Illness Perceptions, Adjustment to Illness, and Depression in a Palliative Care Population

Vol. 43 No. 5 May 2012 Journal of Pain and Symptom Management 819

Original Article

Illness Perceptions, Adjustment to Illness,and Depression in a Palliative Care PopulationAnnabel Price, MBBS, BSc, MRCPsych, Laura Goodwin, BSc, MSc, PhD,Lauren Rayner, BSc, Emma Shaw, MBBS, Penny Hansford, RN, RM, HV, MSc,Nigel Sykes, MA, BM, BCh, FRCGP, FRCP, Barbara Monroe, BA, BPhil, CQSW,Irene Higginson, BMedSci, BMBS, PhD, FFPHM, FRCP, Matthew Hotopf, MBBS,BSc, MSc, PhD, MRCPsych, and William Lee, MBChB, BSc, MSc, MRCPsychDepartment of Psychological Medicine (A.P., L.G., L.R., M.H., W.L.), Institute of Psychiatry, King’s

College London; Department of Palliative Care, Policy and Rehabilitation (L.R., I.H.), Cicely

Saunders Institute, King’s College London; and School of Medicine (E.S.), King’s College London;

and St. Christopher’s Hospice (P.H., N.S., B.M.), London, United Kingdom

Abstract
Context. Representations of illness have been studied in several populations,
but research is limited in palliative care.Objectives. To describe illness representations in a population with advanced

disease receiving palliative care and to examine the relationship between illnessperceptions, adaptive coping, and depression.

Methods. A cross-sectional survey of 301 consecutive eligible patients recruitedfrom a palliative care service in south London, U.K. Measures used included theBrief Illness Perception Questionnaire (Brief IPQ), the Mental Adjustment toCancer (MAC) Scale, and the Primary Care Evaluation of Mental DisordersPatient Health Questionnaire-9.

Results. Scores were not normally distributed for most questions on the BriefIPQ. The correlations found between items on the Brief IPQ were understandablein the context of advanced disease. MAC helplessness-hopelessness and fightingspirit were highly correlated with items on the Brief IPQ in opposite directions.The Brief IPQ domains of consequences, identity, concern, personal control, andemotion were associated with depression, a relationship that was not explained byadaptive coping. Seven causal attribution themes were identified: don’t know,personal responsibility, exposure, pathological process, intrinsic personal factors,chance, fate or luck, and other. Both lung cancer diagnosis and gender werefound to be independently associated with personal responsibility attribution.None of the attribution themes were associated with the presence of depression.

Conclusion. Assessment of illness perceptions in palliative care is likely to yieldimportant information about risk of depression andwill help clinicians topersonalizemanagement of advanced disease. J Pain SymptomManage 2012;43:819e832.� 2012 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved.

Address correspondence to: Annabel Price, MBBS, BSc,MRCPsych, Department of Psychological Medicine,Weston Education Centre, 10 Cutcombe Road,

London SE5 9RJ, United Kingdom. E-mail:[email protected]

Accepted for publication: May 18, 2011.

� 2012 U.S. Cancer Pain Relief CommitteePublished by Elsevier Inc. All rights reserved.

0885-3924/$ - see front matterdoi:10.1016/j.jpainsymman.2011.05.013

mailto:[email protected]

http://dx.doi.org/10.1016/j.jpainsymman.2011.05.013



820 Vol. 43 No. 5 May 2012Price et al.

Key Words
Illness perceptions, illness representations, palliative, depression, adjustment, attribution
IntroductionRepresentations of illness as described in

Leventhal et al.’s Common Sense Model(CSM)1 are the perceptions people form in re-sponse to health threat information. Accord-ing to the CSM, stimuli, such as physicalsymptoms and existing health beliefs, generatecognitive and emotional representations of theillness or health threat.2 The representationsare processed by the individual and result inadoption of behaviors to cope with the healththreat, which are then reappraised in a feed-back loop, with the individual then adjustingtheir illness representations accordingly.

Following qualitative research, Leventhalet al.1 described five dimensions of the cogni-tive representation of illness: identitydthe labelthe patient uses to describe the illness andsymptoms he or she views as being part ofthe disease; consequencesdthe expected effectsand outcome of the illness; causedideas aboutthe cause of illness; timelinedhow long the in-dividual believes the illness will continue; andcure/controldthe extent to which the individualbelieves he or she can recover from or controlthe illness.

Much of the research examining illness per-ceptions has looked at groups with chronic dis-eases, showing associations between illnessperceptions and adherence to treatment,3

functional recovery,4 self-management behav-iors,5 and quality of life.6 Interventions de-signed to change illness perception toinfluence treatment outcomes also are emerg-ing.7 Some studies have been conducted inpopulations with more advanced illnesses, in-cluding chronic obstructive pulmonary dis-ease8 and end-stage renal disease9 but to ourknowledge, there have been none examiningillness perceptions in patients receiving pallia-tive care.10

