The relationship between depression and cognitive function in adults with cardiovascular risk: Evidence from a randomised attention-controlled trial
The relationship between depression and cognitive function in adults with cardiovascular risk: Evidence from a randomised attention-controlled trial
Haley M. LaMonica 0 1
Daniel J. Biddle 0 1
Sharon L. Naismith 0 1
Ian B. Hickie 0 1
Paul Maruff 0
Nicholas Glozier 0 1
0 Editor: Camillo Gualtieri, North Carolina Neuropsychiatry Clinics , UNITED STATES
1 Brain and Mind Centre, University of Sydney , Camperdown , Australia , 2 Charles Perkins Centre, School of Psychology, University of Sydney , Camperdown , Australia , 3 Central Clinical School, Sydney Medical School, University of Sydney , Camperdown , Australia , 4 Cogstate, Melbourne , Australia
Background and aim This study assessed the association between depressive symptom severity and cognition in middle-to-older aged adults with mild-to-moderate depression and cardiovascular risk factors using an online test battery (CogState) and whether changes in depressive symptoms over 3 months were associated with changes in cognition. Participants (mean age = 57.8) with cardiovascular risk and mild±to-moderate depressive symptoms completed measures of psychomotor speed, learning, and executive function prior to (n = 445)_and after (n = 334) online depression or attention control interventions. The symptom severity-cognition relationship was examined both cross-sectionally and
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
Funding: The trial was supported by the
Cardiovascular Disease and Depression Strategic
Research Program (Award Reference No. G08S
4048 to I.H.) funded by the National Heart
Foundation of Australia and beyondblue: the
national depression initiative. The 45 and Up Study
is managed by The Sax Institute in collaboration
Participants exhibited significantly reduced psychomotor speed and variable impairments
on measures of learning and executive functioning relative to normative data. However,
there was no association of depression severity with cognition at baseline or of change in
depressive symptoms with change in cognitive performance.
Participants were well-educated, which may have protected against cognitive decline.
Attrition may limit generalisability, though is unlikely to explain the lack of association between
depression symptoms and cognition.
with major partner Cancer Council New South
Wales; and partners the National Heart Foundation
of Australia (NSW Division); NSW Health;
beyondblue: the national depression initiative;
Ageing, Disability and Home Care, Department of
Human Services NSW; and UnitingCare Ageing.
Competing interests: Professors Naismith, Hickie,
and Glozier received a grant from beyondblue to
support the conduct of the study. Professor Hickie
served as the Commissioner of the National Mental
Health Commission and was a member of the
Medibank Clinical Reference Group and the Bupa
Australia Medical Advisory Board during this study.
Professor Paul Maruff is a full time employee of
Cogstate. All other authors declare that they have
no conflicts of interest. This does not alter our
adherence to PLOS ONE policies on sharing data
Adults with comorbid mild-to-moderate depressive symptoms and cardiovascular risks
performed less well than age-matched normative data on three online cognitive tests; however,
we were unable to show any symptom-cognition association cross-sectionally or
longitudinally, despite significant improvements in depressive symptoms. This challenges the
generalisability of such associations found in more severely unwell clinical samples to those with
a broader depressive symptom profile, or suggests that underlying cardiovascular disease
may account for the association seen in some clinical studies. This has implications for
scaling up selective prevention of cognitive decline.
It is recognized widely that Major Depressive Disorder (MDD) is associated with cognitive
dysfunction [1±3] including impaired learning, working memory, processing speed, and
executive functions [4±6]. The neuropsychological profile is, however, heterogeneous and varies
with depressive symptom severity [7, 8], disease subtype , age of onset [
comorbidities and cerebrovascular disease (CVD) . Regardless of the contributing factors,
cognitive dysfunction in depression is associated with substantial disability and poorer quality
of life , and frequently has been suggested as a selective target for interventions aimed at
preventing cognitive decline and subsequent dementia .
There are three key issues that might limit such a strategy:
1. Is the depression severity-cognition association present only in the small group with
diagnosed MDD or is it applicable to the much larger group with mild to moderate depression
2. Does improvement in depression result in a significant improvement in cognitive function?
The observed cognitive dysfunction of MDD often persists despite reductions in depressive
symptoms [13, 14], and any improvements may not reach the levels of cognition that
existed prior to onset of the depressive episode; the ªscar hypothesisº [13, 15, 16]. Up to
75% of older adults whose symptoms of MDD have reduced following therapy with
antidepressant medication can still be classified as having subtle cognitive dysfunction [
3. Is the observed association between depression and cognition a reflection of an underlying
common cause such as Cerebrovascular Disease (CVD)? `Vascular depression' describes a
syndrome whereby depression occurs for the first time in later life, associated with
underlying CVD, marked cognitive impairment and poor prognosis. In this instance, CVD is
thought to cause structural changes to CNS white matter  which themselves give rise to
slowed processing speed, poor memory and executive dysfunction as well as an increased
likelihood of progression to dementia . Thus, any examination of depression and
cognition in middle-to-older aged adults needs to take into account the presence of CVD and
associated vascular risk factors.
