Somatic and psychiatric comorbidity in the general elderly population: Results from the ZARADEMP Project

earch 65 (2008) 347–355

Somatic and psychiatric comorbidity in the general elderly population:Results from the ZARADEMP Project

Antonio Lobo-Escolara,b,c, Pedro Saza,c,d, Guillermo Marcosc,d,e,f, Miguel Ángel Quintanillac,f,Antonio Campayoc,d,f, Antonio Loboa,c,d,f,⁎

the ZARADEMP Workgroup1

aDepartment of Psychiatry, Universidad de Zaragoza, Zaragoza, SpainbTraumatology Service, Hospital Universitario Miguel Servet, Zaragoza, Spain

cInstituto Aragonés de Ciencias de la Salud (I+CS), Zaragoza, SpaindCentro de Investigación Biomédica en Red de Salud Mental (CIBERSAM), Ministry of Health, Madrid, Spain

eDepartment of Preventive Medicine, Universidad de Zaragoza, Zaragoza, SpainfHospital Clínico Universitario, Zaragoza, Spain

Received 1 June 2007; received in revised form 18 February 2008; accepted 10 March 2008

Abstract

Objective: In a representative sample of the elderly popu-lation in a southern European city, we tested the hypothesis thatthere is an association between general somatic and generalpsychiatric morbidity. Methods: A stratified random sample of4803 individuals aged ≥55 years was selected for the baselinestudy in the ZARADEMP Project. The elderly were assessedwith standardized Spanish versions of instruments, including theGeriatric Mental State (GMS)–AGECAT. Psychiatric cases werediagnosed according to GMS–AGECAT criteria, and somaticmorbidity was documented with the EURODEM Risk FactorsQuestionnaire. Results: General comorbidity clustered in 19.9%of the elderly when hypertension was removed from the somaticconditions category, with 33.5% of the sample remaining freefrom both somatic and psychiatric illnesses. General comorbid-ity was associated with age, female gender, and limitededucation, but did not increase systematically with age. Thefrequency of psychiatric illness was higher among the somatic

⁎ Corresponding author. Servicio de Psiquiatría, Hospital ClínicoUniversitario, Planta 3, Avda. San Juan Bosco, 15, 50009 Zaragoza,Spain. Tel.: +34 976 55 11 67; fax: +34 976 76 17 12.

E-mail address: [email protected] (A. Lobo).1 The following members of the ZARADEMP Workgroup are also

authors of this article: C. de la Cámara, J.L. Día, A. Martín, J.A. Montañés,B. Quetglas, T. Ventura, and M. Zapata.

0022-3999/08/$ – see front matter © 2008 Elsevier Inc. All rights reserved.doi:10.1016/j.jpsychores.2008.03.002

cases than among noncases, and the frequency of somaticmorbidity among the psychiatric cases was higher than amongnoncases. This association between somatic and psychiatricmorbidity remained statistically significant after controlling forage, gender, and education [odds ratio (OR)=1.61; confidenceinterval (CI)=1.38–1.88]. Most somatic categories were asso-ciated with psychiatric illness, but after adjusting for demo-graphic variables and individual somatic illnesses, theassociation remained statistically significant only for cerebro-vascular accidents (CVAs) (OR=1.47; CI=1.09–1.98) andthyroid disease (OR=1.67; CI=1.10–2.54). Conclusion: This isthe first study to document that there is a positive andstatistically significant association between general somaticmorbidity and general psychiatric morbidity in the (predomi-nantly) elderly population. CVAs and thyroid disease may havemore weight in this association.© 2008 Elsevier Inc. All rights reserved.

Keywords: Comorbidity (somatic and psychiatric); Community survey; Elderly; Prevalence; ZARADEMP Project

Introduction

Pioneer studies by authors such as Eastwood andTrevelyan [1] have found that psychiatric and somaticillnesses tend to “cluster” in a limited group of individuals inthe general population. The first author speculated aboutvulnerability to illness, and research in this area was

348 A. Lobo-Escolar et al. / Journal of Psychosomatic Research 65 (2008) 347–355

considered “the main task for epidemiology in the field ofpsychosomatic medicine” [2]. Since then, a considerablenumber of studies have shown associations between somaticand psychiatric morbidity, both in studies of psychiatricpatients identified from inpatient registers [3] and in studiesof patients recognized in different medical settings [4,5].Furthermore, the negative consequences of coexistingpsychiatric morbidity and somatic illness have beendocumented in a number of reports [6,7]. While someauthors argue that the association between somatic andpsychiatric morbidity is now well established, they alsounderline the fact that previous research has been conductedprimarily in clinical samples and/or in a restrictive range ofphysical or mental conditions [8]. Important populationcomorbidity studies have been recently conducted by Ormelet al. [9] and Baune et al. [10], but the former focused onheart conditions and the latter focused on depression.

The relevance of studying general comorbidity has beenrecently shown in the World Mental Health Surveysreported by Scott et al. [8]. Conjoint psychiatric conditionswere more strongly associated with several chronic physicalconditions than were single mental disorders. Nonetheless,this important survey, similarly to the study by Eastwoodand Trevelyan [1], was carried out in predominantlyindividuals. Therefore, the statement of Eastwood [11]suggesting that the association between general psychiatricmorbidity and general somatic morbidity has not beenconvincingly shown in the elderly population is still valid.Given the relationships between comorbidity and frailtydescribed in the elderly, as well as the negative con-sequences [12], studies in the older population shouldbecome a research priority. Minor psychiatric morbidity,which frequently goes undetected and has been shown tohave a negative outcome [13–15], must be included in newinquiries, and the influence of a specific age group in theassociation should also be explored. Moreover, in inquiriesabout general medical comorbidity, the relative weight ofspecific medical conditions should be studied, sincedifferent patterns of association have been reportedaccording to physical disease [10].

Finally, inquiries in new sociocultural settings may giveclues to the influence of environmental factors, includingfactors related to service provision [16]. Roca-Benasar et al.[17] failed to confirm that, in a Spanish island, the fre-quency of general medical comorbidity in subjects diag-nosed with International Classification of Diseases, TenthRevision (ICD-10) mental disorders was significantlyhigher than that in respondents without psychiatric diag-nosis. Furthermore, Braam et al. [18], in a cross-nationalEuropean study, found a consistent association of physicalillness with depression, but differences between countrieswere also suggested.

