INTERNATIONAL JOURNAL OF GERIATRIC PSYCHIATRY

Int J Geriatr Psychiatry 2009; 24: 169–176.

Published online 21 July 2008 in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/gps.2088

Depression and excess mortality: evidence for a doseresponse relation in community living elderly

R. A. Schoevers1,2*, M. I. Geerlings3, D. J. H. Deeg4, T. J. Holwerda1,2,C. Jonker1,4 and A. T. F. Beekman1

1VU University Medical Center, Department of Psychiatry, Amsterdam, The Netherlands2JellinekMentrum Mental Health Care Amsterdam, Amsterdam, The Netherlands3Julius Center for Health Sciences and Primary Care, University Medical Center, Utrecht, The Netherlands4Institute for Research on Extramural Medicine, VU University Medical Center, Amsterdam, The Netherlands

SUMMARY

Context Depression is associated with an increased mortality risk. It is not known to what extent depression characteristicssuch as severity and length of exposure to depression contribute to the association with excess mortality.Objectives To investigate the association between depression severity and duration with mortality in community-livingelderly.Design Two-wave prospective cohort study with 10-year follow-up of vital status. Assessment of depression at baselineand at three year follow-up (GMS-AGECAT). Cox proportional hazards analyses of mortality with depression according toseverity and length of exposure, adjusted for demographic variables, physical illnesses, cognitive decline and functionaldisabilities.Setting and participants Randomly selected cohort of 3 746 non-demented older community-living persons in the city ofAmsterdam.Main outcomemeasures Excess mortality of both the baseline cohort, and of non-demented subjects participating in bothassessments (n¼ 1989).Results Both moderate (MHR 1.29, 95% CI 1.03–1.61) and severe depression (MHR 1.34, 95% CI 1.07–1.68) predicted10-year mortality after multivariate adjustment. Chronic depression was associated with a 41% higher mortality risk in6-year follow-up compared to subjects without depression.Conclusions Severity and chronicity of depression are associated with a higher mortality risk. In combination with otherfindings this is suggestive of a causal relationship and may have implications for both preventive and treatment strategies oflate-life depression. Copyright # 2008 John Wiley & Sons, Ltd.

key words— depression; mortality; elderly; prognosis; epidemiology

INTRODUCTION

Unipolar major depression is expected to become thesecond leading cause of disability world-wide in 2020,with only ischemic heart disease causing higher diseaseburden and economic impact (Murray and Lopez,1997). Depression is also associated with excessmortality (Wulsin et al., 1999), both in the community

*Correspondence to: Dr R. A. Schoevers, JellinekMentrum MentalHealth Care, Amsterdam, WG-plein 416, 1054 SH Amsterdam, TheNetherlands. E-mail: robert.schoevers@mentrum.nl

Copyright # 2008 John Wiley & Sons, Ltd.

(Cuijpers and Smit, 2002), in medical patients (Katon,2003) and in studies that specifically focussed on theelderly (Pennix et al., 1999; Ganguli et al., 2002).

Still, the exact mechanism by which this associationcan be explained remains elusive. Depression isassociated with biological dysregulations such asdysfunctions in the hypothalamic-pituitary-adrenocor-tical (HPA) system (Musselman and Nemeroff, 1996;Holsboer, 2000), autonomic dysregulation (Yeraganietal.,2002),areducedheartratevariability(vanderKooyet al., 2006) and inflammatory processes (Penninx et al.,2003). Depression has been shown to be an independent

Received 11 July 2007Accepted 3 June 2008

170 r. a. schoevers ET AL.

risk factor for the onset of cardiovascular diseases (Vander Kooy et al., 2007). Similarly, depression may lead toexcess mortality through suicidal ideation, althoughpreviousworksuggeststhatsuicidealonecannotaccountfor the excess mortality of depression (Murphy et al.,1987, 1988; Lindesay, 1989; Penninxet al., 2001; Maris,2002). Depression is also associated with unhealthy lifestyles such as smoking, alcohol use (Aneshensel andHuba, 1983; Goodwin and Jamison, 1990), physicalinactivity (Stephens, 1988), and noncompliance withmedical treatment (DiMatteo et al., 2000). Still, theseassociations are often bi-directional and complex(Cuijpers and Schoevers, 2004).

