Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2021 Jun 3;77(2):284–294. doi: 10.1093/geronb/gbab063

The Age-Dependent Association Between Vascular Risk Factors and Depressed Mood

Maria Blöchl 1,2,✉, H Lina Schaare 3,4, Ute Kunzmann 5, Steffen Nestler 1
Editor: Shevaun Neupert
PMCID: PMC8824621  PMID: 34080633

Abstract

Objectives

Cumulative burden of vascular risk factors (VRFs) has been linked to an increased risk of depressed mood. However, the role of age in this association is still unclear. Here, we investigated whether VRF burden is associated with levels and changes in depressed mood and whether these associations become stronger or weaker from mid- to later life.

Method

We used longitudinal data from 5,689 participants (52–89 years) of the English Longitudinal Study of Ageing. A composite score incorporated the presence of 5 VRFs: hypertension, diabetes, smoking, obesity, and hypercholesterolemia. Second-order latent growth models were used to test whether levels and changes of depressed mood differed as a function of baseline VRF burden, and whether these associations were moderated by age.

Results

Baseline VRF burden showed a small association with higher levels of depressed mood (estimate = 0.081; 95% CI: 0.024, 0.138, p = .005). This association varied with age, such that it was stronger in midlife compared to later life (estimate = −0.007; 95% CI: −0.013, −0.002, p = .017). There was no evidence that VRF burden was associated with changes in depressed mood.

Discussion

Our findings suggest that VRF burden in midlife, but less so in later life, predicts individual differences in depressed mood. These findings are consistent with reports on the importance of midlife VRFs and support the idea that promotion of vascular health in this age group or earlier in life may be critical to maintain mental health across adulthood.

Keywords: Aging, Cardiovascular health, Depression, Latent growth models, Vascular depression hypothesis

In the past two decades, modifiable vascular risk factors (VRFs), such as hypertension, obesity, smoking, diabetes, and hypercholesterolemia, have emerged as major targets to promote healthy aging (Lloyd-Jones et al., 2010). Numerous studies show that the presence of VRFs, in particular, the cumulative burden from multiple risk factors in midlife, is associated with an increased risk of adverse health outcomes in later life (Lloyd-Jones et al., 2010; Qiu & Fratiglioni, 2015; Stamler et al., 1999). However, how VRF burden relates to trajectories of depressed mood and whether these relations vary across adulthood is less clear. This issue is key as mental health is an important aspect of healthy aging (World Health Organization, 2019) and to maintain mental health, the timing of public health interventions to improve vascular health may be important. In the current study, we examined the associations of VRF burden with levels and longitudinal changes of depressed mood, and whether these associations are dependent on age.

VRF Burden and Depressive Symptoms

Considerable theoretical work has implicated VRF burden in the pathophysiology of depression, especially in later life (Alexopoulos et al., 1997; Aziz & Steffens, 2013; Gaupp et al., 2000). One of the most prominent theories, the vascular depression hypothesis, postulates that VRFs “predispose, precipitate, or perpetuate” depressive symptoms in older adults (Alexopoulos et al., 1997). This link is thought to arise because VRFs contribute to small vessel damage in the brain, which can compromise the integrity of fronto-subcortical circuits involved in mood regulation (Smagula & Aizenstein, 2016; Taylor et al., 2013). However, while some cross-sectional and longitudinal studies have supported the vascular depression hypothesis (Kivimäki et al., 2012; Luijendijk et al., 2008; Mast et al., 2005), other studies failed to demonstrate an association between VRFs and depressive symptoms (Lyness et al., 2000; Stewart et al., 2001; Valkanova & Ebmeier, 2013). A major drawback of previous studies is that they have rarely taken a life-span approach: Either they only focused on older adults or they did not consider that the association with depressive symptoms may critically depend on the age at which VRFs are assessed.

The Role of Age

Although few studies have explicitly investigated the role of chronological age, most prior research presumed that the association between VRF burden and depression strengthens across the adult life span. VRF burden and associated microvascular damage are typically thought to increase with age (Laina et al., 2018; Lloyd-Jones et al., 2010). In line with the vascular depression hypothesis, it has been hypothesized that mood disturbances may only emerge if vascular burden reaches a certain threshold, above which it compromises the ability to regulate emotions or cope with age-related declines (Alexopoulos et al., 1997; Smagula & Aizenstein, 2016; Taylor et al., 2013). Direct evidence for this hypothesis is, however, limited. Only one cross-sectional study reported a significant association between VRF burden and depressive symptoms in individuals who were older but not younger than 85 years (Mast et al., 2005). Importantly, this finding can only be interpreted with caution, given that the authors’ index of VRF burden included heart diseases. Manifest vascular diseases, such as heart diseases, are more common in later life and might cause depression not because of vascular-related pathology but because of their life-threatening nature and effects on functional abilities (Kivimäki et al., 2012; Lenze et al., 2001; Valkanova & Ebmeier, 2013). Accordingly, manifest vascular diseases have been established as strong risk factor for depression (e.g., Wikman et al., 2011). The distinction between manifest vascular diseases and VRFs is critical, and it would be necessary to establish similar age-dependent effects in the absence of manifest vascular diseases.

Notably, several lines of research point to the alternative prediction, suggesting that VRF burden is a more powerful predictor of depressive symptoms in midlife compared to later life. First, psychological research has shown age-related gains in emotion regulatory abilities and motivational priorities, which arguably help adults to better adapt to losses in daily functioning and physical health as they age (Charles, 2010; Charles & Carstensen, 2010). In line with this notion, mood and well-being are well maintained until old age, despite an age-related increase in VRF burden (Almeida, 2008; Sutin et al., 2013). Moreover, an increase in health problems, such as a high VRF burden, can be seen as almost normative in later life (Barrett & Gumber, 2020; Ubel et al., 2005; Wrosch & Heckhausen, 2005). Middle-aged adults, for whom VRFs are less normative, may have more difficulties to psychologically adjust to the diagnosis and management of such risk factors (e.g., intake of medications) than older adults. Finally, accumulating research on physical and cognitive health taking a life-span perspective suggests that the detrimental effect of VRF burden is particularly pronounced in midlife and tends to decrease with age (Pase et al., 2018; Qiu et al., 2005; Qiu & Fratiglioni, 2015).

Preliminary evidence suggesting the association between VRF burden and depressive symptoms weakens with age exists but is still limited. More specifically, Kivimäki and colleagues (2012) found that in individuals younger than 65 years, but not in those older than 65 years, VRF burden—in the absence of manifest vascular disease—predicted depressive symptoms 5 years later. However, this study did not explicitly and systematically investigate the role of age at which VRFs were assessed. Moreover, it only focused on few follow-ups, which does not allow firm conclusions about longitudinal changes in depressive symptoms to be drawn. Studies with multiple follow-ups provide a crucial inroad into longitudinal changes in depressive symptoms. Importantly, they also allow researchers to address whether VRFs are associated with faster mood declines over time, possibly indicative of slowly progressing damage to fronto-subcortical structures. Only few longitudinal studies have assessed depressive symptoms at multiple time points to test whether trajectories differ as a function of VRFs (Montagnier et al., 2014; Virtanen et al., 2017) and none have focused on cumulative burden of VRFs or investigated age-differential effects.

