Authors: David C. Steffens, Kevin J. Manning, Rong Wu, James J. Grady
Categories: Article, Depression, Stress, Neuroticism, Memory
Source: The American journal of geriatric psychiatry : official journal of the American Association for Geriatric Psychiatry
Authors: David C. Steffens, Kevin J. Manning, Rong Wu, James J. Grady
The relationships among depression, personality factors, stress and cognitive decline in the elderly are complex. Depressed elders score higher in neuroticism than non-depressed older individuals. Independently, the presence of neuroticism and the number of stressful life events are each associated with worsening cognitive decline in depressed older adults. Yet little is known about combined effects of changes in neuroticism and changes in stress on cognitive decline among older depressed adults.
Longitudinal observational study.
Academic Health Center
We examined 62 participants in the Neurobiology of Late-life depression (NBOLD) study to test the hypothesis that, compared with older depressed subjects who experience improved neuroticism and lower psychosocial stressors over time, those with worsening neuroticism and greater psychosocial stressors will demonstrate more cognitive decline.
We measured neuroticism using the NEO-Personality Inventory-Revised at baseline and one year. Study psychiatrists measured depression using the Montgomery-Ǻsberg Depression Rating Scale. At annual assessments, subjects reported the number of psychosocial stressors in the prior year and completed a neuropsychological evaluation. Participants completed a detailed neuropsychological battery at baseline and annually over three years. The battery included a test of delayed story memory (Logical Memory −2 or LMII). The outcome three-year change in cognitive scores was regressed against three-year change scores of neuroticism and number of psychosocial stressors, plus their interaction, while adjusting for sex, age, race, education, baseline cognitive score, and three-year change in MADRS score as covariates.
In multivariable linear regression analysis with the above covariates, the interaction effect of three-year change in Total Neuroticism score and 3-year change in Total Stressors on change in LMII performance was statistically significant (B=−0.080[95%CL: −0.145 to −0.015], T = −2.48, df = 52, p = 0.017). Further exploration of this finding showed that 1) when total stressors increased by 2 or more over three years, LMII change was inversely associated with neuroticism change; and 2) when neuroticism improved less, LMII change score was inversely associated with total stressor change. There were no other significant interactions between stress and neuroticism on cognition.
Our findings document the importance of tracking change in neuroticism and monitoring psychosocial stress over the long-term course of treatment in geriatric depression. Both factors exert important combined effects on memory over time. Future studies in larger samples are needed to confirm our results and to extend them to examine both cognitive change and development of dementia.
The links between late-life depression (LLD), the personality trait of neuroticism, presence of stress and cognitive decline are complex, though some research undertaken in the last two decades has shed light on this area. LLD is associated with later cognitive decline and development of dementia, and presence of neuroticism in older depressed patients has been shown to increase risk of cognitive decline and dementia (1, 2). Similarly, greater number of stressful life events has also been associated with greater cognitive decline (3).
While traditional thinking holds that personality is stable over time, there is evidence that neuroticism may increase as older adults transition across the spectrum of pathological aging (4). In this vein, Caselli et al. found that an increase in neuroticism preceded the transition from cognitively normal to mild cognitive impairment (5). Mindfulness interventions may mitigate negative effects of neuroticism on cognition (6). Antidepressant treatment may also reduce neuroticism in the context of major depression(7). Another study suggested that perceived stress and hypothalamic-pituitary-adrenal axis dysregulation could act as mechanisms underlying the association between neuroticism and cognitive functioning and decline among older men (8). Thus, it may be important to include longitudinal measures of both neuroticism and stressors when examining predictors of cognitive decline. Indeed, changes in neuroticism and changes in stressors have been found to exert independent effects on cognitive decline (9, 10). However, little is known about the combined effects of changes in both neuroticism and stress on cognitive performance over time, especially among individuals with LLD.
In this study, we examined changes in neuroticism and stressors in older enrolled in the Neurobiology of Late-Life Depression (NBOLD) study (11). Specifically, we studied the effects of three-year change in neuroticism as measured on the NEO Personality Inventory and three-year change in reported stressors on measures of global cognitive performance, memory and executive dysfunction. We hypothesized a significant interaction effect on cognition between neuroticism and stress such that those with higher neuroticism over time and those with higher stressors over time would demonstrate worse cognitive performance over three years.