Apart from biological plausibility, and consistencyof an association across different population samples,criteria for a causal association between risk factorsand outcome in epidemiological research include theexistence of a dose-response relationship (Weed,1997). Few studies have explicitly examined suchdose-response associations. Studies that dichotomisedseverity levels as major and minor depression(Penninx et al., 1999), or psychotic and neuroticdepression (Schoevers et al., 2000b) did not show astatistically significant association of severity withexcess mortality after adjustment for other explana-tory variables. More continuous measures of severityhowever did yield an association with mortality in twostudies (Schulz et al., 2000; Geerlings et al., 2002).The association of depression with mortality may alsobe affected by length of exposure to depression. Indepressions with a longer duration, the negativeeffects of depression on both biological and psycho-social functioning have had more time to accumulate.Although this seems plausible, again very few studieshave examined this parameter (Geerlings et al., 2002).

In order to establish a possible dose-response asso-ciation between depression and survival, prospectivecommunity based studies are needed that include acomprehensive set of confounders, large sample sizesandarelativelylongfollow-uptime.TheAMSTELstudyis communitybasedand started in1990, witha follow-upthree years later, and collection of mortality data until 10years after baseline (Schoevers et al., 2003b, 2006). Theresearchquestionforthisstudyistodescribetheimpactofdepression on mortality according to depression severityand length of exposure to depression.

METHODS

Sample

The Amsterdam Study of the Elderly (AMSTEL) is alongitudinal study in a large and representative sample

Copyright # 2008 John Wiley & Sons, Ltd.

of non-institutionalized community living older per-sons on mental health problems, medical diagnosesand demographic characteristics. Informed consentwas asked and obtained from each participant beforestarting the study. The study was approved by theMedical Ethical Committee of the Vrije UniversiteitAmsterdam.

Sampling, data collection procedures and responsehave been described in depth elsewhere (Launer et al.,1993, 1994). In short, a random and representativesample was drawn from the non-institutionalizedAmsterdam population over 65–84 years old. An age-stratified cohort of 4,051 subjects was enrolled for thebaseline interview, which was conducted betweenMay 1990 and November 1991. GMS-AGECATdementia was present in 261 subjects (6.4%).

The study sample for this study consisted of all 3,746subjects participating in the baseline assessment whodid not have GMS-AGECAT dementia, of whom vitalstatus could be ascertained. At 3-year follow-up(median 38 months), data were available in 2,144(57.2%) subjects. At follow-up, 553 (14.8%) personswere deceased, 604 (16.1%) persons refused furtherparticipation, 254 (6.8%) were too ill or cognitivelyimpaired to respond, and 191 (5.1%) were not availablefor interview due to other reasons. Of the responders,155 subjects (7.2%) had developed dementia betweenthe two assessments. These subjects were excludedfrom the survival analyses after 1994. The study groupfor the assessment of the effect of length of exposure todepression thus consisted of 1989 subjects withoutGMS-AGECAT dementia.

Baseline measures

A 1-h interview was developed to gather informationon psychiatric symptoms, demographic informationand medical status. The interview consisted of theDutch translation of the Mini-Mental State Examin-ation (Folstein et al., 1975), all Geriatric Mental StateExamination-items related to organic, affective andanxiety syndromes (Copeland et al., 1986), theActivities of Daily Living (ADL) scale (Katz et al.,1963), the Instrumental Activities of Daily Living(IADL) scale (Lawton and Brody, 1969), and theCAMDEX-interview (Roth et al., 1986). The inter-view was administered during home visits by layinterviewers who were specially trained using videosessions and regularly supervised. The same GMS-AGECAT package with an identical algorithm wasused in the second wave of the study. When re-interviewing, raters were unaware of previous data anddiagnoses.

Int J Geriatr Psychiatry 2009; 24: 169–176.

DOI: 10.1002/gps

depression and excess mortality 171

Depression and dementia. Depression and dementiawere diagnosed using GMS-AGECAT (Copelandet al., 1986, 1988), an established and widely usedinstrument that diagnoses cases of late-life depressionrequiring clinical attention in the community (Cope-land et al., 1990, 1992) and in hospital patients (Amesand Tuckwell, 1994). GMS-AGECAT has provenreliability for epidemiological work in replicationstudies (Copeland et al., 1988; Hooijer et al., 1991).

