Depressive Symptoms Are Associated with Cognitive Function in the Elderly with Type 2 Diabetes
Abstract
Background:
Type 2 diabetes (T2D) is a metabolic condition associated with poor clinical and cognitive outcomes including vascular disease, depressive symptoms, cognitive impairment, and dementia. In the general elderly population, depression has been consistently identified as a risk factor for cognitive impairment/decline. However, the association between depression and cognitive function in T2D has been understudied.
Objective:
We investigated the association between depression and cognitive function in a large sample of cognitively normal elderly with T2D.
Methods:
In this cross-sectional study, we examined 738 participants, aged 65–88 years old, enrolled in the Israel Diabetes and Cognitive Decline study. For each cognitive domain (Episodic Memory, Executive Function, Attention/Working Memory, Language/Semantic Categorization) and Overall Cognition, multiple linear regressions assessed its association with depression (score greater than 5 on the 15-item version of the Geriatric Depression Scale [GDS]), adjusting for age, sex, and education.
Results:
Depression (n = 66, 8.9%) was associated with worse performance on tasks of Executive Function (p = 0.004), Language/Semantic Categorization (p < 0.001), and Overall Cognition (p < 0.002), but not Episodic Memory (p = 0.643) or Attention/Working Memory (p = 0.488). Secondary analyses using GDS as a continuous variable did not substantially change the results. Adjusting also for a history of antidepressant medication use slightly weakened the findings.
Conclusion:
Significant associations of depression with several cognitive domains and Overall Cognition even in cognitively normal elderly with T2D, suggest that depression may have a role in impaired cognitive function in T2D, which may be attenuated by antidepressants.
INTRODUCTION
In the general elderly population, the association between neuropsychiatric symptoms (and in particular depressive symptoms) and cognitive impairment/ decline including dementia has been robustly found [1, 2]. This association has been reported in the elderly from different ethnic backgrounds [3] and has been supported by neuropathological findings [4, 5]. However, longitudinal studies have reported an association that affects only men [6] or those with higher educational attainment [7]. Other studies have found that other characteristics, such as age of onset of depression and prior cognitive impairment [8–10], can affect the association of depression with cognitive function.
In the general elderly population, the associations between depression and cognitive function differ by cognitive domain (for a review, see [11]), and have been reported for all domains [12]. Impairment in specific cognitive domains has also been reported, with substantial deficits observed in attention and information processing speed [13]. Differences may also be evident in populations with specific medical conditions, such as deficits in episodic memory in recurrent depression and deficits in attention and executive function in late-onset depression [14].
Indeed, elderly with type 2 diabetes (T2D) and depression are at higher risk of cognitive impairment/decline [15, 16]. However, the literature reporting on this association is scarce and inconsistent, and little is known about the cognitive domains that are compromised. In a cross-sectional study of elderly people with T2D, Brands and colleagues examined the association of depressive symptoms with five cognitive domains, but no association was found using either a cutoff score for depression, or the total number of symptoms, or using specific depression domain scores [17]. In contrast to this cross-sectional study, a longitudinal study found a significant association of depressive symptoms with cognitive decline in three cognitive tests representing psychomotor function, verbal memory, and executive function [18], which is largely consistent with those domains affected in the general elderly population with depression.
Thus, the present study investigated the cross-sectional association of depression with cognitive function in a well-characterized, cognitively normal (Clinical Dementia Rating [CDR] scale = 0 at the Israel Diabetes and Cognitive Decline [IDCD] study entry) elderly population with T2D. This study is aimed at extending findings from the depression literature in the general elderly population in two ways: (1) by examining the association of depression with cognitive function in elderly with T2D and (2) by identifying specific cognitive domains that are impaired.
METHODS
2.1Participants
The IDCD study design has been previously described in detail [19], for 897 participants with baseline data. Briefly, the IDCD recruited community-dwelling elderly with T2D (65+ years old) living in central Israel, from approximately 11,000 clients enrolled in the diabetes registry of the Maccabi Healthcare Services (MHS). The MHS diabetes registry was established in 1998 to facilitate diabetes management and to improve treatment. Any of the following criteria, primarily based on the American Diabetes Association, were assessed and sufficient for enrollment into the registry: 1) Hemoglobin A1c Hemoglobin A1c (HbA1c) >55.7 mmol/mol; 2) Glucose >11.10 mmol/l on two exams more than three months apart; 3) purchase of diabetic medication twice within three months supported by a HbA1c >47.5 mmol/mol or Glucose >6.94 mmol/l within half a year; 4) diagnosis of type 2 T2D (ICD-9 code [http://www.icd9data.com/2007/Volume1]) by a general practitioner, internist, endocrinologist, ophthalmologist, or type 2 T2D advisor, supported by an HbA1c >47.5 mmol/mol or Glucose >6.94 mmol/l within half a year. These criteria have been validated by twenty physicians in MHS against their own practice record [20]. All blood assessments were performed during fasting. IDCD inclusion criteria were having T2D; normal cognition at entry; being free of any neurological (e.g., Parkinson’s disease, stroke), psychiatric (e.g., schizophrenia), or other diseases (e.g., alcohol or drug abuse) that might affect cognition; and having an informant.
For the initial IDCD screening, the electronic medical records of potential participants were screened by the MHS team for diagnosis of dementia, and its subtypes, and for cholinesterase inhibitors. If any of these criteria were present in the potential participant’s record, the potential participant was excluded from the study. Then, MHS personnel asked potential participants, on the phone, whether a doctor had ever told them that they have a memory problem, or if they had ever been treated for a memory problem. Those who responded positively were excluded from the study.
Potential participants were assessed by an IDCD physician experienced in assessment and diagnosis of dementia, and were administered the CDR scale [21]. The CDR scale assesses the severity of cognitive and functional impairment in 6 domains (memory, orientation, judgment and problem solving, community affairs, home and hobbies, and personal care) through an interview with the participant and an informant. A score of 0 represents normal cognition (an inclusion criterion for the IDCD study), 0.5 represents questionable dementia, and scores of 1 through 3 reflect increasing severity of dementia [21, 22].
