Abstract
Background:
An association between depression and an increased risk for subsequent dementia is well-established. Sexspecific associations are understudied yet.
Objective:
We aimed to investigate sex-specific associations between depressive symptoms and dementia risk.
Methods:
Longitudinal analyses were conducted in a pooled data set (n = 4,255, mean age = 80 years) of two prospective cohort studies (LEILA 75+, AgeCoDe). Depressive symptoms were harmonized by dichotomized scores of two different depression screening scales using established cutoffs. Transition to dementia was used as outcome in Cox proportional hazards models.
Results:
Depressive symptoms at baseline were associated with an increased risk for subsequent dementia, and this association was more pronounced in males (interaction of depressive symptoms × sex: HR = 1.64, 95% CI: 1.02–2.64, p = 0.042) in a model adjusted for study, age, and education. After additional adjustment for subjective and objective cognition, depressive symptoms and their interaction with sex (HR = 1.38, 95% CI: 0.85–2.23, p = 0.188) were no longer significantly associated with the risk for subsequent dementia. Sex-stratified analyses showed stronger and significant associations between depressive symptoms and subsequent dementia in men (e.g., HR= 2.10, 95% CI: 1.36–3.23, p = 0.001, compared to HR= 1.28, 95% CI: 1.04–1.58, p = 0.020, in women).
Conclusions:
Overall, we provide evidence for a stronger association between depression and dementia in men compared to women. Depressive symptoms should be diagnosed, monitored, and treated, not only due to depression, but also with respect to the risk for subsequent dementia, especially in elderly men.
INTRODUCTION
Several studies provided evidence for an association between depression and an increased risk for subsequent dementia [1–3]. Interactions of depression with vascular diseases, changes in glucocorticoid levels leading to hippocampal atrophy, accumulation of amyloid-β plaques, inflammatory processes, and a lack of nerve growth factors might be potential biological pathways to explain the association between depression and dementia [4]. Sex-specific differences, indicated by a higher ratio of women affected compared to men (2:1), were found both for depression [5, 6] and for dementia [7], although regional differences (between the United States, Europe, and other low- and middle-income countries of the world) might exist, and the risk for Alzheimer's disease (AD) might not differ in women and men after taking the higher life expectancy of women and secular trends in younger cohorts into account [7, 8]. According to Mielke [9], depression may stronger affect the risk for AD dementia in women due to the fact that the lifetime prevalence of depression is higher in women. However, the sex-specific association between depression and subsequent dementia in elderly men and women has not sufficiently been addressed in investigations so far.
Regarding the available literature, some studies found no sex-specific differences regarding the association between depression and subsequent dementia [10, 11]. Kessing and Nilsson [10] included 11,741 patients with depression and Chen et al. [11] used a sample of about 5,000 participants. Mild and severe depressive symptoms were associated with an increased risk for incident dementia in women, whereas only mild depressive symptoms were associated with an increased risk for incident dementia in men in another study [12], with n = 2,501 participants considered in the longitudinal analyses. Other studies found an association between depressive symptoms and subsequent dementia (using a sample of n = 2,805) [13] or depressive symptoms and transition from mild cognitive impairment (MCI) to subsequent dementia only in women [14, 15]. Artero et al. [14] included n = 2,879 MCI cases and Kim et al. [15] included n = 294 MCI cases in their analyses. However, there are also studies that found a stronger association in men. A systematic review discussed a possible “gender gap” indicated by a higher risk for subsequent dementia in men with depression [1] compared to women with depression. Fuhrer et al. [16] found that depressive symptoms were associated with an increased risk for subsequent dementia only in men in a sample of 3,777 participants initially aged 65 years or older. Dal Forno et al. [17] also found that only men with elevated depressive symptoms had an increased risk for subsequent AD in a sample of 1,357 participants with about 65 years of initial mean age (ranging from 38 to 98 years at baseline) and an observation period of up to 14 years. Mirza et al. [18] found a stronger short-term association between depressive symptoms and subsequent dementia within up to 5 years in men compared to women in a sample of 4,393 participants. Divergent results regarding sex differences might be partly explained by differences concerning the operationalization of depression and dementia, the set of covariates included, the age of participants, the follow-up duration, and the cognitive status (as cognitively impaired or unimpaired) at baseline.
