Meta-Analysis of the Prevalence of Major Depressive Disorder Among Older Adults With Dementia

This work may not be copied, distributed, displayed, published, reproduced, transmitted, modified, posted, sold, licensed, or used for commercial purposes. By downloading this file, you are agreeing to the publisher’s Terms & Conditions.

Meta-Analysis of the Prevalence of Major Depressive Disorder Among Older Adults With Dementia

M. Selim Asmer, MDa; Julia Kirkham, MDa; Hailey Newton, BSca; Zahinoor Ismail, MDb,c; Heba Elbayoumi, BSc Pharmc; Roxanne H. Leung, MSca; and Dallas P. Seitz, MD, PhD, FRCPCa,*

With population aging, there will be increasing numbers of older adults with Alzheimer’s disease and dementia.1 Dementia currently affects 24 million individuals worldwide, and its prevalence is expected to quadruple by 2050.2 While dementia is associated with cognitive changes, behavioral changes such as depression also frequently occur3 with up to 20% of individuals reporting some degree of clinically significant depressive symptoms.4 Depressive symptoms are even common in mild cognitive impairment (MCI)-a recent meta-analysis found that the prevalence of depression in MCI was an estimated 32%.5 Depressive symptoms can have adverse consequences for patients and their care givers,6 and thus a clear understanding of the prevalence of depression in dementia is warranted.

Individuals with dementia are twice as likely as age-matched controls to be diagnosed with depression7; conversely, the presence of chronic depression increases the risk of dementia later in life.8 As operationalized in the construct mild behavioral impairment,9 the emergence of new depressive symptoms in older adults is associated with increased risk of cognitive decline and dementia, suggesting that depression may be a prodrome of dementia in addition to a factor associated with poorer outcomes when comorbid with dementia.10,11 It is hypothesized that depression and dementia may be linked by common risk factors such as vascular disease, hypothalamic-pituitary-axis dysfunction, and increased expression of inflammatory cytokines.8,12,13 The occurrence of major depressive disorder (MDD) in dementia (ie, “depression in dementia”: DpD) may accelerate cognitive and functional decline, lead to poor medical outcomes, hasten admission to long-term care, and increase mortality.14-21 DpD is also associated with increased burden and depression among caregivers of people with dementia.14,22

Although there are known associations between MDD and dementia, the overall prevalence of DpD has not been well described. An estimated 60% of dementia patients with depressive symptoms meet the criteria for MDD, and 1 study reported a prevalence of MDD in dementia of 12.7%.23 Other studies12,14,24,25 have observed results that range from 8.0% to 40.0%, with many estimates near the lower end of the range. Possible reasons for the wide range may be due to different diagnostic criteria used to define MDD, methodological variation, and differences in the underlying study populations.

The goal of our study was to conduct a systematic review and meta-analysis to determine the prevalence of DpD among older adults (age range 58.7-87.8 years) in studies that used validated criteria for the diagnosis of both depression and dementia. We also investigated factors associated with DpD including subtype and severity of dementia, clinical setting of studies, and clinical criteria for diagnosing dementia and depression. This information may help us better understand the overall burden of DpD and inform screening, clinical evaluation, and management decisions.

METHODS

Literature Search and Inclusion Criteria

Our review was approved by Queen’s University Health Sciences and Affiliated Teaching Hospitals Research Ethics Board. Prior to beginning the literature review, our protocol was registered in the PROSPERO registry (2015:CRD42015020681).26 We searched MEDLINE (1946 to February 2017), Embase (1980 to 2017 week 5) and PsycINFO (1967 to February 2017) databases for relevant articles. We used medical subject headings and free-text search terms to identify studies measuring the prevalence of depression in patients with dementia. See Supplementary Figure 1 for search terms used in the MEDLINE database; these terms were modified slightly for the other databases.