Those with a CDR > 0 (reflecting questionable or higher levels of dementia severity) were excluded from the IDCD study and referred back to their primary physician. Potential participants were also assessed by a neuropsychologist, who administered the broad neuropsychological battery. For descriptive purposes, global cognitive function was assessed with the Mini-Mental State Examination (MMSE) [23]. It assesses various areas of cognitive functions (orientation, concentration, memory, language, and visual construction) and is widely used as a cognitive screening instrument for dementia. Depression was assessed using the Geriatric Depression Scale (GDS).
IDCD eligibility was determined by a diagnostic consensus conference that included neurologists, psychiatrists, and neuropsychologists experienced with dementia, with at least two specialties present. It is important to note that the neuropsychological battery was not used in the process of screening for normal cognition since it was used to calculate the cognitive outcome measures.
This study employed prospective historical diabetes-related data from the MHS, and baseline cognitive and depression data collected by the IDCD study on 897 participants. The sample for this study consisted of the 738 IDCD participants with complete data on cognitive domains, demographic characteristics (age, sex, and education), T2D-related characteristics (HbA1c; number of follow-up years in the registry, a surrogate for duration of disease [24]; and whether medication for T2D was taken [no medication, hypoglycemic medication, and insulin or insulin + hypoglycemic medication]), and cardiovascular risk factors (body mass index, creatinine, total cholesterol, triglycerides, and diastolic and systolic blood pressure).
Study procedures and informed consent were approved by the Icahn School of Medicine at Mount Sinai, Sheba Medical Center, and MHS IRB committees. All participants signed an informed consent.
2.2Predicted cognitive function/outcomes
For the IDCD study (n = 897), cognitive function at entry was assessed using 13 neuropsychological tests, grouped into cognitive domains according to the factor with the highest loading: (1) Episodic Memory: Word List Memory, Word List Recall, and Word List Recognition from the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) neuropsychological battery [25, 26]; (2) Attention/Working Memory: Digit Span (forward and backward) from the Wechsler Memory Scale-Revised (WMS-R) [27]; (3) Language/Semantic Categorization: Similarities from the Wechsler Adult Intelligence Scale-Revised (WAIS-R) [28], Letter Fluency [29], and Animal Fluency [30]; and (4) Executive Function: Trail Making Test (A and B) [31], CERAD-Constructional Praxis, Shape Cancellation, and Digit Symbol from the WAIS-R [28]. Raw scores were converted to z scores using participants’ means and SDs. A composite measure of global cognitive function (Overall Cognition) was created by averaging all the z scores. Scores for the four cognitive domains were calculated as averages of z scores.
2.3Depression predictor
Depression was the predictor variable, which was assessed by the 15-item version of the GDS [32]. The GDS is a widely used screening instrument for depressive symptoms in the elderly population, which follows a yes/no answer format. The cut-off score greater than 5, suggesting clinical depression, was used to define depression as a dichotomous status [33].
2.4Statistical analyses
The association of depression with each cognitive outcome was examined using multiple regression analyses, with Episodic Memory, Attention/Working Memory, Language/Semantic Categorization, Executive Function, and Overall Cognition as the dependent variables. The primary analyses, Model 1, adjusted for the demographic characteristics of age, sex, and education. In secondary analyses, Model 2, we also adjusted for T2D-related characteristics and cardiovascular risk factors, which are potential confounders that may be associated with cognitive function [34]. The association of depression with each cognitive outcome was also examined using the predictor as a continuous variable (GDS-15 total score).
We had information from the MHS that some participants (n = 192) received antidepressant medications at some time point since they entered the Diabetes Registry, suggesting they had depression. Thus, the main analyses were repeated adjusting also for antidepressants. In additional analyses, we applied multivariate analyses of covariance (MANCOVA) separately in those who had or had not taken any antidepressant to test the effects of depression status (non-depression versus depression) on cognitive functioning (the four cognitive domains and Overall Cognition), adjusting for demographics.
For each of the models of Tables 3 and 4, results for the five cognitive outcomes were considered “significant” if they met criteria of the Holm multiple comparisons procedure [35], an enhancement of the Bonferroni inequality, at the 0.05 level of significance.
For comparisons of two groups, the small (0.20), medium (0.50), and large (0.80) conventions for effect sizes of t-tests are widely used, while the effect sizes and conventions for the F-test are not. Therefore, t-test effect sizes are presented for F-tests comparing two groups. Effect size conventions for Pearson’s chi-square are similarly not familiar. For a 2×2 table, the square root of Pearson’s chi-square is the z-test for the difference between two proportions, analogous to the t-test for the difference between means. The corresponding effect size for the difference between proportions has the same small, medium, and large conventions as the t-test.
RESULTS
For the entire sample (n = 738), 66 (8.9%) were depressed and 672 (91.1%) were not depressed. Mean age was 71.9 (SD = 4.7). There were more men (60.3%) than women. The MMSE mean score was 28.1 (SD = 1.7), which is consistent with normal cognition. Table 1 presents the demographic, clinical, and cognitive characteristics of the sample by depression status (i.e., non-depression versus depression groups). The depression group had, on average, fewer years of education (p < 0.001), lower MMSE scores (p < 0.01), by definition had higher depression scores, and were more likely to have taken antidepressants (p < 0.001) than the non-depression group. The depression group had a majority of women, while the non-depression group had a majority of men (p < 0.001). With the exception of Attention/Working Memory (p = 0.643), the depression group performed more poorly on all cognitive domains (p ≤ 0.019) and Overall Cognition (p = 0.002) than the non-depression group.