The Present Study

In the present study, we aimed to investigate the age-dependent relationship between VRF burden and trajectories of depressed mood. More specifically, we sought to examine the extent to which VRF burden is associated with levels and changes in depressed mood, and whether these associations increase or decrease with age. If VRF burden is associated with mood because of underlying brain damage, the association of VRF burden and levels of depressed mood would strengthen with age, and a higher number of VRFs might also predict accelerated changes in depressed mood. If, however, the effects of age-related improvement in emotional and motivational competencies prevail, the association of VRF burden and depressed mood would presumably weaken with age.

To test these competing predictions, we leveraged data from 5,689 individuals (52–89 years) of the English Longitudinal Study of Ageing (ELSA), who had extensive assessments of VRFs at baseline and were followed-up five times at 2-year intervals, covering a period of 10 years. We used second-order latent growth models (LGMs) to model trajectories of depressed mood over time, and test whether levels and changes differed as a function of baseline VRF burden, age, and their interaction. We focus on depressed mood because it lies at the heart of depression and is less susceptible to potential confounding effects of vascular health and age compared to somatic symptoms of depression (Bus et al., 2011; Schaakxs et al., 2017). Importantly, we only include participants who were free of manifest vascular diseases at baseline. Moreover, we statistically account for the presence of physical limitations, which are known to confound the association between VRF burden and depression (Almeida, 2008; Lyness et al., 2000; Valkanova & Ebmeier, 2013), as well as for gender, level of education, and ethnicity.

Method

Sample

The ELSA is a large, multidisciplinary longitudinal study of community-dwelling adults aged 50 or older (i.e., the core sample) and their partners, some of whom were younger (Steptoe et al., 2013). Data collection started in 2002 (Wave 1), with a nationally representative sample being drawn from the Health Survey for England. For the purpose of this study, we used data from Wave 2 (2004–2005) as baseline, as this was the first assessment that included self-reported information of VRFs (diagnosed hypertension, history of smoking) as well as laboratory assessments (blood pressure, fasting glucose, body mass index [BMI], and cholesterol levels) taken during a nurse visit. Participants have been re-examined every 2 years, with repeated measures of depressive symptoms until Wave 7 (2014–2015), spanning a total of 10 years. Participants were excluded if they were not part of the core sample, if they reported a manifest vascular disease (stroke, heart attack, congestive heart failure) or neurodegenerative disorder (Parkinson’s disease, dementia, Alzheimer’s disease) at baseline, or if they did not provide data in the nurse visit. Finally, we excluded 1,146 people with missing data on covariates. Thus, the final sample included 5,689 participants (Figure 1).

Figure 1.

Figure 1.

Open in a new tab

Flow chart of individuals included in the analyses. PD = Parkinson’s disease; AD = Alzheimer’s disease.

As common in longitudinal studies of aging, sample attrition was sizable. Yet, 53% of the initial sample participated in all six waves of data collection and on average, participants contributed 4.64 (SD = 1.79, median = 6) waves of data. Attrition analyses indicated that those who contributed five or six waves of survey data were younger (d = 0.54), had fewer physical limitations (d = 0.23), and were slightly more educated (d = 0.21) compared to those who contributed fewer than five waves of survey data. Differences in gender, ethnicity, depressive symptoms, and VRFs at baseline were negligible (all ds < 0.15).

ELSA was conducted in accordance with the Declaration of Helsinki and informed consent was received from all participants to take part. Ethical approval and experimental protocols were granted by the Multicentre Research and Ethics Committee. The ELSA data sets can be accessed through the UK Data Service (https://www.ukdataservice.ac.uk).

Measures

Depressed mood

Depressed mood was measured using five affective items of the eight-item version of the Centre for Epidemiologic Studies Depression Scale (CES-D-8). The CES-D-8 is a well-established and widely used depression screening instrument for older adults (Turvey et al., 1999), in which participants are asked to report whether they had experienced eight symptoms “much of the time” in the past week (“yes”/“no”). We focused on the five items tapping into depressed mood (i.e., felt depressed, was happy, felt lonely, enjoyed life, felt sad) because they reflect the core diagnostic criteria of major depressive disorder and because somatic symptoms (three items: everything was an effort, sleep was restless, and could not get going) might inflate the observed associations with VRF burden during aging (Bus et al., 2011; Schaakxs et al., 2017). To further evaluate the frequency of elevated depressed mood, we analyzed how many participants reported two (out of five) symptoms at baseline, given that a cutoff value of three or larger was previously proposed for the full CES-D-8 scale (Andresen et al., 1994; Iob et al., 2020).

VRF burden

VRFs were assessed at baseline and included: hypertension, diabetes, current smoking, obesity, and hypercholesterolemia. Hypertension was defined as either reporting the presence of diagnosed hypertension or having systolic blood pressure ≥130 mmHg and diastolic blood pressure ≥80 mmHg (Muntner et al., 2019). Diabetes was defined as either reporting the presence of diagnosed diabetes or having a fasting blood glucose level of ≥7.0 mmol/l (Fox et al., 2015). Smoking was assessed using the core survey questionnaire by asking whether participants are currently smoking. Obesity was defined as a BMI of ≥30 kg/m2 (Jensen et al., 2014). BMI was calculated from measurements of the participants’ height and weight as assessed during the nurse visit. Hypercholesterolemia was defined as either reporting the presence of diagnosed hypercholesterolemia or high cholesterol or low-density lipid cholesterol level of ≥4.1 mmol/L (Grundy et al., 2019).

Cumulative burden from these risk factors has been extensively linked to subsequent vascular disease, beyond the effect of single risk factors (Qiu & Fratiglioni, 2015; Stamler et al., 1999). Thus, we calculated VRF burden as the number of individual risk factors. For this purpose, individual risk factors were coded as present (“1”) or absent (“0”) at baseline and then summed into an overall composite score (total score range 0–5).

Age

Information on participants’ chronological age (in years) was provided by the ELSA team and computed from the difference between participants’ date of birth and their date of the individual interview at baseline.

Covariates

Sociodemographic information included gender, educational attainment, and ethnicity. Educational attainment was defined as an indicator of having completed a higher or tertiary education (“yes”/“no”). Ethnicity was defined as a dichotomous variable (“White”/“non-White”), because the vast majority of participants considered their ethnicity to be White.

Physical limitations were included as a further covariate as they are known to increase the risk of depressed mood (Lenze et al., 2001; Sutin et al., 2013). Physical limitations were assessed by asking participants whether they experienced limitations in 10 basic activities of daily living (e.g., walking 100 yards, sitting about 2 hr, climbing one flight of stairs without resting, picking up a coin from a table). A scale ranging from 0 to 10 indicating the severity of physical limitations was computed (Spector & Fleishman, 1998).

