Journal of Affective Disorders 132 (2011) 383–388

Contents lists available at ScienceDirect

Journal of Affective Disorders

j ourna l homepage: www.e lsev ie r.com/ locate / j ad

Research report

Subtyping depression in the medically ill by cluster analysis

Jenny Guidi a,⁎, Giovanni A. Fava a,b, Angelo Picardi c, Piero Porcelli d, Antonello Bellomo e,Silvana Grandi a, Luigi Grassi f, Paolo Pasquini g, Roberto Quartesan h, Chiara Rafanelli a,Marco Rigatelli i, Nicoletta Sonino b,j

a Laboratory of Psychosomatics and Clinimetrics, Department of Psychology, University of Bologna, Bologna, Italyb Department of Psychiatry, State University of New York at Buffalo, Buffalo, NY, United Statesc Italian National Institute of Health, Roma, Italyd Psychosomatic Unit, IRCSS De Bellis Hospital, Castellana Grotte, Bari, Italye University of Foggia, Foggia, Italyf University of Ferrara, Ferrara, Italyg Clinical Epidemiology Unit, Istituto Dermopatico della Immacolata (IDI-IRCSS), Roma, Italyh University of Perugia, Perugia, Italyi University of Modena and Reggio Emilia, Modena, Italyj University of Padova, Padova, Italy

a r t i c l e i n f o

⁎ Corresponding author at: Department of PsychologyViale Berti Pichat 5, 40127 Bologna, Italy. Tel.: +39 051 2243086.

E-mail address: jenny.guidi2@unibo.it (J. Guidi).

0165-0327/$ – see front matter © 2011 Elsevier B.V.doi:10.1016/j.jad.2011.03.004

a b s t r a c t

Article history:Received 26 November 2010Received in revised form 28 February 2011Accepted 1 March 2011Available online 31 March 2011

Background: There is increasing awareness of the need of subtypingmajor depressive disorder,particularly in the setting of medical disease. The aim of this investigation was to use bothDSM-IV comorbidity and the Diagnostic Criteria for Psychosomatic Research (DCPR) forcharacterizing depression in the medically ill.Methods: 1700 patients were recruited from 8medical centers in the Italian Health System and1560 agreed to participate. They all underwent a cross-sectional assessment with DSM-IV andDCPR structured interviews. 198 patients (12.7%) received a diagnosis of major depressivedisorder. Data were submitted to cluster analysis.Results: Two clusters were identified: depressed somatizers and irritable/anxious depression.The somatizer cluster included 58.6% of the cases and was characterized by DCPR somatizationsyndromes (persistent somatization, functional somatic symptoms secondary to a psychiatricdisorder, conversion symptoms, and anniversary reactions) and DCPR alexithymia. Theanxious/irritable cluster had 41.4% of the total sample and included DCPR irritable mood andtype A behavior and DSM-IV anxiety disorders.Limitations: The study has limitations due to its cross-sectional nature. Further, these findingsrequire additional validation in another sample.Conclusions: The findings indicate the need of expanding clinical assessment in themedically illto include the various manifestations of somatization, irritable mood, type A behavior andalexithymia, as encompassed by the DCPR. Subtyping major depressive disorder may yieldimproved targets for psychosomatic research and treatment trials.

© 2011 Elsevier B.V. All rights reserved.

Keywords:Major depressive disorderDiagnostic Criteria for PsychosomaticResearchMedically illCluster analysis

, University of Bologna,091339; fax: +39 051

All rights reserved.

1. Introduction

Major depressive disorder has emerged as an extremelyimportant source of comorbidity in medical disorders (Katon,2003). It was found to affect quality of life and social func-tioning, to lead to increased health care utilization, to beassociated with higher mortality (particularly in the elderly

384 J. Guidi et al. / Journal of Affective Disorders 132 (2011) 383–388

people), to have an impact on compliance and to increasesusceptibility to medical illness (DiMatteo et al., 2000; Favaand Sonino, 1996; Frasure-Smith and Lespérance, 2003;Katon, 2003; Schulz et al., 2002).