Given these questions, we undertook a secondary analysis of a large trial of an online
treatment that was shown to be effective for depressive symptoms in community-based adults
with mild-to-moderate depression symptoms and self-reported history of CVD or
cardiovascular risk factors who underwent cognitive testing at baseline and post intervention. We
2 / 14
hypothesised that the depressive symptom severity-cognition association would be found in
this group, and that improvements in depressive symptoms would be associated with
improvement on cognitive testing despite the presence of cardiovascular risk for cognitive dysfunction.
Materials and methods
The protocol of the Cardiovascular Risk and E-Couch Depression Outcome (CREDO) trial
has been published elsewhere  and the primary results of the CREDO study have been
published . CREDO was a double-blind, parallel group randomised controlled trial designed
to compare the effectiveness of internet cognitive behavioural therapy (eCBT) relative to an
online attention control in adults, aged 45- to 75-years, with self-reported CVD or significant
CVD risk factors, and repeated evidence of mild±to-moderate depressive symptoms.
The trial complied with the Code of Ethics of the World Medical Association. Ethical
approval was obtained from the University of Sydney Human Research Ethics Committee in
June of 2009 and from the Australian National University Human Research Ethics Committee
in 2010. Additionally, the trial was registered with the Australian and New Zealand Clinical
Trials Registry (ACTRN12610000085077). The 45 and Up Study had primary ethical approval
from the University of New South Wales Human Research Ethics Committee.
Participants (n = 562) were recruited through the 45 and Up Study, a large-scale longitudinal
population-based cohort study comprising over 260,000 men and women aged 45-years and
over in New South Wales (NSW), Australia . Comparative analyses between the 45 and Up
Study and the NSW Population Health Survey revealed that the estimated relative risk of a
range of health-related risk factors generalised to the wider population . From July 2010
and January 2011, participants were randomly selected from the 45 and Up Study database
using an algorithm to identify those that met the eligibility criteria described in detail in
Glozier et al. (2013). In summary, participants were required to have provided a valid email
address and to have a self-reported history of CVD or CVD risk factors and persistent or
recurrent symptoms of depression. Following an online informed consent process,
participants completed online baseline assessments and were randomised to either intervention arm
using a customised, fully automated randomisation facility built into the trial website.
Randomisation was stratified by depressive symptom severity. Participants were blinded to which
programme was the `active' intervention.
The ªactive' intervention was E-couch, an automated software program that offers 12 modules
addressing mental health literacy, cognitive behavioural therapy (CBT), interpersonal therapy
(IPT), relaxation techniques and exercise programs targeting depression. This has been
demonstrated to be effective compared to an attention control in reducing depressive symptoms in
younger groups without comorbidities .
The attention control was HealthWatch, a 12-week online program in which participants
were provided with information on a variety of topics including nutrition, physical activity,
heart health, and pain. The attention control intervention was matched for contact (i.e.,
12 modules) and was used to balance the effects of the expectation of therapeutic benefit
offered in the active treatment arm. In previous trials, participants showed a small, although
3 / 14
significant reduction in depressive symptoms using this program, potentially reflecting natural
remission or some small effect [
Depressive symptoms, the primary pre-specified outcome measure for the trial, were measured
using the PHQ-9 at baseline, 4-, 8-, and 12-weeks. The PHQ-9 is a nine item measure of DSM
depression symptoms which has been found to be a reliable and valid screening measure, is
widely used in previous community studies of people with depression, and is sensitive to
change in clinical status [
Three aspects of cognition were assessed at baseline and 12 weeks using an online version
of Cogstate, a computerised neuropsychological test battery with established validity and
sensitivity for the detection of subtle cognitive change in community cohorts in general and also in
]. For this trial, an internet version was utilised, which was downloaded onto
participants' computers to prevent bandwidth and connectivity interference. The test battery was
chosen for brevity as well as capacity to assess neuropsychological skills commonly affected in
older people with depression [
]. The Detection Task required participants to respond when
a card presented on the screen turned over the course of two minutes. Speed was the primary
outcome, with lower scores reflecting better performance. Visual learning and memory was
assessed with the One Card Learning Task. During this five minute, one-back style task,
participants responded `yes' or `no' if a card presented on-screen was the same as a previous card.
The score reflects accuracy of performance. Executive functioning was assessed with the
Groton Maze Task. In this five minute, spatial problem-solving task, participants were shown a
grid of tiles on-screen and had to use the mouse to find a hidden pathway based on trial and
error feedback. The score reflects the number of errors, again with lower scores reflecting
As part of their baseline assessment through the 45 and Up Study, participants completed a
questionnaire about demographic and social characteristics, personal health behaviours, and
general health-related data . These data, the coding of which is standard and detailed in
 were included in the present study for the purposes of characterising the sample and to
assess attrition bias and potential confounding.