Statistical Analyses

We first fitted a series of longitudinal factor models to verify that stability and changes could be attributed to a common latent variable over time (Ferrer et al., 2008). We tested increasing levels of measurement invariance by successively introducing equality constraints on factor loadings, item thresholds, and factor variances in the measurement models of depressed mood, respectively. Strict measurement invariance constraints were satisfied, indicating that the same latent construct was being measured at each time of measurement (for more information on measurement invariance analyses and results, see Supplementary Results 1 including Tables S1 and S2).

We then fitted an unconditional second-order LGM (Hancock et al., 2001) to estimate the trajectory of depressed mood as a function of time. LGMs are structural equation models that allow for an examination of the initial level of an outcome (i.e., intercept), and its rate of change (i.e., slope), which describe the group-level growth curve of the outcome. Second-order LGMs additionally model the construct at each time point as a latent variable, which allows the constructs free of measurement error to be modeled, and longitudinal measurement invariance to be examined. The unconditional second-order LGM fit the data well (χ2(371) = 1184.1, p < .001; comparative fit index [CFI] = 0.987; root mean square error of approximation [RMSEA] = 0.018; standardized root mean square residual [SRMR] = 0.045), indicating a linear growth trajectory of depressed mood over time.

Finally, our main model examined the extent to which growth trajectories between individuals differed as a function of baseline VRF burden, and how this effect varied by age. Conditional second-order LGMs were fitted that included VRF burden, age, and their interaction as predictors of between-person differences in levels (i.e., intercept) or changes (i.e., slope) in depressed mood (Supplementary Figure S1). We furthermore controlled for the effects of gender, education, ethnicity, and physical limitations. Age, VRF burden, and physical limitations were centered to their respective mean to aid the interpretation of the model parameters.

We fitted the statistical models using R (version 3.6.2) and the “lavaan” package (Rosseel, 2012; R Core Team, 2014). For model estimation, we used a diagonally weighted least squares estimator with robust standard errors and a mean and variance adjusted test statistic. This estimator is denoted as means and variance adjusted weighted least square (WLSMV) in the “lavaan” package. WLSMV accounts for the ordinal nature of the indicator variables and is generally robust to deviations from the assumption of multivariate normally distributed latent responses, especially when using relatively large samples (Flora & Curran, 2004; Li, 2016; Rhemtulla et al., 2012). To handle missing data, we used a pairwise deletion approach.

Global model fit was evaluated based on three absolute fit indices: CFI (values > 0.95 considered acceptable), RMSEA (values < 0.06 considered acceptable), and SRMR (values < 0.08 considered acceptable). To evaluate single parameters, we report point estimates and their 95% confidence interval (CI). For significance testing of parameter estimates, an alpha of .05 was used as a threshold, given that previous studies on VRFs and depression tended to report small effects (Valkanova & Ebmeier, 2013).

Robustness Analyses

Although pairwise deletion maximizes the available information when using WLSMV, we conducted several robustness analyses to rule out any potential biases due to missing data. First, we used data of our main analytic sample (N = 5,689) and refitted the final second-order LGM as a first-order LGM using the sum scores of depressed mood as manifest variables. Because sum scores can be regarded as metric variables, this approach allowed us to leverage robust full information maximum likelihood (robust FIML) estimation that assumes that the missing data are missing at random (Enders, 2001), which is a standard assumption in longitudinal studies with attrition. Second, we calculated another first-order LGM (with FIML) to account for potential biases due to missing covariate data. In this model, we additionally included the data of the 1,146 participants that were excluded from our main analysis due to lack of information on covariates.

A final analysis probed the robustness of our results when accounting for selective nonresponses. Although previous comparisons of sociodemographic characteristics suggested that the sample at Wave 1 was broadly representative of the general population (Steptoe et al., 2013), differential nonresponse at Wave 2 and our exclusion criteria might have led to a sample that is somewhat unrepresentative of the target population (i.e., persons born before March 1, 1952 living in private households in Britain). To account for selective nonresponses in our models, we additionally considered Wave 2 sampling weights, which were provided by the ELSA team, when we fitted the first-order LGM (with FIML).

Results

Sample Characteristics

Sample characteristics are provided in Table 1 and zero-order correlations among all variables are presented in Supplementary Table S4. At baseline, the mean age of participants was 65.4 years (SD = 9.1; range = 52−89) and 3,110 participants (55%) were female. Depressed mood was generally low (M = 0.65, SD = 1.19) but ranged across the full spectrum from zero to five, with 937 participants (17%) reporting two or more symptoms. Overall VRF burden was moderate, with 1.46 (SD = 1.02) risk factors on average.

Table 1.

Participant Characteristics in the Overall Sample and Stratified by Age

Age group
Whole sample 50–59 years 60–69 years 70–79 years >80 years
n at baseline (%)a 5,689 (100) 1,912 (34) 1,979 (35) 1,298 (23) 500 (9)
Age in years, mean ± SD 65.36 ± 9.08 55.87 ± 2.03 64.38 ± 2.94 73.98 ± 2.87 83.20 ± 2.42
Women, n (%) 3,110 (55) 1,032 (54) 1,067 (53) 706 (54) 305 (61)
White ethnicity, n (%) 5,598 (98) 1,868 (98) 1,949 (98) 1,281(99) 500 (100)
No higher education, n (%) 4,892 (86) 1,549 (81) 1,701 (86) 1,173 (90) 469 (94)
Phys. limitations, mean ± SD 1.72 ± 2.32 1.30 ± 2.18 1.61 ± 2.24 2.05 ± 2.34 2.89 ± 2.63
VRF burden, mean ± SD 1.46 ± 1.02 1.43 ± 1.03 1.48 ± 1.03 1.57 ± 1.01 1.22 ± 0.92
Depressed mood, mean ± SDb
Wave 2 (baseline) 0.65 ± 1.19 0.68 ± 1.24 0.54 ± 1.11 0.67 ± 1.19 0.93 ± 1.29
Wave 3 0.58 ± 1.17 0.56 ± 1.17 0.51 ± 1.11 0.62 ± 1.19 0.89 ± 1.27
Wave 4 0.58 ± 1.14 0.51 ± 1.09 0.57 ± 1.15 0.66 ± 1.17 0.82 ± 1.19
Wave 5 0.63 ± 1.19 0.56 ± 1.17 0.57 ± 1.12 0.77 ± 1.27 0.99 ± 1.31
Wave 6 0.57 ± 1.13 0.52 ± 1.13 0.51 ± 1.06 0.69 ± 1.19 0.94 ± 1.36
Wave 7 0.53 ± 1.11 0.45 ± 1.03 0.49 ± 1.06 0.74 ± 1.30 0.95 ± 1.41
Open in a new tab

Notes: ADL = activities of daily living; phys. = physical; VRF = vascular risk factor.

aPercentages do not exactly add up to 100% due to rounding. bA sum score was calculated from participants’ data for descriptive statistics, ranging from 0 to 5 (positive items were reverse-coded). Scores represent the mean and SD across all participants’ sum scores.