We included prospective and retrospective observational studies published in the English language. Depression was defined as MDD diagnosed according to Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised (DSM-III-R) or more recent versions27 or International Classification of Diseases, Ninth Edition (ICD-9) or more recent versions.28 In addition to the generic criteria for MDD, we also included studies that used National Institute of Mental Health provisional criteria for depression in Alzheimer’s disease (NIMH-dAD). These criteria were derived from the DSM-IV criteria for MDD with some modifications to the duration and frequency of depressive symptoms to reflect unique features of DpD.29,30 Dementia was defined as all-cause dementia as diagnosed using DSM-III-R (or more recent) and ICD-9 (or more recent) and as specific types of dementia using National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) criteria for Alzheimer’s disease31 and National Institute of Neurological Disorders and Stroke-Association Internationale pour la Recherche et l’ Enseignement en Neurosciences (NINDS-AIREN) criteria for vascular dementia.32

• Depressive symptoms commonly occur in people with dementia, although the prevalence of major depressive disorder and factors associated with depression in dementia are not well understood.
• For all patients with dementia, particularly those with vascular dementia, clinicians should be alert to the high prevalence of major depressive disorder.

We excluded studies that only measured depressive symptoms without the criteria for MDD, studies of only Parkinson’s disease and frontotemporal dementia, and studies that defined dementia solely using cognitive screening tests. We also excluded studies where the depression or dementia diagnosis was made retrospectively using chart reviews rather than clinical evaluations.

Titles and abstracts of citations that were retrieved from the electronic databases were screened by 2 authors, and relevant studies were retrieved for full-text review. In cases where there was conflict between reviewers, a third author reviewed the citation.

Data Extraction

Two authors used a standardized form to independently extract information from each study regarding the total sample size and proportion of individuals with dementia who met criteria for MDD. Demographic information such as mean age, sex, study setting (eg, community, outpatient, long-term care, inpatient), and study country, when reported, was extracted. Clinical data regarding dementia subtype (eg, Alzheimer’s disease, vascular dementia [VaD], mixed dementia types [ie, VaD and Alzheimer’s disease], or dementia with Lewy bodies [DLB]), diagnostic criteria for the diagnosis of MDD and dementia, mean scores on cognitive testing (eg, Mini-Mental State Examination [MMSE] or other tests), and dementia severity scales (eg, Clinical Dementia Rating), when reported, were also extracted.

Assessment of Study Quality

Study quality was assessed using the Loney criteria.33 These criteria assess study bias through 8 characteristics: representativeness of the study population, sampling method, sample size, use of standardized criteria, and unbiased assessment of diagnoses and prevalence. We also considered the presence and size of confidence intervals in the estimates of prevalence. Each characteristic was rated “yes,” “no,” or “unclear” (see Table 2); “yes” answers indicated a lower risk of bias in the respective characteristics.

Data Synthesis and Meta-Analysis

Studies were assessed qualitatively to determine appropriateness to combine in meta-analysis. Estimates from clinically homogeneous study populations in individual studies were then combined using the logit transformation in a random-effects meta-analysis to arrive at the pooled estimates and 95% confidence intervals for the prevalence of DpD. We assessed the statistical heterogeneity of studies within the meta-analysis using the Q and I2 statistics, with evidence of statistical heterogeneity defined as Q (χ2) statistic P values of < .1 or I2 values > 50%. In all analyses, 2-sided P values < .05 were used as the threshold for statistical significance. Publication bias was visually assessed using funnel plots. The software package R version 3.2.3 was used for all statistical analyses (R Development Core Team and R Foundation for Statistical Computing, Vienna, Austria).

Subgroup Analyses

A comparison of the prevalence of depression by dementia subtype (ie, Alzheimer’s disease, VaD, mixed dementia types, or DLB), the criteria used to diagnose MDD and dementia, study continent, study setting (eg, community, outpatient, long-term care, inpatient), and severity of cognitive impairment was examined in subgroup analyses. Community setting was defined as participants from epidemiologic samples assessed in a general community setting or at home. Outpatient setting was defined as outpatients in a hospital or another health-care outpatient setting. Long-term care was defined as participants living in a long-term care or nursing-home facility. Inpatient setting included participants being admitted to a psychiatric inpatient unit at the time of their assessment. The severity of dementia was also examined in subgroup analyses and categorized as “mild” if the mean MMSE score was between 18-24, “moderate” if the mean MMSE score was between 10-17, and “severe” if the mean MMSE score was < 10. We also reported subgroup analyses on the basis of Clinical Dementia Rating scale scores of 0.5, 1, 2, and 3.