Two reviews on the association between depression and dementia critically discussed a potential reporting bias of sex-specific associations [19, 1], meaning that only significant results might have been published. Due to the small number of studies with heterogeneous findings, we aim to investigate if sex-specific associations between depressive symptoms and subsequent dementia exist in a large pooled elderly sample consisting of data from one study that recruited primary care patients and another mainly population-based study. As depression might be a prodromal feature of subsequent dementia (e.g., [20]), we included subjective and (objective) mild cognitive impairment in addition to study, age, and education as covariates. The results can contribute to an improved prognosis of subsequent dementia risk in men and women with depression and have potential implications for treatment decisions of general practitioners, psychiatrists, and other clinicians (i.e., close cognitive monitoring in addition to depression treatment and sex-specific dementia prevention to approach precision medicine).
METHODS
Sample
We used data from two prospective German elderly cohort studies called the “German Study on Ageing, Cognition, and Dementia in Primary Care Patients (AgeCoDe)” and the “Leipzig Longitudinal Study of the Aged (LEILA 75+)” as part of the study “Healthy Aging: Gender specific trajectories into latest life (AgeDifferent.de)”. More detailed information on both studies has been published elsewhere (e.g., AgeCoDe: [21]; LEILA 75+: [22]). Both studies recruited participants at least 75 years of age at baseline (AgeCoDe: primary care patients in six German cities (Bonn, Düsseldorf, Hamburg, Leipzig, Mannheim, Munich); LEILA 75+: general population via local registry office plus a small portion of institutionalized subjects in Leipzig). All procedures contributing to this work comply with the Helsinki Declaration of 1975, as revised in 2008. The study protocol was approved by the local ethics committees. Written informed consent was obtained from all subjects. The interviews were conducted in-person by trained research assistants. The present study included data of covariates and predictors assessed at baseline and information of dementia status until the last follow-up of both studies (AgeCoDe with 9 follow-up assessments (observation period of up to more than 13 years), M = 6.4 years; LEILA 75+ with 6 follow-up assessments (observation period of up to more than 16 years), M = 4.8 years). Participants with a diagnosis of dementia at baseline were excluded. The present core sample consisted of 4,255 participants. Sample characteristics are given in Table 1. Sex was documented as female (coded as 0) or male (coded as 1).
Sample characteristics with status of predictors and covariates at baseline separated by dementia status until the last follow-up, sex, and depression status at baseline, respectively
Last follow-up: LEILA 75+ = follow-up 6, AgeCoDe = follow-up 9; depression status defined per cutoff scores; M, mean; SD, standard deviation; MCI, mild cognitive impairment (objective cognitive impairment without consideration of subjective cognitive impairment); SCD, subjective cognitive decline; CES-D, Center for Epidemiologic Studies Depression Scale; GDS, Geriatric Depression Scale.
Covariates
The study from which data originated was included as covariate to control for potential differences (LEILA 75+ or AgeCoDe; AgeCoDe = reference). Age at baseline was assessed in years. Education was recoded into low, medium, and high (low = reference), according to the revised version of the international new CASMIN educational classification [23]. MCI (no MCI = reference) was conceptualized as objective cognitive impairment and subjective cognitive decline (SCD) was not considered as a criterion to define MCI. MCI status was determined by the Mini-Mental State Examination (MMSE) [24] score and clinical dementia cases at baseline (see “dementia diagnoses” section below for definition), participants younger than 75 years at baseline, and participants without a Structured Interview for Diagnosis of Dementia of Alzheimer type, Multi-infarct Dementia and Dementia of other Aetiology (SIDAM; e.g., [25]) cognitive section score (SISCO) at baseline were excluded. MCI cases were defined by MMSE scores from 10 to 26 a priori, but MCI cases with scores ranging from 10 to 15 were actually not included in the present analysis (about 70% of MCI cases had MMSE scores > 24, about 30% had MMSE scores between 20 and 24, and about 0.6% of MCI cases had MMSE scores < 20; all MCI cases did not fulfill dementia diagnosis criteria at baseline). Participants with MMSE scores from 27 to 30 were classified as non-MCI. SCD was defined as subjective memory impairment and was assessed before cognitive testing by asking “Do you feel like your memory is becoming worse?” in AgeCoDe and by asking “Do you have problems with your memory?” in LEILA 75+. A dichotomous SCD variable was built (present or absent; no SCD = reference).