Attribution theory is concerned with theways in which people explain the behavior ofthemselves or others and attribute causes toevents.11 The components of attribution are lo-cus: either internal (dispositional or personal)or external (situational); stability: whether the

cause is likely to stay the same or change inthe future; and controllability: whether the per-son has/had control over the cause. There isan extensive research literature on causal attri-bution of disease in different populations in-cluding healthy participants,12,13 the patient’srelatives,14 acute life-threatening illness,15e17

chronic non-life-threatening illness,18 andchronic life-threatening illness, including re-nal disease19 and cancer.20e24 A meta-analyticreview of 27 studies25 investigating causal attri-butions and their influence on coping and psy-chological adjustment to various conditions,including cancer, spinal cord injury, andfertility, found that internal, unstable, and con-trollable attributions facilitated positive psy-chological coping, that is, actively copingwith a situation rather than avoiding it and fo-cusing on dealing with the distressing emo-tions produced by the situation rather thantrying to change the situation itself. In addi-tion, stable and uncontrollable attributionswere associated with avoidance coping andwere, therefore, indirectly associated with neg-ative psychological adjustment. The authorssuggested that although attributions do not ac-count for all the variance in coping and psy-chological adjustment to illness, they do playa role in influencing cognitions and behaviorsin the context of illness and are related to cer-tain coping strategies. In a review of illness at-tribution and adjustment to life-threateningillness, Turnquist et al.26 found equivocal evi-dence for an association between attributionsand psychological adjustment, although re-porting an attribution of any type tended to re-sult in a more positive physical or emotionaloutcome.There has been only one study looking at

causal attribution of illness in a populationwith terminal disease.27 This study comparedbeliefs about the causes of cancer in 120 late-stage cancer patients with patients withchronic disease. The terminal cancer patientshad less firm convictions about factors they be-lieved caused their illness than the chronic ill-ness group and tended to attribute their illness

Vol. 43 No. 5 May 2012 821Illness Perceptions in Palliative Care

to genetics or God’s will, whereas the chronicillness group tended to cite diet or the envi-ronment. The authors suggested that patientswith advanced cancer tend to defend them-selves against self-blame as a means of copingwith their illness. Although several studieshave looked at associations between attribu-tion of illness and adjustment to illness, fewhave looked at the relationship between causalattribution and clinical outcome, including de-pression, with none using standardized criteriafor measuring major depression.

Coping styles in cancer have been shown tobe correlated with psychological outcomes,such as anxiety and depression. In particular,‘‘fighting spirit,’’ as measured by the MentalAdjustment to Cancer (MAC) Scale,28 is corre-lated with lower levels of anxiety and depres-sion. In advanced cancer, the utility of copingstyles is less clear-cut, and methodological lim-itations in studies investigating the relation-ship between coping and adjustment toadvanced cancer mean that it is difficult todraw firm conclusions about what constitutesadaptive coping.29

In this study, we describe illness representa-tions in a population with advanced disease re-ceiving palliative care using the Brief IllnessPerception Questionnaire (Brief IPQ)2 and ex-amine the relationship between illness percep-tions (including causal attribution), adaptivecoping, and depression. In particular, we hy-pothesize that those who perceive greater con-sequences, identity (symptoms), concern, andemotional effects and those who perceivelower control and understanding of their ill-ness are more likely to be depressed; that thesefactors also will be associated with poorer adap-tive coping; and also that those who have ‘‘con-trollable’’ illness attributions, for example,smoking, will be more likely to be depressed.

MethodsStudy Design

This was a cross-sectional study of patientsreceiving palliative care who were newly re-ferred to a hospice.30

SampleAlthough there have been a number of

studies addressing the issue of prevalence of

depression in palliative populations, the rangevaries widely (5%e23%) depending on themethod used to detect depression;31 also,with no previous studies looking at illness per-ceptions in patients with advanced disease,there was little to inform a calculation of thenumbers required to detect a difference in ill-ness perception or coping between the de-pressed and nondepressed groups. The goal,therefore, was to recruit 300 patients to thestudy, which would compare favorably with re-cruitment by other studies within palliativepopulations.

Patients were recruited from St. Christo-pher’s Hospice, Sydenham, a large hospice insoutheast London, with both a home careand an inpatient service serving a large popula-tion across five boroughs. In the U.K. and Re-public of Ireland, hospice services providepalliative care to the local community along-side hospital-based palliative care services.They typically provide medical and nursingcare in conjunction with rehabilitation, a rangeof therapies, access to spiritual support, practi-cal and financial advice, and bereavement carefor people with advanced disease and theirfamilies.

Recruitment commenced in January 2007,and all new patients accepted by the hospicewere eligible for inclusion. Patients were ex-cluded if they lacked capacity to consent, hadvery poor functional status (scoring 4 onthe Eastern Cooperative Oncology Group[ECOG] Scale),32 were too ill to participate,or had sensory impairment or language diffi-culties such that they would not be able to par-ticipate in the research interview.

Potential participants were initially assessedby a clinical nurse specialist who determinedwhether they were eligible for recruitment. Arandom sample of the eligible patients wascontacted by telephone by study personnel toask if they would be willing to participate. Aface-to-face interview was then arranged byone of three researchers, usually at the pa-tient’s home, within a week of recruitment. Re-cruitment was completed in August 2008.

ProcedureAfter gaining written consent to participate,

the interviewer administered the AbbreviatedMental Test Score.33 The interview was termi-nated for those who scored below six of 10,


indicating cognitive impairment. The inter-viewer then administered several question-naires including those described below. Theinterview lasted approximately one hour.

MeasuresIllness Perception. The Brief IPQ is a nine-itemquestionnaire designed for use across illnesspopulations, with the option to adapt questionwording to the specific illness condition.2 Thefirst eight questions are answered on a 0e10interval scale, with the scale for each questionhaving a different meaning; for example, ill-ness consequences were assessed using thequestion ‘‘How much does your illness affectyour life?,’’ with the scale ranging from 0/10‘‘no effect at all’’ to 10/10 ‘‘severely affectsmy life.’’ Answers to the ninth item, addressingcause, ‘‘Please list in rank-order the three mostimportant factors that you believe caused yourillness,’’ were recorded as free text by theinterviewer.