Sensitivity Analysis

To determine if study quality had an impact on the estimates of DpD and heterogeneity, we conducted a sensitivity analysis where the primary analysis of DpD was completed after excluding studies with a quality score of < 7 of 8.

RESULTS

Study Selection and Description of Included Studies

A total of 9,421 nonduplicate citations were found, and 8,724 were excluded after review of title and abstract. We included for full-text review 697 studies of which 642 were excluded. The flow diagram and reasons for exclusion are summarized in Figure 1. A total of 55 studies29,34-87 met inclusion criteria (Table 1). Together, these studies included a total of 13,172 subjects. Thirty studies were conducted in unspecified or mixed outpatient and inpatient settings, 2 studies were conducted in a long-term care setting, 18 were conducted in outpatient settings, 3 were conducted in community settings, and 2 were conducted in inpatient settings.

Assessment of Study Quality

Most studies were found to be of high methodological quality. Of the 55 included studies, 46 studies had a score of 7 or 8 on the quality assessment score. The remaining 9 studies received scores ranging from 4 to 6 (Table 2).

Meta-Analysis of the Prevalence of MDD in Older Adults With Dementia

The prevalence of DpD across all dementia subtypes varied from 0.0% to 91.8% in the individual studies (Figure 2). In meta-analysis, the overall prevalence of DpD for any definition of MDD and all-cause dementia was 15.9% (95% CI, 12.6%-20.1%; Figure 2). There was a high degree of heterogeneity noted in the overall meta-analysis (I2 = 98%, P < .01). There was no evidence of publication bias in this analysis, according to a visual inspection of the funnel plot (Supplementary Figure 2).

Subgroup Analyses

Prevalence of DpD by dementia subtypes. Of the 55 studies, 46 reported on the prevalence of DpD among individuals with Alzheimer’s disease and 9 reported similarly among individuals with VaD. Two studies also reported on individuals with DLB. The pooled prevalence of DpD among older adults with AD was 14.8% (95% CI, 11.5-19.1), 24.7% (95% CI, 17.6-34.6) among those with VaD, and 21.5% (95% CI, 10.5-43.9) among those with DLB (Figure 3). There was a statistically significant difference between the 3 groups (Q = 5.9, P = .05), and pairwise comparisons of each subtype showed a statistical difference between Alzheimer’s disease and VaD (Q = 5.63, P = .02) but none between the other comparisons (Alzheimer’s disease vs DLB: Q = 0.93, P = .34; VaD vs DLB: Q = 0.12, P = .73). There was evidence of statistical heterogeneity within the Alzheimer’s disease (Q = 2,314.2, P < .01, I2 = 98%) and the VaD subgroups (Q = 37.3, P < .01, I2 = 79%).

Prevalence of DpD by diagnostic criteria for MDD. All 55 studies reported at least one diagnostic criterion used to diagnose MDD. In reporting the prevalence of DpD, 34 studies used the DSM-III-R, 11 used the DSM-IV, 3 used the Research Diagnostic Criteria, 3 used the NIMH-dAD, 3 used multiple diagnostic criteria, and 1 used Automated Geriatric Examination for Computer Assisted Taxonomy to define MDD. Within the DSM-III-R and DSM-IV subgroups, the pooled prevalence of DpD was 13.2% (95% CI, 9.4%-18.6%) and 17.3% (95% CI, 9.6%-31.4%), respectively. The prevalence of DpD using NIMH-dAD was 35.6% (95% CI, 27.6%-46.0%). The difference between the 3 subgroups was statistically significant (Q = 22.1, P < .01), and pairwise comparisons of each subgroup showed a statistical difference between DSM-III-R and NIMH-dAD (Q = 20.78, P < .01) and DSM-IV and NIMH-dAD (Q = 4.79, P = .03) but none between DSM-III-R and DSM-IV (Q = 0.61, P = .43) (Supplementary Figure 3).

