primary care patients' personal illness models for depression: relationship to coping behavior...

Available online at www.sciencedirect.com

General Hospital Psychiatry 29 (2007) 492–500

Psychiatry and Primary CareRecent epidemiologic studies have found that most patients with mental illness are seen exclusively in primary care medicine. These patients often present

with medically unexplained somatic symptoms and utilize at least twice as many health care visits as controls. There has been an exponential growth in studies

in this interface between primary care and psychiatry in the last 10 years. This special section, edited by Jürgen Unutzer, M.D., will publish informative

research articles that address primary care-psychiatric issues.

Primary care patients' personal illness models for depression: relationshipto coping behavior and functional disability☆

Charlotte Brown, Ph.D.a,⁎, Deena R. Battista, Ph.D.a, Susan M. Sereika, Ph.D.b,Richard D. Bruehlman, M.D.a,c, Jacqueline Dunbar-Jacob, Ph.D.b, Michael E. Thase, M.D.a

aUniversity of Pittsburgh School of Medicine, Pittsburgh, PA 15213, USAbUniversity of Pittsburgh School of Nursing, Pittsburgh, PA 15213, USA

cRenaissance Family Practice, Pittsburgh, PA 15139, USA

Received 21 July 2006; accepted 30 July 2007

Abstract

Objective: The applicability and clinical utility of Leventhal et al.'s model of illness cognition were evaluated in depressed primary carepatients. The intercorrelations of illness beliefs and the mediational effects of coping behavior on these beliefs were also evaluated.Moderating effects of coping behaviors were explored.Methods: Baseline evaluations of demographic information, depression diagnoses, depressive symptom severity, self-reported psychosocialand physical functioning, medical comorbidity, illness beliefs and depression coping strategies were obtained from 191 primary care patientsreceiving antidepressant medication for the treatment of depression.Results: Patients' beliefs about depressive symptoms, causes, duration as well as controllability and the consequences of these symptoms aredescribed. Leventhal et al.'s mediational model was partially supported for the outcome of psychosocial functioning. Coping behavior did notmediate the relationship between illness beliefs and physical functioning. The relationships between participants' beliefs about the cause,controllability and duration of depressive symptoms were mediated by the use of behavioral disengagement, venting or self-blame as astrategy to cope with depression. In addition, use of acceptance, religious coping or behavioral disengagement moderated the relationshipbetween beliefs about the cause of depression (i.e., environment or chance or medical illness) and psychosocial functioning.Conclusions: Illness models for depression are important determinants of functioning in depressed primary care patients. These beliefs andcoping behaviors are potentially modifiable and could be the target of interventions to decrease functional impairment in depressed patients.© 2007 Elsevier Inc. All rights reserved.

Keywords: Depression; Illness perceptions; Primary care; Coping

1. Introduction

Determining how individuals understand and definedepression and its consequences as a health problem is

An earlier version of this work was presented at the “Complexities ofCo-Occurring Conditions Conference” in Washington, D.C., June 2004.

☆ This research was supported by a grant from the National Institute ofMental Health (R01 MH60763) to Charlotte Brown, Ph.D.

⁎ Corresponding author. Tel.: +1 412 383 5134; fax: +1 412 383 5412.E-mail address: [email protected] (C. Brown).

0163-8343/$ – see front matter © 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.genhosppsych.2007.07.007

critical to understanding how they manage this illness [1].Knowledge about primary care patients' model of theirillness can help in the development of more effectivetreatment engagement strategies, self-management training,adherence interventions and support services appropriate topatients' needs. While patterns of illness behavior amongindividuals with psychiatric disorders have been documented[1,2], little attention has been given to the role of thedepressed person's illness beliefs in identifying and copingwith the disorder and the association of these factors withdisability, care seeking and treatment adherence.

mailto:[email protected]

http://dx.doi.org/10.1016/j.genhosppsych.2007.07.007

493C. Brown et al. / General Hospital Psychiatry 29 (2007) 492–500

Leventhal et al.'s [3,4] self-regulatory model of illnessrepresentation addresses just these relationships. In thismodel, the primary emphasis is on patients' beliefs abouttheir condition. The individual is viewed as an activeproblem solver who develops hypotheses about the natureof symptoms and appropriate ways of coping with them.Thus, the person's conception of the cognitive and affectiveattributes of health threats shapes behaviors for coping withor controlling the illness and may play a critical role inevaluating coping and adaptational outcomes [3,4].Leventhal et al.'s model is a mediational one that proposesthat illness beliefs lead to coping responses that, in turn,influence health outcomes. Thus, the effect of beliefsabout depression on psychological and physical function-ing would be mediated by the intermediate variable ofcoping behaviors.

Illness beliefs have five distinct components: (a)identity, the label the person uses to describe the illnessand the symptoms viewed as being part of the disorder; (b)cause, beliefs about what caused the symptoms or illness;(c) timeline, the expected duration of the symptoms orillness; (d) consequences, expected effects and outcomesof the illness; and (e) perceived controllability, theresponsiveness of the symptoms or illness to treatmentor self-management.