Table 1
Mean ± SD of demographic and clinical characteristics of the participants by their clinical status, non-depression versus depression
| Non-Depression | Depression | p | Effect Size | |
|---|---|---|---|---|
| N | 672 | 66 | – | |
| Age | 72.7 ± 4.6 | 73.1 ± 5.0 | 0.50a | 0.09 |
| Education | 13.4 ± 3.5 | 11.6 ± 2.3 | <0.001a | 0.51 |
| Male (%) | 62.8 | 34.8 | <0.001b | 0.57 |
| Female (%) | 37.2 | 65.2 | ||
| # of follow up years in the registry | 10.4 ± 1.5 | 10.4 ± 1.2 | 0.81a | 0.03 |
| Body mass index (kg/m2) | 28.3 ± 4.5 | 28.8 ± 4.8 | 0.43a | 0.09 |
| Creatinine (mg/dL) | 1.0 ± 0.3 | 0.9 ± 0.3 | 0.07a | 0.23 |
| Total cholesterol (mg/dL) | 180.3 ± 25.2 | 183.7 ± 23.8 | 0.29a | 0.14 |
| Triglycerides (mg/dL) | 157.2 ± 65.0 | 167.3 ± 62.4 | 0.22a | 0.16 |
| Diastolic BP (mmHg) | 76.7 ± 4.9 | 76.4 ± 4.4 | 0.62a | 0.06 |
| Systolic PB (mmHg) | 134.8 ± 9.5 | 135.1 ± 8.8 | 0.87a | 0.02 |
| Hemoglobin A1c (%), (mmol/mol) | 6.8 (0.8), (51) | 6.9 (0.8), (52) | 0.24a | 0.15 |
| Type 2 diabetes medication (%) | ||||
| 1- No medication (n = 93) | 13.1 | 7.6 | 0.053b | 0.09 |
| 2- Hypoglycemic medication (n = 577) | 78.4 | 75.8 | ||
| 3-d Insulin or insulin + hypoglycemic medication (n = 68) | 8.5 | 16.7 | ||
| GDS-15 | 1.2 ± 1.3 | 7.5 ± 1.7 | <0.001a | 4.63 |
| Antidepressant use (n = 715) (%) | 0.45 | |||
| Antidepressant use at any time point in the diabetes registry (n = 192) | 25.1 | 45.9 | <0.001b | |
| MMSE score (n = 65 for depression) | 28.2 ± 1.6 | 27.2 ± 2.2 | <0.001a | 0.58 |
| Attention/Working Memory (SEM) | 0.051 (0.064) | – 0.051 (0.208) | 0.643c | 0.06 |
| Executive Function (SEM) | 0.224 (0.120) | – 0.956 (0.389) | 0.004c | 0.37 |
| Language/Semantic Categorization (SEM) | 0.250 (0.080) | – 0.787 (0.260) | <0.001c | 0.49 |
| Episodic Memory (SEM) | 0.139 (0.080) | – 0.309 (0.261) | 0.103c | 0.21 |
| Overall Cognition (SEM) | 0.500 (0.187) | – 1.480 (0.609) | 0.002c | 0.40 |
BP, blood pressure; GDS, Geriatric Depression Scale; MMSE, Mini-Mental State Examination; SEM, Standard Error of the Mean. ap-value by Student’s t-test, bPearson’s chi-square for percentages, 2 categories. cAnalysis of covariance controlling for age, sex, and education. dSince 63 participants received both hypoglycemic and insulin medication and 5 received only insulin, they were combined. Pearson’s chi-square for percentages, effect size, 10 small, 30 medium, 0.50 large.
For the 738 participants in this study, Table 2 shows the intercorrelations among the four cognitive domains and Overall Cognition, and their correlations with the 13 neuropsychological tests.
Table 2
Correlations among five cognitive outcomes and cognitive tests
| Cognitive Outcomes | n | Episodic Memory | Executive Function | Language/Semantic Categorization | Attention/Working Memory | Overall Cognition |
|---|---|---|---|---|---|---|
| Correlations | ||||||
| Attention/ Working Memory | 738 | 0.234 | 0.466 | 0.487 | – – | 0.657 |
| Executive Function | 738 | 0.259 | – – | 0.558 | 0.466 | 0.844 |
| Language/Semantic Categorization | 738 | 0.344 | 0.558 | – – | 0.487 | 0.804 |
| Episodic Memory | 738 | – – | 0.259 | 0.344 | 0.234 | 0.588 |
| Overall Cognition | 738 | 0.588 | 0.844 | 0.804 | 0.687 | – – |
| Cognitive Tests | ||||||
| Attention/Working Memory | ||||||
| Digit Span Forward | 738 | 0.156 | 0.350 | 0.417 | 0.877 | 0.576 |
| Digit Span Backward | 738 | 0.253 | 0.467 | 0.437 | 0.879 | 0.643 |
| Executive Function | ||||||
| Trails A | 737 | 0.202 | 0.801 | 0.435 | 0.331 | 0.661 |
| Trails B | 712 | 0.252 | 0.811 | 0.481 | 0.484 | 0.739 |
| Praxis | 734 | 0.132 | 0.608 | 0.359 | 0.257 | 0.506 |
| Digit Symbol | 737 | 0.264 | 0.811 | 0.544 | 0.435 | 0.744 |
| Shape Cancellation | 700 | 0.99 | 0.648 | 0.230 | 0.202 | 0.461 |
| Language/ Semantic Categorization | ||||||
| Letter Fluency | 731 | 0.259 | 0.446 | 0.855 | 0.441 | 0.668 |
| Animal Fluency | 738 | 0.293 | 0.409 | 0.794 | 0.331 | 0.616 |
| Similarities | 737 | 0.274 | 0.486 | 0.754 | 0.398 | 0.650 |
| Episodic Memory | ||||||
| Word List Memory | 738 | 0.815 | 0.815 | 0.284 | 0.414 | 0.305 |
| Word List Recall | 738 | 0.815 | 0.200 | 0.235 | 0.109 | 0.440 |
| Word List Recognition | 738 | 0.635 | 0.101 | 0.127 | 0.113 | 0.307 |
**p < 0.001.
Table 3 presents multiple regression results: the primary analyses adjusting for demographics (Model 1), and the secondary analyses adjusting for T2D-related characteristics and cardiovascular risk factors (Model 2). Model 1 analyses showed significant associations of depression with poorer cognitive performance in the Executive Function domain (partial r = – 0.106, t = – 2.893, p = 0.004), Language/Semantic Categorization domain (partial r = – 0.139, t = – 3.79, p < 0.001), and Overall Cognition (partial r = – 0.114, t = – 3.096, p = 0.002). Although strongly significant, these associations (individually) explained ≤ 0.019 of the variability in cognitive performance. These results remained largely unchanged using Model 2.