All subjects were enrolled in NBOLD, a NIMH funded study at the University of Connecticut Health Center approved by its Institutional Review Board. After reviewing study information, all subjects provided written, informed consent to participate.
Depressed subjects were recruited from clinic referrals and newspaper advertisements. Inclusion criteria for depressed subjects were age 60 or above, ability to read and write English, Mini-Mental State Examination (MMSE) score 25 or greater and meeting criteria for major depression, single episode or recurrent. Study exclusion criteria current or recent alcohol or drug dependence; conditions associated with brain MRI abnormalities; physical or intellectual disability that may affect completion of self-rating instruments; established clinical diagnosis of dementia; other major DSM Axis 1 psychiatric disorders; and metal or pacemaker in the body or claustrophobia that might preclude MRI. In addition, current treatment with fluoxetine was an exclusion criterion for the depressed group given its long wash-out period.
The screening and assessment procedures used in NBOLD have been reported previously(11). Briefly, participants were screened for depression using the Center for Epidemiologic Studies-Depression (CES-D) scale, using a score of 16 or greater as a cut-off for depression. Upon enrollment and completion of baseline assessments, each participant was paid $100 for their time completing the MRI, cognitive test battery and experimental computerized measures. The clinical assessment procedures are summarized below.
Trained clinical research assistants administered the Duke Depression Evaluation Schedule (DDES) to each participant via computer-assisted data entry. The DDES contains items covering demographic data, social variables, and the Diagnostic Interview Schedule (DIS) sections for depression, mania, generalized anxiety disorder, somatization symptoms, and alcohol use. It also included 20 life events and asked if the event occurred during the previous year(12). A study psychiatrist interviewed each subject to establish a clinical diagnosis of major depression and then administered the Montgomery-Ǻsberg Depression Rating Scale (MADRS) to document depression symptom severity. Each subject completed several self-report measures, including the NEO PI-R(13) as a measure of personality, from which we report a total neuroticism score. Within the neuroticism subscale, we also calculated the score for facet 6, “Vulnerability to Stress.”
Study psychiatrists followed a treatment protocol that employed both structured and naturalistic components. All depressed subjects were offered open-label treatment with sertraline for 12 weeks. Individuals taking antidepressants at baseline who otherwise met inclusion criteria, underwent a study-related two-week medication washout with weekly telephone contact to assess clinical status and provide in-person assessments as warranted. Those who had prescribers or psychotherapists who did not wish to participate in study-based treatment could continue medication treatment or psychotherapy outside the study.
Subjects were administered a standardized cognitive assessment on an annual basis for three years. Cognitive measures included tests of global cognition measured with the Mini-Mental State Exam (used only for descriptive purposes in the current study) and the Consortium to Establish a Registry in Alzheimer’s Disease (CERAD) neuropsychological battery(14), as well as tests of processing speed, executive functioning, and memory described further below. The CERAD is a composite measure of wordlist learning, wordlist recall, wordlist recognition, confrontation naming, animal naming, and constructional praxis (maximum score = 100)(14). Tests of processing speed and executive functioning included time to complete Part B of the Trail Making Test(15) and total correct from the Symbol Digit Modalities Test (SDMT, maximum score = 110)(16). Tests of verbal memory included the Logical Memory subtest of the Wechsler Memory Scale–Revised(17) including immediate recall (LMI, maximum score 50) and delayed recall (LMII, maximum score 50). Tests of non-verbal memory included total correct from the Benton Visual Retention Test (BVRT, maximum score = 10)(18).
The measures of interest were three-year change scores (Year 3 minus baseline) of psychosocial and neuropsychological variables. These included neuroticism, MADRS, total number of stressors, CERAD, Trail Making Part B, SDMT, BVRT, LMI and LMII. Descriptive statistics of these measures and demographic variables were summarized for the total sample. Bivariate associations between these measures were assessed using Spearman’s correlation analysis.