Age and VRF Burden

Table 1 shows the average VRF burden by 10-year age groups. The prevalence of baseline VRF burden increased across age groups and was highest in individuals aged 70–79 years (mean = 1.57, SD = 1.01); however, VRF burden was again lower in the oldest age group, aged ≥80 years (mean = 1.22, SD = 0.92). A breakpoint regression model confirmed a nonlinear relationship between VRF burden and age: VRF burden increased slightly and linearly up to 75 years, and then dropped again (Supplementary Results 2).

Associations Between VRF Burden and Levels of Depressed Mood

Results of the second-order LGM revealed that VRF burden (estimate = 0.081; 95% CI: 0.024, 0.138, p = .005), but not age (estimate = −0.001; 95% CI: −0.006, 0.005, p = .939), showed a small association with levels of depressed mood (Table 2). Importantly, the association between VRF burden and levels of depressed mood was moderated by age. More specifically, this association was slightly weaker if VRFs were measured at an older age (estimate = −0.007; 95% CI: −0.013, −0.002, p = .017). Predicted results depicted in Figure 2 for specific age groups at baseline (50-, 60-, 70-, and 80-year olds) illustrate that individuals with more risk factors reported more depressed mood, and this effect was stronger in middle-aged adults than those who were older at baseline.

Table 2.

Conditional Growth Curve Model of Depressed Mood Over Time (N = 5,689)

Intercept Slope
Estimate 95% CI, lower bound 95% CI, upper bound Standard estimate Estimate 95% CI, lower bound 95% CI, upper bound Standard estimate
Fixed effectsa
−2.279* −2.720 −1.838 −1.643 −0.076 −0.165 0.012 −0.315
 Age −0.001 −0.006 0.005 −0.001 0.005* 0.003 0.007 0.189
 Genderb −0.454* −0.587 −0.321 −0.163 0.014 −0.023 0.052 0.030
 Educationc −0.111 −0.269 0.047 −0.028 −0.016 −0.065 0.032 −0.023
 Ethnicityd 0.866* 0.370 1.362 0.078 −0.081 −0.199 0.037 −0.042
Phys. limitations 0.179* 0.137 0.221 0.301 0.005 −0.005 0.014 0.017
 VRF burden 0.081* 0.024 0.138 0.060 0.004 −0.014 0.022 0.017
 VRF burden × Age −0.007* −0.013 −0.002 −0.045 0.001 −0.002 0.002 0.013
Random effects
 Variance 1.618* 0.757 2.478 0.841 0.056* 0.024 0.088 0.954
 Covariance −0.085 −0.206 0.036 −0.282
Model fit
  χ 2 (df) 612.33 (567)
 CFI 0.978
 RMSEA 0.004
 SRMR 0.051
Open in a new tab

Notes: ADL = activities of daily living; CFI = comparative fit index; CI = confidence interval; phys. = physical; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual; VRF = vascular risk factor.

aThe fixed effect of the intercept in this model reflects the logit for answering the first items (“felt depressed”) them with “yes”; converting the estimates to probabilities (by P(y1i = 1)= 1/1+exp(−intercept), indicates that the probability of answering the first with “yes” is 1.10, accounting for all covariates. bReference category = female. cReference category = lower education. dReference category = White.

*p < .05.

Figure 2.

Figure 2.

Open in a new tab

Trajectories of depressed mood over time as a function of baseline vascular risk factor (VRF) burden, age, and their interaction. The mean predicted trajectories for different levels of VRF burden are depicted in black (solid = 5 VRFs, dashed = 2 VRFs, dotted = 0 VRFs). Individual predicted trajectories are depicted in light gray. Scores for depressed mood (y-axis) can be converted to the probability of answering the first item of the depression questionnaire (“felt depressed”) with “yes.” Mean trajectories were plotted for a chosen profile of covariates (i.e., White females without a higher degree and mean activities of daily living).

Associations Between VRF Burden and Changes of Depressed Mood

Consistent with previous studies (e.g., Sutin et al., 2013), results of our second-order LGM furthermore revealed small decreases in depressed mood over time (estimate = −0.076; 95% CI: −0.165, 0.012, p = .092; see also Table 2 and Figure 2). This decrease was smaller for individuals who were older at baseline (estimate = 0.005; 95% CI: 0.003, 0.007, p < .001). However, there was no evidence that VRF burden at baseline was associated with changes in depressed mood (estimate = 0.004; 95% CI: −0.014, 0.022, p = .653), or that this association was moderated by age (estimate = 0.001; 95% CI: −0.002, 0.002, p = .717). Predicted results depicted in Figure 2 illustrate that individuals with more VRFs at baseline across all ages did not experience a different rate of change in depressed mood over time.

Robustness Analyses

Findings remained stable when we fitted a first-order LGM with a robust FIML estimator using sum scores of depressed mood as manifest indicators. Higher VRF burden was associated with higher levels of depressed mood (estimate = 0.040; 95% CI: 0.012, 0.068, p = .006), and this association weakened with age (estimate = −0.004; 95% CI: −0.007, −0.001, p = .037; see Supplementary Table S5 for full results). Findings also remained stable, but were slightly attenuated, when we additionally included 1,146 participants with missing data on covariates (N = 6,835). Higher VRF burden was associated with higher levels of depressed mood (estimate = 0.027; 95% CI: 0.001, 0.053, p = .048), and this association weakened with age (estimate = −0.003; 95% CI: −0.006, −0.001, p = .033; see Supplementary Table S6 for full results). Finally, findings also remained robust when employing sampling weights to account for selective nonresponse and sampling bias. Higher VRF burden was associated with higher levels of depressed mood (estimate = 0.043; 95% CI: 0.014, 0.073, p = .003), and this association weakened with age (estimate = −0.004; 95% CI: −0.007, −0.001, p = .036; see Supplementary Table S7 for full results).

Discussion

This study examined how VRF burden relates to levels and changes of depressed mood and whether these associations vary with age. Our analyses revealed that a higher number of VRFs at baseline was associated with elevated levels of depressed mood, and that this association was moderated by age, such that it was stronger in middle-aged, as compared to older, individuals. However, these effects were small when accounting for sociodemographic factors and physical limitations, and there was no evidence that baseline VRF burden was associated with changes in depressed mood over subsequent years.

Age, VRF Burden, and Levels of Depressed Mood

VRFs have long been implicated in the development of depression, while empirical evidence for this hypothesis has been mixed (Alexopoulos et al., 1997; Aziz & Steffens, 2013; Taylor et al., 2013; Valkanova & Ebmeier, 2013). Here, we found that the presence of multiple VRFs shows a small association with depressed mood, a core feature of depression, but that this association critically depends on the age window in which VRFs are assessed. This result could help to explain conflicting findings in previous research. More specifically, it is possible that there are differences in the age ranges of previous studies’ samples and that studies with mainly middle-aged or younger adults found stronger associations than studies that included mainly older adults. In addition, and similar to previous studies, our results suggest that after accounting for vascular diseases and physical limitations, there is limited evidence for an association between VRF burden and depressive symptoms in older adults (Kivimäki et al., 2012; Lyness et al., 2000; Valkanova & Ebmeier, 2013).