Adjustment to Illness. This was assessed usingthe MAC Scale, a 40-item self-rated scale usinga four-point response category ranging from(1) ‘‘definitely does not apply to me’’ to (4)‘‘definitely does apply to me.’’ Respondentswithout a cancer diagnosis were asked to com-plete the MAC Scale, but the word ‘‘illness’’ wassubstituted for cancer where appropriate.

Depression. Presence of depression was as-sessed with the Primary Care Evaluation ofMental Disorders (PRIME-MD�) PatientHealth Questionnaire (PHQ)-934 used to de-tect the presence of psychiatric disorder ingeneral populations, based on the Diagnosticand Statistical Manual of Mental Disorders, 4thEdition diagnostic criteria for depression. Foreach symptom item, the respondent is askedto indicate whether they have experienced itnot at all, for several days, for more than halfthe days, or nearly every day in the last twoweeks. Participants were assessed for the pres-ence of ‘‘any depressive syndrome,’’ definedas the presence of low mood or loss of interestand at least two symptoms in total for morethan half the days in the preceding two weeks.The presence of any suicidal thoughts countedtoward the diagnosis at any frequency.

Additional Data. Demographic data includingage, sex, ethnicity, religion, occupation, andmarital status were recorded from the patientnotes. Data also were gathered on diagnosis,severity of illness (for those with a cancer diag-nosis), and functional status using the ECOGscore.

AnalysisQuantitative Data. Data management andanalysis were conducted using Stata v10 (Stata-Corp LP, College Station, TX).35 Level of statis-tical dependence between items on the BriefIPQ with each other and between items onthe Brief IPQ and the MAC Scale weredetermined by calculating Spearman’s rankcorrelation coefficient because data were non-parametric. Logistic regression was used to ex-amine associations between illness perceptionitems and depression. The Brief IPQ itemswere converted to categorical variables by di-viding scores into quartiles, with the lowestand highest quartiles being compared withthe mid-quartiles (reference category). To ex-amine whether the relationship betweenillness perceptions and depression was ac-counted for in whole or in part by adjustmentto illness, we adjusted for those domains onthe MAC Scale that were independently associ-ated with ‘‘any depressive syndrome’’ in oursample.

Qualitative Data. Data management and cod-ing were conducted using Microsoft Excel2003 (Microsoft Corp., Redmond, WA), andquantitative analysis of the qualitative datawas conducted using Stata v10. Patient re-sponses regarding cause of illness were en-tered into the database. Two independentresearchers assigned codes to these answersand grouped them into subthemes. The sub-themes generated were discussed and finalsubthemes agreed on. The subthemes werethen grouped by agreement into main themes.Coding, subthemes, and themes were reviewedand discussed with a third researcher. Codingand theme generation were atheoretical andgenerated by best fit with the data by agree-ment between researchers after independentanalysis.Associations between qualitative themes and

demographic and clinical factors were ex-plored quantitatively by converting themes


into binary variables, for example, personal re-sponsibility or not, and using Chi-squared andMann-Whitney U tests to determine associa-tions. This was done only for the first answergiven of a possible three. Further analysisused logistic regression to explore possibleconfounding factors.

ResultsSample Characteristics

During the study period, 743 patients wereeligible for inclusion. Forty percent of eligiblepatients entered the study. Of the 443 patientswho were eligible but not interviewed, 278 de-clined an interview, 83 were not contactable,the patient’s clinician declined for 44, the fam-ily declined for 13, and 25 were not interviewedfor other reasons.30 Three hundred one partic-ipants were interviewed, 300 of whom hadcompleted the PRIME-MD� PHQ-9. Just under

Fig. 1. Distribution of scores for the items in the Brief

half were female (49.3%), and the mean agewas 68.5 years. Most participants were of whiteethnicity, and the distribution across occupa-tional groups was fairly even. The most com-mon diagnosis was lung cancer (27%), andmost (59%) had metastatic disease; 8.3% ofthe sample had nonmalignant disease. Mostparticipants were restricted in physical activitybut capable of self-care (80% ECOG 1 or 2).One hundred nine participants (36.3%) metthe criteria for ‘‘any depressive syndrome,’’which includes both major depressive disorderand minor depression.

Distribution of Brief IPQ ScoresFig. 1 shows the distribution of scores for the

first eight domains of the Brief IPQ (scored onan 11-point scale). Scores were not normallydistributed for most questions. The questionsconcerning consequences, timeline, treatmentcontrol, and illness comprehensibility in par-ticular were skewed toward one extreme, with

Illness Perception Questionnaire (Brief IPQ).


median scores of 8 (interquartile range [IQR]5e10), 10 (IQR 8e10), 8 (IQR 5e10), and 8(IQR 6e10), respectively.

Correlation Between Domains of the Brief IPQTable 1 shows how the quantitative domains

of the IPQ correlated with each other in thissample. Positive correlations were shown be-tween consequences and identity, concernand emotion, identity with concern and emo-tion, personal control with treatment controland illness understandability, and betweenconcern and emotion. Negative correlationswere shown between consequences and per-sonal control and timeline and treatmentcontrol.