Depressive symptoms
The German short version of the Geriatric Depression Scale (GDS) [26, 27] was used to assess depressive symptoms in the AgeCoDe study. The GDS contains 15 dichotomous items than can be affirmed or rejected. Higher scores indicate more depressive symptoms (maximum score = 15). The German 20-item version of the Center for Epidemiologic Studies Depression Scale (CES-D) [28, 29] was used to assess depressive symptoms in the LEILA 75+ study. Responses to each CES-D item range from 0 to 3. Higher scores indicate more depressive symptoms (maximum score = 60). A GDS score of≥6 or a CES-D score of≥23 was considered as cutoff for depressive symptoms in the present study, as we considered these as commonly used and accepted thresholds to indicate clinically relevant depression (e.g., [16, 30]).
Dementia diagnoses
Assessment in both studies was based on the SIDAM (e.g., [25]), a structured clinical interview for the diagnosis of dementia based on operationalized diagnostic criteria of the DSM-III-R, DSM-IV, and ICD-10 classification systems. SIDAM includes a cognitive test battery (SISCO) and a section for clinical judgment and third-party information on psychosocial impairment, including a scale for the assessment of activities of daily living with 14 items (SIDAM-ADL scale). SISCO consists of 55 items with several cognitive domains (i.e., orientation, memory, intellectual abilities, verbal ability, calculation, constructional ability, aphasia, apraxia), including the 30 items of the MMSE [24]. In the AgeCoDe study, dementia diagnoses were based on DSM-IV criteria. If SIDAM could not be assessed, a Global Deterioration Scale [31] score of at least 4 and/or a Blessed Dementia Rating Scale [32] score were used as proxy. In the LEILA 75+ study, dementia diagnoses were based on DSM-III-R criteria. In case of proxy interviews only, the Clinical Dementia Rating Scale (CDR) [33] was used to obtain information on cognitive and psychosocial functioning and to diagnose dementia. All interviews were conducted by trained research assistants (psychologists or physicians). In both studies, positive dementia diagnoses were validated in consensus conferences of interviewers and geriatric experts (psychologists, geriatricians or geriatric psychiatrists), resulting in a consensus diagnosis for each incident dementia case.
Statistical analyses
Participants with a dementia diagnosis at baseline were excluded. Baseline variables were used to predict the time of transition to all-cause dementia until the last follow-up of both studies in Cox proportional hazards models. Time (not age) was used as time scale in Cox proportional hazards models. To avoid over- or underestimation of the time until dementia was manifest, the onset of incident dementia was estimated by dividing the time between the last dementia-free follow-up assessment and the follow-up assessment when dementia was first diagnosed in addition to all dementia-free follow-ups before (e.g., if dementia was first diagnosed in follow-up 2, the midpoint of the interval between follow-up 1 and follow-up 2 was estimated and this value was added to the time until follow-up 1 to obtain the time until dementia onset). In a first set of analyses, covariates (model 1: study, age, education; model 2: model 1 + MCI, SCD) were entered at first, followed by sex (female = reference) and depressive symptoms (harmonized via dichotomized cutoff scores of both scales) in the second step, and the interaction term of sex (female = 0, male = 1) multiplied by depressive symptoms (no depressive symptoms = 0, depressive symptoms = 1) in the third step. The proportional hazard assumption was examined for both of these cox regression models and it was sufficiently met, as Schoenfeld residuals were not significantly correlated with survival time (except those of MCI and SCD in model 2), indicating that we are able to interpret the effects of depressive symptoms. In a second set of analyses, the analyses were stratified by sex and models were adjusted for covariates that were entered in the first step (model 1 and model 2 as reported above) followed by depressive symptoms (dichotomized as reported above) entered in the second step. Additional sensitivity analyses were conducted where depressive symptoms were considered as continuous scores harmonized by Z-standardization of the sum scores of each scale (GDS, CES-D). Multicollinearity was tested. Significant, but low correlation coefficients between SCD and depressive symptoms, ranging from 0.10 to 0.15, were found. Additionally, collinearity statistics revealed VIF values of all predictor variables (i.e., depressive symptoms, SCD, MCI, sex, education, age, and study) close to 1 (range: 1.011 to 1.083), which is low enough to assume that multicollinearity was not present and did not affect our results. Level of significance was set to α< 0.05. The analyses were performed using SPSS version 24 for Windows.