Depression severity. In order to study the associationof different severity levels of depression withmortality, a sum score was composed out of the totalnumber of 31 items for depression in GMS (seeTable 1 for items and item scores). According to thescores of subjects with depression at baseline, a severityvariable was calculated that classified the lower tertile ofthe scale as ‘mild depression’ (2–12 points), the middletertile as ‘moderate depression’ (13–19 points), and theupper tertile as ‘severe depression’ (20–48 points). In asecond approach, a continuously scaled variable of thetotal depression item score (0–48) to was used toindicate severity.

Length of exposure. Longitudinal patterns of theexposure to depression were assessed in the subjectsthat participated in both waves of the study (n¼1,989). To investigate whether a longer period ofexposure to depression would show a higher mortalityrisk, two measurements of depression allowed fourdifferent categories. Subjects were classified as ‘nodepression’, ‘remitted depression’ (depressed at base-line and no longer depressed at follow-up), ‘incidentdepression’ (newly depressed at follow-up) and‘chronic depression’ (depressed at both assessments).

Covariates. Potentially confounding covariatesincluded age, sex and level of education. Educationalstatus was assessed as the number of years ofeducation (5–18 years). Marital status was assessedbased on the questions in GMS-AGECAT. Thepresence of chronic diseases was assessed with thepertinent Camdex questions on cardiovascular dis-eases, cancer, lung disease, diabetes, Parkinson’sdisease and epilepsy. Cognitive status was assessed byMMSE-score. Subjects were considered to havefunctional disability if their ADL scores were twoor more points below the maximum score on the scale.This indicates that subjects were ‘in need of help’ toperform at least two of the tasks mentioned, or wereunable to perform at least one task.

Copyright # 2008 John Wiley & Sons, Ltd.

Mortality

The follow-up for recording deaths extended fromthe date of the baseline examination until the31 December 2000. Dates of death were ascertainedthrough the registers of the municipality of Amster-dam, or the municipalities where subjects had movedto during the study period. Data on vital status weremissing in 47 subjects (98.8% complete).

Data analysis

Baseline sample characteristics and participation inthe follow-up study were analyzed using Chi2

statistics. Cox proportional hazards regression modelswere used to examine the association betweendepression and time to deaths. The mortality riskwas expressed as the mortality hazard ratio (MHR).When the 95% Confidence Interval (CI) of the latterdid not include one, the association was regarded to bestatistically significant.

In multivariate analysis, 10-year mortality from1990–2000 according to depression severity (mild,moderate or severe depression, each compared with‘no depression’, and a separate analysis using thecontinuous depression severity score) was studied in aCox regression model, with stepwise adjustment forpotential confounding factors. In the first step,sociodemographic factors (age, gender, years ofeducation, marital status) were entered. The secondstep adjusted for cardiovascular diseases, the third stepfor all other diseases and in the fourth step the MMSEscore was entered. As functional limitations arestrongly associated with depression (Penninx et al.,1998), and overcorrection for the effect of depressionmay result when entering this factor in a multivariatelogistic regression model, this factor was adjusted for inthe last step. Results are presented for each step inthe multivariate model. Differences between strata wereconsidered statistically significant when the confidenceintervals mutually excluded the opposite MHR value.

Mortality according to length of exposure todepression (remitted depression, incident depressionand chronic depression vs no depression, and remitteddepression vs chronic depression) was studied in asimilar Cox regression model, with survival timecounted from date of 1994 interview until 2000.

RESULTS

Sample characteristics

Baseline characteristics and the associations withmortality are shown in Table 1. At baseline, depression

Int J Geriatr Psychiatry 2009; 24: 169–176.