A recent meta-analysis [5] reported on 45 studies thatexamined this model in 23 illnesses and conditions (e.g.,diabetes, arthritis and irritable bowel syndrome); however,no study of psychiatric disorders was identified. Inaddition, the mediational effect of coping behavior hasbeen demonstrated in such illnesses as hypercholesterole-mia [6] and Huntington's disease [7] but not in psychiatricconditions. Given the model's demonstrated usefulness inunderstanding self-care behaviors, adherence and func-tional adaptation in physical illnesses, it would beimportant to know whether it can be applied to a commonpsychiatric disorder such as depression. In fact, one of thekey challenges of applying Leventhal et al.'s self-regulatory model to depression is that it specifies that(a) somatic, affective and cognitive symptoms have animpact on illness beliefs and (b) depression itself ischaracterized by changes in these symptoms. Of particularsignificance is the fact that depression is characterized bycognitive distortions, and this may be a confoundingfactor. However, to the extent that people view theirphysical and emotional well-being in similar ways, thismodel should be applicable to depression and otherpsychiatric disorders.

This article describes personal illness models fordepression in a sample of 191 primary care patientsreceiving antidepressant medication. We also evaluated theapplicability and clinical utility of this model in depressionby (a) examining whether the intercorrelations of illnesscognition dimensions are consistent with those reported byother physical conditions and (b) determining the nature ofthe relationship between individuals' beliefs about depres-

sion and their functional disability. We hypothesized that(a) the magnitude of the intercorrelations of illnesscognition dimensions would be within the low to moderaterange and (b) behaviors used by patients to cope withdepression would mediate the relationship between illnessbeliefs and functional disability.

2. Methods

2.1. Participants

Study participants were family practice patients enrolledin a yearlong observational study of adherence and personalillness models for depression. All participants were referredto the study by their primary care physician. Primary carepatients were eligible for referral if they (a) were 18 yearsor older, (b) had been prescribed an antidepressantmedication for treatment of depression in the preceding2 weeks or had been switched to a new antidepressantwithin the preceding 2 weeks, (c) had no lifetime history ofbipolar disorder, (d) had no substance dependence or abusein the preceding 6 months and (e) had no current psychoticsymptom or history of psychotic disorder. The studyprotocol was approved by the University of PittsburghInstitutional Review Board and the Veterans AffairsInstitutional Review Board, and all participants providedwritten informed consent.

Participants were recruited from two urban family practicehealth centers, a single specialty group practice and twoprimary care clinics in the Veterans Health Administration.The health centers are the outpatient clinical settings of theUniversity of Pittsburgh Medical Center Family PracticeResidency Program. The group practice has eight officelocations and 23 board-certified family physicians. In addition,patients were recruited from two primary care clinics located atthe Veterans Affairs (VA) Pittsburgh Healthcare SystemOakland site (VA Pittsburgh–Oakland). The VA Pittsburgh–Oakland is a major medical center located in VISN 4 (theVA Stars and Stripes Healthcare Network) that coversPennsylvania, West Virginia, Eastern Ohio and Delaware.

Four hundred fifty-six patients were referred to thestudy (8 were ineligible, 172 were unable to be reachedwithin 2 weeks of the referral and 60 refused toparticipate in the study), 216 completed the baselineassessment (25 were ineligible) and 191 were eligible andenrolled in the study.

2.2. Procedure

Participants completed five evaluations over a 1-yearperiod. Semistructured interviews and self-report question-naires were used to ascertain demographic information,clinical status (depression and anxiety diagnoses, depressionseverity), psychosocial and physical functioning, medicalcomorbidity and beliefs about depression. This articlefocuses on data obtained from the baseline evaluation ofthe 191 patients enrolled in the study.

494 C. Brown et al. / General Hospital Psychiatry 29 (2007) 492–500

2.3. Measures

2.3.1. Demographic informationSelf-reported data on race, sex, education, employment,

marital status and income were obtained from each participant.

2.3.2. Clinical characteristicsThe structured interview guide Primary Care Evaluation

of Mental Disorders (PRIME-MD) was used to assess thepresence of current mood, anxiety, substance use andsomatoform disorders [8].

The Beck Depression Inventory is a widely used 21-itemself-report inventory designed to assess the severity ofdepressive symptoms [9,10].

The 36-item Medical Outcomes Study (MOS) Short-Form General Health Survey, a self-administered question-naire, was developed for use with primary care patients in theRand MOS [11]. The eight MOS scales were weighted andaggregated into the Physical Component Scale score and theMental Component Scale score [12]. These scales havestandardized scores, which have a mean of 50 and a standarddeviation of 10.

Medical comorbidity was assessed with the MedicalComorbidity Checklist. Number of current medical condi-tions was assessed with a modified version of a chronicmedical condition checklist developed byWells et al. [13] foruse in the MOS. The checklist was modified by adding thefollowing conditions: thyroid disease, eye disease (cataracts,glaucoma, detached retina), lupus and HIV/AIDS.