Table 3
Associations of depression, as assessed by the dichotomized GDS-15, with performance on cognitive domains and overall cognition
| Partial ra | β (95% CI) | p | Partial rb | β (95% CI) | p | |
|---|---|---|---|---|---|---|
| Attention/Working Memory | – 0.017 | – 0.102 | 0.643 | – 0.016 | – 0.094 | 0.668 |
| (– 0.531, 0.328) | (– 0.525, 0.336) | |||||
| Executive Function | – 0.106 | – 1.180 | 0.004c | – 0.103 | – 1.123 | 0.006c |
| (– 1.981, – 0.379) | (– 1.918, – 0.329) | |||||
| Language/Semantic Categorization | – 0.139 | – 1.037 | < 0.001c | – 0.138 | – 1.025 | < 0.001c |
| (– 1.573, 0.500) | (– 1.562, – 0.487) | |||||
| Episodic Memory | – 0.060 | – 0.448 | 0.103 | – 0.058 | – 0.431 | 0.119 |
| (– 0.987, 0.090) | (– 0.973, 0.111) | |||||
| Overall Cognition | – 0.114 | – 1.980 | 0.002c | – 0.112 | – 1.954 | 0.002c |
| (– 3.235, – 0.724) | (– 3.216, – 0.692) |
T2D, type 2 diabetes; HbA1c, hemoglobin A1c; BMI, body mass index; BP, blood pressure; β, Average difference in cognitive performance (dependent variable) of participants with depression compared to those without depression. aAdjusting for demographics (age, sex, and education). bAdjusting for demographics, T2D-related characteristics (HbA1c, number of follow up years in the registry, and type 2 diabetes medication), and cardiovascular risk factors (BMI, creatinine, total cholesterol, triglycerides, and diastolic and systolic BP). cSignificant at 0.05 level by Holm multiple comparisons procedure.
We repeated the Model 1 analyses by examining depression as a continuous variable (GDS-15 total score) and adjusting for demographics. Results remained significant, with a slight increase in effect size, for the same three cognitive outcomes, Executive Function, Language/Semantic Categorization, and Overall Cognition (partial r≤– 0.175, t≤– 4.818, p < 0.001), and one additional domain, Episodic Memory, became significant (partial r = – 0.093, t = – 2.538, p = 0.011). These results remained largely unchanged using Model 2. Of note, since GDS-15 total score was not normally distributed, when the Model 1 analyses were repeated using logarithmic transformation (which reduced kurtosis to – 0.8 from its original value of 3.6), findings remained largely consistent, but Episodic Memory was no longer significant (p = 0.062).
In the subsample of 715 participants with antidepressant use information, the results for the association of depression with poorer Executive Function (p = 0.026) and Overall Cognition (p = 0.040) lost significance by the Holm criterion, but the Language/Semantic Categorization domain (p = 0.002) remained significant. The MANCOVA in this subsample showed that among the 26.9% (n = 192) who had taken any antidepressant, cognitive functioning was similar in the depressed and non-depressed groups (Table 4). The latter result remained non-significant when cognitive outcomes were compared by antidepressant classes (i.e., Selective-Serotonin Reuptake Inhibitors [SSRIs], Serotonin-Norepinephrine Reuptake Inhibitors [SNRIs], and tricyclics). However, due to the small sample size, the antidepressant class result should be interpreted with caution since the largest comparison consisted of those who were taking SSRIs (n = 36) versus those who were not (n = 30). In contrast, among those who had not taken antidepressants, test performance was significantly worse in the depressed subgroup in the Language/Semantic Categorization (p = 0.007) domain, as compared to the non-depressed group (Table 4).
Table 4
Means and standard error of the mean (SEM) of Z scores of cognitive performance by antidepressant use and depression status
| Non-Antidepressant | Antidepressant | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Cognitive domain | Non-depression | Depression | F (df = 1, 518) | p | Effect Size | Non-depression | Depression | F (df = 1, 187) | p | Effect Size |
| n | 490 | 33 | 164 | 28 | ||||||
| Attention/Working Memory | 0.145 (0.075) | 0.014 (0.296) | 0.184 | 0.668 | 0.08 | – 0.170 (0.127) | 0.099 (0.312) | 0.632 | 0.428 | 0.16 |
| Executive Function | 0.496 (0.139) | – 0.686 (0.544) | 4.415 | 0.036 | 0.38 | – 0.429 (0.245) | – 0.880 (0.600) | 0.478 | 0.490 | 0.14 |
| Language/Semantic Categorization | 0.382 (0.092) | – 0.630 (0.361) | 7.335 | 0.007a | 0.49 | – 0.006 (0.169) | – 0.604 (0.412) | 1.785 | 0.183 | 0.27 |
| Episodic Memory | 0.193 (0.089) | 0.324 (0.348) | 0.134 | 0.715 | 0.07 | 0.058 (0.176) | – 0.339 (0.430) | 0.724 | 0.396 | 0.17 |
| Overall Cognition | 0.840 (0.212) | – 0.795 0.830) | 3.627 | 0.057 | 0.34 | – 0.184 (0.395) | – 0.865 (0.967) | 0.422 | 0.517 | 0.13 |
Note: Adjusting for age, sex, and education. Effect size is for t-test comparing two groups. aSignificant at 0.05 level by Holm multiple comparisons procedure.