Our results strengthen and extend findings from one previous study reporting preliminary evidence for age-dependent associations in depression (Kivimäki et al., 2012) by systematically investigating the moderating role of age in the association between VRF burden and depressed mood in individuals across a large age span (52–89 years). The current results are also in line with other studies showing the importance of midlife VRF burden for predicting vascular diseases, dementia, and mortality (Pase et al., 2018; Qiu et al., 2005; Qiu & Fratiglioni, 2015).

Notably, our results contradict findings from a cross-sectional study by Mast and colleagues (2005), who reported that unique associations between VRF burden and depressive symptoms, beyond physical limitations, emerged only in later life. These divergent findings may be explained by differences in study design. Most importantly, Mast and colleagues used a risk index that incorporated the presence of heart diseases, which may have confounded the association with depression, especially in older adults. In contrast, people with vascular diseases were excluded in the present study. This approach allowed us to disentangle the effect of VRF burden and manifest vascular diseases, although we note that it may also have selected slightly healthier-than-average older participants. The results of our robustness analyses, however, suggest that this is unlikely to explain our results.

More generally, our findings challenge predictions from the vascular depression hypothesis. While the vascular depression hypothesis has posited links between VRFs and depression in later life (Alexopoulos et al., 1997; Taylor et al., 2013), we show that these associations are already present and even stronger earlier in life. Our findings therefore underline the importance of taking a life-span perspective to further elucidate the association between vascular health and depression and clarify the mechanisms underlying this association. They furthermore bear potential relevance for public health, as they suggest that preventing the accumulation of VRFs in midlife or even earlier, rather than in later life, may be important to maintain mental health.

Why precisely the association between VRF burden and levels of depressed mood weakens with age remains an open question for future research. It is possible that in later life, a decline in VRF burden is associated with looming pathological processes and general health deterioration (Qiu et al., 2005). These processes may lead to general increases in levels of depressed mood, which we also observed in our study, and weaken the association between VRFs and depressed mood in older age.

It is also plausible that psychological mechanisms explain the current pattern of results. Specifically, an increasing body of evidence suggests that older adults place a greater emphasis on emotion regulation strategies that allow for the preservation of well-being and reduction of stress (Charles, 2010; Charles & Carstensen, 2010). Such age-related benefits may also protect older adults’ mood in the presence of VRFs. For example, when faced with diabetes, recent studies showed that older adults report fewer diabetes-related stressors (Berg et al., 2020) or experience such stressors as less negative (Helgeson et al., 2020) compared to younger adults. Therein, expectations about aging may also play a role. The presence of some VRFs may be considered normative in later life, as it is more in keeping with the expected patterns of aging. Moreover, people tend to compare themselves to their same-age peers when asked to rate their physical health (Jylhä, 2009; Ubel et al., 2005). This comparison may be more detrimental for middle-aged adults with multiple VRFs, whose peers are generally healthier, than for older adults, whose peers tend to have a higher number and more severe health problems. Alternatively, medical care and treatment required by some risk factors may influence age- and health-related self-perceptions, and in turn take a stronger toll on mental health in middle-aged adults than in older adults (Barrett & Gumber, 2020).

A final possibility refers to personality changes with age. More specifically, age-related decreases in neuroticism (e.g., Steunenberg et al., 2005) might explain why the association between VRF burden and depressed mood weakens with age. As individuals become more emotionally stable, they might worry less about their health and report fewer depressive symptoms (Lahey, 2009). Decreases in neuroticism may facilitate age-related improvement in emotion regulation, which could help older individuals to cope with VRFs. Future research is needed to test the relative or unique contributions of different psychological mechanisms underlying the association between VRF burden and depressive symptoms. To disambiguate different hypotheses, it would be important to assess potential mediating and predictive factors (e.g., emotion regulation, neuroticism, daily stress processes, self-perceptions of health and aging, and treatment information) separately and ideally on multiple levels (self- and other-reports, performance-based).

Age, VRF Burden, and Changes of Depressed Mood

Our analyses furthermore showed no evidence that changes in depressed mood over time varied as a function of baseline VRF burden, neither in midlife nor later life. Importantly, this suggests that the link between VRF burden in midlife and subsequent depression observed in previous longitudinal studies (Kivimäki et al., 2012) may reflect the persistence of earlier differences, rather than differential rates of age-related mood decline. This result also calls into question previously hypothesized mediating pathways. If midlife VRF burden relates to depression in later life because of vascular pathology or progressing brain damage, as frequently hypothesized (Smagula & Aizenstein, 2016; Taylor et al., 2013), we arguably should have observed that VRF burden predicts age-related mood declines over time.

However, we acknowledge that the lack of associations with changes in depressed mood could be due to limitations of our study design. More specifically, the present sample did not include information on VRF burden in earlier adulthood. It is possible that the adverse effects of VRF burden on brain structure and function were already present in at least some middle-aged individuals before the study begun, especially if they had suffered from VRFs for some time (Maillard et al., 2012; Schaare et al., 2019). Future studies should investigate this possibility, for example, by recruiting younger adults and taking the time of onset of VRFs into account.

Limitations and Future Directions

The present study had several strengths, including the longitudinal approach with a focus on trajectories of depressed mood using second-order LGMs, the use of a large, age-heterogeneous sample, and the exclusion of individuals with manifest vascular diseases. These strengths need to be put in the context of some limitations. First, the design of the study is not ideally suited to make firm conclusions about the causality and directionality between VRF burden and depressed mood. For example, there is also evidence that depression in earlier life can promote the development of VRFs (Matthews et al., 2019). Repeated measurements of VRF burden would allow for a stronger design allowing estimation of time-ordered, reciprocal effects and clarification of potential bidirectional relationships across the life span. Alternatively, randomized controlled trials could test whether targeted public health efforts and prevention of VRFs in midlife, or earlier, will ultimately reduce depressive symptoms across adulthood. Second, our sample only included people from the fifth decade of their life onward. In the light of our results and given that VRFs typically develop earlier in life (Karmali & Lloyd-Jones, 2014; Lloyd-Jones et al., 2010), more research including young and young middle-aged adults is needed to elucidate when precisely the observed associations arise. Third, we operationalized some VRFs by combining self-reported clinical diagnosis with laboratory assessments. While this approach improves the correct classification of people with VRFs such as hypertension or hypercholesterolemia, it does not allow us to distinguish between the undetected presence, diagnosis, and treatment of risk factors. How much of the observed effects might be attributable to changes in self-views due to a diagnosis or perhaps side effects of medications require further research. Finally, generalizations of our findings are furthermore constrained by a rather homogeneous sample with respect to race and ethnicity. Future studies extending our findings to more diverse samples are therefore warranted.