Prevalence of DpD by severity of cognitive impairment. A total of 10 studies reported on the prevalence of DpD among individuals with mild dementia (MMSE, 18-24) and 9 studies reported on individuals with moderate dementia (MMSE, 10-17). The pooled prevalence of DpD among those with mild dementia was 22.1% (95% CI, 15.7%-30.9%). The pooled prevalence of DpD among those with moderate dementia was 11.6% (95% CI, 6.9%-19.7%). The difference between the 2 subgroups did not meet the threshold for statistical significance (Q = 4.03, P = .04) (Supplementary Figure 4). There were also no differences in prevalence of DpD according to the Clinical Dementia Rating scale scores in the 4 studies reporting this information (Supplementary Figure 5).

Prevalence of DpD by study setting. A total of 18 studies reported on the prevalence of DpD in outpatient settings, 2 reported on those examined in long-term care, 3 reported on those in the community, and 2 reported on those sampled from inpatient settings. Studies in outpatient clinical settings reported a pooled prevalence estimate of DpD of 18.4% (95% CI, 13.9%-24.4%). The pooled prevalence of DpD among individuals in long-term care was 10.6% (95% CI, 3.4%-33.3%), in the community was 20.0% (95% CI, 8.3%-48.0%), and in inpatient samples was 16.7% (95% CI, 10.7%-26.3%). There was no significant difference between subgroups (Q = 1.0, P = .81) (Supplementary Figure 6.

Prevalence of DpD by continent of study sample. We pooled the studies by continent for subgroup analysis: 5 were from Asia, 1 from Australia, 19 from Europe, 25 from North America, and 5 from South America. The pooled prevalence of DpD from studies in Asia was 17.0% (95% CI, 8.8%-32.7%); in Australian studies, 17.7% (95% CI, 13.7%-22.8%); in European studies, 18.1% (95% CI, 12.4%-26.5%); in North American studies, 12.2% (95% CI, 7.5%-19.8%); and in South American studies, 29.5% (95% CI, 16.1%-54.1%). The differences between subgroups were not statistically significant (Q = 5.1, P = .28) (Supplementary Figure 7).

Sensitivity Analysis

After excluding studies with a quality score of less than 7, the prevalence of DpD from the remaining 46 studies was 14.1% (95% CI, 10.9%-18.3%), and heterogeneity remained high (Q = 2,364.5, P < .01, I2 = 98%).

DISCUSSION

Our review found that MDD is common among older adults with dementia with a prevalence of 15.9%. There was a high degree of heterogeneity in the estimates of the prevalence of DpD. Some of the factors that were associated with the prevalence of DpD included type of dementia as well as the diagnostic criteria used for dementia. We did not observe that study setting or dementia severity had a significant impact on the prevalence of DpD. The findings from our review are consistent with a previous meta-analysis23 of DpD, which reported a prevalence of 12.7%, although this study included only 25 studies as compared to the 55 studies that were included in our review. Overall, our review provides evidence supporting a high prevalence of DpD, highlighting its clinical importance in this population.

Our review indicates that individuals with dementia have an increased prevalence of MDD when compared to individuals without dementia, where depression prevalence is approximately 1.8%.88 The reasons for this difference in prevalence are most likely multifactorial. Neurodegenerative changes associated with dementia may lead to alterations in neurotransmitters that contribute to the increased prevalence of depression.79,89 Depression or depressive symptoms, common among individuals with MCI,5 may also be a prodrome of dementia in many individuals,10,12 and the presence of symptoms may be associated with an increased risk of subsequent conversion to dementia.90 Chronic depression has also been identified as a risk factor for the development of dementia91,92 and may contribute to the high prevalence of DpD. Individuals with dementia also frequently experience stressful life events and limited social supports that further contribute to the increased risk of depression.93 There are quite likely multiple pathways that lead to the development of DpD, and further studies are required to understand how these different factors contribute.