There is increasing awareness of the fact that diagnosis ofmajor depressive disorder is too broad and of the need ofsubtyping it for selecting the right treatment for eachindividual patient (Baumeister and Parker, 2010; Bech,2010; Lichtenberg and Belmaker, 2010). This particularlyapplies to depression in the setting of medical disease. Thecategory adjustment disorder with depressive mood iscommonly used but has not attracted adequate researchand offers little to our understanding (Semprini et al., 2010).The majority of patients with major depressive disorder donot qualify for one, but for several Axis I and Axis II disorders(Zimmerman et al., 2002). A source of subtyping may lie inthis comorbidity. Feinstein, however, when he introduced theconcept of comorbidity, referred to any “additional co-existing ailment” separate from the primary disease, even inthe case this secondary phenomenon does not qualify as adisease per se (Feinstein, 1970). Indeed, in clinical medicinethe many methods that are available for measuring comor-bidity are not limited to disease entities (de Groot et al.,2003). Current emphasis in psychiatry is about assessment ofsymptoms resulting in syndromes identified by diagnosticcriteria (DSM). However, there is emerging awareness thatalso psychological symptoms that do not reach the thresholdof a psychiatric disorder may affect quality of life and entailpathophysiological and therapeutic implications (Fava et al.,in press). The Diagnostic Criteria for Psychosomatic Research(DCPR) were developed by an international group ofinvestigators to translate the large body of evidence accu-mulated in psychosomatic medicine in operational tools(Porcelli and Rafanelli, 2010; Porcelli and Sonino, 2007;Wise, 2009). The DCPR allow to translate clinically illnessbehavior (the ways inwhich individuals experience, perceive,evaluate and respond to their health status), the variousmodalities of somatization and constructs such as demoral-ization, irritable mood and alexithymia (Porcelli and Sonino,2007;Wise, 2009; Porcelli and Rafanelli, 2010; Cockram et al.,2009; Fava and Sonino, 2009). Whenever the DCPR have beenused in conjunction with the DSM, they have been found tocarry additional clinical information (Porcelli and Rafanelli,2010).

The aim of this investigation was to use both DSM andDCPR for examining the feasibility of subtyping in a highlyheterogeneous group of medical patients diagnosed as suffer-ing from a major depressive disorder by a cluster analysistechnique.

2. Methods

2.1. Design, procedures and subjects

Patients were recruited from different medical settings inan ongoing multicenter project concerned with the psycho-social dimensions of medical patients (Porcelli and Sonino,2007). Although studies involved in the research project haddifferent aims and sample sizes, they shared a commonmethodology in the assessment of psychopathology andpsychosocial syndromes. Patients were recruited consecu-

tively, with the intent of being representative of their res-pective patient populations:

1. Consecutive outpatients with functional gastrointestinaldisorders (N=190, 12.2% of the total sample) from theFunctional Gastrointestinal Disorders Outpatient Clinic ofthe Scientific Institute of Gastroenterology at CastellanaGrotte, Italy.

2. Consecutive outpatients with heart diseases (N=351,22.5%) from 3 different sources: 1) 198 patients whounderwent heart transplantation from the Heart Trans-plantation Unit of the Institute of Cardiology at S. OrsolaHospital of Bologna, Italy; 2) 61 consecutive patients witha recent (within 1 month) first myocardial infarctiondiagnosis from the Cardiac Rehabilitation Program of theBellaria Hospital in Bologna, Italy; and 3) 92 consecutivepatients with a recent (within 1 month) first myocardialinfarction diagnosis, from the Institute of Cardiology ofUniversity Hospital in Modena, Italy.

3. Consecutive outpatients with endocrine disorders(N=162, 10.4%) from the Division of Endocrinology ofthe University of Padova Medical Center, Padova, Italy.

4. Consecutive outpatients who had received a diagnosis ofcancer within the past 18 months (N=104, 6.7%) from theS. Anna University Hospital in Ferrara, Italy.

5. Consecutive outpatients with skin disorders (N=545,34.9%) from the Dermopathic Institute of the Immaculate(IDI-IRCCS), Rome, Italy.

6. Consecutive inpatients referred for psychiatric consulta-tion in 2 large university-based general hospitals (N=208,13.3%) from the University of Perugia and University ofFoggia, Italy.

The study was approved by institutional review boardsand local ethics committees, and written informed consentwas obtained from all patients. The patients who wereapproached were 1700; 140 (8.2%) declined to participate.The most common reason for refusal was lack of time. Thetotal sample included 1560 patients (712men, 45.6%, and 848women, 54.4%), with a mean age of 45 (SD=15.02) years,and a mean of 10.6 (SD=3.85) years of education. Therewere no significant differences in terms of sociodemographicvariables between the patients who accepted and those whorefused.