Descriptive statistics were used to summarise the baseline demographic and health
characteristics of the study sample, and their association with the measures of cognition. One-sample
ttests were then used to compare performance on the cognitive tests between the CREDO
sample and age-matched normative data based on a healthy population of subjects enrolled in
clinical trials as well as research and academic studies [
Hypothesis one: Depressive symptom severity-cognition association. Cognitive
function values were then converted to normalised scores based on the above-mentioned
agematched normative data [
], and Spearman correlations were used to estimate the
relationship between depression severity and cognition at baseline. Pearson product-moment
correlations were used to assess the relationship between normally-distributed cognitive and binary
sociodemographic and clinical characteristics. Linear regression using the enter method was
also used to explore this relationship adjusting for potential confounders. At step 1 depression
was entered alone, at step 2 age, gender, education (post-school versus no post-school),
psychotropic medication, and intervention arm were entered, and finally diagnosed and treated
cardiovascular disease was entered at step 3.
4 / 14
Before conducting linear regression models, major assumptions of the approach were
assessed. Homoscedasticity and normality of residuals was confirmed through inspection of
scatterplots and graphs, respectively. Lack of association between independent variables and
residuals was confirmed through examining correlations for continuous independent variables
and t-tests for binary independent variables. Potential multicollinearity was assessed via
examining variance inflation factors (VIFs). As residuals for the cross-sectional analysis of executive
function were non-normal, executive function was further transformed into its natural
logarithm for this analysis.
Outliers, defined as cases scoring +/-3 standard deviations from the mean on the outcome
or independent variable of interest for univariate analyses (t-tests and correlations), and
standardized residuals <-3 or >3 for multivariate analyses (regressions), were excluded.
Hypothesis two: Change in cognitive function. For those with baseline and follow-up
cognitive function data, paired t-tests were used to assess change in cognitive variables and
depression over the course of the study. Change scores were then calculated for each
normalised cognitive function variable, along with depressive symptoms, by subtracting follow-up
scores from baseline scores. Linear regression was used to estimate associations between
changes in depressive symptoms and changes in cognitive function and then adjusted, in a
stepwise fashion as above, for age, gender, education, depression symptoms, medication and
intervention arm, and then the presence of diagnosed CVD and risk factors. Outliers (cases
with standardized residuals <-3 or >3, were excluded from these analyses. In sensitivity
analyses the regression models were re-run with MDD, current depression symptom severity
(centered), and their interaction, entered at step 1, to gauge whether the depression-cognition
relationship was apparent for those with probable MDD at baseline. Probable MDD was
defined as scores of 2 or more on either questions 1 or 2 of the PHQ-9, and a total of 5 or more
items with scores of 2 or 3 [
]. In a further sensitivity analysis designed to take into account
status at baseline, latent change scores were created for depression and cognitive function
tasks by regressing status at baseline on status post-intervention. The residuals, which equate
to the estimated change in each variable not explained by status at baseline, were entered into
regression models in the same manner described above. Results are reported such that positive
coefficients equate to better cognitive performance or improvement (i.e. faster reaction times,
greater learning accuracy, fewer executive function task errors).
As referenced above, the primary results of the CREDO trial have been published previously
. For the purposes of this study, the data from both intervention arms was pooled.
Baseline demographics and health characteristics
Of the 562 recruited participants, 445 (79.2%) completed at least one baseline cognitive task.
Baseline characteristics for these participants are presented in Table 1. Participants had a mean
age of 58-years. The majority of participants were female (63.6%), spoke English at home
(94.8%), and had more than a high school level of education (73.2%). Over half the sample
(55.5%) had a prior diagnosis of depression. At baseline, the mean PHQ-9 score was 11.93
(SD: 3.4) and 24% (n = 107) of the participants met criteria for probable MDD. There were
no statistically significant differences in any variable between arms indicating adequate
randomisation . There were few significant differences between those who completed at least
one baseline cognitive task and those who did not. Those with baseline cognitive data were
more likely to be female (63.6% vs 53.0%), partnered (74.5% vs 64.0%), taking psychotropic
5 / 14
+ n = 1 missing education.
++ Probable MDD was defined as scores of 2 or more on either questions 1 or 2 of the PHQ-9, and a total of 5 or more items with scores of 2 or 3.
~ Prior Diagnosis of Cardiovascular Disease includes doctor diagnosis of any one of Heart Disease, Stroke or Hypertension.
^ Optimal screening cut points for the AUDIT-C, a 3-item alcohol screening questionnaire, for alcohol dependence based on a sample of individuals with a history of
psychopathology in the past year. NB n = 88 missing AUDIT scores.
$ At least 150 mins of self-report moderate activity over at least 5 sessions each week
# Treatment for any Cardiovascular Disease includes: any one of heart attack/angina, other heart disease hypertension or high blood cholesterol.