DISCUSSION
This study investigated the association of depression with performance on four different cognitive domains and on Overall Cognition in participants with T2D. Depression was associated with poorer cognitive performance on Executive Function and Language/Semantic Categorization, and with poorer Overall Cognition, but not on Episodic Memory or Attention/Working Memory, after adjusting for demographics (age, sex, and education). With the exception of Episodic Memory, which was not significant, these results are largely consistent with those on the association between depression and cognition in the general elderly population. The Episodic Memory result suggests that impaired memory may be less frequently observed in the depressed diabetic population as compared to the depressed general elderly population. T2D is associated with vascular risks and diseases, which in turn are associated with impairment in information processing speed. Most of the tasks in the Language/Semantic Categorization and the Executive Function domains were timed, for which good performance relied heavily on how fast participants processed the information on which they were being tested.
Depression explained≤1.9% of the variability in each cognitive domain and in Overall Cognition. These results remained largely unchanged after further adjusting for T2D-associated characteristics and cardiovascular risk factors. Although, this might be a result of the relatively small prevalence (8.9%, as assessed by the GDS) of depression in this sample, the associations were less significant when antidepressant use (which yielded a higher rate: 26.9%, n = 715) was used as the predictor of cognition. It is noteworthy that the 8.9% prevalence is similar to that reported in non-T2D population-based studies [36], but smaller than that reported in T2D studies [37]. This discrepancy can be at least partially explained by the inclusion of a population-based sample as compared to a clinical sample, the eligibility criteria of the study, which required normal cognition, and the depression screening method. It is also possible that the potential effects of T2D on cognitive function masked the association of depression with cognitive function. This latter possibility is not the case in studies of depression in the general elderly population since these older adults may or may not have T2D.
The lack of report of the effect sizes of the association make comparisons of these results with those of other studies challenging. One cross-sectional study with a small sample (n = 52 white non-diabetic older adults) reported larger effect sizes (correlations ranging from 0.311 to 0.497) for the associations between depressive symptoms (as measured by the total score or two individual items from the Hamilton Depression Rating Scale) [38] and cognitive status, after controlling for demographics [3].
However, our results are in line with studies that included a T2D population and found either no association or small effects. For example, Brands and colleagues failed to find an association of cognitive performance with the cut-off score, the total score, or subdomain scores from the Beck Depression Inventory [17]. The longitudinal study by Sullivan and colleagues with a sample size of over 2,700 participants reported a link between depression and cognitive decline, with results remaining significant after adjusting for important confounders such as demographics, lipid treatment, and cardiovascular disease, and when the depression instrument was used as a continuous variable [18]. Although, Sullivan et al. addressed a different question from ours on the role of depression in cognition, their reported effect sizes were also small. The discrepancy in research findings across different studies may reflect, at least partially, differences in research design/methodology (e.g., longitudinal versus cross-sectional design, sample size and characteristics of participants, instruments used to assess depression and cognitive performance, and operational definition of depression: cutoff score, total depression score, or clinical diagnosis based on established diagnostic criteria). Results for the dichotomized GDS-15 and total score were largely similar. The IDCD sample is by eligibility criteria cognitively normal at baseline. Although within normal cognition the sample showed substantial variability in cognitive performance, it is possible that exclusion of cognitively impaired individuals explains the small association of depression with cognition in this sample. Consistent with this argument, elderly who were hospitalized for major depression and had symptoms of dementia that subsided after treatment for depression had a 4.7-times higher risk of developing frank dementia after a period of 3 years compared to hospitalized patients whose depression was not accompanied by dementia symptoms [9].
Psychological distress like depression can be an emotional reaction to a chronic medical condition such as T2D and may have different biological etiologies. In the elderly, depression that is associated with cerebrovascular disease has been referred to as “vascular depression.” [39]. This concept has also been supported by the presence of microbleeds in late-onset depression as compared to early onset depression [40]. The small effect size of the association between depression and cognition in T2D is surprising in view of the strong link between vascular disease and T2D (for a review, see [41]). Indeed, our results showed that participants with depression were not more likely to present with vascular risks (body mass index, total cholesterol, triglycerides, and diastolic and systolic blood pressure) than those without. This is consistent with Brands and colleagues’ findings showing that depression was neither linked to hypertension, nor to micro- or macro-vascular events, nor to white matter abnormalities [17]. Overall, these findings suggest that mechanisms underlying depression in T2D might originate from neurobiological substrates other than cerebrovascular risks/diseases [42–45], or be linked to cerebrovascular disease [46] such as brain atrophy, interleukins, and cortisol levels. In the general elderly population with major depressive disorder, Smagula and colleagues reported significant findings from pilot data on the associations among immunological markers, brain structure, and executive function [47]. Similarly, Charlton and colleagues reported an association between pro-inflammatory cytokines and memory function in late-life depression [48].
Within the depression group, relatively more participants took some diabetes medication and received more intensive treatment including insulin, which may suggest that depressed elderly with diabetes are at heightened risk of worse disease or poor diabetes management (due to difficulty following a diet or medication adherence), and of poorer cognitive performance before receiving intensive T2D treatment.
Because depression is strongly associated with psychomotor retardation [49], we reorganized the neuropsychological tests into composites of timed and untimed tests (data not shown). Partial correlations with depression showed that the depression group tended to perform worse on both types of tests (partial r controlling for demographics: timed, r = – 0.073, p = 0.06; untimed, r = – 0.100, p = 0.007), but the effects were small.