2.2. Assessment

All patients underwent two detailed semistructuredinterviews by clinical psychologists or psychiatrists withextensive experience, including psychosomatic research.Psychiatric disorders were investigated with the StructuredClinical Interview for DSM-IV (SCID; First et al., 2000).Diagnosis of Major Depressive Disorder (MDD) was consid-ered individually, while other diagnoses were groupedaccording to diagnostic categories such as anxiety disorders,somatoform disorders, adjustment disorders, and otherdisorders (including psychotic disorders, eating disorders,sexual dysfunctions and substance use related disorders).Psychosomatic syndromes were diagnosed with the Struc-tured Interview for DCPR (Porcelli and Sonino, 2007). TheDCPR encompass various diagnostic rubrics: abnormal illnessbehavior (disease phobia, thanatophobia, health anxiety, and

able 1requencies of diagnostic categories of psychiatric disorders and psychoso-atic syndromes.

Diagnostic category FrequencyN (%)

DSM anxiety disorders 25 (12.6)DSM somatoform disorders 10 (5.0)DSM adjustment disorders 7 (3.5)Other DSM disorders 5 (2.5)DCPR somatization 95 (48.0)DCPR abnormal illness behavior 50 (25.2)DCPR irritable mood and type A behavior 65 (32.8)DCPR demoralization 111 (56.1)DCPR alexithymia 43 (21.7)No DCPR syndromes 13 (6.6)

385J. Guidi et al. / Journal of Affective Disorders 132 (2011) 383–388

illness denial), somatization syndromes (persistent somati-zation, functional somatic symptoms secondary to a psychi-atric disorder, conversion symptoms, and anniversaryreactions), irritability (irritable mood and type A behavior),demoralization, and alexithymia. The interview for DCPRconsists of 58 items scored in a yes/no response formatevaluating the presence of one or more of the twelve psycho-somatic syndromes mentioned above. The interview hasshown excellent inter-rater reliability, construct validity, andpredictive validity for psychosocial functioning and treatmentoutcome (Galeazzi et al., 2004).

2.3. Data analysis

Data were entered in SPSS (SPSS Inc., USA), after whichdescriptive statistics were calculated. Two-step clusteranalysis was performed to organize observations into twoor more mutually exclusive groups, where members ofthe groups shared properties in common (Kaufman andRousseeuw, 1990). The following variables were included inthe analysis: DSM anxiety disorders, somatoform disorders,adjustment disorders, other disorders (psychotic disorders,eating disorders, sexual dysfunctions and substance usedisorders), DCPR abnormal illness behavior, somatization,irritability, demoralization, alexithymia, and absence of anyDCPR syndrome.

The two-step cluster method is a scalable cluster analysisalgorithm designed to handle large data sets. It can handleboth continuous and categorical variables. The two steps are:1) pre-cluster the cases into many small sub-clusters; and2) cluster the sub-clusters resulting from pre-cluster step intothe desired number of clusters. The log-likelihood distancemeasure was used, with subjects assigned to the clusterleading to the largest likelihood. No prescribed number ofclusters was suggested. The Bayesian Information Criterion(BIC) was used to judge adequacy of the final solution.Differences in sample characteristics were compared accord-ing to cluster membership using independent sample t-testsand chi squared tests for continuous and categorical variables,respectively. For all tests performed, the significance levelwas set at .05, two-tailed.

3. Results

A total of 198 patients (12.7%; 62.6% female) received adiagnosis of Major Depressive Disorder (MDD) according toDSM-IV criteria, with a mean age of 45.79 (SD=14.39) years,and a mean of 9.73 (SD=3.99) years of education. Of these,fifty-one (25.8%) had at least 1 comorbid Axis I disorder(particularly, anxiety disorders), and 185 (93.4%) presentedat least 1 comorbid DCPR syndrome (mainly demoralization,somatization syndromes, and irritability). Frequencies foreach of the diagnostic categories of psychiatric disorders andpsychosomatic syndromes are shown in Table 1.

Two-step cluster analysis yielded 2 clusters, with noexclusion of cases. The composition of the clusters (Fig. 1) andthe importance of variables within a cluster were thenexamined.