One challenge with accurately diagnosing DpD is that depression itself is associated with cognitive impairment.94 Individuals with major depression can display clinically significant cognitive deficits, particularly in measures of executive functioning and attention, both during depressive episodes and in the euthymic state.94 The presence of cognitive deficits in depression can also be a marker of poor response to antidepressant treatment.95 Our review focused on the diagnosis of MDD and dementia using validated diagnostic criteria, and as such, misclassification of individuals as having dementia when these cognitive deficits may have been caused by depression should have been minimal. However, additional studies are required to identify both neuropsychological profiles94,96 and imaging biomarkers,97 which may be useful in distinguishing between cognitive deficits caused by depression versus those of early dementia. In clinical practice, actively treating depression and other comorbidities that may be contributing to cognitive impairment would be recommended for individuals presenting with clinically significant depressive symptoms and cognitive impairment prior to establishing a diagnosis of dementia.

In our review, the prevalence of MDD in older adults with VaD (24.7%) was significantly higher than in those with Alzheimer’s disease (14.8%). While there may be multiple reasons for this difference, it is most likely due to the direct relationship between cerebrovascular disease pathology and the risk of developing depression. Cerebrovascular disease is known to be an independent risk factor for depression among individuals without dementia; therefore, the higher rate of DpD in VaD is consistent with the increased risk of depression.98 VaD is also a heterogeneous disease, and specific patterns of cerebrovascular disease in VaD (eg, multiinfarct dementia vs large cortical infarct) may be associated with different rates of MDD prevalence.99 Our study suggests that individuals with VaD may be at particularly high risk of depression, and clinicians should carefully monitor mood symptoms in this population.

We found that the NIMH-dAD provisional diagnostic criteria for DpD were associated with higher prevalence estimates for DpD when compared to MDD criteria that are not specific to dementia. The NIMH-dAD criteria were derived from DSM-IV criteria with modifications that include reducing the number of core depressive symptoms from 5 to 3 symptoms, omitting the concentration criteria, and allowing symptoms to be present at any point within a 2-week period, which is in contrast to the DSM-IV requirement that all the symptoms be present for most of the time in a 2-week period.30 Our review found that the NIMH-dAD criteria are quite likely more sensitive to detecting depression in older adults with dementia than the DSM-IV criteria, although the increased sensitivity may be associated with a loss of specificity.29 Whether the NIMH-dAD is susceptible to false positives or the DSM-III-R or DSM-IV is susceptible to false negatives is unknown, and whether 1 set of criteria is preferred or more accurate than another requires further study. Clinicians should be aware that the diagnostic criteria used to identify DpD may impact the case definition and prevalence estimates, and the criteria most relevant for clinical application may depend on the specific situation and intended clinical purpose such as screening or confirmation of diagnosis.

The estimated prevalence of DpD according to dementia severity was numerically higher among those with milder dementia than moderate dementia, although this was not statistically significant. While a recent meta-analysis100 showed no association between dementia severity and frequency of depressive symptoms, another review23 of DpD found a higher prevalence of MDD among individuals with milder severity of dementia. It is likely that depression criteria are more accurate among individuals with mild dementia where awareness of symptoms is preserved to a greater extent than among individuals with more advanced dementia.101 Whether the accuracy of depression criteria differs across severities of dementia requires further study. Although not statistically significant, the prevalence of depression among individuals in long-term care was lower than in other settings. This may in part be related to the severity of dementia, which tends to be higher among long-term care residents, although other factors may also play a potential role in these differences as well.

The high prevalence of DpD has important clinical implications14-21 and supports recommendations that clinicians should be vigilant for MDD in dementia.102 Given the relatively high baseline probability of depression in dementia and the availability of screening tools with good diagnostic test properties,103 screening for depression in dementia may be clinically feasible. Once depression is identified, guidelines recommend nonpharmacologic and supportive measures for all individuals with dementia,104,105 and there is increasing evidence for specific psychotherapies for treatment of DpD.106 Guidelines also recommend a trial of antidepressants when there is an inadequate response to nonpharmacologic management of MDD,104,105 although the efficacy of antidepressants in treating DpD is controversial.107

Our study used rigorous review methods and included only studies that used standardized criteria for both MDD and dementia. Most of the included studies were of high methodological quality. However, significant statistical heterogeneity was observed in our meta-analyses. While we could identify some potential contributors to this heterogeneity, additional differences in underlying methodologies and study populations very likely contributed to this heterogeneity. There are some limitations to our review. Although our review was focused on MDD among individuals with dementia, subsyndromal symptoms of depression are also common in dementia and associated with poor outcomes.4 The prevalence of subsyndromal depression and factors associated with these symptoms have been evaluated in mild cognitive impairment,5 and examination of depressive symptoms more broadly among individuals with dementia should also be considered as a separate topic for meta-analysis. Information on the use of antidepressants was not reported in the majority of studies included in our review, although antidepressant use has been reported to be common among individuals with dementia.108 The effectiveness of antidepressants for treating DpD is currently questionable,18 and whether antidepressant treatment impacts the prevalence estimates is unknown.