An area of research that remains largely unresolved in the general and T2D populations, is the effect of antidepressant medication on cognition. In the general elderly population, at least two classes of antidepressants, the SNRIs (specifically duloxetine) and the atypical antidepressants (specifically vortioxetine), have been reported to improve cognitive function, in particular memory and executive function [50, 51]. However, another class of antidepressant, the SSRIs (specifically citalopram), has been reported to decrease overall cognitive function in patients with AD [52]. Similarly, another study found MMSE scores declined after treatment with SSRIs for depression or Obsessive-Compulsive Disorder [53]. In this study, we examined associations of antidepressant classes with cognitive functioning in participants who had taken antidepressants at any time point since entering the MHS diabetes registry (n = 192). Our results showed that there were no significant differences in cognitive functioning between depressed and non-depressed participants who were on antidepressants of any class, or in those who were on a specific antidepressant class. However, the latter has to be interpreted with caution due to the small sample size of the subgroups.
This study had several limitations including the cross-sectional design, so causation cannot be inferred. The study lacked a non-T2D control group that would permit examining whether the associations found and their strengths are generalizable to non-T2D elderly. Similarly, this study excluded cognitively impaired T2D participants (such as those with mild cognitive impairment), thus preventing examination of the association in those who are at heightened risk of dementia. The use of follow up years in the diabetes registry as a surrogate for duration of T2D was a truncated estimate for all those who already had T2D when entering the registry, so the reported values are underestimates. Neuroimaging data that were not available could have helped explore the biological basis of depression and of diminished cognition in this sample (e.g., white matter hyperintensities and silent brain infarcts). To the extent that cerebrovascular disease may be a biological mechanism linking the association between depression and poorer cognition in T2D, excluding participants with stroke (an eligibility criterion of the IDCD study) could have diminished the magnitude of the associations. This study did not include as predictors age of depression onset or duration since then. Recent longitudinal studies finding age or duration modulation of dementia risk associated with depression suggest that depression may be an early sign of dementia, a condition that shares a common cause, or even dementia that was misdiagnosed [10, 54–56].
Strengths of this study included a large sample size, a well-characterized diagnosis of T2D, well-defined and numerous measurements of T2D characteristics, and a comprehensive neuropsychological battery. This battery permitted the evaluation of Overall Cognition and specific cognitive domains. Availability of antidepressant treatment information through the MHS registry permitted exploration of its contribution to the relationship of depression with cognition.
In summary, these results showed a significant but small association between depression and cognitive function; depression and impaired cognition may have common but also other distinct underlying neurobiological mechanisms. Investigation of these mechanisms may be an important goal for future research that includes T2D and non-T2D elderly.
ACKNOWLEDGMENTS
This study was supported by NIA grants R01 AG034087 to Dr. Beeri and P50 AG05138 to Dr. Sano, the Helen Bader Foundation and the Leroy Schecter Foundation to Dr. Beeri, and the Alzheimer’s Association grant MNIRGD-14-321113 to Dr. Guerrero-Berroa.
Authors’ disclosures are available online (https://www.j-alz.com/manuscript-disclosures/17-0778r2).
REFERENCES
[1] | Wilson RS , Barnes LL , Mendes de Leon CF , Aggarwal NT , Schneider JS , Bach J , Pilat J , Beckett LA , Arnold SE , Evans DA , Bennett DA ((2002) ) Depressive symptoms, cognitive decline, and risk of AD in older persons. Neurology 59: , 364–370. |
[2] | Ismail Z , Malick A , Smith EE , Schweizer T , Fischer C ((2014) ) Depression versus dementia: Is this construct still relevant?. Neurodegener Dis Manag 4: , 119–126. |
[3] | Guerrero-Berroa E , Kluger A , Schmeidler J , Sailor K , Lizardi H , Golomb J , Ferris S , Reisberg B ((2014) ) Neuropsychological and neuropsychiatric prediction of global cognitive status among older Spanish-speaking Hispanics and English-speaking whites. J Geriatr Psychiatry Neurol 27: , 266–275. |
[4] | Rapp MA , Schnaider-Beeri M , Grossman HT , Sano M , Perl DP , Purohit DP , Gorman JM , Haroutunian V ((2006) ) Increased hippocampal plaques and tangles in patients with Alzheimer disease with a lifetime history of major depression. Arch Gen Psychiatry 63: , 161–167. |
[5] | Rapp MA , Schnaider-Beeri M , Purohit DP , Perl DP , Haroutunian V , Sano M ((2008) ) Increased neurofibrillary tangles in patients with Alzheimer disease with comorbid depression. Am J Geriatr Psychiatry 16: , 168–174. |
[6] | Cervilla JA , Prince M , Joels S , Mann A ((2000) ) Does depression predict cognitive outcome 9 to 12 years later? Evidence from a prospective study of elderly hypertensives. Psychol Med 30: , 1017–1023. |
