these potentially confounding variables. Further, these results cannot be attributed to differences in cell size or variability as no significant differences were revealed when partial correlation coefficients between age and neuropsychological performance across disease stage were compared.

Although our results are consistent with previous studies that found both older age^4–7 and disease status to significantly impact cognitive impairment in HIV-seropositive subjects, this investigation did not find the hypothesized synergistic effect between aging and HIV serostatus when education was controlled. These findings are consistent with a previous study,⁸ which concluded that older seropositive individuals are not at an increased risk for HIV-related cognitive impairment when normal age-related cognitive changes are considered. The study included 76 subjects, 29 of which were over age 55 (range=21–69), and found a trend toward a significant age by serostatus interaction on only one measure. In contrast, another study⁹ found an age by serostatus interaction, in that AIDS patients over age 50 had higher rates of impairment than younger patients at earlier stages of disease progression. The study had a somewhat larger sample of older subjects and stratified subjects according to age (less than 40, 40–49, and greater than 50). The mean age of our older subjects (47 years) is comparable to the mean age of the older subjects in these studies (47.2 and 44.5), although the current study had only 25 subjects older than age 45 in the seropositive group. The latter⁹ study had a substantially larger subgroup of subjects older than age 50.

In addition to the current findings, data from the two previous investigations^8,9 help to identify critical factors that will be needed to resolve the question of an age-serostatus interaction. One study⁹ that demonstrated an age by serostatus interaction was based on a clinical sample, whereas the current study was based on a research cohort. In contrast, the study that utilized a less extensive test battery⁸ and/or relatively smaller numbers of older subjects did not find evidence of this interaction. Previous studies have stratified subjects according to age, although the criteria have been arbitrary and inconsistent across studies. When the impact of age was examined as a continuous variable in the current study, there was no difference in the relationship of age to cognitive function. Furthermore, the disease stratifications have been based on disease classification criteria rather than on virologically based criteria such as viral load. As the epidemic continues to evolve and advances in treatment continue, there will be an increased number of older individuals living with HIV and AIDS.² The generalizability of the current results is limited since the cohort was composed of primarily white homosexual/bisexual males. Other samples with greater gender and ethnic diversity should also be studied since differences in access to health care as well as differences in diagnostic recognition could influence the interaction of age and HIV serostatus. Future studies with larger cohorts of older subjects that employ comprehensive evaluations will be necessary in order to clearly identify this question. If such an interaction exists, it will be essential to determine whether there is an age or serostatus threshold at which this occurs.

This study was supported by grants from the NIMH (MH45649), the National Institute on Drug Abuse (DA10248), and the National Institute on Alcohol Abuse and Alcoholism (AA11720).