The present study is part of the ZARADEMP Project, anepidemiological inquiry that aims to document in the elderlycommunity the prevalence, incidence, and risk factors ofdementia, depression, and psychiatric morbidity, as well as

their association with somatic morbidity [19]. The mainobjective in this study was to try to confirm in a southernEuropean elderly population the tendency of generalpsychiatric morbidity and general somatic morbidity(specifically conditions considered to be risk factors fordementia) to cluster in some individuals, and to findsupport for the general hypothesis that there is a positiveassociation between them. If this was confirmed, otherobjectives were: (a) to document the influence of demo-graphic characteristics in the association, and (b) to explorewhich specific somatic conditions have more weight inthe association.

Methods

Background, design overview, sampling, and instruments

The site of the study was Zaragoza, a capital concentrat-ing 622,371 inhabitants (fifth city in the country) or 51% ofthe population of the historical kingdom of Aragón. Theobjectives and general methodology of the ZARADEMPProject have been described in more detail in a previousarticle [19]. It is a longitudinal epidemiological study withthree waves, and Wave I (ZARADEMP I) is relevant for thisreport. It is the baseline cross-sectional study that intended todocument the prevalence and distribution of dementia,depression, and psychiatric morbidity, as well as somaticmorbidity and associated variables. Participating individualswill be followed up in Waves II and III (or ZARADEMP IIand III) to study the influence of hypothesized risk factors inincident cases.

Sampling size was decided as a function of Type I andType II errors. To ensure an adequate sample of the elderly ina 10-year follow-up study, a large stratified random sampleof individuals (aged ≥55 years) from the census lists wasneeded, since one of the main objectives of the ZAR-ADEMP Project was to obtain reliable measures of thecause–effect association in the case–control study inincident cases. The response rate in Wave I, after theremoval of expected losses (those who moved away andthe dead), was 63.6%. The refusal rate was 20.5%. Finally,4803 people were interviewed in ZARADEMP I [19].Table 1 summarizes relevant demographic characteristics.Residents of elderly people's homes and other institutionswere included in the community sample.

Several international instruments previously standardizedin Spain by the same research group have been used and areincorporated into the ZARADEMP Interview. They includethe Mini-Mental Status Examination (MMSE) [20,21]; forthe purpose of this report, the following instruments willbe described:

Geriatric Mental State (GMS): a semistructured standar-dized clinical interview used for assessing the mental stateof elderly people [22]. A computerized diagnostic

Table 1Demographic characteristics of the full sample (n=4803) and the samplewith full somatic morbidity data (n=4227)

Full sampleSample with somaticdata

n % (CI) n % (CI)

GenderWomen 2771 57.7 (56.3–59.1) 2466 58.3 (56.8–59.8)Marital statusMarried/common law 2796 58.2 (56.8–59.6) 2477 58.6 (57.1–60.1)Widowed 1494 31.1 (29.8–32.4) 1308 30.9 (29.5–32.3)Educational levelIlliterate/some primary 2250 46.8 (45.4–48.2) 1981 46.9 (45.4–48.4)Primary 1778 37.0 (35.6–38.4) 1567 37.1 (35.6–38.6)Age (years)55–64 1088 22.6 (21.4–23.8) 944 22.3 (21.1–23.6)65–74 1702 35.4 (34.0–36.8) 1521 36.0 (34.6–37.5)75–84 1097 22.8 (21.6–24.0) 981 23.2 (21.9–24.5)≥85 916 19.1 (18.0–20.2) 781 18.5 (17.3–19.7)

349A. Lobo-Escolar et al. / Journal of Psychosomatic Research 65 (2008) 347–355

program, AGECAT, can be applied to it [23]. The GMS-B, a shortened community version that may be used bylay interviewers and in international comparisons, wasselected [24]. This interview is also a syndrome case-finding instrument, with the GMS-B threshold scoresdiscriminating between “noncases,” “subcases,” and“cases.” The elderly people receive a probable diagnosisof “dementia,” “depression,” and “anxiety,” which isrecorded by the examiner in the final step of theGMS administration.History and Etiology Schedule: a standardized method ofcollecting history and etiology data from an informant, ordirectly from the respondent when he or she is judged tobe reliable [25]. It concentrates on those features that areexpected to be relevant to psychiatric diagnosis in olderpeople, and is crucial to completing the GMS and tofacilitating a diagnostic process such as the one performedwith the Diagnostic and Statistical Manual of MentalDisorders (DSM) system in these studies.

The ZARADEMP Interview incorporates standardizedSpanish versions of disability scales [19]. It alsoincorporates the Risk Factors Questionnaire designed bythe EURODEM Workgroup [26]. This instrument includesinformation related to medical diseases that are consideredto be potential risk factors for dementia, and may be usedby trained lay interviewers. For the purpose of this report,the history of the following medical diseases is relevant:hypertension, angina, myocardial infarct, cerebrovascularaccident (CVA), epilepsy, head trauma, Parkinson'sdisease (PD), diabetes mellitus, and thyroid disease.Each item in the interview has been operationally definedaccording to previously agreed EURODEM criteria. TheZARADEMP Interview also contains a section onphysical examination, which includes the checking ofblood pressure following World Health Organizationstandards [27].

Procedure

Modern epidemiological procedures in field studies werevigorously completed to ensure the participation of theelderly [19]. An epidemiological screening design was usedin Wave I. Lay interviewers, who were senior medicalstudents and had been previously standardized in themethods, administered the ZARADEMP Interview, with allthe incorporated instruments. Caregivers were interviewedwhen the selected elderly person was considered to beunreliable. Information on medication use was collected, andmedical reports, which are commonly kept at home by theSpanish elderly, were used to verify medications and to helpin the diagnostic process.