CONCLUSION

MDD is common among individuals with dementia, and the prevalence of DpD varies with the type of dementia and depression criteria. Persons with dementia, caregivers, and health care providers should be aware of the high prevalence of MDD in dementia. Further research is needed to clarify the relationships between dementia and depression as well as to develop optimal strategies to identify and manage depression in individuals with dementia once it develops.

Submitted: June 28, 2017; accepted December 7, 2017.

Published online: July 31, 2018.

Potential conflicts of interest: Dr Seitz has been involved as a site investigator for clinical research trials sponsored by Roche. Drs Asmer, Kirkham, Ismail, and Mss Newton, Elbayoumi, and Leung have no conflicts to disclose.

Funding/support: This work was supported by the Canadian Institutes of Health Research Knowledge Synthesis Grant KRS-141000 and the Canadian Consortium on Neurodegeneration in Aging CNA-137794.

Role of the sponsor: The supporters had no role in the design, analysis, interpretation, or publication of this study.

Previous presentation: Poster presented at the 2017 annual meeting of the American Association for Geriatric Psychiatry; March 24-27; Dallas, Texas and the Canadian Academy of Geriatric Psychiatry; November 4-5; Toronto, Ontario.

Supplementary material: Available at PSYCHIATRIST.COM.

REFERENCES

100. Verkaik R, Nuyen J, Schellevis F, et al. The relationship between severity of Alzheimer’s disease and prevalence of comorbid depressive symptoms and depression: a systematic review. Int J Geriatr Psychiatry. 2007;22(11):1063-1086. PubMed CrossRef

101. van Vliet D, de Vugt ME, Köhler S, et al. Awareness and its association with affective symptoms in young-onset and late-onset Alzheimer disease: a prospective study. Alzheimer Dis Assoc Disord. 2013;27(3):265-271. PubMed CrossRef

102. Knopman DS, DeKosky ST, Cummings JL, et al. Practice parameter: diagnosis of dementia (an evidence-based review): report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology. 2001;56(9):1143-1153. PubMed CrossRef

103. Goodarzi ZS, Mele BS, Roberts DJ, et al. Depression case finding in individuals with dementia: a systematic review and meta-analysis. J Am Geriatr Soc. 2017;65(5):937-948. PubMed CrossRef

104. Moore A, Patterson C, Lee L, et al. Fourth Canadian Consensus Conference on the Diagnosis and Treatment of Dementia: recommendations for family physicians. Can Fam Physician. 2014;60(5):433-438. PubMed

105. Rabins P, Robner B, Rummans T, et al. Guideline Watch (October 2014): Practice Guideline for the Treatment of Patients with Alzheimer’s Disease and Other Dementias. Psychiatryonline website. http://psychiatryonline.org/pb/assets/raw/sitewide/practice_guidelines/guidelines/​alzheimerwatch.pdf. 2014. Accessed July 31, 2016.

106. Orgeta V, Qazi A, Spector A, et al. Psychological treatments for depression and anxiety in dementia and mild cognitive impairment: systematic review and meta-analysis. Br J Psychiatry. 2015;207(4):293-298. PubMed CrossRef

107. Leong C. Antidepressants for depression in patients with dementia: a review of the literature. Consult Pharm. 2014;29(4):254-263. PubMed CrossRef

108. Laitinen M-L, Lönnroos E, Bell JS, et al. Use of antidepressants among community-dwelling persons with Alzheimer’s disease: a nationwide register-based study. Int Psychogeriatr. 2015;27(4):669-672. PubMed CrossRef