## Other comorbid conditions include: cancer (skin, prostate, breast or other cancer), blood clot (thrombosis), asthma, Parkinson's disease, osteoarthritis, and/or thyroid
p < .05.
p < .01.
p < .001
NB 4 participants did not complete psychomotor speed task (DET) at baseline. Outliers (cases with scores scores +/- 3 SD from the mean) were excluded from
correlations for continuous variables psychomotor speed (n = 6), visual learning and memory (n = 5), executive function (n = 3), and depression (n = 2).
PLOS ONE | https://doi.org/10.1371/journal.pone.0203343
6 / 14
medication (35.96% vs 25.86%) and younger (M = 57.54 vs 59.48), than those who did not
complete any baseline cognitive measures.
Although cognitive function was worse in the older participants (slower psychomotor
speed (ρ = -.194, p <.001), worse learning (ρ = -.104, p <.05), and worse executive function (ρ
= -.123, p <.01)) there were few other associations of demographic, illness or social variables
with the cognitive function measures. Of the 54 associations analysed in addition to age, only
three were associated, and then with only one of the measures each, a number quite likely due
Cognitive function in mild-to-moderate depression compared to normative
Compared to aged-matched groups from the CogState normative data (Table 2), the study
participants showed a significant impairment in psychomotor speed in each age group. Similar
results were seen in relation to learning accuracy, with significant deficits noted for
participants aged 50 to 59 and 60 to 69, but not for those in their 70s. Those in their 60s made fewer
errors on the executive functioning tasks than age-matched peers, however when outliers
(n = 2) we included this difference was no longer significant. There were no significant
differences in performance level on the executive functioning task for those subjects in their 50s,
and 70s relative to age-matched peers. A one-way ANOVA suggested significant differences in
depressive symptoms as a function of age (F(3,439) = 5.396, p = .001), and post-hoc contrasts
indicated participants 70±79 years of age had less depressive symptoms at baseline than those
50±59 years of age (t = 2.15, p = 0.032).
Course of depressive symptoms and cognition over the study
Over the 12 weeks of the study, participants in both arms showed improvements in depressive
symptoms of 3.66 points (95% CI: 3.05±4.27) with eCBT and 2.60 points (95% CI: 2.05±3.16)
in the control group, with a significantly greater decline in the PHQ-9 for eCBT compared to
control (1.06; 95% CI: 0.23±1.89; time by arm interaction p = .012).
The complete results relating to the effects of the web-based intervention on depression
symptom severity were published previously . Of participants with baseline cognitive
function data (n = 445), 75% (n = 334) completed at least one of the cognitive measures at week 12
(see Table 3). Those who were missing week 12 cognitive data had lower baseline scores for the
learning task (M = -.59, SD = .76) than completers (M = -.37, SD = .79, t(443) = -2.623, p =
.009), and worse scores for executive function (M = -.05, SD = .95) than completers (M = .16,
SD = .89, p = .035), but did not differ in baseline psychomotor speed or PHQ-9 scores. Those
who provided only baseline data did not differ on most characteristics, but were less likely to
speak English at home (89.2% vs 96.7%, X2 = 9.606, p = .002) and be obese (56.9% vs 44.4%, X2
= 5.078, p = .024), neither of which showed any association with baseline cognition. Of the
cognitive measures only executive function showed any change over the 12 weeks of the study.
Cross-sectional association of depression severity, and cognition
Baseline depression severity was not associated with performance on any of the cognitive
tasks, with correlations ranging from -.038 to .027, and p-values from .425 to .768 (see
Table 1). In the linear regression models there were no associations between depression
severity and cognitive function either with or without adjustment for potential confounders (see
7 / 14
Executive function: total number of errors made in attempting to learn the same hidden pathway on five
consecutive trials at a single session on Groton Maze Task
PHQ-9 continuous: depressive symptom severity
Outliers (cases with scores scores +/- 3 SD from the mean) were excluded from these analyses for psychomotor speed (n = 6), visual learning and memory (n = 5),
executive function (n = 3), and depression (n = 2). Comparisons with cognitive function norms for the 45-49-year-olds are not displayed as norms were only available
for those 35±49 years of age.
Association of change in cognitive function and change in depressive
Change in depression symptoms was not associated with change in any of the cognitive
function variables (Table 4). There was no effect on this lack of association when adjusting for
potential confounders. However, in the sensitivity analysis including an interaction between
symptom severity and probable MDD, a significant interaction was observed for change in
psychomotor speed at the first step of the model (t = 2.746, p = .006). The model became
non8 / 14
significant when covariates were added, and no covariates were significant predictors of
change in psychomotor speed, so the first model (with change in depression, probable MDD,
and their interaction) was retained. In a post-hoc split-file regression analysis it was found that
only for those with probable MDD was change in depressive symptom severity associated
positively with change in psychomotor speed. For these participants, change in depressive
symptoms explained 12.6% of the variability in change in psychomotor speed, with each standard
deviation increase in depression change associated with a .355 standard deviation increase in
psychomotor speed change (t = 3.285, p = .002).
In the latent change sensitivity analysis, the results mirrored those of the initial analysis i.e.
no prospective associations between depression and cognitive function were found.