DOI: 10.1002/gps

Table 1. Baseline sample and 10-year mortality

Variables Baseline Deceased (%) Statistics X2 (df) P

N 3746 1844 (49.2)Age

65–69 803 219 (27.3) 453.3 (3)***70–74 929 338 (36.4)75–79 974 533 (54.7)80–86 1040 754 (72.5)

GenderMen 1425 831 (58.3) 76.0 (1)***Women 2321 1013 (43.6)

EducationMore than Primary School 2231 1017 (45.6) 29.3 (1)***Primary School or less 1515 827 (54.6)

Marital statusMarried 1853 859 (46.4) 12.0 (1)**Not/no longer married 1890 983 (52.0)

Myocard.infarction: No/never 3346 1586 (47.1) 57.1 (1)***Yes 382 258 (67.5)Stroke: No/never 3545 1709 (48.2) 27.3 (1)***Yes 201 135 (67.2)Cancer: No/never 3332 1609 (48.3) 10.6 (1)**Yes 414 235 (56.8)Lung disease: No 3136 1480 (47.2) 31.8 (1)***Yes 610 364 (59.7)Diabetes: No 3417 1624 (47.5) 44.9 (1)***Yes 329 220 (66.9)Epilepsy: No/never 3691 1810 (49.0) 3.5 (1) nsYes 55 34 (61.8)Parkinson’s: No 3692 1798 (48.7) 28.2 (1)***Yes 54 46 (85.2)Arthritis: No 3107 1517 (48.8) 0.28 (1) n.s.Yes 639 327 (51.2)

MMSE26–30 3195 1476 (46.2) 89.9 (2)***22–25 465 297 (63.9)0–21 86 71 (82.6)

ADL disability: No 3492 1656 (47.4) 70.0 (1)***Yes 254 188 (74.0)

DepressionNo 3291 1586 (48.2) 11.6 (1)**Yes 455 258 (56.7)

Depression severity scoreNo depression 3291 1586 (48.2) 13.8 (3)**Mild (2–12) 165 86 (52.1)Moderate (13–19) 146 88 (60.3)Severe (20–48) 144 84 (58.3)

Deceased at 01–01–2001.p< 0.05; **p< 0.01; ***p< 0.001; ns¼ not significant with p> 0.05.

172 r. a. schoevers ET AL.

was present in 455 subjects (12.1%). One hundred andsixty-five subjects had mild depression, 146 haddepression of moderate severity, and 144 severedepression. The total follow-up time of the sample was

Copyright # 2008 John Wiley & Sons, Ltd.

28156.5 person years. One thousand eight hundredand forty-four (49.2%) subjects had died during thestudy period. The crude mortality rate was 65.5 deathsper 1000 person-years. The average follow-up period

Int J Geriatr Psychiatry 2009; 24: 169–176.

DOI: 10.1002/gps

Figure 1. Kaplan-Meyer survival curve according to depressionseverity level.

depression and excess mortality 173

from the baseline assessment of subjects who werestill alive at 1/1/2001 was 117.8 months, ranging from109.2 to 127.2 months. The average follow-up periodof subjects participating in the 1994 assessment whowere still alive at 1/1/2001 79.6 months, ranging from72.3 to 95.9 months. Univariate comparison of base-line characteristics with mortality showed that allcharacteristics, except arthritis and epilepsy, wereassociated with 10-year mortality.

Severity

Subjects with a depression of moderate or highseverity showed an excess mortality risk of 46% and35% respectively, when compared to subjects withoutdepression. This pattern is visually illustrated by theKaplan-Meyer survival curve in Figure 1. Afteradjustment for all of the other variables in the model,both moderate (MHR 1.29, 95% CI 1.03–1.61) andsevere depression (MHR 1.34, 95% CI 1.07–1.68) stillshowed a statistically significant association withexcess mortality. At the last step, entering functionalimpairment, these Risk Ratios just failed to reachstatistical significance, most likely due to over-correction of the effect of depression (Table 2). Thedifferences between each of the mild, moderate andsevere strata were not statistically significant, as the95% CI were not mutually exclusive. A severity effectwas confirmed in a similar analysis with the continu-ously scaled total depression item score with an MHR of1.02 (95% CI 1.01–1.02) per scalepoint incrementafter multivariate adjustment (data not shown).

Length of exposure

Of the 229 subjects with baseline depression whoparticipated at follow-up, 95 (41.5%) had remitted,and 134 (58.5%) were still depressed at follow-up.

Table 2. Depression severity and 10-year mortality

MHR Mild vs no depression

Unadjusted ratio 1.13 (0.91–1.40)Stepwise adjustment for:Demographics 1.24 (1.00–1.55)þ Cardiovascular disease 1.23 (0.98–1.53)þ Other diseases 1.15 (0.92–1.43)þ MMSE 1.13 (0.90–1.40)þ Functional (ADL) impairment 1.09 (0.88–1.37)

Deceased by 01–01–2001.Mortality Hazard Ratio with 95% Confidence Intervals with stepw‘moderate’/‘severe’ depression with ‘no depression’, n¼ 3746.