[7] | Geerlings MI , Schmand B , Braam AW , Jonker C , Bouter LM , van Tilburg W ((2000) ) Depressive symptoms and risk of Alzheimer’s disease in more highly educated older people. J Am Geriatr Soc 48: , 1092–1097. |
[8] | van Reekum R , Simard M , Clarke D , Binns MA , Conn D ((1999) ) Late-life depression as a possible predictor of dementia: Cross-sectional and short-term follow-up results. Am J Geriatr Psychiatry 7: , 151–159. |
[9] | Alexopoulos GS , Meyers BS , Young RC , Mattis S , Kakuma T ((1993) ) The course of geriatric depression with “reversible dementia”: A controlled study. Am J Psychiatry 150: , 1693–1699. |
[10] | Singh-Manoux A , Dugravot A , Fournier A , Abell J , Ebmeier K , Kivimäki M , Sabia S ((2017) ) Trajectories of depressive symptoms before diagnosis of dementia: A 28-year follow-up study. JAMA Psychiatry 74: , 712–718. |
[11] | Koenig AM , Bhalla RK , Butters MA ((2014) ) Cognitive functioning and late-life depression. J Int Neuropsychol Soc 20: , 461–467. |
[12] | Butters MA , Whyte EM , Nebes RD , Begley AE , Dew MA , Mulsant BH , Zmuda MD , Bhalla R , Meltzer CC , Pollock BG , Reynolds CF 3rd , Becker JT ((2004) ) The nature and determinants of neuropsychological functioning in late-life depression. Arch Gen Psychiatry 61: , 587–595. |
[13] | Koenig AM , DeLozier IJ , Zmuda MD , Marron MM , Begley AE , Anderson SJ , Reynolds CF 3rd, Arnold SE , Becker JT , Butters MA ((2015) ) Neuropsychological functioning in the acute and remitted States of late-life depression. J Alzheimers Dis 45: , 175–185. |
[14] | Rapp MA , Dahlman K , Sano M , Grossman HT , Haroutunian V , Gorman JM ((2005) ) Neuropsychological differences between late-onset and recurrent geriatric major depression. Am J Psychiatry 162: , 691–698. |
[15] | Downer B , Vickers BN , Al Snih S , Raji M , Markides KS ((2016) ) Effects of comorbid depression and diabetes mellitus on cognitive decline in older Mexican Americans. J Am Geriatr Soc 64: , 109–117. |
[16] | Demakakos P , Muniz-Terrera G , Nouwen A ((2017) ) Type 2 diabetes, depressive symptoms and trajectories of cognitive decline in a national sample of community-dwellers: A prospective cohort study. PLoS One 12: , e0175827. |
[17] | Brands AM , Van den Beng E , Manschot SM , Biessels GJ , Kappelle LJ , De Haan EH , Kessels RP ((2007) ) A detailed profile of cognitive dysfunction and its relation to psychological distress in patients with type 2 diabetes mellitus. J Int Neuropsychol Soc 13: , 288–297. |
[18] | Sullivan MD , Katon WJ , Lovato LC , Miller ME , Murray AM , Horowitz KR , Bryan RN , Gerstein HC , Marcovina S , Akpunonu BE , Johnson J , Yale JF , Williamson J , Launer LJ ((2013) ) Association of depression with accelerated cognitive decline among patients with type 2 diabetes in the ACCORD-MIND trial. JAMA Psychiatry 70: , 1041–1047. |
[19] | Beeri MS , Ravona-Springer R , Moshier E , Schmeidler J , Godbold J , Karpati T , Leroith D , Koifman K , Kravitz E , Price R , Hoffman H , Silverman JM , Heymann A ((2014) ) The Israel Diabetes and Cognitive Decline (IDCD) study: Design and baseline characteristics. Alzheimers Dement 10: , 769–778. |
[20] | Heymann AD , Chodick G , Halkin H , Karasik A , Shalev V , Shemer J , Kokia E ((2006) ) The implementation of managed care for diabetes using medical informatics in a large Preferred Provider Organization. Diabetes Res Clin Pract 71: , 290–298. |
[21] | Hughes CP , Berg L , Danziger WL , Coben LA , Martin RL ((1982) ) A new clinical scale for the staging of dementia. Br J Psychiatry 140: , 566–572. |
[22] | Fillenbaum GG , Peterson B , Morris JC ((1996) ) Estimating the validity of the clinical Dementia Rating Scale: The CERAD experience. Consortium to Establish a Registry for Alzheimer’s Disease. Aging (Milano) 8: , 379–385. |
[23] | Folstein MF , Folstein SE , McHugh PR ((1975) ) “Mini-mental state.” A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12: , 189–198. |
[24] | West RK , Ravona-Springer R , Schmeidler J , Leroith D , Koifman K , Guerrero-Berroa E , Preiss R , Hoffman H , Silverman JM , Heymann A , Schnaider-Beeri M ((2014) ) The association of duration of type 2 diabetes with cognitive performance is modulated by long-term glycemic control. Am J Geriatr Psychiatry 22: , 1055–1059. |
[25] | Beeri MS , Schmeidler J , Sano M , Wang J , Lally R , Grossman H , Silverman JM ((2006) ) Age, gender, and education norms on the CERAD neuropsychological battery in the oldest old. Neurology 67: , 1006–1010. |
[26] | Welsh KA , Butters N , Mohs RC , Beekly D , Edland S , Fillenbaum G , Heyman A ((1994) ) The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD). Part V. A normative study of the neuropsychological battery. Neurology 44: , 609–614. |
[27] | Wechsler D ((1987) ) Wechsler Memory Scale- Revised Manual, Psychological Corporation, San Antonio, TX. |
[28] | Wechsler D ((1981) ) Wechsler Adult Intelligence Scale-Revised Manual, Psychological Corporation, San Antonio, TX. |
[29] | Spreen O , Benton AL ((1977) ) Neurosensory Center Comprehensive Examination for Aphasia (NCCEA), 1977 Revision: Manual of Instructions, Neuropsychology Laboratory, University of Victoria, Victoria BC. |
[30] | Newcombe F ((1969) ) Missile Wounds of the Brain: A Study of Psychological Deficits, Oxford University Press Oxford. |
[31] | Reitan RM ((1958) ) Validity of the Trail Making Test as an indicator of organic brain damage. Percept Mot Skills 8: , 271–276. |