We further conducted individual multiple regression analyses for the outcomes of three-year change in each cognitive score. The regression models included the predictor variables of three-year change in total neuroticism score (ΔNeuroticism), three-year change in total numbers of stressors (ΔTotstress) and a (ΔNeuroticismΔTotstress) interaction term. All models controlled for baseline measure of the outcome variable, three-year change in MADRS score, education, sex, and race. To better understand the (ΔNeuroticismΔTotstress) interaction term for cognitive scores, we computed and graphed the slopes of one predictor variable on the outcome variable, which were conditioned on the quartile values (i.e., 25^th^, 50^th^, and 75^th^ percentiles) of another predictor variable.
All the analyses were performed using SAS 9.4 and a two-sided significance level of 0.05 was used.
The sample consisted of 62 older depressed subjects with complete three-year data. Demographic and clinical characteristics of the sample are shown in Table 1. Spearman correlation analysis (see Table 2) showed that three-year change in LMII was not statistically significantly associated with three-year change in Total Neuroticism (r(S)=−0.22, P=0.090), three-year change in total number of stressors (r(S)=−0.19, P=0.15) or three-year change in MADRS (r(S)=−0.21, P=0.097). Three-year change in Total Neuroticism was associated with three-year change in MADRS (r(S) = 0.36, P=0.004), but not significantly associated with three-year change in total number of stressors (r(S)=−0.12, P=0.37).
In multivariable linear regression analysis controlling for age, sex, education, race, baseline LMII and three-year change in MADRS score (see Table 3), the interaction effect of three-year change in Total Neuroticism score and three-year change in Total Stressors on change in LMII performance was statistically significant (B=−0.080[95%CL: −0.145 to −0.015], T = −2.485, df = 52, p = 0.017). We found a similar association when we examined the interaction of three-year change in the Vulnerability to Stress facet of Neuroticism and three-year change in Total Stressors on change in LMII performance, although this association did not reach α=0.05 significant level.
To better understand the interaction effect, we estimated from the model the slopes of neuroticism change on the outcome of LMII change score, conditioned on quartile values of total stressor change score (See Figure 1a). The association between three-year change in LMII score and three-year change in neuroticism was not statistically significant when total stressors decreased by 1 or when total stressors did not change. However, when total stressors increased by 2 or more over three years, LMII change was inversely associated with neuroticism change. Similarly, we estimated the effect of total stressor change on LMII change score, at different levels of neuroticism change (see Figure 1b). When neuroticism score decreased more (shown as 25th percentile of neuroticism change score, i.e., −23), total stressor change had no effect on LMII change. However, when neuroticism improved less, LMII change score was inversely associated with total stressor change (for example, when neuroticism score decreased by 5).
A similar relationship between LMII change, total neuroticism change and total stressors change was also observed when three-year change in LMI score (immediate memory) was used as outcome and/or three-year change in vulnerability to stress was used as predictor. There was no significant interaction effect between stress and neuroticism on total CERAD score. (See supplemental tables S1 and S2). Finally, there were no other significant interaction effects between stress and neuroticism on other cognition tests (data not shown).
The major finding of this study is that the interaction of three-year changes in neuroticism and stressful life events was associated with three-year change in delayed memory performance. As expected in this study that included standardized pharmacological treatment(19), neuroticism generally decreased over three years of follow up. While our main hypothesis focused on an interaction effect on cognition, expecting that those with higher neuroticism over time and those with higher stressors over time would demonstrate worse cognitive performance over three years, our analyses of the interaction, while statistically significant, yielded a slightly different finding related to the extent to which neuroticism decreased over time. That is, further exploration of our interaction analyses revealed important effects on longitudinal cognitive performance based on extent of decrease in neuroticism scores such that when neuroticism improved less, LMII change score was inversely associated with total stressor change. For example, when neuroticism score decreased by 5, one-unit decrease of stressors would result in an improvement of 1.63 points on LMII. The partial correlation coefficient between stressor change and LMII change in this scenario, was 0.40, which represents a medium-to-large effect size. Among patients with greater improvement in neuroticism, change in stressors was not associated with LMII change. Interestingly, when total stressors increased by 2 over three years, LMII change was inversely associated with neuroticism change, with 10-point improvement on neuroticism corresponding to 2.20-point improvements on LMII. This represents a medium effect size with a partial correlation coefficient of 0.35. When there was little or no change in stressors, neuroticism change was not associated with LMII.