RESULTS
At first, we included covariates, sex, depressive symptoms, and an interaction term of depressive symptoms multiplied by sex in Cox proportional hazards models. In the model adjusted for study, age, and education as covariates, sex, depressive symptoms, and the interaction of depressive symptoms multiplied by sex (HR = 1.64, 95% CI: 1.02–2.64, p = .042) were significantly associated with the risk for subsequent all-cause dementia. Male sex was associated with a lower risk (HR = 0.84, 95% CI: 0.72–0.97, p = 0.020) and depressive symptoms with a higher risk (HR = 1.31, 95% CI: 1.07–1.62, p = 0.011) for subsequent dementia. The direction of the interaction coefficient indicated that depressive symptoms in men were associated with a significantly higher risk for subsequent dementia compared to women (see Table 2).
Prediction of subsequent all-cause dementia until the last follow-up (LEILA 75+ = follow-up 6, AgeCoDe = follow-up 9) by covariates (study, age, education), sex, depressive symptoms (indicated by cutoff scores), and interaction term (depressive symptoms×sex) assessed at baseline in Cox proportional hazards model
HR, hazard ratio; CI, confidence interval, depressive symptoms dichotomized by scale-specific cutoff scores of GDS≥6 and CES-D≥23; GDS, Geriatric Depression Scale; CES-D, Center for Epidemiologic Studies Depression Scale, interaction term: depressive symptoms×sex = sex (female = 0, male = 1) multiplied by depressive symptoms (no depressive symptoms = 0, depressive symptoms = 1).
After an additional adjustment for MCI and SCD (beside study, age, and education), depressive symptoms (HR = 1.18, 95% CI: 0.96–1.47, p = 0.124) and the interaction of depressive symptoms multiplied by sex (HR = 1.38, 95% CI: 0.85–2.23, p = 0.188) were not significantly associated with the risk for subsequent dementia anymore, whereas a lower risk for subsequent dementia in men remained significant (HR = 0.83, 95% CI: 0.71–0.97, p = 0.016; see Table 3).
Prediction of subsequent all-cause dementia until the last follow-up (LEILA 75+ = follow-up 6, AgeCoDe = follow-up 9) by covariates (study, age, education, MCI, SCD), sex, depressive symptoms (indicated by cutoff scores), and interaction term (depressive symptoms×sex) assessed at baseline in Cox proportional hazards model
HR, hazard ratio; CI, confidence interval, depressive symptoms dichotomized by scale-specific cutoff scores of GDS≥6 and CES-D≥23; GDS, Geriatric Depression Scale; CES-D, Center for Epidemiologic Studies Depression Scale, interaction term: depressive symptoms×sex = sex (female = 0, male = 1) multiplied by depressive symptoms (no depressive symptoms = 0, depressive symptoms = 1).
Sex-stratified Cox proportional hazards models indicated a higher risk to develop subsequent all-cause dementia in men with depressive symptoms compared to women with depressive symptoms in two adjusted models, although associations were significant in both sexes in model 1. In the first model adjusted for study, age, and education (see Table 4, model 1), depressive symptoms in women were associated with an increased risk for subsequent all-cause dementia by 28% (HR = 1.28, 95% CI: 1.04–1.58, p = 0.020), whereas depressive symptoms in men were associated with a more than two-fold increased risk for subsequent all-cause dementia (HR = 2.10, 95% CI: 1.36–3.23, p = 0.001). After additional adjustment for MCI and SCD (beside study, age, and education; see Table 4, model 2), depressive symptoms were not associated with a significantly increased risk for subsequent dementia in women (HR = 1.15, 95% CI: 0.93–1.43, p = 0.193), but a significant association between depressive symptoms and an increased risk for subsequent dementia remained in men (HR = 1.67, 95% CI: 1.08–2.60, p = 0.022).