The individuals were nominated as “probable cases” (or“subcases”) on the basis of GMS threshold “global” scoresand/or MMSE standard cutoff points, which we havepreviously reported to have good validity coefficients [21].However, the data on each elderly person were thoroughlyreviewed by the research psychiatrists individually super-vising the lay interviewers. The cases considered to be“doubtful” (n=159) according to predetermined criteria werereassessed by the supervising psychiatrists in the elderlyperson's home. This reassessment was completed on thefollowing day when possible, and there were no dropouts. Inthe final step of Wave I, the psychiatrists recorded thepsychiatric diagnosis of the identified cases. For the purposeof this report, AGECAT diagnosis was used. Our previousstudies have supported the validity of this diagnostic processperformed by research psychiatrists in the elderly commu-nity [21].

The following operational definitions were used inthis study:

Somatic morbidity: the presence in a given individual ofone or more somatic diseases as previously defined.Psychiatric morbidity: The presence in a given individualof symptoms fulfilling the criteria of AGECAT “case” or“subcase” in the following categories: “dementia,”“depression” (both “severe” or “psychotic” in the originalterm; and “nonsevere” or “neurotic” in the original term),and “anxiety.”Somatic and psychiatric comorbidity: the coexistence inthe same individual of one or more somatic diseases andone or more psychiatric conditions, as defined above.

Quality control and ethics

Systematic control of the reliability of interviewers wasprogrammed to prevent “reliability decay.” Every 6 months,lay interviewers who failed to pass a reliability test in videointerviews and/or were considered by the research psychia-trists to have difficulties supervising the process wererestandardized. The same process was also used to ensurereliability between the psychiatrists, under the supervision ofthe principal investigators (A.L. and P.S.). Standard ethical

350 A. Lobo-Escolar et al. / Journal of Psychosomatic Research 65 (2008) 347–355

principles, as well as the compromise of information to theelderly, written consent, privacy, confidentiality, and secur-ity, have been maintained throughout the study, according toSpanish Law 5/1992.

Statistical analyses

The 95% confidence intervals (CIs) were calculated forthe frequency of sociodemographic characteristics and forthe prevalence of somatic and psychiatric conditions.

To study the specific hypothesis that there is anassociation between somatic and psychiatric morbidity,univariate logistic regressions were performed in the firststep. In a second step, based on both literature review and theresults of previous univariate analyses, a number of multi-variate analyses were performed to confirm the associationafter controlling for the main demographic variables. In athird step, the strength of the association of each somaticillness with psychiatric morbidity was calculated aftercontrolling for the demographic variables and the remainingsomatic illness (as covariates). For the sake of simplicity, themost explicative models in this sample are shown.

Results

Reliable information related to psychiatric illness wascollected in the full sample, and complete informationrelated to somatic illness was collected for 4227 individuals.Demographic characteristics in these patients did not differsignificantly from those in the full sample (Table 1). Somedata regarding the following somatic illnesses were incom-plete or unreliable: hypertension, 426 individuals; angorpectoris, 57 individuals; myocardial infarction, 79 indivi-duals; CVA, 145 individuals; epilepsy, 42 individuals; headtrauma, 479 individuals; PD, 252 individuals; diabetesmellitus, 226 individuals, thyroid disease, 66 individuals.

Most of the elderly had at least one medical condition(3430 individuals, 81.1%), but the frequency would be33.1% (1590 individuals) if hypertension, which is fre-quently asymptomatic, were excluded. As expected, hyper-tension was the most prevalent medical condition (61.7%;

Table 2Prevalence of somatic illness among community-dwelling individuals aged ≥55 y

Nonpsychiatric morbidity (n=2211)

Cases (n) Prevalence (%) 95

Hypertension 1357 61.4 59Angor pectoris 95 4.3 3Myocardial infarction 65 2.9 2CVA 112 5.1 4Epilepsy 15 0.7 0Head trauma 116 5.2 4PD 16 0.7 0Diabetes mellitus 169 7.6 6Thyroid disease 46 2.1 1

95% CI=60.3–63.1), but other medical categories, such asdiabetes mellitus (8.7%) and CVAs (6.9%), were relativelyfrequent. Most categories of somatic morbidity were morefrequent in the elderly with psychiatric morbidity, whencompared to the elderly without psychiatric illness, with thedifferences being significant in CVAs, PD, and thyroiddisease (Table 2). The distribution of somatic morbidity inmales was different from that in females, and the followingcategories were all significantly more frequent in males:angor pectoris [6.5% (CI=5.5–7.7) vs. 4.3% (CI=3.6–5.2)],myocardial infarction [4.8% (CI=3.9–5.8) vs. 1.5%(CI=1.1–2.0)], and head trauma [6.6% (CI=5.6–7.8) vs.3.9% (CI=3.2–4.7)]. On the contrary, both hypertension[65.1% (CI=63.3–66.9) vs. 57.0% (CI=54.8–59.2)] andthyroid disease [4.9% (CI=4.1–5.8) vs. 0.7% (CI=0.4–1.2)]were significantly more frequent among women.

Two thousand five hundred ninety-two individuals in thissample (54%; 95% CI=52.6–55.4) were considered to havepsychiatric morbidity according to the operational criteriaused, although the morbidity would decrease to 966 (20.1%;95% CI=19.0–21.2) if subcases were excluded. Depression(severe and nonsevere) was the most frequent categoryamong the cases (554 cases, 11.5%), and anxiety was themost frequent diagnosis among the subcases (1079 subcases,22.5%). Most categories of psychiatric morbidity were morefrequent in the elderly with somatic illness, when comparedto the elderly without somatic conditions, but the differenceswere not significant (Table 3). Similarly, no statisticallysignificant differences were observed in the frequency ofpsychiatric morbidity in the elderly eliminated from thisanalysis, when compared with either group, with the onlyexception being nonsevere depression, which was signifi-cantly higher in the somatic morbidity group (Table 3).