Copyright # 2008 John Wiley & Sons, Ltd.

Incident depression was present in 302 subjects(14.4% of subjects that participated in both wavesof the study) (Schoevers et al., 2000a, 2003a).

The unadjusted mortality risks of remitted andincident depression were not statistically significant.Only chronic depression showed a significant associ-ation with excess mortality (MHR 1.65, 95% C.I.1.25–2.17) (Figure 2). This association remainedstatistically significant in stepwise multivariateanalysis (MHR 1.41, 95% CI 1.05–1.89), also at thelast step, entering functional limitations (MHR 1.38,95% CI 1.03–1.85)(Table 3).

Moderate vs no depression Severe vs no depression

1.46 (1.18–1.81) 1.35 (1.08–1.68)

1.52 (1.22–1.89) 1.65 (1.32–2.06)1.48 (1.19–1.84) 1.60 (1.28–2.00)1.36 (1.09–1.69) 1.41 (1.12–1.77)1.29 (1.03–1.61) 1.34 (1.07–1.68)1.24 (0.99–1.55) 1.24 (0.98–1.56)

ise correction for other explanatory variables, comparing ‘mild’/

Int J Geriatr Psychiatry 2009; 24: 169–176.

DOI: 10.1002/gps

Figure 2. Kaplan-Meyer survival curve according to length ofexposure.

174 r. a. schoevers ET AL.

DISCUSSION

This study systematically examined whether severityand duration of exposure to depression are associatedwith a higher mortality risk. The findings may haveimplications for both prevention and intervention inlate life depression, and provide cues for furtherresearch on the mechanisms linking depression andexcess mortality.

Studying survival according to different levels ofdepression severity is based on the hypothesis that ahigher ‘dosage’ of depression would result in a highermortality risk for the individual patient. This would bein line with cross-sectional studies that have demon-strated the detrimental effects of depression with

Table 3. Length of exposure from T0–T1 with mortality 1994–2000

MHR Remitted vs no depression

Unadjusted ratio 0.97 (0.66–1.42)Stepwise adjustment for:Demographics 1.14 (0.77–1.67)þ Cardiovascular disease 1.12 (0.76–1.64)þ Other diseases 1.06 (0.72–1.56)þ MMSE 1.05 (0.71–1.54)þ Functional (ADL) impairment 0.98 (0.66–1.45)

Deceased between follow-up 1994 and 01–01–2001.Mortality Hazard Ratio with stepwise adjustment for other variables,

Copyright # 2008 John Wiley & Sons, Ltd.

regard to physical health and various aspects offunctioning to be directly related to the number ofdepressive symptoms (Hays et al., 1995; Ormel et al.,1998). The current study confirms this hypothesis, asmoderate to severe levels of depression wereassociated with a higher mortality risk.

Similarly, the dosage of depression can bequantified as the length of exposure to depression.Subjects with chronic depression had a 41% highermortality risk in 6-year follow-up than subjects whowere not depressed at either assessment. Subjects witheither incident or remitted depression did not have ahigher mortality risk. Length of exposure thus plays avital role with respect to longevity. This once moreunderlines the detrimental effects of chronic depres-sion, that has already been shown to have a profoundnegative influence for late-life physical and socialadjustment (Vaillant et al., 1996), health service use(Beekman et al., 1997; Unutzer et al., 1997) and well-being (Hays et al., 1995; Ormel et al., 1998). Withchronicity ranging around 50% (Cole et al., 1999), andeven higher at more severe levels of depression(Beekman et al., 2002), the clinical importance ofthese findings is considerable.