Previousmental health treatment was assessedwith a singlequestion: “At any time in the past, have you ever visited anymental health professional (e.g., psychiatrist, psychologist,social worker, psychiatric nurse, other mental health profes-sional or counselor) for a problem with your emotional ormental health? If yes, was it (a) in the past 12 months, (b) past6 months, (c) past 3 months or (d) past month?”

2.3.3. Illness perceptionsThe 57-item Illness Perception Questionnaire was used to

measure the components of patients' illness beliefs based onLeventhal et al.'s [4] self-regulatory model of illnessbehavior. The five scales assess illness components (i.e.,identity, cause, timeline, consequences and perceivedcontrollability) that underlie a patient's representation ofillness. Scale items are scored on a 4-point (all of the time tonever) or 5-point (strongly agree to strongly disagree) Likertscale. The scale was developed on diabetes, rheumatoidarthritis and chronic pain samples and has demonstratedvalidity and reliability in these chronic illnesses [14]. Theauthors noted that the scale can be adapted for use in avariety of illnesses. For the present study, an adapted versionof the original scale was used. Several components of theoriginal scale were modified for the assessment of illnessbeliefs in a depressed sample and piloted in 40 primary carepatients [15]. In that study, illness beliefs were found to besignificantly associated with depression coping strategiesand illness management behaviors such as prior mental

health treatment, current antidepressant treatment andmedication adherence.

The illness identity subscale is composed of 23 items thatassess the frequency of symptoms endorsed as part ofdepression on a 4-point Likert scale (all of the time to never).The timeline subscale assesses the duration of depressivesymptoms endorsed on a 5-point Likert scale (strongly agreeto strongly disagree; e.g., “My symptoms will last a shorttime.”). Perceived consequences of depressive symptoms areassessed with 7 items scored on a 5-point Likert scale(strongly agree to strongly disagree; e.g., “My symptomshave had major consequences on my life.”). Perceivedcontrollability of depressive symptoms is assessed with 7items scored on a 5-point Likert scale (strongly agree tostrongly disagree; e.g., “My symptoms will improve intime.”). The cause subscale is composed of 14 items, eachassessing a specific cause of depression. In order to use theseitems in bivariate and multiple regression analyses, wecompleted an exploratory factor analysis to determinewhether items constituted a unidimensional or multidimen-sional scale. A principal components extraction withorthogonal rotation was completed on the 14 items. Aneigenvalue of at least 1.0 was the criterion used to retainprincipal component factors [16], and the analysis yielded afive-factor solution. Factor items were limited to those with aloading of 0.40 or greater.

The five factors identified by principal componentsanalysis accounted for 57% of the total item variance.These factors were environment/chance, stress/interpersonalrelationships, medical illness, self-care and heredity/death.Items constituting each factor were used to develop subscalescores, using unit weighting.

2.3.4. Coping strategiesThe 28-item Brief COPE Scale was used to assess coping

styles in response to depressive symptoms. The scale has 14conceptually distinct subscales with 2 items per scale. Thisscale is a brief version of the COPE Inventory, which has beenused extensively in health-related research [17,18]. Subscalesinclude active coping (taking action to remove or circumventthe stressors), planning (thinking about how to confront thestressor, planning one's active coping efforts), seekinginstrumental support (seeking assistance, information oradvice about what to do), seeking emotional support (gettingsympathy or emotional support from someone), religion(increased engagement in religious activities), positive refram-ing (making the best of a situation by growing from it orviewing it in a more favorable light), acceptance (accepting thefact that the stressor has occurred and is real), humor (makingjokes about the stressor), self-distraction (focusing on doingthings to take one's mind off the stressor), denial (an attempt toreject the reality of the stressor), venting (to ventilate ordischarge emotional distress), substance use (turning to alcoholor drugs as a way of disengaging from the stressor), behavioraldisengagement (giving up or withdrawing from coping withthe stressor) and self-blame (blaming oneself for the

Table 1Baseline clinical and demographic characteristics of the study participants(N=191)

Characteristic Distribution

DemographicAge [years, mean (S.D.)] 45.1 (15.9)Female [n (%)] 135 (70.7)White [n (%)] 179 (93.7)Employed full/part time [n (%)] 101 (52.9)Married or living with partner [n (%)] 103 (53.9)Education greater than high school [n (%)] 106 (55.5)ClinicalBeck Depression Inventory [mean (S.D.)] 19.5 (10.3)MOS Physical Component Scale [mean (S.D.)] 44.6 (11.5)MOS Mental Component Scale [mean (S.D.)] 30.7 (10.7)No. of current medical conditions [mean (S.D.)] 2.0 (1.6)With mental health visit in the past 12 months [n (%)] 38 (19.9)


occurrence of the stressor). For this study, scale items thatmade reference to coping with situations or stressors (e.g., “Itry to grow as a result of the experience.”) were modified toassess coping with symptoms (e.g., “I try to grow as a result ofhaving experienced these symptoms.”).

2.4. Statistical analyses

The participants' clinical characteristics, demographiccharacteristics and illness beliefs (identity, timeline, cause,consequences and controllability) are described using means,standard deviations and frequency counts. Bivariate associa-tions among illness beliefs were examined using Pearson'scorrelation coefficients.