Correlation Between the Brief IPQ and theMAC Scale

Table 2 shows how the domains of the MACScale correlate with the quantitative domainsof the Brief IPQ. For MAC helplessness-hopelessness, positive correlations were shownwith consequences, timeline, identity, concern,and emotion. Negative correlations wereshown with personal control, treatment con-trol, and illness understandability. For MACfighting spirit, positive correlations wereshown with personal control and treatmentcontrol, and negative correlations with conse-quences, timeline, identity, and emotion. ForMAC anxious preoccupation, positive correla-tions were shown with concern and emotion.For avoidance, a positive correlation wasshown with personal control and a negativecorrelation with consequences. For MAC fatal-ism, a positive correlation was shown with time-line and negative correlations with personalcontrol and treatment control.

Table 1Correlation Matrix for All the Brief

Consequences TimelinePersonalControl

TreCo

Consequences 1.00Timeline 0.10 1.00Personal control �0.24a �0.10 1.00Treatment control �0.09 �0.24a 0.31a

Identity 0.47a 0.06 �0.07Concern 0.35a �0.04 �0.10Illness understandability 0.04 0.07 0.17a

Emotion 0.44a 0.0005 �0.07 �aP< 0.05.

Illness Perceptions and Risk of DepressionTable 3 shows the associations between

items on the Brief IPQ and depression. Thosewho scored in the lowest quartile for conse-quences, identity, concern, and emotion hada significantly lower frequency of depressionthan those with mid-quartile scores, as didthose who scored in the highest quartile forpersonal control. Those who scored in thehighest quartile for consequences, concernand emotion had significantly greater risk ofdepression than those with mid-quartilescores. The frequency of depression was great-est for those scoring in the highest quartile foremotion (odds ratio [OR] 4.06, 95% confi-dence interval [CI] 1.98, 8.31). Timeline,treatment control, and illness comprehensibil-ity were not associated with depression in oursample.Functional status and disease status were

found to be independently associated with de-pression in our sample and were, therefore,adjusted for in the analysis. Adjustment forthese factors in the analysis did not changethe findings.Items on the MAC Scale that were inde-

pendently associated with depression werehelplessness-hopelessness, anxious preoccupa-tion (both positively associated), and fightingspirit (negatively associated). Fatalism was notassociated with ‘‘any depressive syndrome’’ inour sample. The associations between conse-quences of illness, personal control, identity,emotions, and depression remained after ad-justing for adjustment to illness. The asso-ciation between concern and depression,however, was no longer significant after adjust-ment for helplessness-hopelessness or anxiouspreoccupation.

IPQ Items With Each Other

atmentntrol Identity Concern

IllnessUnderstandability Emotions

1.000.05 1.000.004 0.30a 1.000.11 0.04 �0.12a 1.000.004 0.35a 0.63a �0.13 1.00

Table 2Correlation Matrix of Brief IPQ vs. MAC Scale Domains

MAC HH MAC FS MAC AP MAC AV MAC FA

Consequences 0.28a �0.26a 0.13 �0.17a 0.05Timeline 0.16a �0.23a �0.13 �0.13 0.19a

Personal control �0.31a 0.28a 0.03 0.15a �0.29a

Treatment control �0.35a 0.42a �0.06 �0.04 �0.20a

Identity 0.26a �0.18a 0.06 �0.07 0.01Concern 0.25a �0.07 0.49a 0.02 0.08Illness understandability �0.22a 0.11 �0.14 �0.08 �0.09Emotion 0.30a �0.16a 0.48a �0.04 0.09

HH¼ helplessness-hopelessness; FS¼ fighting spirit; AP¼ anxious preoccupation; AV¼ avoidance; FA¼ fatalism.aP< 0.05.


Perceived Cause of IllnessResponses to the final question on the Brief

IPQ, ‘‘Please list in rank-order the three mostimportant factors that you believe causedyour illness,’’ asked to determine representa-tions of illness cause, were analyzed qualita-tively. All respondents (n¼ 301) gave at leastone response, 63 gave at least two responses,and 11 gave three responses. On analysis ofall the responses, 25 subthemes emerged.These were then grouped into seven mainthemes. Six of the seven identified mainthemes were represented in the first responsegiven by the participants. The seventh maintheme, ‘‘other,’’ contained only two sub-themes, which were only represented in sec-ond or third responses. Further analysis usingquantitative methods was performed on firstresponses only.

Table 4 shows the participants’ first re-sponses grouped by theme and subtheme,with examples of responses for each subthemegiven in the final column. Most respondents(51%) said that they did not know what hadcaused their illness, followed by attributionsto personal responsibility (22%), an exposure(10%), a pathological process (9%), intrinsicpersonal factors (6%), and chance, fate orluck (1%).

Associations of Perceived Cause of IllnessMale participants were more likely to attri-

bute their illness to personal responsibilitythan females (P¼ 0.002). There were no otherstatistically significant associations between ill-ness attribution and gender or with other de-mographic variables. Subjects with lungcancer were more likely to attribute their ill-ness to personal responsibility than subjects

with other diagnoses (P< 0.0001). Of thosewith lung cancer, 42 of 79 (53%) gave smokingas the primary cause of illness, with smokinggiven as the cause in all personal responsibilityattribution responses.

When we examined the relationship be-tween lung cancer, gender, and personalresponsibility attribution using logistic regres-sion, both lung cancer diagnosis and genderwere found to be independently associatedwith personal responsibility attribution: fe-males vs. males: OR 0.44 (95% CI 0.23, 0.82),P¼ 0.01; lung cancer vs. other diagnoses: OR8.75 (95% CI 4.72, 16.20), P# 0.0001.