The current study is a naturalistic study in thecommunity with a 10-year follow-up of mortality data.One limitation is, that a number of possible influencesand changes affecting the prognosis of the almost4,000 subjects that participated in the study were notaccounted for in these analyses. Depression is adisorder with a variety of course types. Subjects thatwere depressed at baseline may have recovered,remitted, and/or subsequently relapsed, or may havebecome chronically depressed. Likewise, subjectswithout depression at baseline may have experienceddepressive episodes during the study period, whichmay also have influenced their survival. Still, studieswith a limited number of assessments tend to overrepresent chronic depression. When comparing course

Incident vs no depression Chronic vs no depression

1.07 (0.86–1.34) 1.65 (1.25–2.17)

1.11 (0.89–1.38) 1.81 (1.36–2.39)1.06 (0.85–1.32) 1.76 (1.33–2.33)1.05 (0.84–1.31) 1.54 (1.15–2.07)1.04 (0.83–1.30) 1.41 (1.05–1.89)1.03 (0.83–1.28) 1.38 (1.03–1.85)

organic cases at follow-up excluded, n¼ 1989.

Int J Geriatr Psychiatry 2009; 24: 169–176.

DOI: 10.1002/gps

depression and excess mortality 175

types, misclassification may therefore, if anything,have led to an underestimation of the strength of theassociation between length of exposure and mortality.

A second limitation is that subjects may have hadtreatment for their depression, which may have affectedprognosis. Unfortunately the impact of treatment isprobably very limited. Studies in comparable samples ofcommunity living elderly across different countrieshave consistently shown that case-finding is poor, andadequate treatment of depression even more so(German et al., 1987; Sonnenberg et al., 2003). Ourstudy is no exception to this. Although the effects oftreatment were probably very limited, this may againhave lead to an underestimation of the strength of theassociation between depression and mortality.

A strength of the current study is that it was able toexamine both severity and duration with regard to theirimplications for survival, using a design that allowedto control for a large number of possibly confoundingvariables. Secondly, the rigorous follow-up of vitalstatistics provides evidence that the impact ofdepression on survival is still measurable over aperiod of 10 years.

In conclusion, the current study shows that severalcharacteristics of depression are of importance whenassessing the risk of mortality. Severity and chronicityof depression are directly related to an increasedmortality risk, findings that are supportive of a causalassociation. Apart from further examining possiblebehavioral pathways explaining the depression-mortality association, more active casefinding andtreatment of persons at risk may yield importantbenefits from a clinical perspective.

CONFLICT OF INTEREST

None known.

ACKNOWLEDGEMENTS

The Amsterdam Study of the Elderly (AMSTEL) wassupported by grants from the Netherlands HealthResearch Programme (SGO) and the NetherlandsMental Health Fund (NFGV). The AGECAT compu-ter program was installed with the kind co-operationand final authorization of Michael E. Dewey, Depart-ment of Psychiatry, University of Liverpool.

REFERENCES

Ames D, Tuckwell V. 1994. Psychiatric disorders among elderlypatients in a general hospital. Med J Aust 160(11): 671–675.

Copyright # 2008 John Wiley & Sons, Ltd.

Aneshensel CS, Huba GJ. 1983. Depression, alcohol use, andsmoking over one year: a four-wave longitudinal causal model.J Abnorm Psychol 92(2): 134–150.

Beekman AT, Deeg DJ, Braam AW, et al. 1997. Consequences ofmajor and minor depression in later life: a study of disability,well-being and service utilization. Psychol Med 27(6): 1397–1409.

Beekman AT, Geerlings SW, Deeg DJ, et al. 2002. The naturalhistory of late-life depression: a 6-year prospective study in thecommunity. Arch Gen Psychiatry 59(7): 605–611.

Cole MG, Bellavance F, Mansour A. 1999. Prognosis of depressionin elderly community and primary care populations: a systematicreview and meta-analysis. Am J Psychiatry 156(8): 1182–1189.

Copeland JR, Davidson IA, Dewey ME, et al. 1992. Alzheimer’sdisease, other dementias, depression and pseudodementia: preva-lence, incidence and three-year outcome in Liverpool. Br JPsychiatry 161: 230–239.

Copeland JR, Dewey ME, Griffiths-Jones HM. 1986.A computerized psychiatric diagnostic system and case nomen-clature for elderly subjects: GMS and AGECAT. Psychol Med16(1): 89–99.

Copeland JR, Dewey ME, Griffith-Jones HM. 1990. Dementia anddepression in elderly persons: AGECAT compred with DSM IIIand pervasive illness. Int J Geriatr Psychiatry 5: 47–51.

Copeland JR, Dewey ME, Henderson AS, et al. 1988. The GeriatricMental State (GMS) used in the community: replication studies ofthe computerized diagnosis AGECAT. Psychol Med 18(1): 219–223.