Four regression models were used to test for full statisticalmediation and partial statistical mediation [19]. The firstmodel determined whether there was a significant associa-tion between illness beliefs and coping behavior. The secondregression model examined the association between copingbehavior and functional disability (physical and mentalhealth functioning). The third regression model evaluated theassociation between illness beliefs and physical as well aspsychosocial functioning. The final regression modeldetermined whether the statistical significance of illnessbelief components was eliminated or reduced when copingbehavior was controlled in a model predicting physical orpsychosocial functioning. A separate set of regressionmodels was used to evaluate each combination of indepen-dent, mediator and dependent variables. Recruitment sitewas always included as a covariate. The INDIRECT SPSSmacro [20] was utilized to assess the indirect effects withinthe multiple mediator models described above. This macroprovides regression coefficients in addition to bootstrapconfidence intervals for mediator models with more than onemediator or for models with statistical controls (as usedhere). Due to the number of models evaluated, a conservativealpha level of .01 was used to test for significance.

When a statistical mediation effect was not found,regression models were constructed to explore the potentialmoderating effect of the coping behavior on the relationshipbetween functional impairment and depression illnessbeliefs. Recruitment site was used as a covariate in eachmodel. Sequential regression modeling was utilized wherecoping behavior (moderator) and illness belief (predictor)components were included in the first step and the interactionbetween coping behavior and illness belief components wasincluded in the second step to investigate statisticalmoderation. Multicollinearity was also assessed. Again,due to the number of models evaluated, a conservative alphalevel of .01 was used to test for significance.

3. Results

3.1. Sample characteristics

Table 1 presents the participants' demographic character-istics. On average, the study participants were 45 years old;

the sample was predominantly white and female. More thanhalf of the participants were employed at least part time,were married or living with a partner and had an educationlevel greater than high school.

On average, participants reported two current medicalconditions, mild to moderate symptoms of depression, mildimpairment in physical functioning and significant impair-ment in psychosocial functioning. Approximately 20% of thesample had visited a mental health professional in thepreceding year.

According to the PRIME-MD interview, two thirds of theparticipants were diagnosed with major depression or majordepression plus dysthymia. Approximately 15% of thesample received no depression diagnosis on the PRIME-MD. The remaining diagnoses (19%) included partialremission of major depression, dysthymia and minordepression. Comorbid anxiety disorder or subsyndromalanxiety disorder was common in this sample, with 26–49%of patients endorsing an anxiety disorder or clinicallysignificant anxiety symptoms.

3.2. Description of personal illness models for depression

3.2.1. IdentityAs shown in Table 2, more than two thirds of the sample

reported experiencing mood disturbance (sadness or anhe-donia) frequently in the past month, while approximatelyhalf endorsed irritability. Fatigue and sleep disturbance werealso commonly reported, with at least 75% of participantsendorsing these symptoms. Few participants endorsedfrequent suicidal thoughts, but frequent feelings of hope-lessness, worthlessness, psychomotor retardation or agita-tion and difficulty concentrating were more common. Othertypes of co-occurring symptoms included anxiety, pain,gastrointestinal symptoms and shortness of breath or aracing heart.

3.2.2. Perceived cause of depressive symptomsThe five most commonly reported causes of depressive

symptoms are shown in Table 2. Participants endorsed

Table 2Illness cognition dimensions for depression

n %

Illness identity a

Sadness 73Loss of interest 64Irritability 51Fatigue 80Sleep disturbance 75Hopelessness 32Worthlessness 39Suicidal thoughts 10Psychomotor retardation or agitation 65Trouble concentrating 51Anxiety 45Pain (headaches, general aches and pains, stomachpain or chest pain)

46

Gastrointestinal symptoms (nausea, gas or constipation) 34Shortness of breath or racing heart 16

Perceived cause of depressive symptoms b

Stress 155 81Other people 88 46Marriage/relationship problems 79 41Hereditary 77 40Own behavior 74 39

Expected duration of depressive symptomsb

Chronic and intermittent 45 25Acute and intermittent 38 21Intermittent only 36 20Chronic only 30 16None of the above 30 16Acute only 4 2

Perceived controllability of depressive symptomsb

What I do can determine whether my symptoms getbetter or worse

156 82

Treatment will be effective in improving my symptoms 154 81My symptoms will improve in time 131 69There is a lot that I can do to control my symptoms 114 60Recovery from my symptoms depends on my faith in God 92 48Recovery from my symptoms depends on chance or fate 27 14There is very little that can be done to improve my symptoms 22 12

Perceived consequences of depressive symptomsb

My symptoms have strongly affected the way I see myselfas a person

142 74

My symptoms have had major consequences on my life 125 65My symptoms have strongly affected the way others see me 98 51My symptoms are part of a serious condition 94 49My symptoms have serious financial consequences 78 41My symptoms have become easier to live with 63 33My symptoms have not had much effect on my life 12 6

a Symptoms endorsed frequently in the past month.b Agreement or strong agreement with the statement.


multiple causes, with “stress” being the most commonlyreported. Most (97%) respondents identified at least twocauses of depressive symptoms. Relationship or interperso-nal issues were also commonly reported, as were “heredity”and “own behavior.”