Conclusion

Cumulative burden of VRFs in midlife appears to be a stronger predictor of levels of depressed mood compared to VRF burden in later life. This suggests that taking a life-span perspective and focusing on age spans prior to reaching old age may be critical to understand the associations between risk factors for vascular diseases and depressed mood. As midlife is an important period in life to pave the way for healthy aging, preventing and managing VRFs in middle-aged or younger adults, rather than older adults, may be important to maintain mental health.

Supplementary Material

gbab063_suppl_Supplementary_Materials
Click here for additional data file. (863.4KB, pdf)

Acknowledgments

The authors are grateful for the immense effort of the ELSA team as well as their participants for collecting and providing the original data. As such, the investigators within the ELSA contributed to the design and implementation of but did not participate in analysis or writing of this manuscript. All R scripts used in the analysis and accompanying information to reproduce the results are available at https://osf.io/epvk4/. The data are open access upon registration at https://www.ukdataservice.ac.uk. More information on the materials used in ELSA can be found at https://www.elsa-project.ac.uk. This study was not preregistered.

Funding

Financial support for the English Longitudinal Study of Ageing (ELSA) was provided by the National Institute on Aging (grant RO1AG7644) and by a consortium of UK government departments coordinated by the UK Economic and Social Research Council. The funders had no role in the study design, data collection, data analysis, data interpretation, or writing of this manuscript.

Conflict of Interest

None declared.