The first cluster had 58.6% of the cases (N=116) andcontained primarily DCPR somatization syndromes (i.e.,persistent somatization, functional somatic symptoms sec-

TFm

ondary to a psychiatric disorder, conversion symptoms, andanniversary reactions) and DCPR alexithymia; this clusterwas named somatizer.

The second cluster had 41.4% (N=82) of the total sampleand was mainly characterized by patients who displayedDCPR irritability (i.e., irritable mood and type A behavior),and DSM-IV anxiety disorders; this cluster was namedirritable and anxious.

Both DCPR demoralization and abnormal illness behaviorwere highly represented in both clusters (demoralization:51.4% vs. 48.6% and abnormal illness behavior: 46% vs. 54%),even though these diagnostic categories were approximatelyequally distributed and did not make a substantial contribu-tion to cluster formation. Similarly, the frequency and/or theimportance of the remaining variables (e.g., DSM-IV somato-form disorders, adjustment disorders, other disorders listedin DSM-IV, and absence of any DCPR diagnosis) were com-parable among the groups and/or did not appear important indistinguishing between the two clusters.

When differences among the cluster groups were exam-ined, no significant differences were found with regard toboth gender and age. Years of education differed among theclusters (t130=−2.805; pb .01), with somatizers receivinglower education (mean 9.16 years; SD=3.72) compared toirritable/anxious patients (mean 11.31 years; SD=4.32).

With regard to cluster composition across differentmedical settings, the somatizer cluster primarily includedpatients from the Functional Gastrointestinal DisordersOutpatient Clinic (N=39; 33.6%), as well as inpatients frompsychiatric consultation services (N=21; 18.1%) and patientswith skin diseases (N=21; 18.1%). A considerable proportionof patients with a recent first myocardial infarction diagnosiswere also found within somatizers (N=13; 11.2%). Theanxious/irritable cluster wasmainly characterized by patientswith endocrine disorders (N=33; 40.2%), followed bypatients with skin diseases (N=17; 20.7%) and patientswho underwent heart transplantation (N=13; 15.9%).

4. Discussion

This study has indicated that major depressive disorder inthe medically ill may be classified into two clusters.

The first cluster (named somatizer) encompassed about60% of cases and was characterized by DCPR somatizationsyndromes (persistent somatization, functional somatic

Fig. 1. Distribution of diagnostic categories within each cluster.

386 J. Guidi et al. / Journal of Affective Disorders 132 (2011) 383–388

symptoms secondary to a psychiatric disorder, conversionsymptoms, and anniversary reactions) and alexithymia. TheDCPR category of persistent somatization is conceptualized asa grouping of functional symptoms involving different organsystems and is issued from the concept of symptomsclustering developed by Kellner (1994). The parent categoryof functional somatic symptoms secondary to a psychiatricdisorder refers to symptoms of autonomic arousal or func-tional medical disorder in patients with a psychiatric disorderpreceding the onset of functional somatic symptoms. Accord-ing to Engel's (1970) stringent criteria, conversion symptomsinvolve features such as ambivalence, histrionic personality,and precipitation of symptoms by psychological stress ofwhich the patient is unaware. Anniversary reactions refer tosymptoms of autonomic arousal occurring at the same time asan important date for the patient or at the same age that aclose relative developed a life-threatening illness or died.

The depressed somatizer cluster is in line with severalresearch findings. Kellner (1990) remarked that the researchevidence for the association between depression and soma-tization has been consistent regardless of the design of thestudy: depressed patients tend to have more somatic symp-toms than nondepressed individuals, and somatizers tend tobe more depressed than patients with physical disease.Further, the close relationship that we found between depre-ssive and somatic symptoms by interviewmethods replicatesthe findings obtained with self-rating scales (Fava et al., 1982,1984). The link between alexithymia and depression alsoreflects an increasing amount of literature (Taylor, 2010).Vanheule et al. (2007) used cluster analysis in a sample of 404patients in mental health centers and found support for acategorical differentiation between high and low alexithymicdepressed patients. Such differentiation was found to entailclinical implications in an elderly population (Honkalampiet al., 2010). Not surprisingly, depressed somatizers werefound to characterize functional gastrointestinal disorders,patients for whom consultation psychiatry was requested, inaddition to some of the patients who has a recent first myo-cardial infarction.