This study suggests that people with mild-to-moderate depressive symptoms and self-reported
CVD or cardiovascular risk factors generally perform below an aged matched normative
Variables entered at Step 1 = depression, Step 2 = age, gender, education, psychotropic medication, treatment arm, Step 3 = cardio-vascular disease. B = standardized
beta coefficients. Residual outliers removed for cross-sectional analysis of psychomotor speed (n = 8), learning and memory (n = 5), and executive function (n = 5).
Residual outliers also removed for change score analysis of psychomotor speed (n = 7), learning and memory (n = 6), and executive function (n = 3).
9 / 14
database on measures of psychomotor speed and learning accuracy. Participants aged 60 to
69-years and 70-to-79 years also exhibited executive dysfunction, but there was no evidence of
executive dysfunction for those participants in their 50s. Contrary to our hypotheses there was
no association at baseline between depression symptom severity and cognitive performance
and no suggestion of potential (reverse) confounding by variables including age, gender,
education, psychotropic medications, treatment arm, and diagnosed CVD. This null finding
suggests that the symptom severity cognition association commonly observed in clinical samples
 may not be evident in community settings where individuals may have a more
heterogeneous and milder pattern of symptoms. Given that cognitive dysfunction is associated with
greater disability in depression [
] and that functional impairment can drive help seeking,
such a discrepancy between settings might be expected. This does, however, have implications
for the generalisation of findings from clinical settings.
As previously reported , over the course of the study, participants in both arms showed
improvements in depressive symptoms, significantly greater in the active arm, although with a
small effect size. However, only executive function improved over the course of the study and
there was no association of depressive symptom reduction with cognitive improvement over
the 12 weeks overall. Amongst the small group with probable MDD at baseline, the change in
depressive symptom severity was associated with change in psychomotor speed, but not in
learning or executive functions. These results may indicate a real lack of association between
changes in depressive symptom and cognition in this group with predominantly
mild-to-moderate depression, especially given there was no association of symptoms and cognition at
baseline. The lack of association may also reflect the small size of the change in depression and that
of the cognitive measures only executive function showed statistical improvement. However,
although the mean changes were small, their marked variance could have reflected
intra-individual correlations. It may be that in this study a mean change of only 3 or 4 on the PHQ-9
may be inadequate to lead to observable changes in neuropsychological test performance, or
that there may be a time lag between changes in affective symptoms and cognition as has been
observed with function [
In contrast to previous studies, we did not find that elevated symptoms of depression were
associated with greater cognitive dysfunction nor did we find that a reduction in depression
symptom severity was associated with cognitive improvement . Furthermore, based on
available data, we cannot conclude that the relationship between depression symptom severity
and cognition was masked by a common cause, such as diagnosed CVD; however, it is possible
that unmeasured confounders such as age of onset (early vs. late onset depression), duration of
illness, episode frequency [
], comorbid anxiety, underlying white matter change or other
medical comorbidities may have mediated any relationship between depressive symptom
severity and cognition, thus potentially concealing an association in this instance. The
suggestion that there was however a symptom-cognition association in those with probable MDD is
intriguing. Could it be that the frequently observed association is only present in those with a
ªdisorderº and that they are somehow qualitatively different, as there was not much greater
variance in the measures in this subgroup which was of a small sample size?
A further reason for the lack of association between depression symptoms and cognition in
this study may be that any association is specific to certain cognitive domains. The
meta-analysis  did suggest that the association was stronger in executive function than in processing
speed, something observed previously in a cross-sectional study of older adults where
increasing severity of depression was related to lower semantic fluency scores and poorer ability to
shift cognitive set, but not to worsening performances on measures of learning and memory,
phonemic fluency, or complex problem-solving [
]. Additionally, contradictory results have
been found with regard to the relationship between depression severity and speed of
10 / 14
information processing [
]. Therefore, it is plausible that the online cognitive tests may not
have been sensitive to the cognitive deficits typical of individuals with depression, or those that
might change with symptoms improvement. It is important to recognise that a
neuropsychological evaluation in a clinical setting is a more comprehensive assessment of cognition and
may be more sensitive to cognitive decrement than the brief, unsupervised online assessment
used in this study.
There are some study limitations. Our measure of education was binary, and normative
data was age-matched only, which may have masked some education-related variability that
could affect cognitive performance. Also, this was an educated, computer-literate sample,
which may have impacted on the relationships between depressive symptom severity and
CVD risk with cognitive performance [
]. As such, they likely had greater cognitive reserve
(usually assessed by educational attainment or occupational complexity) and thus have had
some `buffer' against cognitive decline due to depressive symptoms or CVD, although despite
this they performed below an aged matched normative sample. Previous studies have shown
that higher rates of cognitive reserve protect against deficits in verbal fluency [
] and verbal
] in individuals with elevated symptoms of depression. The use of tools such as
Cogstate in community-based research or primary care settings will be subject to similar
selection bias compared to intensive testing in secondary or tertiary care samples. The same is also
true of the attrition, whereby more of those we are most interested in following, i.e. with
poorer cognition, were lost to follow-up, although aging in only one of the three domains. For
many outcomes loss of the most severe might bias results to the null but with cognition, which
tends to show a consistent decline over time, this may not be true. Further, this attrition may
limit generalisability but does not explain the lack of association between depressive symptoms
and cognitive function in this large sample. Finally, the measures of depressive symptoms [
and cognition [
27, 36, 37
] are well-validated although our classification of probable MDD was
based upon self-report [
] rather clinical interview, potentially leading to a misclassification
bias, although more likely random misclassification. Nevertheless, it may be worthwhile to
examine these relationships further in a more clinically robust sample of diagnosed patients
with CVD and depression.