All psychiatric categories were significantly morefrequent in women when compared to men, with the onlyexception being anxiety subcases. The prevalence ofdementia was 4.2% (CI=3.4–5.2) in men and 8.6%(CI=7.6–9.7) in women. Corresponding figures were asfollows—dementia subcases: 4.5% (CI=3.6–5.5) in men and6.6% (CI=5.7–7.6) in women; depression (severe): 1.1%(CI=0.6–1.5) in men and 3.2% (CI=2.6–3.9) in women;depression (nonsevere): 4.3% (CI=3.5–5.3) in men and

ears (distribution by psychiatric morbidity)

Psychiatric morbidity (n=2592)

% CI Cases (n) Prevalence (%) 95% CI

.3–63.4 1606 62.0 60.1–63.9

.5–5.2 156 6.0 5.1–7.0

.2–3.7 74 2.9 2.3–3.6

.2–6.1 220 8.5 7.4–9.6

.4–1.1 15 0.6 0.3–1.0

.3–6.2 126 4.9 4.1–5.8

.4–1.1 44 1.7 1.2–2.3

.5–8.8 248 9.6 8.5–10.8

.5–2.8 104 4.0 3.3–4.8

Table 3Prevalence of psychiatric morbidity (AGECAT criteria) among community-dwelling individuals aged ≥55 years (distribution by somatic morbidity)

Somatic morbidity (n=3430) No somatic morbidity (n=797) Incomplete somatic information (n=576)

Cases (n) Prevalence (%) 95% CI Cases (n) Prevalence (%) 95% CI Cases (n) Prevalence (%) 95% CI

Dementia 227 6.6 5.8–7.5 42 5.3 3.9–7.1 54 9.4 7.1–12.1Dementia subcase 198 5.8 5.0–6.6 41 5.1 3.7–6.9 37 6.4 4.5–8.7Depression (severe) 79 2.3 1.8–2.9 16 2.0 1.2–3.2 17 3.0 1.8–4.7Depression (nonsevere) 342 10.0 9.0–11.1 66 8.3 6.5–10.4 34 5.9 4.1–8.2Depression subcase 188 5.5 4.8–6.3 39 4.9 3.5–6.6 44 7.6 5.6–10.1Anxiety 61 1.8 1.4–2.3 15 1.9 1.1–3.1 13 2.3 1.2–3.9Anxiety subcases 788 23.0 21.6–24.5 182 22.8 19.9–25.9 109 18.9 15.8–22.3Noncases 1547 45.1 43.4–46.8 396 49.7 46.2–53.2 268 46.5 42.4–50.7

351A. Lobo-Escolar et al. / Journal of Psychosomatic Research 65 (2008) 347–355

12.8% (CI=11.6–14.1) in women; depression subcases:3.8% (CI=3.0–4.7) in men and 7.0% (CI=6.1–8.0) inwomen; anxiety: 1.1% (CI=0.6–1.5) in men and 2.4%(CI=1.9–3.0) in women; anxiety subcases: 20.7% (CI=19.0–22.5) in men and 23.8% (CI=22.2–25.4) in women.

As expected, the prevalence of somatic disease, inparticular PD, tends to increase with age in most categories(Table 4). However, it decreases after the age of 84 years inthe following categories: angina, myocardial infarction,epilepsy, diabetes, thyroid disease, and hypertension, withthe differences being significant in this last category. Theprevalence of psychiatric morbidity is also influenced byage (Table 4). Dementia, and also subcases of dementia,increase steadily with age. Nonsevere depression tends toincrease with age, but the prevalence of severe depressionremains stable, and the frequency of anxiety categoriestends to decrease.

Among the somatic cases, 1883 individuals (54.9%) hadpsychiatric morbidity, with the proportion being higher thanamong noncases (401 individuals, 50.3%). Similarly, among

Table 4Prevalence of somatic illness and prevalence of psychiatric illness (distribution by

55–64 years (n=1088) 65–74 years (n=1702)

Cases(n)

Prevalence(%) 95% CI

Cases(n)

Prevalence(%) 9

Hypertension 584 53.7 50.7–56.7 1081 63.5 6Angor pectoris 37 3.4 2.4–4.7 83 4.9Myocardial infarction 27 2.5 1.7–3.6 40 2.4CVA 34 3.1 2.1–4.3 89 5.2Epilepsy 6 0.6 0.2–1.3 15 0.9Head trauma 59 5.4 4.1–6.9 86 5.1PD 1 0.1 0.03–0.5 20 1.2Diabetes mellitus 63 5.8 4.5–7.4 177 10.4Thyroid disease 29 2.7 1.8–3.8 66 3.9Dementia 9 0.8 0.4–1.5 44 2.6Dementia subcase 31 2.8 1.9–4.0 70 4.1Depression (severe) 25 2.3 1.5–3.4 40 2.4Depression (nonsevere) 84 7.7 6.2–9.4 155 9.1Depression subcase 51 4.7 3.5–6.1 93 5.5Anxiety 25 2.3 1.5–3.4 34 2.0Anxiety subcases 283 26.0 23.4–28.7 425 25.0 2Noncases 580 53.3 50.3–56.3 841 49.4 4

the psychiatric cases with full information available, theproportion of individuals with somatic morbidity (1883individuals, 82.4%) was higher than among noncases (1547individuals, 79.6%). If hypertension were removed from thecategory of somatic conditions, the association betweengeneral somatic morbidity and general psychiatric morbiditywould seem stronger. Among the somatic cases, 792individuals (60.4%) had psychiatric morbidity, with theproportion being higher than among the noncases (1333individuals, 50.1%). Similarly, among the psychiatric cases,792 individuals (37.3%) had somatic morbidity, with theproportion being higher than among the noncases (520individuals, 28.1%). In unadjusted analysis, psychiatricmorbidity is associated with somatic morbidity, either withor without hypertension. However, in the final adjustedmodel, which includes sociodemographic variables, theassociation remains statistically significant only whenhypertension is removed from the somatic conditionscategory (OR=1.61; CI=1.38–1.88) (Table 5). Most somaticcategories were associated with psychiatric illness, but after

age group)

75–84 years (n=1097) ≥85 years (n=916)

5% CICases(n)

Prevalence(%) 95% CI

Cases(n)