3.2.3. Expected duration of depressive symptoms (timeline)The majority (66%) of participants characterized

depressive symptoms as fluctuating or cyclic. As shownin Table 2, this aspect of depression was commonly

endorsed even when symptoms were described as acuteor chronic in nature. Furthermore, a substantial minorityof patients (41%) described depressive symptoms aschronic, while only 23% of patients described symptomsas acute.

3.2.4. Perceived controllability of depressive symptomsPerceived controllability was high in this sample, with at

least 80% of participants indicating that their behaviordetermined whether symptoms got better or worse and thattreatment would be effective in improving depressivesymptoms. Similarly, approximately two thirds believedthat their symptoms would improve with time and that theycould control depressive symptoms. The participants weresplit in their belief that recovery from depressive symptomsdepended on faith in God, while very few participantsbelieved that recovery depended on chance or fate or thatlittle could be done to improve symptoms.

3.2.5. Perceived consequences of depressive symptomsMost participants viewed depressive symptoms as having

significant negative consequences on their lives, while veryfew indicated that symptoms had little impact on their lives(Table 2). However, only about half viewed depression as aserious condition. Approximately one third of the respon-dents reported that depressive symptoms had significantfinancial consequences, and more than half indicated that itaffected the way they viewed themselves as well as the wayothers viewed them. Despite the negative impact ofdepressive symptoms, a number of participants reportedthat living with depressive symptoms had become easier.

3.3. Intercorrelations among illness beliefs

Significant low to moderate correlations were foundamong illness beliefs (Table 3). Thus, endorsement offrequent symptoms (identity) was negatively associated withperceived control and positively associated with perceivedconsequences of depression and the belief that depression iscaused by stress or interpersonal relationships. Greaterperceived control was associated with fewer perceivedconsequences, less illness chronicity and lower endorsementof beliefs that depression was caused by the environment orchance or medical illness. Greater perceived consequenceswere associated with greater perceived chronicity ofdepression and the belief that depression is caused bymedical illness, self-care or stress. Greater perceivedchronicity of depression was associated with endorsementof heredity or death as a cause of depression.

3.4. Mediation effects

Coping behavior did not mediate the relationship betweenillness beliefs and physical functioning. However, whenpredicting mental health functioning, illness beliefs aremediated by coping behavior. Of the 14 coping subscalesevaluated, venting and self-blame were found to have asignificant mediating effect (Table 4). Venting and self-blame

Table 3Intercorrelations among illness belief subscales ⁎

Illness cognition Identity Perceivedcontrollability

Consequences Timeline Environment/Chance

Heredity/Death

Medicalillness

Self-care

Stress/Interpersonalrelationships

IdentityPerceived controllability −.16 ⁎Consequences .38 ⁎⁎ −.18 ⁎Timeline .14 −.29 ⁎⁎ .21 ⁎⁎

CauseEnvironment/Chance −.04 −.16 ⁎ −.07 −.04Heredity/Death .11 −.13 .07 .17 ⁎ .14Medical illness .06 −.17 ⁎ .24 ⁎⁎ .11 .15 ⁎ .19 ⁎⁎

Self-care .06 .08 .16 ⁎ .12 .03 .13 .06Stress/Interpersonal relationships .27 ⁎⁎ −.08 .22 ⁎⁎ .05 −.01 .28 ⁎⁎ −.07 .19 ⁎⁎

⁎ Pb .05 level; ⁎⁎ Pb .01.


both statistically mediated the relationship between stress andinterpersonal problems as a perceived cause of depressivesymptoms and psychosocial functioning. Thus, stress andinterpersonal problems as perceived causes of depressivesymptoms are associated with psychosocial functioningbecause of their effect on venting and self-blame. Forexample, if individuals believe that their depressive symp-toms are caused by stress or interpersonal problems, they aremore likely to vent or blame themselves and are thereforelikely to exhibit poorer psychosocial functioning.

Behavioral disengagement and self-blame partiallymediated the relationship between illness beliefs andpsychosocial functioning (Table 4). Both behavioral disen-gagement and self-blame partially mediated the relationshipbetween perceived duration of depressive symptoms (time-line) and psychosocial functioning. Similarly, behavioraldisengagement partially mediated the relationship between

Table 4Illness beliefs, coping behaviors and psychosocial functioning: mediation effects