This large study of cognitive function suggests that even when technologically literate and
well-educated, adults with mild-to-moderate depression and CVD risks in the community
have poorer cognitive function than well peers, particularly in relation to psychomotor speed
and learning. However, we were unable to find any of the hypothesised cross sectional or
prospective associations of depressive symptoms and cognition in this community-based sample
with milder depressive symptoms. This suggests that any such associations may be present
only in those with actual MDD as seen in such secondary care samples. As such, alleviating
depressive symptoms may lead to improvement in cognitive function only among those with
probable MDD (as opposed to more mild symptoms); however, large-scale efforts to prevent
progression from more mild symptoms to diagnosed MDD still seem worthwhile as a means
to reduce the risk of future cognitive decline [
S1 Data. De-identified data set.
11 / 14
S2 Data. Marginal distribution of outcomes and tests of assumptions.
the trial participants.
This trial was funded by the Cardiovascular Disease and Depression Strategic Research
Program (Award Reference No. G08S 4048) funded by the National Heart Foundation of Australia
and beyondblue: the national depression and anxiety initiative. Neither of the funders had any
role in the study design, analysis or interpretation or the submission decision. The researchers
are independent of the funders.
The 45 and Up Study is managed by the Sax Institute in collaboration with major partner
Cancer Council New South Wales; and partners the Heart Foundation (NSW Division); NSW
Ministry of Health; beyondblue: the national depression and anxiety initiative; Ageing,
Disability and Home Care, NSW Family and Community Services; and the Australian Red Cross
The authors wish to thank the men and women participating in the 45 and Up Study and
Conceptualization: Sharon L. Naismith, Ian B. Hickie, Nicholas Glozier.
Data curation: Daniel J. Biddle, Nicholas Glozier.
Formal analysis: Haley M. LaMonica, Daniel J. Biddle.
Funding acquisition: Sharon L. Naismith, Ian B. Hickie, Nicholas Glozier.
Investigation: Sharon L. Naismith, Ian B. Hickie, Nicholas Glozier.
Methodology: Sharon L. Naismith, Ian B. Hickie, Nicholas Glozier.
Project administration: Sharon L. Naismith, Nicholas Glozier.
Software: Paul Maruff.
Supervision: Nicholas Glozier.
Writing ± original draft: Haley M. LaMonica, Daniel J. Biddle.
Writing ± review & editing: Haley M. LaMonica, Daniel J. Biddle, Sharon L. Naismith, Ian B.
Hickie, Paul Maruff, Nicholas Glozier.
12 / 14
13 / 14
pair study. Psychiatry research. 2014; 215(1):87±94. https://doi.org/10.1016/j.psychres.2013.10.037
1. Hermens DF , Hodge MAR , Naismith SL , Kaur M , Scott E , Hickie IB . Neuropsychological clustering highlights cognitive differences in young people presenting with depressive symptoms . Journal of the International Neuropsychological Society . 2011 ; 17 ( 02 ): 267 ± 76 .
2. McIntyre RS , Cha DS , Soczynska JK , Woldeyohannes HO , Gallaugher LA , Kudlow P , et al. Cognitive deficits and functional outcomes in major depressive disorder: determinants, substrates, and treatment interventions . Depression and Anxiety . 2013 ; 30 ( 6 ): 515 ± 27 . https://doi.org/10.1002/da.22063 PMID: 23468126
3. Naismith SL , Hickie IB , Turner K , Little CL , Winter V , Ward PB , et al. Neuropsychological performance in patients with depression is associated with clinical, etiological and genetic risk factors . Journal of clinical and experimental neuropsychology . 2003 ; 25 ( 6 ): 866 ± 77 . https://doi.org/10.1076/jcen. 25.6.866. 16472 PMID: 13680463
4. Hsu KJ , Young-Wolff KC , Kendler KS , Halberstadt LJ , Prescott CA. Neuropsychological deficits in major depression reflect genetic/familial risk more than clinical history: a monozygotic discordant twin5.
Jaeger J , Berns S , Uzelac S , Davis-Conway S . Neurocognitive deficits and disability in major depressive disorder . Psychiatry research . 2006 ; 145 ( 1 ): 39 ± 48 . https://doi.org/10.1016/j.psychres. 2005 . 11 .011 PMID: 17045658
Naismith SL , Hickie IB , Ward PB , Scott E , Little C . Impaired implicit sequence learning in depression: a probe for frontostriatal dysfunction? Psychological Medicine . 2006 ; 36 ( 03 ): 313 ± 23 .