Prevalence(%) 95% CI

1.2–65.8 746 68.0 65.1–70.7 552 60.3 57.0–63.53.9–6.0 75 6.8 5.4–8.4 56 6.1 4.6–7.81.7–3.2 42 3.8 2.7–5.1 30 3.3 2.2–4.74.2–6.4 84 7.7 6.2–9.4 125 13.6 11.4–15.90.5–1.5 6 0.5 0.2–1.1 3 0.3 0.1–0.94.1–6.2 60 5.5 4.2–7.0 37 4.0 2.8–5.50.7–1.8 18 1.6 0.9–2.5 21 2.3 1.4–3.58.9–11.9 107 9.8 8.1–11.7 70 7.6 5.9–9.53.0–4.9 34 3.1 2.2–4.3 21 2.3 1.4–3.51.9–3.5 83 7.6 6.1–9.3 187 20.4 17.8–23.23.2–5.1 76 6.9 5.5–8.6 99 10.8 8.9–13.01.7–3.2 26 2.4 1.6–3.5 21 2.3 1.4–3.57.8–10.6 108 9.8 8.1–11.7 95 10.4 8.5–12.64.5–6.7 71 6.5 5.1–8.1 56 6.1 4.6–7.81.4–2.8 20 1.8 1.1–2.8 10 1.1 0.5–2.03.0–27.1 221 20.1 17.8–22.6 150 16.4 14.1–19.07.0–51.8 492 44.8 41.8–47.8 298 32.5 29.5–35.6

Table 5Unadjusted and adjusted logistic regression models for predictingpsychiatric morbidity

Unadjusted analysis Adjusted analysis

P OR (95% CI) P OR (95% CI)

Somatic morbidity(hypertensionincluded)

.019 1.20 (1.03–1.40) .080 0.85 (0.71–1.02)

Somatic morbidity(hypertensionnot included)

.000 1.52 (1.33–1.74) .000 1.61 (1.38–1.88)

Female gender .000 2.75 (2.44–3.09) .000 2.97 (2.59–3.41)Age .000 1.03 (1.02–1.03) .000 1.02 (1.01–1.03)Illiterate/some primaryeducation

.000 1.71 (1.52–1.92) .000 1.43 (1.24–1.64)

352 A. Lobo-Escolar et al. / Journal of Psychosomatic Research 65 (2008) 347–355

adjusting for demographic variables and individual somaticillnesses, the association remained statistically significantonly for CVAs (OR=1.47; CI=1.09–1.98) and thyroiddisease (OR=1.67; CI=1.10–2.54).

Table 6 describes the distribution of somatic andpsychiatric comorbidity in this population sample. Thefrequency of comorbidity was 44.5%, and only 9.4% of theelderly were free from both physical and psychiatricillnesses. However, comorbidity was significantly morefrequent among women than among men, and the category“no morbidity” was significantly more frequent among men.As expected, the distribution was also influenced by age.Comorbidity increases with age and is 57.1% at the age of≥85 years. “Only psychiatric” morbidity is relatively higherin the youngest age stratum, but also tends to increase in theolder age strata. On the contrary, “only somatic” morbiditydecreases in the oldest stratum, when compared to theyounger strata (Table 6). If the hypertension category wereremoved from the group of somatic conditions, substantialchanges would have been observed. Comorbidity clusteredin 19.9% of the sample and did not increase systematicallywith age: it was 12.5% in the age group 55–64 years,increased to 22.1% in the age group 75–84 years, but was20.9% in the oldest age group. At the other extreme, 33.5%

Table 6Frequency of somatic and psychiatric comorbidity among community-dwelling in

General(community-dwellingindividuals) (N=4803)

Male(n=2032)

Female(n=2771)

Comorbidity 44.5 (43.1–45.9) 31.7 (29.7–33.8) 53.7 (51.8–55.6)Only somaticmorbidity oronly psychiatricmorbidity

46.1 (44.7–47.5) 55.0 (52.8–57.2) 39.8 (38.0–41.6)

Only somaticmorbidity

36.6 (35.2–38.0) 48.0 (45.8–50.2) 28.5 (26.8–30.2)

Only psychiatricmorbidity

9.5 (8.7–10.4) 7.0 (5.9–8.2) 11.3 (10.1–12.5)

No morbidity 9.4 (8.6–10.3) 13.3 (11.8–14.8) 6.5 (5.7–7.6)

Distribution by gender and age groups.

of the sample was free from both somatic and psychiatricillnesses, with the proportion decreasing steadily with age, to20.9% in the oldest age group.

Discussion

To our knowledge, this is the first study to confirm in the(predominantly) elderly population the clustering of generalsomatic morbidity and general psychiatric morbidityreported in classical studies in nonelderly adult samples[1]. If hypertension, a condition compatible with a goodquality of life, were removed from the somatic conditions,this comorbidity would be found in 19.9% of the elderly,and, at the other extreme, 33.5% of them would be free fromeither somatic or psychiatric illness. While research into ageneral concept such as comorbidity is much more difficultthan research into specific illnesses, the ultimate publichealth utility may be greater and has been considered aresearch priority [2,8]. It has been suggested that psychiatriccomorbidity is a different and more severe and chronic formof psychological disorder than the morbidity of isolatedconditions [28], and that it might be the interaction of generalmedical and behavioral health disorders that leads to pooroutcomes [8]. Important studies on physical and psychiatriccomorbidity have recently been reported, but they wereconducted in predominantly nonelderly adult populations[8,10,17]. In addition, relevant comorbidity studies reportedin the elderly population have been conducted in a restrictiverange of physical [9] or psychiatric [18] conditions.

Assumptions about the prevalence of disorder in a newsociocultural setting, such as a southern European city, couldnot be made in advance. Differences in the frequency ofspecific disorder [29] or in the association between physicaland psychiatric illness [18] have been reported in elderlypopulations in European cities. Furthermore, the prevalenceof disorder, both physical and mental, may be influenced bythe availability and organization of medical services [16].Psychogeriatric services are very uncommon in Spain, andspecifically so in Zaragoza [30].

dividuals [prevalence (CI)]

55–64 years(n=1088)

65–64 years(n=1702)

75–84 years(n=1097)

≥85 years(n=916)

33.9 (31.1–36.8) 42.6 (40.2–45.0) 47.8 (44.8–50.8) 57.1 (53.8–60.3)50.4 (47.4–53.4) 47.9 (45.5–50.3) 45.7 (42.7–48.7) 37.8 (34.6–41.0)