Mediation effect

FullCause: stress/interpersonal to venting (a)Direct effects of venting on psychosocial functioning (b)Total effect of cause: stress/interpersonal on psychosocial functioning (c)Direct effect of cause: stress/interpersonal on psychosocial functioning (c′)Cause: stress/interpersonal to self-blame (a)Direct effects of self-blame on psychosocial functioning (b)Total effect of cause: stress/interpersonal on psychosocial functioning (c)Direct effect of cause: stress/interpersonal on psychosocial functioning (c′)PartialControl to behavioral disengagement (a)Direct effects of behavioral disengagement on psychosocial functioning (b)Total effect of control on psychosocial functioning (c)Direct effect of control on psychosocial functioning (c′)Timeline to behavioral disengagement (a)Direct effects of behavioral disengagement on psychosocial functioning (b)Total effect of timeline on psychosocial functioning (c)Direct effect of timeline on psychosocial functioning (c′)Timeline to self-blame (a)Direct effect of self-blame on psychosocial functioning (b)Total effect of timeline on psychosocial functioning (c)Direct effect of timeline on psychosocial functioning (c′)

perceived control of depressive symptoms and psychosocialfunctioning. Thus, if individuals believe that their depres-sive symptoms will last longer, they are more likely todisengage from meaningful behaviors and have lowerpsychosocial functioning. Similarly, if individuals believethat their depressive symptoms are more chronic, they arealso more likely to blame themselves and therefore toreport lower psychosocial functioning. Also, if individualsbelieve that they have control of their depressivesymptoms, they are less likely to disengage from mean-ingful behaviors and are therefore more likely to reporthigher psychosocial functioning.

3.5. Moderation effects

Several moderating effects of coping behaviors on therelationship between illness beliefs and psychosocialfunctioning were found (Table 5). The coping behaviors

B SEB P

0.4931 0.1583 .0021−1.3697 0.4297 .0017−2.1355 0.9500 .0258−1.4602 0.9514 .12650.4918 0.1809 .0072

−1.9115 0.3595 .0000−2.1355 0.9500 .0258−1.1955 0.9047 .1880

−0.6651 0.1851 .0004−2.3459 0.4864 .00003.7923 1.2996 .00402.2319 1.2700 .08050.7275 0.1706 .0000

−2.1936 0.4885 .0000−4.6184 1.1937 .0002−3.0225 1.1908 .01201.0560 0.2248 .0000

−1.7233 0.3693 .0000−4.6184 1.1937 .0002−2.7986 1.1974 .0205

Table 5Illness beliefs, coping behaviors and psychosocial functioning: significant moderation effects

B SEB β R2 ΔR2

Predicting psychosocial functioningStep 1 0.044 ⁎

Site Variable 1 0.088 1.966 0.003Site Variable 2 7.179 2.437 0.211 ⁎

Step 2 0.051 0.007Cause: Environment/Chance 1.523 1.185 0.091Acceptance 0.014 0.546 0.002Step 3 0.094 ⁎ 0.043Cause: Environment/Chance×Acceptance −2.562 0.866 −0.209 ⁎



Step 2 0.180 ⁎ 0.136Cause: Medical Illness 0.519 0.879 0.040Behavioral Disengagement −2.755 0.471 −0.396 ⁎Step 3 0.210 ⁎ 0.030Cause: Medical Illness×Behavioral Disengagement 1.274 0.485 0.179 ⁎



Step 2 0.063 0.019Cause: Medical Illness 0.718 0.926 0.055Religious 0.614 0.340 0.127Step 3 0.099 ⁎ 0.036Cause: Medical Illness×Religious −1.146 0.423 −0.192 ⁎

⁎ Pb .01.


included acceptance, behavioral disengagement and religiouscoping. Acceptance moderated the relationship betweenenvironment/chance as a perceived cause of depression andpsychosocial functioning. Among those who stronglybelieved that depressive symptoms were caused by externalfactors (i.e., environment/chance), greater acceptance ofdepressive symptoms was associated with poorer psychoso-cial functioning, while lower acceptance was associated withhigher psychosocial functioning. Behavioral disengagementand religious coping bothmoderated the relationship betweenmedical illness as a perceived cause of depression andpsychosocial functioning. Among those who stronglyendorsed medical illness as a cause of depression, greateruse of behavioral disengagement was associated with poorerpsychosocial functioning, while less behavioral disengage-ment was associated with higher psychosocial functioning.Finally, greater use of religious coping among participantswho strongly endorsed medical illness as a cause ofdepression was associated with poorer psychosocial func-tioning, while less religious coping was associated withhigher psychosocial functioning.

4. Discussion

Although many studies have examined illness beliefs inphysical conditions, to date, few have examined personalillness models in depressed persons. This investigation

presents data on primary care patients' beliefs aboutdepression and the nature of the relationships betweenstrategies used to cope with depression, depression beliefsand physical and psychosocial functioning.

4.1. Illness beliefs about depression

Primary care patients most commonly endorsed multiplecauses of depression, with “stress” and “interpersonalconflict” being the most common causes. While patientsvaried in terms of how long they expected their symptoms tolast, the majority of the patients characterized theirsymptoms as cyclic or fluctuating in nature. Most patientsbelieved that their symptoms were controllable eitherthrough their own behavior or through treatment, which isnot surprising given that all study patients had agreed to takeantidepressant medication prescribed by their primary carephysician. While most participants also indicated thatdepressive symptoms had a negative impact on their lives,interestingly, only about half of the sample vieweddepression as a serious condition.