When we examined the relationship be-tween attribution themes and adjustment to ill-ness (using the MAC), the only strongassociation found was between pathologicalprocess and MAC fighting spirit; respondentswho gave a pathological process as a causal at-tribution had higher scores than expected forfighting spirit (P¼ 0.01). A very weak associa-tion was found between pathological processand MAC helplessness-hopelessness; respon-dents who gave pathological process as a causalattribution had lower than expected scores forhelplessness-hopelessness (P¼ 0.06). None ofthe attribution themes were associated withthe presence of depression.

DiscussionDistribution of Scores on the Brief IPQ

In our sample, the scores onmost of the itemswere not normally distributed. However, ina population with advanced disease, one mightexpect more homogeneity in illness percep-tions than was demonstrated in our sample.For example, although 30% of the sample rated

Table 3Comparison of Illness Perceptions Between Depressed and Nondepressed Participants

Not Depressed at T1(n¼ 191, n [%])

Depressed at T1(n¼ 109, n [%])

Unadjusted OR(95% CI)

Adjusted forFunctional Statusand Disease Status(OR [95% CI])

Adjusted for FunctionalStatus, Disease Status,and MAC HH (OR

[95% CI])

Adjusted forFunctional Status,Disease Status, and

MAC FS (OR [95% CI])

Adjusted for FunctionalStatus, Disease Status,and MAC AP (OR

[95% CI])

Consequences of illnessfrom Brief IPQ

Lowest quartile 47 (24.7) 4 (3.67) 0.17 (0.06, 0.50) 0.14 (0.04, 0.48) 0.12 (0.03, 0.55) 0.18 (0.05, 0.64) 0.18 (0.05, 0.63)Mid-quartiles 109 (57.4) 54 (4.96) 1.00 1.00 1.00 1.00 1.00Highest quartile 34 (17.9) 51 (46.8) 3.03 (1.76, 5.21) 2.85 (1.56, 5.20) 2.89 (1.46, 5.72) 3.05 (1.57, 5.91) 2.47 (1.28, 4.77)

Timeline of illness fromBrief IPQ

Lowest quartile 38 (25.3) 21 (23.6) 0.77 (0.22, 2.74) 0.89 (0.23, 3.40)Mid-quartiles 7 (4.7) 5 (5.6) 1.00 1.00Highest quartile 105 (70.0) 63 (70.8) 0.84 (0.26, 2.76) 0.90 (0.26, 3.15)

Personal control fromBrief IPQ


Treatment control fromBrief IPQ


Identity from Brief IPQLowest quartile 54 (28.6) 11 (10.5) 0.33 (0.16, 0.66) 0.29 (0.13, 0.67) 0.20 (0.07, 0.61) 0.24 (0.09, 0.64) 0.19 (0.07, 0.52)Mid-quartiles 115 (60.9) 72 (68.6) 1.00 1.00 1.00 1.00 1.00Highest quartile 20 (10.6) 22 (21.0) 1.76 (0.90, 3.45) 1.52 (0.74, 3.16) 1.53 (0.69, 3.42) 1.50 (0.67, 3.38) 1.24 (0.55, 2.77)

Concern from BriefIPQ


IllnessComprehensibilityfrom Brief IPQ


Emotions from BriefIPQ


T1 ¼ Time 1; HH¼ helplessness-hopelessness; FS¼ fighting spirit; AP¼ anxious preoccupation.Items in bold are statistically significant.

826

Vol.

43No.

5May

2012

Price

etal.

Table 4Causal Attributions of Illness

Themen (%) for

First Answer Subtheme Example of Response

Don’t know 154 (51)Personal

responsibility67 (22) � Failure to report significant sign or

symptomIgnoring breast lump

� Failure to attend routine healthchecks

Failure to attend a routine health check-up

� Poor life choices Bad life choices (jobs/relationships)� Lifestyle, for example, smoking,

weight, and dietSmoking and chewing tobacco

Exposure 31 (10) � Environmental exposure Exposure to sunshine� Modern life Modern diet-full of chemicals� Past traumatic event or experience:

physical or emotionalShock of losing husbandA fall onto my back

� Exposure to specific harmfulsubstance: not specificallyoccupational or occupational

Bacon brine exposureExposure to asbestos

� Work related Chairbound at work, caused deep veinthrombosis

� Stress Stress� Social isolation Social isolation� Iatrogenic: medical mismanagement,

medical intervention, side effect ofmedication, stopping medication

Doctor ignored back pain

Pathologicalprocess

27 (9) � Specified causative pathology Cancer� Related to previous specific causative

pathologySevere constipation many years ago

� Biological malfunction Body malfunction� Genetic malfunction Cancer gene gone wrong� Infectious process Chest infection

Intrinsicpersonal factors

18 (6) � Advancing age Old age� Genetic factors Genetic factors� Genetic predisposition Cancer runs in dad’s family� Immunity Poor immunity

Chance/fate/luck 4 (1)


themselves as ‘‘extremely concerned’’ abouttheir illness, 15% rated themselves as ‘‘not atall concerned.’’ This shows that even withina sample of patients with advanced disease, theappraisal of illness varies greatly between indi-viduals, challenging assumptions about how ad-vanced disease is perceived by patients.

On the timeline item, however, the scoreswere distributed at such extremes that the me-dian score was 10 on an 11-point scale. A ques-tion assessing timeline in a population witha fairly short prognosis presents difficulties.Respondents were unsure whether their an-swer should reflect that they had only a shorttime to live, leading them to answer witha low number or that their illness would lastforever (i.e., they would die with the disease),leading them to answer with a very high num-ber. The great majority of respondents indi-cated that they believed their illness wouldlast forever, but the reduced sample size forthat item compared with the other items

reflects the difficulty many had in coming toa satisfactory answer.