37.9 (35.0–40.9) 39.0 (36.7–41.4) 38.0 (35.1–40.9) 28.6 (25.7–31.6)

12.5 (10.6–14.6) 8.9 (7.6–10.3) 7.7 (6.2–9.4) 9.2 (7.4–11.3)

15.7 (13.6–18.0) 9.5 (8.1–11.0) 6.5 (5.0–8.0) 5.1 (3.8–6.7)

353A. Lobo-Escolar et al. / Journal of Psychosomatic Research 65 (2008) 347–355

The findings about the substantial degree of comorbidityfound in this elderly population have public health implica-tions, but also clinical implications, in particular for liaisonpsychiatry. A number of studies show that a considerableproportion of patients with physical morbidity do not receivetreatment for their mental illness [7], but the somatic diseasein such cases may require special consideration [31].Furthermore, comorbidity has been considered to beclinically consequential: both somatic and psychiatricillnesses have poorer outcomes [7,8]. In contrast, there aresuggestions that novel specialized interventions in patientswith somatic and psychiatric comorbidity lead to signifi-cant outcome improvement [32]. We also suggest thatpsychiatric subcases, which were included in the morbiditycriteria, may have clinical implications in the elderly inview of the negative outcome documented in subcases ofboth “mild cognitive deficit” and depression or anxietydisorders [14,33].

Contrary to Roca-Benasar et al. [17], we have foundsupport for the hypothesis of a positive and statisticallysignificant association between general somatic and psy-chiatric morbidity, particularly when hypertension wasremoved from the somatic conditions category. In Zaragoza,when hypertension was removed from the group of medicalconditions, the proportion of the elderly with psychiatricmorbidity among the somatic cases (60.4%) was higher thanamong the noncases (50.1%). Similarly, among the psychia-tric cases, the proportion of individuals with somaticmorbidity (37.3%) was higher than among the noncases(28.1%). While differences in rates of somatic morbiditybetween psychiatric cases and noncases and vice versa arerelatively small, and comorbidity has been found to beinfluenced by age, gender, and education, the statisticallysignificant association remained after controlling for thesedemographic factors (OR=1.61; CI=1.38–1.88). This cross-sectional study does not inform on the direction of theassociation between physical illness and mental illness. Inline with previous interpretations of similar findings incommunity studies [11,31], our data may support the“susceptibility” to illness conjectures. However, longitudinalstudies should test causal hypotheses in this field, as wehave done to study the long-term effect of diabetes ondepression [34].

Comorbidity was significantly more frequent in women(OR=2.75; CI=2.44–3.09). This may be influenced by thehigher rate of psychiatric morbidity found in women in thisstudy. Nevertheless, a consistent finding in the literature isthat women report symptoms of both physical and mentalillnesses at higher rates than men [35]. Comorbidity was alsosignificantly associated with limited education (OR=1.71;CI=1.52–1.92). The relationship of both psychiatric [21] andsomatic [36] morbidity with a deficient educational back-ground has been shown previously, but we know no previousreports of this association with comorbidity. The findingshould be emphasized, since a substantial proportion of theelderly in this population (46.8%), and probably in other

southern European cities, did not complete primary educa-tion. The positive association of comorbidity with age foundin this study was weak (OR=1.03; CI=1.02–1.03). We do notknow of previous reports documenting this specific associa-tion, but stronger links might be expected. While depressionhas not been observed to increase with age in some largepopulation studies [29], dementia increases systematically[37], and a similar increase in the rate of somatic diseasemight be anticipated. In fact, it is remarkable that mostcategories of somatic morbidity here, and similarly comor-bidity (when hypertension is excluded from the somaticconditions), tend to decrease after the age of 84 years. Thisfinding requires replication, since it might be related toconjectures about vulnerability being concentrated in someindividuals, and resistance in others [2].

This study also explores the association between specificsomatic diseases and psychiatric morbidity, in view ofconsiderable variability reported in the literature [8]. Whileunadjusted logistic regression models support the associationof most somatic categories with psychiatric illness, afteradjusting for demographic variables and individual somaticconditions, the association remained statistically significantonly for thyroid disease (OR=1.67; CI=1.10–2.54) andCVAs (OR=1.47; CI=1.09–1.98). This suggests that bothconditions have more specific weight in the associationfound. We have previously shown strong linear correlationsbetween psychopathology and hormonal levels in thyroiddisease patients [38,39], a pertinent finding in Spain andspecifically in the Zaragoza area where hypothyroidism isendemic [40]. Similarly, the association with CVAs was notunexpected; it has been shown in several studies and hasimportant clinical implications [41].

We believe that this study, which was conducted in a largerepresentative community sample of the elderly, fulfillspresent-day requirements in psychiatric epidemiology [24].However, some limitations should also be addressed. Whilewe have previously argued in support of the response rate inWave I of the ZARADEMP Project [19], some analyses forthis article could be completed in only 4227 subjects, withthe losses being due to the lack of reliable data related tosome medical conditions. We believe that this has notseriously biased the main results in the study, since thefrequency of psychiatric morbidity in the individualseliminated from this analysis did not differ significantlyfrom that in the remaining individuals, with the onlyexception being nonsevere depression. Questions may alsobe raised about the selection of diseases for this inquiry,which was limited to medical diseases considered to bepotential risk factors for dementia. However, we consider itremarkable that hypotheses put forward by authors such asEastwood [11] for nonelderly adults might be sustained inthe elderly upon entry of only a limited amount of somaticillnesses into the comorbidity criteria. It has been shown, forexample, that conditions not considered here, such asarthritis or rheumatic illnesses, are commonly associatedwith psychiatric morbidity [42]. We suspect that the

354 A. Lobo-Escolar et al. / Journal of Psychosomatic Research 65 (2008) 347–355

inclusion of such conditions in the analysis would increasethe rate of comorbidity and would probably reinforce themain conclusions in this study.

A limitation might be the fact that adjustment for dailyfunction was not included in the analysis. However, thepurpose of this article was descriptive rather than explana-tory, and different studies will be needed to address thisissue, since disability may be part of the vulnerability tosomatic and psychiatric comorbidity.