The second cluster (named irritable and anxious) encom-passed about 40% of cases and was characterized by DCPRirritability (irritable mood and type A behavior) and DSM-IVanxiety disorders. Irritability refers to a feeling state that maybe experienced as brief episodes in particular circumstances,or may be prolonged and generalized. It requires an increasedeffort of control over temper by the individual or may resultin irascible verbal or behavioral outbursts. Type A behaviorrefers to the presence of at least 5 of 9 characteristics thatwere described in coronary artery disease: excessive degreeof involvement in work and other activities; steady andpervasive sense of time urgency; display of motor-expressivefeatures; hostility and cynicism; irritable mood; tendency tospeed up physical activities; tendency to speed up mentalactivities; high intensity of desire for achievements andrecognition; and high competitiveness (Porcelli and Sonino,2007). Hostility and irritable mood are fairly common indepressed patients (Fava et al., 1986, 1991; Picardi et al.,2004) and are consistent with factor analysis studies sugges-ting that 3 dimensions, corresponding to the basic humanemotions of sadness, fear and anger, underlie symptomatol-ogy in psychopathology (Biondi et al., 2005; Pancheri et al.,2002; Pasquini et al., 2004). Hostility was found to explain theassociation between depressive mood and mortality(Lemogne et al., 2010). The presence of anxiety disorders inthe setting of depressive illness is also well established inthe literature (Paykel, 1971; Zimmerman et al., 2002). Thefindings of this investigation indicate that a combination ofdepressed mood and anxiety disturbances is common in themedically ill, particularly in the setting of endocrine disease,and heart transplantation. In disorders such as hyperthyroid-ism and Cushing's syndrome, affective disturbances (anxiety,depression and irritability) may be, at least in part, linked tothe severity of hormonal imbalances (Sonino et al., 2007). Inheart transplantation and after myocardial infarction,patients are experiencing life-threatening condition thatmay affect white matter integrity (Rapp et al., 2010). It isdifficult therefore to know from these data whether severityof illness plays a part in producing this cluster.

387J. Guidi et al. / Journal of Affective Disorders 132 (2011) 383–388

The DSM-IV constructs of somatoform and adjustmentdisorders were not found to be important in distinguishingthe clusters. Indeed, both diagnostic rubrics have recentlyundergone considerable criticism as to their clinical useful-ness (Wise, 2009; Semprini et al., 2010). In particular, theDSM-IV category of somatoform disorders appears to beeither too restrictive or too broad, rarely fits with clinicalreality and cannot be used in conjunction with organicdisorders (Fava et al., 2007).

The study has limitations due to its cross-sectional nature.We have no way to know longitudinal course of theseclusters. In addition, the classification found has not yet beenvalidated in another sample. Nonetheless, cluster analysis indepression has yielded results that have survived the test oftime (Paykel, 1971), and there is some evidence that subtypesof depression, such as those identified by means of clusteranalysis, might be predictive of treatment response (Bech,2010; Lichtenberg and Belmaker, 2010). Examples are provi-ded by the poor response of anxious or atypical depression topharmacotherapy; the need of combination therapy inpsychotic depression; the risk of cerebrovascular disease inlate-life depression. Both in research settings and clinicalpractice, assessment of depression in the medically ill gene-rally does not encompass consideration of irritable mood,somatization, alexithymia and abnormal illness behavior.Exclusive reliance on diagnostic criteria has impoverished theclinical process and does not reflect the complex thinking thatunderlies decisions in psychiatric practice (Fava et al., inpress). Subtyping major depressive disorders in the setting ofmedical disease may yield improved targets for psychoso-matic research and treatment trials.

The results indicate the importance of adding supplemen-tary clinical information to the customary clinical taxonomy(DSM), that includes illness behavior, modalities of somatiza-tion, and subclinical distress. Such information may demar-cate prognostic and therapeutic differences among groupsof patients who otherwise seem to be deceptively similarbecause they share the same diagnosis of major depressivedisorder.

Role of funding sourceThis study was supported in part by a grant from Compagnia di San

Paolo, Torino, Italy to Dr. Rafanelli. Compagnia di San Paolo had no furtherrole in the study design; in the collection, analysis and interpretation of data;in the writing of the report; and in the decision to submit the paper forpublication.