[32] | Yesavage JA , Brink TL , Rose TL , Lum O , Huang V , Adey M , Leirer VO , ((1982) –(1983) ) Development and validation of a geriatric depression screening scale: A preliminary report. J Psychiatr Res 17: , 37–49. |
[33] | Shimada H , Park H , Makizako H , Doi T , Lee S , Suzuki T ((2014) ) Depressive symptoms and cognitive performance in older adults. J Psychiatr Res 57: , 149–156. |
[34] | Ravona-Springer R , Heymann A , Schmeidler J , Guerrero-Berroa E , Sano M , Preiss R , Koifman K , Hoffman H , Levy A , Silverman JM , Schnaider-Beeri M ((2013) ) Haptoglobin 1-1 genotype is associated with poorer cognitive functioning in the elderly with type 2 diabetes. Diabetes Care 36: , 3139–3145. |
[35] | Holm S ((1979) ) A simple sequentially rejective multiple test procedure. Scand J Stat 6: , 65–70. |
[36] | Sjöberg L , Karlsson B , Atti AR , Skoog I , Fratiglioni L , Wang HX ((2017) ) Prevalence of depression: Comparions of different depression definitions in population-based samples of older adults. J Affect Disord 221: , 123–131. |
[37] | Anderson RJ , Freedland KE , Clouse RE , Lustman PJ ((2001) ) The prevalence of comorbid depression in adults with diabetes: A meta-analysis. Diabetes Care 24: , 1069–1078. |
[38] | Hamilton M ((1960) ) A rating scale for depression. J Neurol Neurosurg Psychiatry 23: , 56–62. |
[39] | Alexopoulos GS , Meyers BS , Young RC , Campbell S , Silbersweig D , Charlson M ((1997) ) ‘Vascular depression’ hypothesis. Arch Gen Psychiatry 54: , 915–922. |
[40] | Feng C , Fang M , Xu Y , Hua T , Liu XY ((2014) ) Microbleeds in late-life depression: Comparison of early- and late-onset depression. BioMed Res Int 2014: , 692092. |
[41] | Guerrero-Berroa E , Schmeidler J, Beeri MS ((2014) ) Neuropathology of type 2 diabetes: A short review on insulin-related mechanisms. Eur Neuropsychopharmacol 24: , 1961–1966. |
[42] | Elderkin-Thompson V , Irwin MR , Hellemann G , Kumar A ((2012) ) Interleukin-6 and memory functions of encoding and recall in healthy and depressed elderly adults. Am J Geriatr Psychiatry 20: , 753–763. |
[43] | Lee GJ , Lu PH , Hua X , Lee S , Wu S , Nguyen K , Teng E , Leow AD , Jack CR Jr , Toga AW , Weiner MW , Bartzokis G , Thompson PM; Alzheimer’s Disease Neuroimaging Initiative ((2012) ) Depressive symptoms in mild cognitive impairment predict greater atrophy in Alzheimer’s disease-related regions. Biol Psychiatry 71: , 814–821. |
[44] | Lee BK , Glass TA , McAtee MJ , Wand GS , Bandeen-Roche K , Bolla KI , Schwartz BS ((2007) ) Associations of salivary cortisol with cognitive function in the Baltimore memory study. Arch Gen Psychiatry 64: , 810–818. |
[45] | Byers AL , Yaffe K ((2011) ) Depression and risk of developing dementia. Nat Rev Neurol 7: , 323–331. |
[46] | Taylor WD , Aizenstein HJ , Alexopoulos GS ((2013) ) The vascular depression hypothesis: Mechanisms linking vascular disease with depression. Mol Psychiatry 18: , 963–974. |
[47] | Smagula SF , Lotrich FE , Aizenstein HJ , Diniz BS , Krystek J , Wu GF , Mulsant BH , Butters MA , Reynolds CF 3rd , Lenze EJ ((2017) ) Immunological biomarkers associated with brain structure and executive function in late-life depression: Exploratory pilot study. Int J Geriatr Psychiatry 32: , 692–699. |
[48] | Charlton RA , Lamar M , Zhang A , Ren X , Ajilore O , Pandey GN , Kumar A ((2018) ) Associations between pro-inflammatory cytokines, learning, and memory in late-life depression and healthy aging. Int J Geriatr Psychiatry 33: , 104–112. |
[49] | Goldsmith DR , Haroon E , Woolwine BJ , Jung MY , Wommack EC , Harvey PD , Treadway MT , Felger JC , Miller AH ((2016) ) Inflammatory markers are associated with decreased psychomotor speed in patients with major depressive disorder. Brain Behav Immun 56: , 281–288. |
[50] | Papakostas GI ((2015) ) Antidepressants and their effect on cognition in major depressive disorder. J Clin Psychiatry 76: , e1046. |
[51] | Raskin J , Wiltse CG , Siegal A , Sheikh J , Xu J , Dinkel JJ , Rotz BT , Mohs RC ((2007) ) Efficacy of duloxetine on cognition, depression, and pain in elderly patients with major depressive disorder: An 8-week, double-blind, placebo-controlled trial. Am J Psychiatry 164: , 900–909. |
[52] | Porsteinsson AP , Drye LT , Pollock BG , Devanand DP , Frangakis C , Ismail Z , Marano C , Meinert CL , Mintzer JE , Munro CA , Pelton G , Rabins PV , Rosenberg PB , Schneider LS , Shade DM , Weintraub D , Yesavage J , Lyketsos CG ; CitAD Research Group ((2014) ) Effect of citalopram on agitation in Alzheimer disease: The CitAD randomized clinical trial. JAMA 311: , 682–691. |
[53] | Sayyah M , Eslami K , AlaiShehni S , Kouti L ((2016) ) Cognitive function before and during treatment with selective serotonin reuptake inhibitors in patients with depression or obsessive-compulsive disorder. Psychiatry J 2016: , 5480391. |
[54] | Almeida OP , Hankey GJ , Yeap BB , Golledge J , Flicker L ((2017) ) Depression as a modifiable factor to decrease the risk of dementia. Transl Psychiatry 7: , e1117. |
[55] | Tapiainen V , Hartikainen S , Taipale H , Tiihonen J , Tolppanen AM ((2017) ) Hospital-treated mental and behavioral disorders and risk of Alzheimer’s disease: A nationwide nested case-control study. Eur Psychiatry 43: , 92–98. |
[56] | Ismail Z , Gatchel J , Bateman DR , Barcelos-Ferreira R , Chantillon M , Jaeger J , Donovan NJ , Mortby ME ((2018) ) Affective and emotional dysregulation as pre-dementia risk markers: Exploring the mild behavioral impairment symptoms of depression, anxiety, irritability, and euphoria. Int Psychogeriatr 30: , 185–196. |