It might be argued that dementia is considered by some tobe a medical–neurological condition, but for this particularstudy, we included it in the mental diseases group, followingthe ICD-10 classification. We are also aware that not allGMS–AGECAT cases of depression fulfill ICD-10 orDiagnostic and Statistical Manual of Mental Disorders,Fourth Edition psychiatric criteria [19]. However, AGECATdepression diagnoses have been validated against DSMcriteria for depression and have been shown to correspondwith what psychiatrists recognize as cases for intervention[43]. Finally, since the instruments and methods used herewere originally designed for the study of the main categoriesof psychiatric disturbance found in the elderly community,categories such as substance abuse were not thoroughlyassessed here. However, we have data showing that alcoholabuse amounts to 3.3% in this population [44] and wouldtherefore explain a small proportion of the psychiatricmorbidity in this sample. Furthermore, since alcohol abuse iscommonly associated with medical morbidity, we suspectthat its inclusion here would also reinforce the mainconclusions in this study.

Acknowledgments

This work was supported by grants 94-1562, 97-1321E,98-0103, 01-0255, 03-0815, and G03/128 from the Fondode Investigación Sanitaria and the Spanish Ministry ofHealth, Instituto de Salud Carlos III, Red de EnfermedadesMentales (REM-TAP Network; Madrid, Spain); CIBERSAMCB07/09/0016 from the Spanish Ministry of Health, Institutode Salud Carlos III; CICYT SAF93-0453 from the DirecciónGeneral de Investigación Científica y Técnica, SecretaríaGeneral de Universidades, Madrid, Spain; and FundaciónCAI (Zaragoza, Spain).

References

[1] Eastwood MR, Trevelyan MH. Relationship between physical andpsychiatric disorder. Psychol Med 1972;2:363–72.

[2] Eastwood R. Epidemiological studies in psychosomatic medicine. In:Lipowski ZJ, Lipsitt DR, Whybrow PC, editors. Psychosomaticmedicine. Current trends and clinical applications. New York: OxfordUniversity Press, 1977. pp. 411–20.

[3] Osby U, Brandt L, Correia N, Ekbom A, Sparén P. Excess mortality inbipolar and unipolar disorder in Sweden. Arch Gen Psychiatry 2001;58:844–50.

[4] Koenig HG, Vandermeer J, Chambers A, Burr-Crutchfield L, JohnsonJL. Minor depression and physical outcome trajectories in heart failureand pulmonary disease. J Nerv Ment Dis 2006;194:209–17.

[5] Lobo A, Campos R, Marcos G, García-Campayo J, Campayo A,Lopez-Antón R, Pérez-Echeverría MJ. Somatic and psychiatric co-morbidity in primary care patients in Spain. Eur J Psychiatr 2007;21:71–8.

[6] Kessler RC, Ormel J, Demler O, Stang PE. Comorbid mental disordersaccount for the role impairment of commonly occurring chronicphysical disorders: results from the National Comorbidity Survey.J Occup Environ Med 2003;45:1257–66.

[7] Kathol R, Saravay S, Lobo A, Ormel J. Epidemiologic trends and costsof fragmentation. Med Clin North Am 2006;90:549–72.

[8] Scott KM, Bruffaerts R, Tsang A, Ormel J, Alonso J, et al. Depression–anxiety relationships with chronic physical conditions: results from theWorld Mental Health Surveys. J Affect Disord 2007;103:113–20.

[9] Ormel J, Von Korff M, Burger H, Scott K, Demyttenaere K, HuangYQ, et al. Mental disorders among persons with heart disease—resultsfrom World Mental Health Surveys. Gen Hosp Psychiatry 2007;29:325–34.

[10] Baune BT, Adrian I, Arolt V, Berger K. Associations between majordepression, bipolar disorders, dysthymia and cardiovascular diseases inthe general adult population. Psychother Psychosom 2006;75:319–26.

[11] Eastwood R. Relationship between physical and psychologicalmorbidity. In: Williams P, Wilkinson G, Rawnsley K, editors. Thescope of epidemiological psychiatry. London: Routledge, 1989.pp. 210–21.

[12] Slaets JPJ. Vulnerability in the elderly: frailty. Med Clin North Am2006;90:593–601.

[13] de la Cámara C. Estudio de prevalencia y seguimiento de depresión enel anciano en una muestra comunitaria [Doctoral thesis]. Zaragoza,Spain: Universidad de Zaragoza, 1999.

[14] Saz P, Launer LJ, de la Cámara C, Marcos G, Lobo A. Mortality andmental disorders in a Spanish elderly population. Int J GeriatrPsychiatry 1999;14:1031–8.

[15] Beekman ATF, Penninx B, Deeg DJ, Ormel J, Braam AV, van TilburgW. Depression and physical health in later life: results from theLongitudinal Study Amsterdam (LASA). J Affect Dis 1997;46:219–31.

[16] Bramesfeld A. Service provision for elderly depressed persons andpolitical and professional awareness for this subject: a comparison ofsix European countries. Int J Geriatr Psychiatry 2003;18:392–401.

[17] Roca-Bennasar M, Gili-Planas M, Ferrer-Pérez V, Bernardo-ArroyoM.Mental disorders and medical conditions. A community study in asmall island in Spain. J Psychosom Res 2001;50:39–44.

[18] Braam AW, Prince MJ, Beekman AT, Delespaul P, Dewey ME,Geerlings SW, et al. Physical health and depressive symptoms in olderEuropeans. Results from EURODEP. Br J Psychiatry 2005;187:35–42.

[19] Lobo A, Saz P, Marcos G, Día JL, De La Cámara C, Ventura T, et al.The ZARADEMP Project on the Incidence, Prevalence and RiskFactors of Dementia (and Depression) in the Elderly Community: II.Methods and First Results. Eur J Psychiatry 2005;19:40–54.

[20] Folstein MF, Folstein SE, McHugh PR. “Mini-mental state.” Apractical method for grading the cognitive state of patients for theclinician. J Psychiatr Res 1975;12:189–98.

[21] Lobo A, Saz P, Marcos G, Día JL, de la Cámara C. The prevalence ofdementia and depression in the elderly community in a southernEuropean population. The Zaragoza Study. Arch Gen Psychiatry 1995;52:497–506.