Range of Causal AttributionsWe found that, despite an atheoretical ap-

proach to the coding of responses to the itemas-sessing causal attribution, our categories talliedclosely with the domains of causal attributionproposed by Heider.11 The themes personal re-sponsibility andpathological process fit concep-tually at opposite ends of the controllabilitydomain, with exposure and intrinsic personalfactors appearing to be related to the externaland internal loci, respectively.

The answers respondents gave regardingcausality showed that they interpreted thequestion in different ways. When asked whatcaused their illness, many responded with ananswer indicating they interpreted the ques-tion in terms of what caused the disease, for ex-ample, giving smoking as a cause of lungcancer. However, many respondents stated


the disease itself as the cause, indicating thatthey interpreted the question as asking whatcaused their symptoms. Again, this finding leadsus to question assumptions about how patientsconceptualize their illness and that concepts ofcausation may differ between individuals.

Correlation Between Brief IPQ DomainsThe Brief IPQ items that correlated with

each other in our sample did so strongly andsignificantly but not strongly enough to sug-gest conceptual overlap, and the correlationswe found were understandable in the contextof advanced disease. A previous meta-analyticreview of the CSM model of illness representa-tion36 looked at correlations between items onthe original IPQ in 45 studies, mostly ofchronic diseases. Although several of our find-ings agreed with the findings in that review,a particular area of deviation was in the rela-tionship between timeline and other items,possibly reflecting the different meaning oftimeline in a palliative population comparedwith other medically ill populations.

It is also of note that, in our sample, concernwas negatively correlated with illness under-standability. How well the illness is understoodin the context of terminal disease is unlikely tojust reflect how well apprised the sufferer iswith the clinical facts. It is also likely to reflectwider representation of cause and prognosis.Feeling that one does not understand whythe illness has happened and fear of what islikely to happen in the future may understand-ably increase concern about the illness and itslikely outcome.

Correlation Between Illness Representationand Adjustment to Cancer

Leventhal et al.1 made an explicit link be-tween illness representations and coping be-haviors and strategies and proposed that thelink is causal, that is, illness representationswill lead directly to coping responses andadoption of coping strategies. Hagger andOrbell36 found that perceptions of greatercontrol were associated with problem-focusedcoping but not avoidance or denial; however,perceptions of greater consequences andsymptoms (illness identity) were associatedwith avoidance/denial. Timeline was positivelyassociated with avoidance/denial, that is, those

who perceived that their illness would last lon-ger were more likely to be avoidant.In our sample, consequences were nega-

tively associated with avoidance and personalcontrol was positively associated with avoid-ance. There was no significant correlation be-tween avoidance and other Brief IPQdomains. Although our findings contrastedwith those of the study above, it is possiblethat patients with advanced disease who per-ceive their illness affecting their lives moreare less avoidant of the effects of the illnessand that those who perceive control over theillness use an avoidant coping strategy to main-tain the sense of control, thus using avoidanceadaptively.All the quantitative items of the Brief IPQ

correlated strongly with the helplessness-hopelessness domain of the MAC Scale. Direc-tions of association were consistent with whatmight be expected, with the strongest correla-tion being positive between emotion andhelplessness-hopelessness; those who feelmore emotionally affected by their illness aremore likely to adjust maladaptively to it.Fighting spirit as measured by the MAC

Scale also was strongly correlated with mostquantitative domains of the Brief IPQ. In par-ticular, fighting spirit was strongly positivelycorrelated with the domain of treatment con-trol; therefore, those who felt that their treat-ment could help the illness adjusted to theillness with continued optimism.That concern and emotion were strongly

positively correlated with anxious preoccupa-tion in our sample was unsurprising. In ad-vanced disease, concern may center oncurrent illness but is also likely to focuson fear of future suffering and the end oflife, leading to anxiety and worry and poten-tially predisposing to negative psychologicaloutcomes.When comparing the associations of control

items across the domains on the MAC Scale, itis interesting to note negative correlations withboth helplessness-hopelessness and fatalism. Itmay be that different emphasis is placed onthe importance of control by different groupsand that a low sense of control does not neces-sarily lead to maladaptation to illness, depend-ing on how it is appraised and integrated withother perceptions of illness. Also of note arethat directions of correlation among all the


Brief IPQ items with helplessness-hopelessnessand fighting spirit are in opposition, suggest-ing that, in advanced disease, these domainsof adjustment to illness are at different endsof a spectrum of coping.

In contrast to the findings of previous stud-ies, causal attribution had little associationwith adjustment to illness. The only attributionthat showed any association was that of patho-logical process, which was positively associatedwith fighting spirit and very weakly and nonsig-nificantly negatively associated with helpless-ness-hopelessness.

Although the most common response to thequestion of causal attribution was ‘‘don’tknow,’’ the second most common was onewhich could be categorized as ‘‘personal re-sponsibility,’’ of which smoking was the majorfactor cited and was by far most frequentlyseen in patients with lung cancer. This wouldlead us to question the findings of Linnet al.27 that advanced cancer patients need todefend themselves against self-blame asa means of coping with their terminal illness.

Illness Representation and DepressionIllness representations appear to be associ-

ated with depression independent of currentfunctional or disease status, and we have iden-tified both risk and protective factors withinrepresentations of illness. Given that the CSMis proposed as a mediational model where cop-ing mediates the relationship between illnessrepresentations and psychological distress,37

it is surprising that adjustment to illness did lit-tle to account for the relationship between ill-ness representations and depression in oursample. A number of studies also have testedthis mediation hypothesis, with similar nega-tive outcomes,38e43 although other studieshave found significant mediation effects44e46

but only in cross-sectional rather than longitu-dinal analyses.41,47

Causal attribution was not associated withdepression, contrary to our hypothesis thatthose who attributed their illness to personalresponsibility would be more likely to bedepressed.