Prediction of subsequent all-cause dementia until the last follow-up (LEILA 75+ = follow-up 6, AgeCoDe = follow-up 9) by covariates (model 1: study, age, education; model 2: model 1 + MCI, SCD) and depressive symptoms (indicated by cutoff scores) assessed at baseline in Cox proportional hazards models separately for women and men
HR, hazard ratio; CI, confidence interval; MCI, mild cognitive impairment; SCD, subjective cognitive decline, depressive symptoms dichotomized by scale-specific cutoff scores of GDS≥6 and CES-D≥23; GDS, Geriatric Depression Scale; CES-D, Center for Epidemiologic Studies Depression Scale.
Prediction of subsequent all-cause dementia until the last follow-up (LEILA 75+ = follow-up 6, AgeCoDe = follow-up 9) by covariates (model 1: study, age, education; model 2: model 1 + MCI, SCD) and depressive symptoms (indicated by continuous Z-scores) assessed at baseline in Cox proportional hazards models separately for women and men
HR, hazard ratio; CI, confidence interval; MCI, mild cognitive impairment; SCD, subjective cognitive decline, depressive symptoms harmonized by Z-standardization of continuous sum scores of each scale (GDS, CES-D); GDS, Geriatric Depression Scale; CES-D, Center for Epidemiologic Studies Depression Scale.
We tried to replicate the results that we obtained from the pooled data set separately in the LEILA 75+ and in the AgeCoDe study. Interaction effects were not significant in both single studies, although the direction of the interaction coefficients also indicated a stronger association in men with depressive symptoms (see Supplementary Material: LEILA 75+ (cf. Supplementary Tables 1 and 2): HR = 2.68, 95% CI: 0.78–9.24, p = 0.118 (adjusted for age, education, sex, and depressive symptoms); AgeCoDe (cf. Supplementary Tables 5 and 6): HR = 1.26, 95% CI: 0.74–2.12, p = 0.396 (adjusted for age, education, sex, and depressive symptoms)). Sex-stratified analyses within the LEILA 75+ study did not show an association between depressive symptoms and subsequent dementia in both sexes, whereas the AgeCoDe study showed an association in both men and women. However, hazard ratios were higher descriptively for men compared to women in both studies (see Supplementary Material: LEILA 75+ (cf. Supplementary Tables 3 and 4): men: HR = 2.11, 95% CI: 0.63–7.02, p = 0.226, women: HR = 0.79, 95% CI: 0.55–1.15, p = 0.217 (adjusted for age and education); AgeCoDe (cf. Supplementary Tables 7 and 8): men: HR = 2.12, 95% CI: 1.33–3.38, p = 0.001, women: HR = 1.70, 95% CI: 1.32–2.19, p < 0.001 (adjusted for age and education)).
Additionally, sensitivity analyses using depressive symptoms as continuous Z-standardized scores were conducted. Sex-stratified Cox proportional hazards models revealed higher hazard ratios to develop subsequent all-cause dementia in men with depressive symptoms compared to women with depressive symptoms in two adjusted models, although associations were significant in both sexes (see Table 5). In the first model adjusted for study, age, and education (see Table 5, model 1), continuous depressive symptom scores were associated with an increased risk for subsequent all-cause dementia by 21% (HR = 1.21, 95% CI: 1.13–1.30, p < 0.001) in women and by 33% (HR = 1.33, 95% CI: 1.17–1.50, p < 0.001) in men. Hazard ratios decreased after additional adjustment for MCI and SCD (beside study, age, and education; see Table 5, model 2), but remained significant in both sexes, with a higher hazard ratio in men (women: HR = 1.13, 95% CI: 1.05–1.22, p = 0.001; men: HR = 1.27, 95% CI: 1.11–1.44, p < 0.001).