[22] Copeland JR, Kelleher MJ, Kellett JM, Gourlay AJ, Gurland BJ, FleissJL, et al. A semi-structured clinical interview for the assessment ofdiagnosis and mental state in the elderly: the Geriatric Mental StateSchedule: I. Development and reliability. Psychol Med 1976;6:439–49.

[23] Copeland JR, Dewey ME, Griffiths-Jones HM. A computerizedpsychiatric diagnostic system and case nomenclature for elderlysubjects: GMS and AGECAT. Psychol Med 1986;16:89–99.

[24] Copeland JR, Chen R, Dewey ME, McCracken CF, Gilmore C, LarkinB, et al. Community-based case–control study of depression in older

355A. Lobo-Escolar et al. / Journal of Psychosomatic Research 65 (2008) 347–355

people. Cases and sub-cases from the MRC-ALPHA Study. Br JPsychiatry 1999;175:340–7.

[25] Dewey ME, Copeland JRM, Lobo A, Saz P, Día JL. Computerizeddiagnosis from a standardized history schedule—a preliminarycommunication about the organic section of the HAS–AGECATsystem. Int J Geriatr Psychiatry 1992;7:443–6.

[26] Launer LJ, Andersen K, Dewey ME, Lettenneur L, Ott A, AmaducciLA, and the EURODEM Incidence Research Group and Workgroups.Rates and risk factors for dementia and Alzheimer's disease. Resultsfrom EURODEM pooled analysis. Neurology 1999;52:78–84.

[27] World Health Organization. Guidelines for the management of mildhypertension. Memorandum from a World Health Organization/International Society of Hypertension meeting. Guidelines Subcom-mittee of the WHO/ISH Mild Hypertension Liaison Committee.Hypertension 1993;22:392–403.

[28] Angst J. Depression and anxiety: implications for nosology, course,and treatment. J Clin Psychiatry 1997;58(Suppl 8):3–5.

[29] Copeland JR, Beekman AT, Braam AW, Dewey ME, Delespaul P,Fuhrer R, et al. Depression among older people in Europe: theEURODEP Studies. World Psychiatry 2004;3:45–9.

[30] Departamento de Salud, Consumo y Servicios Sociales. PlanEstratégico 2002–2010 de atención a la SaludMental en la ComunidadAutonóma de Aragón. Departamento de Salud, Consumo y ServiciosSociales. Gobierno de Aragón. Zaragoza, 2002. Available at http://portal.aragob.es/pls/portal30/docs/FOLDER/CA_SCBS/FD_SCBS_DOCUMENTACION/FD_SCBS_DOCUMENTACIONINF/FD_SCBS_PLANESTRATEGICO/SALUDMENTAL.PDF [accessed onJanuary 2008].

[31] Fink P. Physical disorders associated with mental illness. A registerinvestigation. Psychol Med 1990;20:829–34.

[32] Katon WJ, Von Korff M, Lin EH, Simon G, Ludman E, Russo J, et al.The Pathaways Study: a randomized trial of collaborative care inpatients with diabetes and depression. Arch Gen Psychiatry 2004;61:1042–9.

[33] Sharma VK, Copeland JR, Dewey ME, Lowe D, Davidson I. Outcomeof the depressed elderly living in the community in Liverpool: a 5-yearfollow-up. Psychol Med 1998;28:1329–37.

[34] de Jonge P, Roy JF, Saz P, Marcos G, Lobo A. Prevalent and incidentdepression in community-dwelling elderly persons with diabetes

mellitus: results from the ZARADEMP Project. Diabetologia 2006;49:2627–33.

[35] Jenkins R, Clare A. Women and mental illness. In: Williams P,Wilkinson G, Rawnsley K, editors. The scope of epidemiologicalpsychiatry. London: Routledge, 1989. pp. 287–305.

[36] Pincus T, Callahan LF, Burkhauser RV. Most chronic diseases arereported more frequently by individuals with fewer than 12 years offormal education in the age 18–64 United States population. J ChronicDis 1987;40:865–74.

[37] Lobo A, Launer LJ, Fratiglioni L, Andersen K, Di Carlo A, BretelerMMB, et al, for the Neurologic Disease in the Elderly Research Group.Prevalence of dementia and major subtypes in Europe: a collaborativestudy of population-based cohorts. Neurology 2000;54(Suppl 5):4–9.

[38] Pérez-Echeverría MJ. Correlaciones entre trastornos endocrinológicos,niveles hormonales en sangre, variables de personalidad y alteracionespsicopatológicas [Doctoral thesis]. Zaragoza, Spain: Universidad deZaragoza, 1985.

[39] Lobo A, Pérez-Echeverría MJ, Jiménez-Aznárez A, Sancho MA.Emotional disturbances in endocrine patients: validity of the scaledversion of the General Health Questionnaire (GHQ-28). Br JPsychiatry 1988;152:807–12.

[40] Peris Roig B, Atienzar Herráez N, Merchante Alfaro AA, Calvo RigualF, Tenías Burillo JM, Selfa Moreno S, et al. Bocio endémico y déficitde yodo: ¿sigue siendo una realidad en España? An Pediatr (Barc)2006;65:234–40.

[41] Robinson RG. Poststroke depression: prevalence, diagnosis, treatment,and disease progression. Biol Psychiatry 2003;54:376–87.

[42] Stang PE, Brandenburg NA, Lane MC, Merikangas KR, Von KorffMR, Kessler RC. Mental and physical comorbid conditions and days inrole among persons with arthritis. Psychosom Med 2006;68:152–8.

[43] Copeland JRM, Dewey ME, Griffiths-Jones HM. Dementia anddepression in elderly persons—AGECAT compared with DSM-III andpervasive illness. Int J Geriatr Psychiatry 1990;5:47–51.

[44] Quetglas B. Consumo de alcohol (y vino en particular) en una muestrarepresentativa de la población de 55 ó más años de la ciudad deZaragoza: relación con el rendimiento cognoscitivo (y con otraspsicopatologías) y estudio caso–control como posible factor de riesgode demencia [Doctoral thesis]. Zaragoza, Spain: Universidad deZaragoza, 2004.