Clinical ImplicationsWe have shown that particular illness per-

ceptions are strongly associated with depres-sion in a population with advanced disease

and also that patients may not perceive the ill-ness in the ways one might expect. Explorationof a patient’s illness representations may revealimportant information that will help the clini-cian to personalize the patient’s management.Given that our data were cross-sectional, it isimpossible to determine whether depressionarose from the illness perceptions or vice ver-sa; however, there is the potential for develop-ing interventions designed to address illnessperceptions, with the aim of reducing risk ofdepression and optimizing adaptive coping.

The Utility of the Brief IPQ in Palliative Popula-tions. Limitations in the use of the IPQ havealready been identified, although these arepartially negated by the authors’ encourage-ment of researchers to adapt the wording ofthe questionnaire to meet the needs of specificpopulations.48 This study has raised questionsabout the applicability of some of the itemsin the generic Brief IPQ to palliative popula-tions, and, in particular, we would questionhow appropriate the current concept of time-line is to patients with a short prognosis. Itmay be that some adaptations to the BriefIPQ would improve the utility of the question-naire for this population.

Strengths and LimitationsResearch using samples of patients with ad-

vanced disease presents methodological chal-lenges. Recruitment can be affected byseveral factors including individuals feelingtoo ill or overwhelmed and family and clini-cians acting as gatekeepers,49,50 but this studyachieved a large representative sample in spiteof these factors, with an enrollment rate simi-lar to that reported in previous surveys in pal-liative care.51

The main limitation in this study was theuse of cross-sectional methodology. TheCSM model of illness considers the relation-ship between illness representations, copingbehaviors, and outcomes longitudinally ina feedback loop. Although we were able to ex-amine relationships between different aspectsof the model, we were unable to prospectivelyexamine the relationships between these fac-tors. Further research in this area might usea longitudinal study design with sequentialmeasures of illness perception, adaptive


coping, and depression to fully explore the re-lationships between these areas over time.

ConclusionAssessment of illness perceptions in pallia-

tive care is likely to yield important informationnot just about how patients conceptualize theirillness but also about their risk of depression.Although the Brief IPQ has limitations foruse in this population, with appropriate adap-tation it may be a useful tool for guiding assess-ment and is short enough to be sufficientlytolerable. Future directions for research mightinclude more in-depth exploration of how pa-tients answer the Brief IPQ and developmentof interventions designed to address illnessperceptions.

Disclosures and AckowledgmentsThis study was supported by St. Christo-

pher’s Hospice.L. R. is supported by the COMPASS research

collaborative and the European Commission’sSixth Framework Programme (contract no.LSHT-CT-2006-037777, EPCRC). W. L. is sup-ported by the Medical Research Council andis a Clinical Training Fellow. A. P. is supportedby St. Christopher’s Hospice. I. J. H. is a Na-tional Institute for Health Research (NIHR)senior investigator. M. H. is supported by theNIHR Biomedical Research Centre for MentalHealth at the South London and MaudsleyNHS Foundation Trust and the Institute ofPsychiatry, King’s College London and isa NIHR senior investigator. The authors haveno competing interests.

The authors are grateful to the patients andtheir families who generously gave their timeto participate in this study. They thank the staffat St. Christopher’s for making recruitmentpossible.

The authors thank Professor John Weinmanfor his expert contribution.

References1. Leventhal H, Meyer D, Nerenz D. The commonsense representation of illness danger. In:Rachman S, ed. Medical psychology. New York: Per-gamon Press, 1980:7e30.

2. Broadbent E, Petrie KJ, Main J, Weinman J. TheBrief Illness Perception Questionnaire. J PsychosomRes 2006;60:631e637.

3. French DP, Cooper A, Weinman J. Illness per-ceptions predict attendance at cardiac rehabilita-tion following acute myocardial infarction:a systematic review with meta-analysis. J PsychosomRes 2006;61:757e767.

4. Scharloo M, Kaptein AA, Weinman J, et al. Pa-tients’ illness perceptions and coping as predictorsof functional status in psoriasis: a 1-year follow-up.Br J Dermatol 2000;142:899e907.

5. Horne R, Weinman J. Self-regulation and selfmanagement in asthma: exploring the role of illnessperceptions and treatment beliefs in explainingnon-adherence to preventer medication. PsycholHealth 2002;17:17e32.

6. Scharloo M, Baatenburg de Jong RJ,Langeveld TPM, et al. Quality of life and illness per-ceptions in patients with recently diagnosed headand neck cancer. Head Neck 2005;27:857e863.

7. Petrie KJ, Cameron LD, Ellis CJ, Buick D,Weinman J. Changing illness perceptions aftermyocardial infarction: an early intervention ran-domized controlled trial. Psychosom Med 2002;64:580e586.

8. Scharloo M, Kaptein AA, Weinman J, et al.Illness perceptions, coping and functioning inpatients with rheumatoid arthritis, chronic obstruc-tive pulmonary disease and psoriasis. J PsychosomRes 1998;44:573e585.

9. Kim Y, Evangelista LS. Relationship between ill-ness perceptions, treatment adherence, and clinicaloutcomes in patients on maintenance hemodialysis.Nephrol Nurs J 2010;37:271e280.