Cuijpers P, Schoevers RA. 2004. Increased mortality in depressivedisorders: a review. Curr Psychiatry Rep 6(6): 430–437.

Cuijpers P, Smit F. 2002. Excess mortality in depression: a meta-analysis of community studies. J. Affect Disord 72(3): 227–236.

Dimatteo MR, Lepper HS, Croghan TW. 2000. Depression is a riskfactor for noncompliance with medical treatment: meta-analysisof the effects of anxiety and depression on patient adherence.ArchIntern Med 160(14): 2101–2107.

Folstein MF, Folstein SE, McHugh PR. 1975. ‘Mini-Mental State’.A practical method for grading the cognitive state of patients forthe clinician. J Psychiatr Res 12(3): 189–198.

Ganguli M, Dodge HH, Mulsant BH. 2002. Rates and predictors ofmortality in an aging, rural, community-based cohort: the role ofdepression. Arch Gen Psychiatry 59(11): 1046–1052.

Geerlings SW, Beekman AT, Deeg DJ, et al. 2002. Duration andseverity of depression predict mortality in older adults in thecommunity. Psychol Med 32(4): 609–618.

German PS, Shapiro S, Skinner EA, et al. 1987. Detection andmanagement of mental health problems of older patients byprimary care providers. JAMA 257(4): 489–493.

Goodwin FK, Jamison KR. 1990. Manic-Depressive Illness. OxfordUniversity Press: New York.

Hays RD, Wells KB, Sherbourne CD, et al. 1995. Functioning andwell-being outcomes of patients with depression compared withchronic general medical illnesses.ArchGen Psychiatry 52(1): 11–19.

Holsboer F. 2000. The corticosteroid receptor hypothesis of depres-sion. Neuropsychopharmacol 23(5): 477–501.

Hooijer C, Jonker C, Dewey ME. 1991. A standardized interview forthe elderly (GMS): reliability studies comparing the Dutchlanguage version with the original. Int J Geriatr Psychiatry 6:71–79.

Katon WJ. 2003. Clinical and health services relationships betweenmajor depression, depressive symptoms, and general medicalillness. Biol Psychiatry 54(3): 216–226.

Int J Geriatr Psychiatry 2009; 24: 169–176.

DOI: 10.1002/gps

176 r. a. schoevers ET AL.

Katz S, Ford AB, Moskowitz RW, et al. 1963. Studies of illness inthe aged. The index of ADL: a standardized measure of biologicaland psychosocial functioning. J Am Med Assoc 185: 914–919.

Launer LJ, Dinkgreve MA, Jonker C, et al. 1993. Are age andeducation independent correlates of the mini-mental state examperformance of community-dwelling elderly? J Gerontol 48(6):271–277.

Launer LJ, Wind AW, Deeg DJ. 1994. Nonresponse pattern and biasin a community-based cross-sectional study of cognitive func-tioning among the elderly. Am J Epidemiol 139(8): 803–812.

Lawton MP, Brody EM. 1969. Assessment of older people: self-maintaining and Instrumental Activities of Daily Living. Ger-ontologist 9(3): 179–186.

Lindesay J. 1989. Nonsuicidal mortality in late-life depression.J Geriatr Psychiatry 22(1): 53–65.

Maris RW. 2002. Suicide. Lancet 360(9329): 319–326.Murphy JM, Monson RR, Olivier DC, et al. 1987. Affective

disorders and mortality. A. general population study. Arch GenPsychiatry 44(5): 473–480.

Murphy E, Smith R, Lindesay J, Slattery J. 1988. Increasedmortality rates in late-life depression. Br J Psychiatry 152:347–353.

Murray CJ, Lopez AD. 1997. Alternative projections of mortalityand disability by cause 1990-2020: Global Burden Of DiseaseStudy. Lancet 349(9064): 1498–1504.

Musselman DL, Nemeroff CB. 1996. Depression and endocrinedisorders: focus on the thyroid and adrenal system. Br J Psy-chiatry 30 (Suppl): 123–128.

Ormel J, Kempen GI, Deeg DJ, et al. 1998. Functioning, well-being,and health perception in late middle-aged and older people:comparing the effects of depressive symptoms and chronicmedical conditions. J Am Geriatr Soc 46(1): 39–48.