McDermott LM , Ebmeier KP . A meta-analysis of depression severity and cognitive function . Journal of affective disorders . 2009 ; 119 ( 1 ):1± 8 .
Naismith SL , Norrie LM , Mowszowski L , Hickie IB . The neurobiology of depression in later-life: clinical, neuropsychological, neuroimaging and pathophysiological features . Progress in neurobiology. 2012 ; 98 ( 1 ): 99 ± 143 . https://doi.org/10.1016/j.pneurobio. 2012 . 05 .009 PMID: 22609700
Lin K , Xu G , Lu W , Ouyang H , Dang Y , Lorenzo-Seva U , et al. Neuropsychological performance in melancholic, atypical and undifferentiated major depression during depressed and remitted states: a prospective longitudinal study . Journal of affective disorders . 2014 ; 168 : 184 ± 91 . https://doi.org/10.1016/j.
jad. 2014 . 06 .032 PMID: 25061955
Psychological medicine . 2007 ; 37 ( 12 ): 1693 ± 702 . https://doi.org/10.1017/S0033291707001134 PMID: 17610767
Naismith SL , Longley WA , Scott EM , Hickie IB . Disability in major depression related to self-rated and objectively-measured cognitive deficits: a preliminary study . BMC psychiatry . 2007 ; 7 ( 1 ): 1 .
Norton S , Matthews FE , Barnes DE , Yaffe K , Brayne C . Potential for primary prevention of Alzheimer's disease: an analysis of population-based data . The Lancet Neurology . 2014 ; 13 ( 8 ): 788 ± 94 . https://doi.
org/10 .1016/S1474- 4422 ( 14 ) 70136 -X PMID: 25030513
Psychiatry research . 2010 ; 177 ( 3 ): 323 ±9. https://doi.org/10.1016/j.psychres. 2010 . 03 .006 PMID: 20385412
The Lancet Psychiatry . 2016 ; 3 ( 5 ): 425 ± 35 . https://doi.org/10.1016/S2215- 0366 ( 16 ) 00012 - 2 PMID: 26995298
Butters MA , Becker JT , Nebes RD , Zmuda MD , Mulsant BH , Pollock BG , et al. Changes in cognitive functioning following treatment of late-life depression . American Journal of Psychiatry . 2000 ; 157 ( 12 ): 1949 ±54. https://doi.org/10.1176/appi.ajp. 157 .12. 1949 PMID: 11097959
Doraiswamy PM , Krishnan KRR , Oxman T , Jenkyn LR , Coffey DJ , Burt T , et al. Does antidepressant therapy improve cognition in elderly depressed patients? The Journals of Gerontology Series A: Biological Sciences and Medical Sciences . 2003 ; 58 ( 12 ):M1137± M44 .
17. Jayaweera HK , Hickie IB , Duffy SL , Hermens DF , Mowszowski L , Diamond K , et al. Mild Cognitive Impairment Subtypes in Older People With Depressive Symptoms Relationship With Clinical Variables and Hippocampal Change . Journal of geriatric psychiatry and neurology . 2015 : 0891988715573535 .
HICKIE I , SCOTT E. Late-onset depressive disorders: a preventable variant of cerebrovascular disease? Psychological Medicine . 1998 ; 28 ( 05 ): 1007 ± 13 .
Cockayne NL , Glozier N , Naismith SL , Christensen H , Neal B , Hickie IB . Internet-based treatment for older adults with depression and co-morbid cardiovascular disease: protocol for a randomised, doubleblind, placebo controlled trial . BMC psychiatry . 2011 ; 11 ( 1 ): 10 .
Glozier N , Christensen H , Naismith S , Cockayne N , Donkin L , Neal B , et al. Internet-delivered cognitive behavioural therapy for adults with mild to moderate depression and high cardiovascular disease risks: a randomised attention-controlled trial . PLoS One . 2013 ; 8 ( 3 ):e59139. https://doi.org/10.1371/journal.
pone.0059139 PMID: 23555624
Collaborators US . Cohort profile: the 45 and up study . International Journal of Epidemiology . 2008 ; 37 ( 5 ): 941 . https://doi.org/10.1093/ije/dym184 PMID: 17881411
Christensen H , Griffiths KM , Korten A . Web-based cognitive behavior therapy: analysis of site usage and changes in depression and anxiety scores . Journal of Medical Internet Research . 2002 ; 4 ( 1 ): e3 .