References

  1. Alexopoulos, G. S., Meyers, B. S., Young, R. C., Campbell, S., Silbersweig, D., & Charlson, M. (1997). ‘Vascular depression’ hypothesis. Archives of General Psychiatry, 54(10), 915–922. doi: 10.1001/archpsyc.1997.01830220033006 [DOI] [PubMed] [Google Scholar]
  2. Almeida, O. P. (2008). Vascular depression: Myth or reality? International Psychogeriatrics, 20(4), 645–652. doi: 10.1017/S1041610207006473 [DOI] [PubMed] [Google Scholar]
  3. Andresen, E. M., Malmgren, J. A., Carter, W. B., & Patrick, D. L. (1994). Screening for depression in well older adults: Evaluation of a short form of the CES-D (Center for Epidemiologic Studies Depression Scale). American Journal of Preventive Medicine, 10(2), 77–84. doi: 10.1016/S0749-3797(18)30622-6 [DOI] [PubMed] [Google Scholar]
  4. Aziz, R., & Steffens, D. C. (2013). What are the causes of late-life depression? The Psychiatric Clinics of North America, 36(4), 497–516. doi: 10.1016/j.psc.2013.08.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Barrett, A. E., & Gumber, C. (2020). Feeling old, body and soul: The effect of aging body reminders on age identity. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 75(3), 625–629. doi: 10.1093/geronb/gby085 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Berg, C. A., Helgeson, V. S., Kelly, C. S., Tracy, E. L., Litchman, M. L., & Butner, J. E. (2020). Age differences in reactivity to daily general and type 1 diabetes stressors. Psychology and Aging, 35(8), 1115–1126. doi: 10.1037/pag0000519 [DOI] [PubMed] [Google Scholar]
  7. Bus, B. A., Marijnissen, R. M., Holewijn, S., Franke, B., Purandare, N., de Graaf, J., den Heijer, M., Buitelaar, J. K., & Voshaar, R. C. (2011). Depressive symptom clusters are differentially associated with atherosclerotic disease. Psychological Medicine, 41(7), 1419–1428. doi: 10.1017/S0033291710002151 [DOI] [PubMed] [Google Scholar]
  8. Charles, S. T. (2010). Strength and vulnerability integration: A model of emotional well-being across adulthood. Psychological Bulletin, 136(6), 1068–1091. doi: 10.1037/a0021232 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Charles, S. T., & Carstensen, L. L. (2010). Social and emotional aging. Annual Review of Psychology, 61, 383–409. doi: 10.1146/annurev.psych.093008.100448 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Enders, C. K. (2001). The performance of the Full Information Maximum Likelihood estimator in multiple regression models with missing data. Educational and Psychological Measurement, 61(5), 713–740. doi: 10.1177/0013164401615001 [DOI] [Google Scholar]
  11. Ferrer, E., Balluerka, N., & Widaman, K. F. (2008). Factorial invariance and the specification of second-order latent growth models. Methodology: European Journal of Research Methods for the Behavioral & Social Sciences, 4(1), 22–36. doi: 10.1027/1614-2241.4.1.22 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Flora, D. B., & Curran, P. J. (2004). An empirical evaluation of alternative methods of estimation for confirmatory factor analysis with ordinal data. Psychological Methods, 9(4), 466–491. doi: 10.1037/1082-989X.9.4.466 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Fox, C. S., Golden, S. H., Anderson, C., Bray, G. A., Burke, L. E., de Boer, I. H., Deedwania, P., Eckel, R. H., Ershow, A. G., Fradkin, J., Inzucchi, S. E., Kosiborod, M., Nelson, R. G., Patel, M. J., Pignone, M., Quinn, L., Schauer, P. R., Selvin, E., & Vafiadis, D. K.; American Heart Association Diabetes Committee of the Council on Lifestyle and Cardiometabolic Health; Council on Clinical Cardiology, Council on Cardiovascular and Stroke Nursing, Council on Cardiovascular Surgery and Anesthesia, Council on Quality of Care and Outcomes Research; American Diabetes Association . (2015). Update on prevention of cardiovascular disease in adults with type 2 diabetes mellitus in light of recent evidence: A scientific statement from the American Heart Association and the American Diabetes Association. Diabetes Care, 38(9), 1777–1803. doi: 10.2337/dci15-0012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gaupp, R., Berrios, G. E., & Pomarol-Clotet, E. (2000). Depressive states in old age. (Classic Text No. 42). History of Psychiatry, 11(42 Pt 2), 213–225. doi: 10.1177/0957154X0001104205 [DOI] [PubMed] [Google Scholar]
  15. Grundy, S. M., Stone, N. J., Bailey, A. L., Craig, B., Birtcher, K. K., Blumenthal, R. S., Braun, L. T., Sarah, de F., Joseph, F.-T., Forman, D. E., Ronald, G., Heidenreich, P. A., Hlatky, M. A., Jones, D. W., Donald, L.-J., Nuria, L.-P., Ndumele, C. E., Orringer, C. E., Peralta, C. A., … Joseph, Y. (2019). Guideline on the management of blood cholesterol: Executive summary: A report of the American College of Cardiology/American Heart Association task force on clinical practice guidelines. Circulation, 139(25), e1046–e1081. doi: 10.1161/CIR.0000000000000624 [DOI] [PubMed] [Google Scholar]
  16. Hancock, G. R., Kuo, W.-L., & Lawrence, F. R. (2001). An illustration of second-order latent growth models. Structural Equation Modeling: A Multidisciplinary Journal, 8(3), 470–489. doi: 10.1207/S15328007SEM0803_7 [DOI] [Google Scholar]
  17. Helgeson, V. S., Van Vleet, M., & Zajdel, M. (2020). Diabetes stress and health: Is aging a strength or a vulnerability? Journal of Behavioral Medicine, 43(3), 426–436. doi: 10.1007/s10865-019-00106-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Iob, E., Kirschbaum, C., & Steptoe, A. (2020). Persistent depressive symptoms, HPA-axis hyperactivity, and inflammation: The role of cognitive-affective and somatic symptoms. Molecular Psychiatry, 25(5), 1130–1140. doi: 10.1038/s41380-019-0501-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Jensen, M. D., Ryan, D. H., Apovian, C. M., Ard, J. D., Comuzzie, A. G., Donato, K. A., Hu, F. B., Hubbard, V. S., Jakicic, J. M., Kushner, R. F., Loria, C. M., Millen, B. E., Nonas, C. A., Pi-Sunyer, F. X., Stevens, J., Stevens, V. J., Wadden, T. A., Wolfe, B. M., Yanovski, S. Z. (2014). 2013 AHA/ACC/TOS guideline for the management of overweight and obesity in adults: A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and The Obesity Society. Circulation, 129(25 suppl. 2), S102–S138. doi: 10.1161/01.cir.0000437739.71477.ee [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Jylhä, M. (2009). What is self-rated health and why does it predict mortality? Towards a unified conceptual model. Social Science & Medicine (1982), 69(3), 307–316. doi: 10.1016/j.socscimed.2009.05.013 [DOI] [PubMed] [Google Scholar]
  21. Karmali, K. N., & Lloyd-Jones, D. M. (2014). Achieving and maintaining cardiovascular health across the lifespan. Current Epidemiology Reports, 1(2), 75–81. doi: 10.1007/s40471-014-0011-7 [DOI] [Google Scholar]
  22. Kivimäki, M., Shipley, M. J., Allan, C. L., Sexton, C. E., Jokela, M., Virtanen, M., Tiemeier, H., Ebmeier, K. P., & Singh-Manoux, A. (2012). Vascular risk status as a predictor of later-life depressive symptoms: A cohort study. Biological Psychiatry, 72(4), 324–330. doi: 10.1016/j.biopsych.2012.02.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lahey, B. B. (2009). Public health significance of neuroticism. The American Psychologist, 64(4), 241–256. doi: 10.1037/a0015309 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Laina, A., Stellos, K., & Stamatelopoulos, K. (2018). Vascular ageing: Underlying mechanisms and clinical implications. Experimental Gerontology, 109, 16–30. doi: 10.1016/j.exger.2017.06.007 [DOI] [PubMed] [Google Scholar]
  25. Lenze, E. J., Rogers, J. C., Martire, L. M., Mulsant, B. H., Rollman, B. L., Dew, M. A., Schulz, R., & Reynolds, C. F.3rd. (2001). The association of late-life depression and anxiety with physical disability: A review of the literature and prospectus for future research. The American Journal of Geriatric Psychiatry, 9(2), 113–135. doi: 10.1097/00019442-200105000-00004 [DOI] [PubMed] [Google Scholar]
  26. Li, C. H. (2016). Confirmatory factor analysis with ordinal data: Comparing robust maximum likelihood and diagonally weighted least squares. Behavior Research Methods, 48(3), 936–949. doi: 10.3758/s13428-015-0619-7 [DOI] [PubMed] [Google Scholar]
  27. Lloyd-Jones, D. M., Hong, Y., Labarthe, D., Mozaffarian, D., Appel, L. J., Van Horn, L., Greenlund, K., Daniels, S., Nichol, G., Tomaselli, G. F., Arnett, D. K., Fonarow, G. C., Ho, M. P., Lauer, M. S., Masoudi, F. A., Robertson, R. M., Roger, V., Schwamm, L. H., Sorlie, P., … Rosamond, W. D.; American Heart Association Strategic Planning Task Force and Statistics Committee . (2010). Defining and setting national goals for cardiovascular health promotion and disease reduction: The American Heart Association’s strategic Impact Goal through 2020 and beyond. Circulation, 121(4), 586–613. doi: 10.1161/CIRCULATIONAHA.109.192703 [DOI] [PubMed] [Google Scholar]