DISCUSSION
We provided further evidence that elevated depressive symptoms in late life might be stronger associated with subsequent dementia in men than in women. After the inclusion of subjective and objective cognitive impairment, depressive symptoms were not associated with an increased risk for subsequent dementia in women, but in men, in sex-stratified analyses. Hence, depressive symptoms might hold additional predictive power for subsequent dementia beyond subjective and objective cognitive impairment in men. Interestingly, we found a stronger association between SCD and subsequent dementia in women compared to men in previous analyses in the AgeCoDe study [34] and it is therefore tempting to speculate that men might express rather global and affective complaints through depressive symptoms during the pre-dementia stage, whereas women might express more specific worries via subjective cognitive deficits. Additionally to or instead of SCD, the inclusion of MCI might also have attenuated the association between depressive symptoms and subsequent dementia in women, in the present study.
Our results are in accordance with other studies. Dal Forno et al. [17] found an association between depressive symptoms and subsequent dementia, especially AD, in men, but not in women, in a younger sample compared to ours. Fuhrer et al. [16] found an association between depressive symptoms and subsequent dementia and AD only in men. Their results suggested that vascular diseases might be relevant as the association between depression and dementia was significant only in the male hypertensive group in stratified analyses [16]. Interestingly, Dotson et al. [35] showed that the CES-D “depressed mood” subscale predicted longitudinal increases in white matter hyperintensities in men, but not in women, without dementia. Interactions with vascular diseases have been suggested as one potential biological pathway to explain the association between depression and dementia [4] and stronger vascular relations in men might also explain sex-specific differences of the association between depression and subsequent dementia. Further relevant biological factors explaining the sex difference observed in our study might be hormonal and neurotransmitter differences [17, 36]. Others found a pronounced association between depression and subsequent dementia (especially) in men who carried the Apolipoprotein E ɛ4 (ApoE4) allele [37]. Although these results do not directly relate to the results on sex differences found in our study, they might help to more precisely define groups (of men) with depression at higher risk for subsequent dementia.
Beside biological factors, men might also be more reluctant to report depressive symptoms due to psychosocial or societal reasons [17, 36]. Therefore, using the same cutoff score to define depressive symptoms for both sexes might actually result in more severe male cases. Hence, a greater severity of depressive symptoms reported by men might be one explanation for the stronger association with subsequent dementia in males. Noale et al. [12] also found a low prevalence of severe depressive symptoms in men, which was related to increased mortality in males additionally. Further sensitivity analyses in the present study showed that a continuous measure of Z-standardized depressive symptoms was associated with subsequent dementia also in women. Hence, more severe depressive symptoms in women (also indicated by a higher cutoff score) are probably associated with subsequent dementia in the present sample, in accordance with the findings of others [38]. Interestingly, two studies that found no sex-specific associations between depression and dementia used clinical depression syndromes or diagnoses [10, 11], which might substantiate the relevance of severity. Thus, severe depressive symptoms might be associated with subsequent dementia irrespective of sex. Our results contradict with other studies, which found an association only in women [13–15], but Artero et al. [14] and Kim et al. [15] examined the transition from MCI to dementia, whereas we included participants with and without MCI at baseline.
Recent studies (e.g., [20]) found that late-onset depression is associated with an increased risk for subsequent dementia particularly. Hence, late-onset depression might be considered as a prodromal feature of dementia in some cases, although other studies also found an increased risk for dementia with lifetime depression and depression with an earlier onset [19]. Unfortunately, we did not differentiate between lifetime versus late-onset depression in our present analyses, but it might be interesting to investigate if sex-specific associations between depression onset and dementia exist in future studies.
Strengths and limitations
The present study has several strengths. We used a large sample to obtain a sufficiently high number of cases regarding both depressive symptoms and dementia. Standard procedures to diagnose dementia were used and consensus of experts regarding dementia diagnoses was reached. Depressive symptoms were assessed using two valid and commonly used screening scales. Rates of depression and dementia were comparable to other studies. The external validity and the generalizability of our results might be increased due to the inclusion of different original studies. Additionally, we conducted the analyses separately in both original studies and found similar, although predominantly non-significant, results (cf. Supplementary Material).