4.1.1. Clinical implicationsThese beliefs about depression can be used to inform

psychoeducational efforts with primary care patients. Forexample, patients accurately describe the cyclical natureof depression, treatability of depression and negativeconsequences of depression. Educational efforts might


focus on the complex causes of depression, that depression isa serious illness that should be treated and that although itmay appear that depressive symptoms may be easier to livewith over time, the goal of treatment is full recovery.

4.2. Intercorrelations among illness beliefs

The intercorrelations among illness cognition dimensionswere consistent with those reported in studies of physicalillness in which associations are significant but not of themagnitude that would indicate conceptual overlap [5]. Aspredicted by the model, illness beliefs were significantlyassociated with functional outcomes. This is consistent withprior research on medical conditions and extends them to acommon psychiatric disorder — depression.

4.3. Mediation and moderation of illness beliefs bycoping behaviors

Our findings provide partial support for Leventhal et al.'smediational model. We identified full or partial mediatingrelationships between illness beliefs, coping behaviors andpsychosocial functioning. We did not find support forLeventhal et al.'s mediational model when examiningphysical functioning. Key illness beliefs associated withpsychosocial functioning include perceived control ofdepressive symptoms, perceived duration of depressivesymptoms and perceived cause of depressive symptoms(i.e., stress/interpersonal, environment/chance, medical ill-ness). Coping behaviors that mediated or moderated therelationship between illness beliefs and psychosocial func-tioning include behavioral disengagement, self-blame, vent-ing, acceptance and religious coping. Due to the number ofregression analyses conducted to examine both mediationaland moderating effects, our findings must be considered to bepreliminary and warrant replication in future studies.

In relation to psychosocial functioning, beliefs aboutstress as a cause of depression, the controllability ofdepressive symptoms and expected duration of symptomswere either fully or partially mediated by coping behaviors.In terms of coping behaviors, behavioral disengagement,self-blame and venting were key coping behaviors that eitherfully or partially mediated the relationship between illnessbeliefs and psychosocial functioning.

4.3.1. Clinical implicationsThese findings identify key illness beliefs and coping

behaviors that can be the target of clinical interventions toimprove psychosocial functioning in depressed patients. Ithas been documented that psychosocial impairment is animportant feature of this illness [21] and that functionalimprovement among treated individuals is slower thansymptomatic improvement [22]. Interventions designed toimprove psychosocial functioning in depressed primarycare patients might benefit by specifically focusing onstrategies to help increase perceived control of depressionand decrease behavioral disengagement, venting and self-

blame in depressed patients. Psychoeducational strategiesmight focus on educating depressed persons about theexpected duration of depression and the importance oftreatment in attaining recovery from a specific episode andin preventing future episodes. Specific focus on thesebeliefs and coping behaviors may readily be incorporatedinto existing interventions and may lead to more rapidfunctional improvement during depression treatment.

4.4. Moderating effects of coping behaviors onillness beliefs

Leventhal et al.'s mediational model was not supportedwith respect to physical functioning in depressed patients.However, exploratory analyses indicated that perceivedcause of depression is an important belief whose relationshipto psychosocial functioning is moderated by copingbehaviors such as acceptance, behavioral disengagementand religious coping.

4.4.1. Clinical implicationsPatients' beliefs about the cause of their depressive

symptoms are important determinants of how they managethese symptoms. Our findings suggest that specific copingstrategies (e.g., acceptance) can have a different impact onfunctioning, depending on the perceived cause of depression.Thus, for the patient who believes that depressive symptomsare caused by factors outside of his or her control (i.e.,environment/chance), acceptance of these symptoms leads togreater functional impairment.

It is not uncommon for primary care patients to endorsemedical illness as a cause of depression. Our findingssuggest that when patients endorse this belief, their copingstrategies have an important influence on their psychosocialfunctioning. Thus, increased withdrawal or increasedreligious coping is associated with poorer functioning.Clinicians might inquire about patients' perceived causesof depression. Psychoeducation can be used to provideaccurate information, and behavioral activation strategiesmight help patients decrease behavioral withdrawal. Manypatients in our sample used religious coping to managedepressive symptoms. As noted by Organista and Munoz[23], religiosity can be viewed as a cognition-related domainin which clinicians can work with patients. In their work,churchgoing and praying were reinforced as behavioral andcognitive activities that help patients cope with stress andnegative mood states. However, they recommended thatforms of prayer that serve to decrease active problem solvingbe explored and challenged by the clinician.

In summary, our findings indicate that Leventhal et al.'smodel allows for the identification of potentially modifiablebeliefs and coping behaviors that can be targeted to improvepsychosocial functioning in depression interventions. Ourfindings can be generalized only to primary care patients whoagree to be treated for depression and who are similar to ourstudy patients in demographic and clinical characteristics. Itwill be important to replicate these findings in patient


populations that are more diverse in terms of race/ethnicity,age, socioeconomic status, clinical severity and treatmenthistory (e.g., no prior treatment and treatment withpsychotherapy only) as these are factors that may stronglyinfluence illness beliefs. Future studies should also examinethe relationships between illness beliefs, coping behaviorsand other outcomes such as medication adherence, treat-ment seeking and treatment initiation. This model can beuseful in designing or refining interventions designed toimprove functional outcomes by targeting specific beliefsand coping strategies.