https://doi.org/10.2196/jmir.4.1.e3 PMID: 11956035
PLoS One . 2012 ; 7 ( 12 ):e53244. https://doi.org/10.1371/journal.pone. 0053244 PMID: 23285271
24. Mackinnon A , Griffiths KM , Christensen H . Comparative randomised trial of online cognitive±behavioural therapy and an information website for depression: 12-month outcomes . The British Journal of Psychiatry . 2008 ; 192 ( 2 ): 130 ±4. https://doi.org/10.1192/bjp.bp. 106 .032078 PMID: 18245031
25. LoÈwe B , Spitzer RL , GraÈfe K , Kroenke K , Quenter A , Zipfel S , et al. Comparative validity of three screening questionnaires for DSM-IV depressive disorders and physicians' diagnoses . Journal of affective disorders . 2004 ; 78 ( 2 ): 131 ± 40 . PMID: 14706723
26. LoÈwe B , UnuÈtzer J , Callahan CM , Perkins AJ , Kroenke K. Monitoring depression treatment outcomes with the patient health questionnaire-9 . Medical care. 2004 ; 42 ( 12 ): 1194 ± 201 . PMID: 15550799
27. Falleti MG , Maruff P , Collie A , Darby DG . Practice effects associated with the repeated assessment of cognitive function using the CogState battery at 10-minute, one week and one month test-retest intervals . Journal of clinical and experimental neuropsychology . 2006 ; 28 ( 7 ): 1095 ± 112 . https://doi.org/10. 1080/13803390500205718 PMID: 16840238
28. Cogstate . Cogstate Normative Data . Melbourne, Australia: Cogstate; 2013 .
29. Kroenke K , Spitzer RL , Williams JB . The Phq-9. Journal of general internal medicine . 2001 ; 16 ( 9 ): 606 ± 13 .
30. Dawson DA , Grant BF , Stinson FS . The AUDIT-C: screening for alcohol use disorders and risk drinking in the presence of other psychiatric disorders . Comprehensive psychiatry . 2005 ; 46 ( 6 ): 405 ± 16 . https:// doi.org/10.1016/j.comppsych. 2005 . 01 .006 PMID: 16275207
31. Velligan DI , Alphs L , Lancaster S , Morlock R , Mintz J . Association between changes on the Negative Symptom Assessment scale (NSA-16) and measures of functional outcome in schizophrenia . Psychiatry research . 2009 ; 169 ( 2 ): 97 ± 100 . https://doi.org/10.1016/j.psychres. 2008 . 10 .009 PMID: 19664826
32. Woo YS , Rosenblat JD , Kakar R , Bahk W-M , McIntyre RS . Cognitive deficits as a mediator of poor occupational function in remitted major depressive disorder patients . Clinical Psychopharmacology and Neuroscience . 2016 ; 14 ( 1 ):1. https://doi.org/10.9758/cpn. 2016 . 14 . 1 .1 PMID: 26792035
33. Lee RS , Hermens DF , Porter MA , Redoblado-Hodge MA . A meta-analysis of cognitive deficits in firstepisode major depressive disorder . Journal of affective disorders . 2012 ; 140 ( 2 ): 113 ± 24 . https://doi.org/ 10.1016/j.jad. 2011 . 10 .023 PMID: 22088608
34. Opdebeeck C , Nelis SM , Quinn C , Clare L. How does cognitive reserve impact on the relationships between mood, rumination, and cognitive function in later life? Aging & mental health . 2015 ; 19 ( 8 ): 705 ± 12 .
35. McLaren ME , Szymkowicz SM , Kirton JW , Dotson VM . Impact of Education on Memory Deficits in Subclinical Depression . Archives of Clinical Neuropsychology . 2015 ; 30 ( 5 ): 387 ± 93 . https://doi.org/10.1093/ arclin/acv038 PMID: 26109434
36. Lim YY , Ellis KA , Harrington K , Ames D , Martins RN , Masters CL , et al. Use of the CogState Brief Battery in the assessment of Alzheimer's disease related cognitive impairment in the Australian Imaging, Biomarkers and Lifestyle (AIBL) study . Journal of Clinical and Experimental Neuropsychology . 2012 ; 34 ( 4 ): 345 ± 58 . https://doi.org/10.1080/13803395. 2011 .643227 PMID: 22248010
37. Lim YY , Jaeger J , Harrington K , Ashwood T , Ellis KA , StoÈffler A , et al. Three-month stability of the CogState brief battery in healthy older adults, mild cognitive impairment, and Alzheimer's disease: results from the Australian Imaging, Biomarkers, and Lifestyle-rate of change substudy (AIBL-ROCS) . Archives of clinical neuropsychology . 2013 ; 28 ( 4 ): 320 ± 30 . https://doi.org/10.1093/arclin/act021 PMID: 23552802
38. Royall DR , Palmer RF . Alzheimer's disease pathology does not mediate the association between depressive symptoms and subsequent cognitive decline . Alzheimer's & Dementia . 2013 ; 9 ( 3 ): 318 ± 25 .
39. Wilson RS , Capuano AW , Boyle PA , Hoganson GM , Hizel LP , Shah RC , et al. Clinical-pathologic study of depressive symptoms and cognitive decline in old age . Neurology . 2014 ; 83 ( 8 ): 702 ±9. https://doi.org/ 10.1212/WNL.0000000000000715 PMID: 25080520