Beside these strengths, there are also several limitations. Although the GDS and the CES-D can detect major depression in late-life in primary care with good accuracy, frequently used cutoff scores of GDS and CES-D poorly detected major and minor depression in a very old sample [39]. Although we consider GDS and CES-D as valid instruments, they might not measure completely identical constructs of depressive symptoms. Hence, both scales might predict subsequent dementia with different validity and accuracy. Moreover, we did not include clinical depression diagnoses or age of depression onset in addition to the assessment of depressive symptoms. Different cutoff points for GDS and CES-D to indicate clinically relevant depression have been reported (e.g., [40]). Although we conservatively used rather high cutoff points, other thresholds might have produced different results. The methods used to assess dementia diagnoses differed across and partly within both studies, which must be considered as a clear limitation. However, the cognitive testing part of SIDAM (SISCO) could be assessed at baseline in almost all participants (97.4% of participants without subsequent dementia, 96.7% of participants with subsequent dementia). Although we consider the diagnostic group classification and definition of dementia diagnoses of both included studies as adequate and reasonable, we have to admit that the range of MMSE scores in MCI cases included might correspond to the range of at least mild dementia in other studies. The risk for subsequent dementia was higher in the LEILA 75+ study, which might represent real differences, e.g., due to the longer follow-up and the initial inclusion of a subsample of institutionalized subjects, which was not realized in the AgeCoDe study, or otherwise might represent different diagnostic thresholds for dementia, although we consider this unlikely due to similar diagnostic procedures. The LEILA 75+ study included a longer follow-up than the AgeCoDe study, but there were also differences within each study and mean observation durations were similar in both studies. Our results only refer to all-cause dementia and different etiologies such as AD were not considered. Additional covariates, such as comorbidities, vascular risk factors, and (antidepressant) medication, which are related to depression and dementia, were not consistently assessed in both studies and were thus not incorporated in the pooled data set.
Implications of the study
Depressive symptoms, possibly indicating a clinically relevant depression, should be taken seriously due to the depression itself, but also due to the possible association with the development of later dementia both in women and in men. Depressive symptoms in late life might be a stronger predictor for subsequent dementia in men compared to women. Our results were in accordance with findings from other studies. However, the effect seemed to be small and interaction effects were only significant in the basically adjusted, but not in the fully adjusted interaction model. Nonetheless, additional or frequent cognitive assessments might be indicated in depressed men independent of their objective or subjective cognitive status, whereas the risk for subsequent dementia was not predicted by depressive symptoms defined by common cutoffs beyond cognitive and subjective cognitive impairment in women. However, severe depressive symptoms in women indicated by higher than common cutoffs might also increase the risk for subsequent dementia and should not be dismissed. Therefore, diagnostic investigations including dementia biomarker and eventually ApoE4 assessment, clinical monitoring, and treatment might be initiated in men who exceed common cutoff scores and in women who report severe depressive symptoms. Members of the AgeDifferent.de Study Group: Steffi G. Riedel-Heller (Principal Investigator), Franziska Förster, Johannes Golchert, André Hajek, Kathrin Heser, Hans-Helmut König, Wolfgang Maier, Alexander Pabst, Elzbieta Buczak-Stec, Michael Wagner, Birgitt Wiese
This publication is part of the study “Healthy Aging: Gender specific trajectories into latest life” (AgeDifferent.de) and was funded by the German Federal Ministry of Education and Research (Funding program “Gesund –ein Leben lang", grant numbers 01GL1714A, 01GL1714B, 01GL1714C, 01GL1714D).
Findings from this study will be presented at the congress of Deutsche Gesellschaft für Psychiatrie und Psychotherapie, Psychosomatik und Nervenheilkunde (DGPPN) 2019.
Authors’ disclosures available online (https://www.j-alz.com/manuscript-disclosures/19-0770r2).