Conflict of interestDrs. Guidi, Fava, Picardi, Porcelli, Grandi, Grassi, Pasquini, Quartesan,

Rafanelli, Rigatelli, and Sonino have no conflict of interest to declare. DrBellomo has been sponsored as speaker at meetings and congresses by Ely-Lilly, Janssen-Cilag, AstraZeneca, Bristol Meyers Squibb, Lundbeck, Pfeizer,and Glaxo-SmithKline in the previous year.

Acknowledgement

Nothing to report.

References

Baumeister, H., Parker, G., 2010. A second thought on subtyping majordepression. Psychother. Psychosom. 79 (6), 388–389.

Bech, P., 2010. Struggle for subtypes in primary and secondary depressionand their mode-specific treatment or healing. Psychother. Psychosom.79 (6), 331–338.

Biondi, M., Picardi, A., Pasquini, M., Gaetano, P., Pancheri, P., 2005.Dimensional psychopathology of depression: detection of an ‘activation’dimension in unipolar depressed outpatients. J. Affect. Disord. 84 (2–3),133–139.

Cockram, C.A., Doros, G., de Figueiredo, J.M., 2009. Diagnosis and measure-ment of subjective incompetence: the clinical hallmark of demoraliza-tion. Psychother. Psychosom. 78 (6), 342–345.

de Groot, V., Beckerman, H., Lankhorst, G.J., Bouter, L.M., 2003. How tomeasure comorbidity. A critical review of available methods. J. Clin.Epidemiol. 56 (3), 221–229.

DiMatteo, M.R., Lepper, H.S., Croghan, T.W., 2000. Depression is a risk factorfor noncompliance with medical treatment: meta-analysis of the effectsof anxiety and depression on patient adherence. Arch. Intern. Med. 160(14), 2101–2107.

Engel, G.L., 1970. Conversion symptoms. In: Mac Bride, C.M., Blacklow, R.S.(Eds.), Signs and Symptoms, pp. 650–669. Lippincott, Philadelphia.

Fava, G.A., Sonino, N., 1996. Depression associated with medical illness.Treatment considerations. CNS Drugs 5 (3), 175–189.

Fava, G.A., Sonino, N., 2009. Psychosomatic assessment. Psychother.Psychosom. 78 (6), 333–341.

Fava, G.A., Pilowsky, I., Pierfederici, A., Bernardi, M., Pathak, D., 1982. Depres-sion and illness behavior in a general hospital. Psychother. Psychosom.38 (1), 141–153.

Fava, G.A., Zielezny, M., Pilowsky, I., Trombini, G., 1984. Patterns ofdepression and illness behaviour in general hospital patients. Psycho-pathology 17 (3), 105–109.

Fava, G.A., Kellner, R., Lisansky, J., Park, S., Perini, G.I., Zielezny, M., 1986.Hostility and recovery from melancholia. J. Nerv. Ment. Dis. 174 (7),414–417.

Fava, M., Rosenbaum, J.F., McCarthy, M., Pava, J., Steingard, R., Bless, E., 1991.Anger attacks in depressed outpatients and their response to fluoxetine.Psychopharmacol. Bull. 27 (3), 275–279.

Fava, G.A., Fabbri, S., Sirri, L., Wise, T.N., 2007. Psychological factors affectingmedical conditions: a new proposal for DSM-V. Psychosomatics 48 (2),103–111.

Fava, G.A., Rafanelli, C., Tomba, E. (in press). The clinical process in psy-chiatry. A clinimetric approach. J. Clin. Psychiatry

Feinstein, A.R., 1970. The pre-therapeutic classification of co-morbidity inchronic disease. J. Chron. Dis. 23, 455–469.

First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B.W., 2000. SCID-I —Structured Clinical Interview for DSM-IV Axis I Disorders (Italian versionby Mazzi, F., Morosini, P., de Girolamo, G., Lussetti, M., Guaraldi, G.P.).Organizzazioni Speciali, Firenze, IT.

Frasure-Smith, N., Lespérance, F., 2003. Depression and other psychologicalrisks following myocardial infarction. Arch. Gen. Psychiatry 60 (6),627–636.

Galeazzi, G.M., Ferrari, S., Mackinnon, A., Rigatelli, M., 2004. Interraterreliability, prevalence, and relation to ICD-10 diagnoses of the DiagnosticCriteria for Psychosomatic Research in consultation-liaison psychiatrypatients. Psychosomatics 45 (5), 386–393.