  28. Luijendijk, H. J., Stricker, B. H., Hofman, A., Witteman, J. C., & Tiemeier, H. (2008). Cerebrovascular risk factors and incident depression in community-dwelling elderly. Acta Psychiatrica Scandinavica, 118(2), 139–148. doi: 10.1111/j.1600-0447.2008.01189.x [DOI] [PubMed] [Google Scholar]
  29. Lyness, J. M., King, D. A., Conwell, Y., Cox, C., & Caine, E. D. (2000). Cerebrovascular risk factors and 1-year depression outcome in older primary care patients. The American Journal of Psychiatry, 157(9), 1499–1501. doi: 10.1176/appi.ajp.157.9.1499 [DOI] [PubMed] [Google Scholar]
  30. Maillard, P., Seshadri, S., Beiser, A., Himali, J. J., Au, R., Fletcher, E., Carmichael, O., Wolf, P. A., & DeCarli, C. (2012). Effects of systolic blood pressure on white-matter integrity in young adults in the Framingham Heart Study: A cross-sectional study. The Lancet. Neurology, 11(12), 1039–1047. doi: 10.1016/S1474-4422(12)70241-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Mast, B. T., Azar, A. R., & Murrell, S. A. (2005). The vascular depression hypothesis: The influence of age on the relationship between cerebrovascular risk factors and depressive symptoms in community dwelling elders. Aging & Mental Health, 9(2), 146–152. doi: 10.1080/13607860412331336832 [DOI] [PubMed] [Google Scholar]
  32. Matthews, K. A., Jennings, J. R., Lee, L., & Pardini, D. (2019). Depressive symptoms during childhood and cardiovascular risk factors in black and white men. Psychosomatic Medicine, 81(2), 176–183. doi: 10.1097/PSY.0000000000000652 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Montagnier, D., Dartigues, J. F., Rouillon, F., Pérès, K., Falissard, B., & Onen, F. (2014). Ageing and trajectories of depressive symptoms in community-dwelling men and women. International Journal of Geriatric Psychiatry, 29(7), 720–729. doi: 10.1002/gps.4054 [DOI] [PubMed] [Google Scholar]
  34. Muntner, P., Shimbo, D., Carey, R. M., Charleston, J. B., Gaillard, T., Misra, S., Myers, M. G., Ogedegbe, G., Schwartz, J. E., Townsend, R. R., Urbina, E. M., Viera, A. J., White, W. B., & Wright, J. T.Jr. (2019). Measurement of blood pressure in humans: A scientific statement from the American Heart Association. Hypertension (Dallas, Tex.: 1979), 73(5), e35–e66. doi: 10.1161/HYP.0000000000000087 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Pase, M. P., Davis-Plourde, K., Himali, J. J., Satizabal, C. L., Aparicio, H., Seshadri, S., Beiser, A. S., & DeCarli, C. (2018). Vascular risk at younger ages most strongly associates with current and future brain volume. Neurology, 91(16), e1479–e1486. doi: 10.1212/WNL.0000000000006360 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Qiu, C., & Fratiglioni, L. (2015). A major role for cardiovascular burden in age-related cognitive decline. Nature Reviews. Cardiology, 12(5), 267–277. doi: 10.1038/nrcardio.2014.223 [DOI] [PubMed] [Google Scholar]
  37. Qiu, C., Winblad, B., & Fratiglioni, L. (2005). The age-dependent relation of blood pressure to cognitive function and dementia. The Lancet. Neurology, 4(8), 487–499. doi: 10.1016/S1474-4422(05)70141-1 [DOI] [PubMed] [Google Scholar]
  38. R Core Team . (2014). R: A language and environment for statistical computing.R Foundation for Statistical Computing. http://www.R-project.org/ [Google Scholar]
  39. Rhemtulla, M., Brosseau-Liard, P. É., & Savalei, V. (2012). When can categorical variables be treated as continuous? A comparison of robust continuous and categorical SEM estimation methods under suboptimal conditions. Psychological Methods, 17(3), 354–373. doi: 10.1037/a0029315 [DOI] [PubMed] [Google Scholar]
  40. Rosseel, Y. (2012). lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48(2), 1–36. doi: 10.18637/jss.v048.i02 [DOI] [Google Scholar]
  41. Schaakxs, R., Comijs, H. C., Lamers, F., Beekman, A. T., & Penninx, B. W. (2017). Age-related variability in the presentation of symptoms of major depressive disorder. Psychological Medicine, 47(3), 543–552. doi: 10.1017/S0033291716002579 [DOI] [PubMed] [Google Scholar]
  42. Schaare, H. L., Kharabian Masouleh, S., Beyer, F., Kumral, D., Uhlig, M., Reinelt, J. D., Reiter, A. M. F., Lampe, L., Babayan, A., Erbey, M., Roebbig, J., Schroeter, M. L., Okon-Singer, H., Müller, K., Mendes, N., Margulies, D. S., Witte, A. V., Gaebler, M., & Villringer, A. (2019). Association of peripheral blood pressure with gray matter volume in 19- to 40-year-old adults. Neurology, 92(8), e758–e773. doi: 10.1212/WNL.0000000000006947 [DOI] [PubMed] [Google Scholar]
  43. Smagula, S. F., & Aizenstein, H. J. (2016). Brain structural connectivity in late-life major depressive disorder. Biological Psychiatry. Cognitive Neuroscience and Neuroimaging, 1(3), 271–277. doi: 10.1016/j.bpsc.2015.11.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Spector, W. D., & Fleishman, J. A. (1998). Combining activities of daily living with instrumental activities of daily living to measure functional disability. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 53(1), S46–S57. doi: 10.1093/geronb/53b.1.s46 [DOI] [PubMed] [Google Scholar]
  45. Stamler, J., Stamler, R., Neaton, J. D., Wentworth, D., Daviglus, M. L., Garside, D., Dyer, A. R., Liu, K., & Greenland, P. (1999). Low risk-factor profile and long-term cardiovascular and noncardiovascular mortality and life expectancy: Findings for 5 large cohorts of young adult and middle-aged men and women. Journal of the American Medical Association, 282(21), 2012–2018. doi: 10.1001/jama.282.21.2012 [DOI] [PubMed] [Google Scholar]
  46. Steptoe, A., Breeze, E., Banks, J., & Nazroo, J. (2013). Cohort profile: The English Longitudinal Study of Ageing. International Journal of Epidemiology, 42(6), 1640–1648. doi: 10.1093/ije/dys168 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Steunenberg, B., Twisk, J. W., Beekman, A. T., Deeg, D. J., & Kerkhof, A. J. (2005). Stability and change of neuroticism in aging. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 60(1), P27–P33. doi: 10.1093/geronb/60.1.p27 [DOI] [PubMed] [Google Scholar]
  48. Stewart, R., Prince, M., Richards, M., Brayne, C., & Mann, A. (2001). Stroke, vascular risk factors and depression. British Journal of Psychiatry, 178(01), 23–28. doi: 10.1192/bjp.178.1.23 [DOI] [PubMed] [Google Scholar]
  49. Sutin, A. R., Terracciano, A., Milaneschi, Y., An, Y., Ferrucci, L., & Zonderman, A. B. (2013). The trajectory of depressive symptoms across the adult life span. JAMA Psychiatry, 70(8), 803–811. doi: 10.1001/jamapsychiatry.2013.193 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Taylor, W. D., Aizenstein, H. J., & Alexopoulos, G. S. (2013). The vascular depression hypothesis: Mechanisms linking vascular disease with depression. Molecular Psychiatry, 18(9), 963–974. doi: 10.1038/mp.2013.20 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Turvey, C. L., Wallace, R. B., & Herzog, R. (1999). A revised CES-D measure of depressive symptoms and a DSM-based measure of major depressive episodes in the elderly. International Psychogeriatrics, 11(2), 139–148. doi: 10.1017/s1041610299005694 [DOI] [PubMed] [Google Scholar]
  52. Ubel, P. A., Jankovic, A., Smith, D., Langa, K. M., & Fagerlin, A. (2005). What is perfect health to an 85-year-old? Evidence for scale recalibration in subjective health ratings. Medical Care, 43(10), 1054–1057. doi: 10.1097/01.mlr.0000178193.38413.70 [DOI] [PubMed] [Google Scholar]
  53. Valkanova, V., & Ebmeier, K. P. (2013). Vascular risk factors and depression in later life: A systematic review and meta-analysis. Biological Psychiatry, 73(5), 406–413. doi: 10.1016/j.biopsych.2012.10.028 [DOI] [PubMed] [Google Scholar]
  54. Virtanen, M., Elovainio, M., Josefsson, K., Batty, G. D., Singh-Manoux, A., & Kivimäki, M. (2017). Coronary heart disease and risk factors as predictors of trajectories of psychological distress from midlife to old age. Heart (British Cardiac Society), 103(9), 659–665. doi: 10.1136/heartjnl-2016-310207 [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Wikman, A., Wardle, J., & Steptoe, A. (2011). Quality of life and affective well-being in middle-aged and older people with chronic medical illnesses: A cross-sectional population based study. PLoS One, 6(4), e18952. doi: 10.1371/journal.pone.0018952 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. World Health Organization . (2019). 10 priorities. Towards a decade of healthy ageing. https://www.who.int/ageing/WHO-ALC-10-priorities.pdf
  57. Wrosch, C., & Heckhausen, J. (2005). Being on-time or off-time: Developmental deadlines for regulating one’s own development. In A.-N. Perret-Clermont (Ed.), Thinking time: A multidisciplinary perspective on time (pp. 110–123). Hogrefe & Huber Publishers. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

gbab063_suppl_Supplementary_Materials
Click here for additional data file. (863.4KB, pdf)

Articles from The Journals of Gerontology Series B: Psychological Sciences and Social Sciences are provided here courtesy of Oxford University Press

RESOURCES