Penninx BW, Beekman AT, Honig A, et al. 2001. Depression andcardiac mortality: results from a community-based longitudinalstudy. Arch Gen Psychiatry 58(3): 221–227.

Penninx BW, Geerlings SW, Deeg DJ, et al. 1999. Minor and majordepression and the risk of death in older persons. Arch GenPsychiatry 56(10): 889–895.

Penninx BW, Guralnik JM, Ferrucci L, et al. 1998. Depressivesymptoms and physical decline in community-dwelling olderpersons. JAMA 279(21): 1720–1726.

Penninx BW, Kritchevsky SB, Yaffe K, et al. 2003. Inflammatorymarkers and depressed mood in older persons: results from theHealth, Aging and Body Composition Study. Biol Psychiatry54(5): 566–572.

Roth M, Tym E, Mountjoy CQ, et al. 1986. CAMDEX. A. stan-dardised instrument for the diagnosis of mental disorder in the

Copyright # 2008 John Wiley & Sons, Ltd.

elderly with special reference to the early detection of dementia.Br J Psychiatry 149: 698–709.

Schoevers RA, Beekman AT, Deeg DJ, et al. 2000a. Risk factors fordepression in later life: results of a prospective community basedstudy (AMSTEL). J Affect Disord 59(2): 127–137.

Schoevers RA, Beekman AT, Deeg DJ, et al. 2003a. The naturalhistory of late-life depression: results from the Amsterdam Studyof the Elderly (AMSTEL). J Affect Disord 76(1–3): 5–14.

Schoevers RA, Beekman AT, Deeg DJ, et al. 2003b. Comorbidityand risk-patterns of depression, generalised anxiety disorder andmixed anxiety-depression in later life: results from the AMSTELStudy. Int J Geriatr Psychiatry 18(11): 994–1001.

Schoevers RA, Geerlings MI, Beekman AT, et al. 2000b. Associ-ation of depression and gender with mortality in old age. resultsfrom the Amsterdam Study of the Elderly (AMSTEL). Br JPsychiatry 177: 336–342.

Schoevers RA, Smit F, Deeg DJ, et al. 2006. Prevention of late-lifedepression in primary care: do we know where to begin? Am JPsychiatry 163(9): 1611–1621.

Schulz R, Beach SR, Ives DG, et al. 2000. Association betweendepression and mortality in older adults: The CardiovascularHealth Study. Arch Intern Med 160(12): 1761–1768.

Sonnenberg CM, Beekman AT, Deeg DJ, Van Tilburg W. 2003. Drugtreatment in depressed elderly in the Dutch community. Int JGeriatr Psychiatry 18(2): 99–104.

Stephens T. 1988. Physical activity and mental health in the UnitedStates and Canada: evidence from four population surveys. PrevMed 17(1): 35–47.

Unutzer J, Patrick DL, Simon G, et al. 1997. Depressive symptomsand the cost of health services in HMO patients aged 65 years andolder. A. 4-year prospective study. JAMA 277(20): 1618–1623.

Vaillant GE, Orav J, Meyer SE, et al. 1996. 1995 IPA/BayerResearch Awards In Psychogeriatrics. Late-life consequencesof affective spectrum disorder. Int Psychogeriatr 8(1): 13–32.

Van Der Kooy KG, Van Hout HP, Van Marwijk HW, et al. 2006.Differences in heart rate variability between depressed and non-depressed elderly. Int J Geriatr Psychiatry 21(2): 147–150.

Van Der Kooy K, Van Hout H, Marwijk H, et al. 2007. Depressionand the risk for cardiovascular diseases: systematic review andmeta analysis. Int J Geriatr Psychiatry.

Weed DL. 1997. On the use of causal criteria. Int J Epidemiol 26(6):1137–1141.

Wulsin LR, Vaillant GE, Wells VE. 1999. A systematic review of themortality of depression. Psychosom Med 61(1): 6–17.

Yeragani VK, Rao KA, Smitha MR, et al. 2002. Diminished chaos ofheart rate time series in patients with major depression. BiolPsychiatry 51(9): 733–744.

Int J Geriatr Psychiatry 2009; 24: 169–176.

DOI: 10.1002/gps