Honkalampi, K., Lehto, S.M., Hintikka, J., Koivumaa-Honkanen, H., Niskanen,L., Viinamäki, H., 2010. Symptoms of depression and alexithymic burdenin middle-aged men. Psychother. Psychosom. 79 (4), 259–261.

Katon, W., 2003. Clinical and health services relationships between majordepression, depressive symptoms, and general medical illness. Biol.Psychiatry 54 (3), 216–226.

Kaufman, L., Rousseeuw, P.J., 1990. Finding Groups in Data: an Introductionto Cluster Analysis. Wiley, New York, NY.

Kellner, R., 1990. Somatization. Theories and research. J. Nerv. Ment. Dis. 178(3), 150–160.

Kellner, R., 1994. Psychosomatic syndromes, somatization and somatoformdisorders. Psychother. Psychosom. 61 (1–2), 4–24.

Lemogne, C., Nabi, H., Zins, M., Cordier, S., Ducimetière, P., Goldberg, M.,Consoli, S.M., 2010. Hostility may explain the association betweendepressive mood and mortality: evidence from the French GAZEL cohortstudy. Psychother. Psychosom. 79 (3), 164–171.

Lichtenberg, P., Belmaker, R.H., 2010. Subtyping major depressive disorder.Psychother. Psychosom. 79 (3), 131–135.

Pancheri, P., Picardi, A., Pasquini, M., Gaetano, P., Biondi, M., 2002.Psychopathological dimensions of depression: a factor study of the17-itemHamilton depression rating scale in unipolar depressed outpatients.J. Affect. Disord. 68 (1), 41–47.

Pasquini, M., Picardi, A., Biondi, M., Gaetano, P., Morosini, P., 2004. Relevanceof anger and irritability in outpatients with major depressive disorder.Psychopathology 37 (4), 155–160.

Paykel, E.S., 1971. Classification of depressed patients: a cluster analysisderived grouping. Br. J. Psychiatry 118 (544), 275–288.

388 J. Guidi et al. / Journal of Affective Disorders 132 (2011) 383–388

Picardi, A., Morosini, P., Gaetano, P., Pasquini, M., Biondi, M., 2004. Higherlevels of anger and aggressiveness in major depressive disorder than inanxiety and somatoform disorders. J. Clin. Psychiatry 65 (3), 442–443.

Porcelli, P., Rafanelli, C., 2010. Criteria for psychosomatic research (DCPR) inthe medical setting. Curr. Psychiatry Rep. 12 (3), 246–254.

Porcelli, P., Sonino, N., 2007. Psychological Factors affecting MedicalConditions. Karger, Basel, CH.

Rapp, M.A., Rieckmann, N., Lessman, D.A., Tang, C.Y., Paulino, R., Burg, M.M.,Davidson, K.W., 2010. Persistent depressive symptoms after acute coro-nary syndrome are associated with compromised white matter integrityin the anterior cingulate: a pilot study. Psychother. Psychosom. 79 (3),149–155.

Schulz, R., Drayer, R.A., Rollman, B.L., 2002. Depression as a risk factor fornon-suicide mortality in the elderly. Biol. Psychiatry 52 (2), 205–225.

Semprini, F., Fava, G.A., Sonino, N., 2010. The spectrum of adjustment disor-ders: too broad to be clinically helpful. CNS Spectr. 15 (6), 382–388.

Sonino, N., Tomba, E., Fava, G.A., 2007. Psychosocial approach to endocrinedisease. Adv. Psychosom. Med. 28, 21–33.

Taylor, G.J., 2010. Affects, trauma, and mechanisms of symptom formation:a tribute to John C. Nemiah, MD (1918–2009). Psychother. Psychosom.79 (6), 339–349.

Vanheule, S., Desmet, M., Verhaeghe, P., Bogaerts, S., 2007. Alexithymicdepression: evidence for a depression subtype? Psychother. Psychosom.76 (5), 315–316.

Wise, T.N., 2009. Diagnostic criteria for psychosomatic research are necessaryfor DSM V. Psychother. Psychosom. 78 (6), 330–332.

Zimmerman,M., Chelminski, I.,McDermut,W., 2002.Major depressive disorderand axis I diagnostic comorbidity. J. Clin. Psychiatry 63 (3), 187–193.