These findings add to a growing literature on the effects of neuroticism and stress on cognition over time among older depressed patients. Previously, presence of higher neuroticism was shown to be associated with greater cognitive decline among older depressed individuals(1) and with incident dementia in a general population(2). Our results support inclusion of repeated assessment of neuroticism and consideration of life stressors in the management of depressed older adults. For example, the NEO Five Factor Inventory provides a reliable assessment of neuroticism and other personality domains in 10–15 minutes(20). High scores on neuroticism in older depressed at baseline would indicate an increased likelihood of poor acute antidepressant response and worse long-term depression illness course (1, 21). Moreover, depressed patients in treatment who experience less decline in neuroticism over a one-year period also appear to be at greater risk of three-year cognitive decline(22). The current findings add to this literature and suggest memory abilities in older depressed patients with stable neuroticism scores are particularly susceptible to the detrimental effects of life stressors.
Mechanistically, one can view these results through both a purely clinical prism and also from the standpoint of clinical neuroscience. Clinicians will recognize in the interplay between changes in neuroticism and changes in total number of stressors the notion that greater stress experienced in an individual who is chronically poorly equipped to deal with stress leads to greater cognitive decline. Treated depressed patients who continue to score high on measures of neuroticism despite improvement in mood symptoms and note worsening of psychosocial stressors risk cognitive decline over three years. It is important to remember that while NEO-PI Neuroticism includes “facets” of depression and anxiety, these represent a “tendency to experience psychological distress” rather than requiring syndromal levels of depression or anxiety.(13) It follows that reductions in neuroticism reduce stress reactivity and should engender a more robust response to stressors. Thus, the clinician should develop a treatment plan aimed at helping the patient better manage stress related to events that are largely uncontrollable, such as those we inquired about for this study(12), while seeking to help the patient reduce the load of stressors that are controllable. On this latter point, our finding that change in neuroticism scores and change in stressors are not related suggests that improving neuroticism and reducing stress load may be independent points of clinical intervention.
Clinical neuroscience may also inform the interpretation of our findings. Neuroticism in adults is associated with elevated levels of inflammatory markers such as C-reactive protein and interleukin-6 (23). There is also an emerging literature on neuroimaging findings related to neuroticism in older depressed adults that focuses on frontal circuitry. For example, recent studies found that older depressed individuals high in neuroticism had smaller frontal lobe volumes than depressed subjects low in neuroticism and never-depressed subjects (24), and that older depressed patients, compared with a never-depressed group, showed a higher correlation between negative affectivity (a component of neuroticism and functional connectivity between ventromedial prefrontal cortex and amygdala(25). These findings point to vulnerability of frontal pathways to stress life events. With this vulnerability, the neurobiological effects of increases in stress may be magnified, e.g., through glucocorticoid-related neurotoxic, effects on the hippocampus (26), and pro-inflammatory immunological processes(27). Along these lines, prior research linking depression and stress to hippocampal volume loss(28) supports future investigations of neuroticism, hippocampal measures, and cognitive change.
It is important to note that while our study focused on the personality domain of neuroticism, other domains of the NEO-PI have been examined, such as Openness, Agreeableness Extraversion and Conscientiousness, with lower Conscientiousness being more consistently associated with dementia risk(29, 30). Future studies should examine personality domains more broadly, as well as their relationship to stress, on risk of cognitive decline and development of dementia.
This study has several limitations that should be noted. While few studies in late life depression report three-year clinical, psychosocial, and cognitive data, our sample size of 62 is still relatively small and limited the number of covariates we could include in our modelling. Reliance on self-report of stressors may introduced a bias towards greater stress among individuals with ongoing depression or high neuroticism. Also, as this study was correlational, we could not demonstrate causal or directional relationships. Finally, our findings may not generalize to a broad population of older adults given our cohort was predominantly white and fairly well educated. Future studies in larger, more diverse populations should take all of these limitations into account in order to replicate and extend our findings. In addition, more research is needed on longitudinal assessment of factors correlate with change in neuroticism itself in the context of depression, as well as studies examining how neuroticism and depression change together or distinctly in pharmacologic treatment for depression among older depressed patients.