Acknowledgments

We thank the physicians and medical staff of theRenaissance Family Practice, the University of Pittsburgh–St. Margaret's Family Practice Residency Program, theVeterans Health Administration Primary Care Clinics andDr. Amy Kilbourne for their assistance in recruiting studyparticipants. We also thank Sarah Stephenson, M.S.W.,L.S.W., Nicole Ryan, B.A., and Janine Smith, J.D., for theirtechnical assistance.

References

[1] Mechanic D. Sociological dimensions of illness behavior. Soc Sci Med1995;41(9):1207–16.

[2] Pescosolido BA, Boyer CA. How do people come to use mental healthservices? Current knowledge and changing perspectives. In: HorwitzAV, Scheid TL, editors. A handbook for the study of mental health:social contexts, theories, and systems. Cambridge: CambridgeUniversity Press; 1999. p. 392–411.

[3] Leventhal H, Diefenbach M, Leventhal EA. Illness cognition: usingcommon sense to understand treatment adherence and affect cognitioninteractions. Cogn Ther Res 1992;16(2):143–63.

[4] Leventhal HE, Brissette I, Leventhal EA. The common-sense model ofself-regulation of health and illness. In: Cameron LD, Leventhal H,editors. The self-regulation of health and illness behaviour. New York:Routledge; 2003. p. 42–65.

[5] Hagger MS, Orbell S. A meta-analytic review of the common-sensemodel of illness representations. Psychol Health 2003;18(2):141–84.

[6] Brewer NT, Chapman GB, Brownlee S, Leventhal EA. Cholesterolcontrol, medication adherence and illness cognition. Br J HealthPsychol 2002;7:433–47.

[7] Helder DI, Kaptein AA, Kempen GMJ, Weinman J, van HouwelingenHC, Roos RAC. Living with Huntington's disease: illness perceptions,coping mechanisms, and patient's well-being. Br J Health Psychol2002;7:449–62.

[8] Spitzer RL, Williams JB, Kroenke K, Linzer M, deGruy III FV,Hahn SR, et al. Utility of a new procedure for diagnosing mentaldisorders in primary care: the PRIME-MD 1000 study. J Am MedAssoc 1994;272:1749–56.

[9] Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventoryfor measuring depression. Arch Gen Psychiatry 1961;4:561–71.

[10] Beck AT, Steer RA, Garbin MG. Psychometric properties of the BeckDepression Inventory: twenty-five years of evaluation. Clin PsycholRev 1988;8:77–100.

[11] Stewart AL, Hays RD, Ware Jr JE. The MOS Short-Form GeneralHealth Survey. Med Care 1988;26:724–35.

[12] Ware Jr JE, Kosinski M, Keller SD. SF-36 physical and mental healthsummary scales: a user's manual. Boston (Mass): The Health Institute,New England Medical Center; 1994.

[13] Wells KB, Stewart A, Hays RD, Burnam MA, Rogers W, Daniels M,et al. The functioning and well-being of depressed patients: resultsfrom theMedical Outcomes Study. JAmMedAssoc 1989;262(7):914–9.

[14] Weinman J, Petrie KJ, Moss-Morris R, Horne R. The Illness PerceptionQuestionnaire: a new method for assessing the cognitive representationof illness. Psychol Health 1996;11:431–45.

[15] Brown C, Dunbar-Jacob J, Palenchar DR, Kelleher KJ, BruehlmanRD, Sereika S, et al. Primary care patients' personal illness models fordepression: a preliminary investigation. Fam Pract 2001;18(3):314–20.

[16] Gorsuch R. Factor analysis. 2nd ed. Hillsdale (NJ): Erlbaum; 1983.[17] Carver C. You want to measure coping but your protocol's too long:

consider the Brief COPE. Int J Behav Med 1997;4:92–100.[18] Carver CS, Scheier MF, Weintraub JK. Assessing coping strategies:

a theoretically based approach. J Person Social Psychol 1989;56(2):267–83.

[19] Baron RM, Kenny DA. The moderator–mediator distinction in socialpsychological research: conceptual, strategic, and statistical considera-tions. J Person Social Psychol 1986;51:1173–82.

[20] Preacher K, Hayes A. Asymptotic and resampling strategies forassessing and comparing indirect effects in simple and multiplemediator models. [under review].

[21] Keller MB. Past, present, and future directions for defining optimaltreatment outcome in depression. JAMA 2003;289(23):3152–60.

[22] Hirschfeld RMA, Dunner DL, Keitner G, Klein DN, Koran LM,Kornstein SG, et al. Does psychosocial functioning improveindependently of depressive symptoms? A comparison of nefazo-done, psychotherapy and their combination. Biol Psychiatry 2002;51:123–33.

[23] Organista KC, Munoz RF. Cognitive behavioral therapy with Latinos.Cognitive and behavioral practice: association for the advancement ofbehavior therapy 1996;3(2):255–70.

primary care patients' personal illness models for depression: relationship to coping behavior...

Documents