10. Petrie KJ, Jago LA, Devcich DA. The role of ill-ness perceptions in patients with medical condi-tions. Curr Opin Psychiatry 2007;20:163e167.

11. Heider F. The psychology of interpersonal rela-tions. New York: John Wiley & Sons, 1958.

12. Landrine H, Klonoff EA. Cultural diversity incausal attributions for illness: the role of the super-natural. J Behav Med 1994;17:181e193.

13. KwateNOA, ThompsonHS, Valdimarsdottir HB,Bovbjerg DH. Brief report: etiological attributionsfor breast cancer among healthy African Americanand European American women. Psychooncology2005;14:421e425.

14. Nelson J, O’Leary C, Weinman J. Causal attribu-tion in babies with a cleft lip and/or palate andtheir association with psychological well-being. CleftPalate Craniofac J 2009;46:425e434.

15. Fukuoka Y, Dracup K, Kobayashi F, et al. Illnessattribution among Japanese patients with acutemyocardial infarction. Heart Lung 2004;33:146e153.


16. French D, Maissi E, Marteau TM. The purposeof attributing cause: beliefs about the causes of myo-cardial infarction. Soc Sci Med 2005;60:1411e1421.

17. Weinman J, Petrie KJ, Sharpe N, Walker S.Causal attributions in patients and spouses followingfirst-time myocardial infarction and subsequent life-style changes. Br J Health Psychol 2000;5:263e273.

18. Cho HJ, Bhugra D, Wessely S. Physical or psy-chological? A comparative study of causal attribu-tion for chronic fatigue in Brazilian and Britishprimary care patients. Acta Psychiatr Scand 2008;118:34e41.

19. Pucheu S, Consoli SM, D’Auzac C, Francais P,Issad B. Do health causal attributions and copingstrategies act as moderators of quality of life in peri-toneal dialysis patients? J Psychosom Res 2004;56:317e322.

20. Lavery JF, Clarke VA. Causal attributions, copingstrategies, and adjustment to breast cancer. CancerNurs 1996;19:20e28.

21. Cousson-G�elie F, Irachabal S, Bruchon-Schweitzer M, Dilhydy JM, Lakdja F. Dimensions ofcancer locus of control scale as predictors of psycho-logical adjustment and survival in breast cancer pa-tients. Psychol Rep 2005;97:699e711.

22. Dias OM, Turato ER. Cigarette smokers views ontheir habit and the causes of their illness followinglung cancer diagnosis: a clinical qualitative study.Sao Paulo Med J 2006;124:125e129.

23. Faller H, Schilling S, Lang H. Casual attributionand adaptation among lung cancer patients.J Psychosom Res 1995;39:619e627.

24. Costanzo ES, Lutgendorf SK, Bradley SL,Rose SL, Anderson B. Cancer attributions, distress,and health practices among gynecologic cancer sur-vivors. Psychosom Med 2005;67:972e980.

25. Roesch SC, Weiner B. A meta-analytic review ofcoping with illness: do causal attributions matter?J Psychosom Res 2001;50:205e219.

26. Turnquist DC, Harvey JH, Andersen BL. Attri-butions and adjustment to life-threatening illness.Br J Clin Psychol 1988;27:55e65.

27. Linn MW, Linn BS, Stein SR. Beliefs aboutcauses of cancer in cancer patients. Soc Sci Med1982;16:835e839.

28. Watson M, Homewood J. Mental Adjustment toCancer Scale�: psychometric properties in a largecancer cohort. Psychooncology 2008;17:1146e1151.

29. Wattebot-O’Brien C, Moorey S. Outlook and ad-aptation in advanced cancer: a systematic review.Psychooncology 2010;19:1239e1249.

30. Rayner L, Lee W, Price A, et al. The clinical ep-idemiology of depression in palliative care and thepredictive value of somatic symptoms: cross-sectional survey with 4-week follow-up. Palliat Med2011;25:229e241.

31. Hotopf M, Chidgey J, Addington-Hall J, Ly KL.Depression in advanced disease: a systematic reviewPart 1. Prevalence and case finding. Palliat Med2002;16:81e97.

32. Aaronson NK, Ahmedzai S, Bergman B, et al.The European Organization for Research andTreatment of Cancer QLQ-C30: a quality-of-life in-strument for use in international clinical trials inoncology. J Natl Cancer Inst 1993;85:365e376.

33. Hodkinson HM. Evaluation of a mental testscore for assessment of mental impairment in the el-derly. Age Ageing 1972;1:233e238.

34. Spitzer R, Kroenke K, Williams J. Validation andutility of a self-report version of PRIME-MD:the PHQ primary care study. JAMA 1999;282:1737e1744.

35. StataCorp. Statistical software: Release 10.0. Col-lege Station, TX: StataCorp LP, 2007.

36. Hagger MS, Orbell S. A meta-analytic review ofthe common-sense model of illness representations.Psychol Health 2003;18:141e184.

37. Leventhal H, Nerenz DR, Steele DJ. Illness rep-resentations and coping with health threats. In:Baum A, Taylor SE, Singer JE, eds. Handbook of psy-chology and health. Hillsdale, NJ: Erlbaum, 1984:219e252.

38. Heijmans MJWM. Coping and adaptive out-come in chronic fatigue syndrome: importance ofillness cognitions. J Psychosom Res 1998;45:39e51.

39. Kaptein AA, Helder DI, Scharloo M, et al. Ill-ness perceptions and coping explain well-being inpatients with Huntington’s disease. Psychol Health2006;21:431e446.

40. Kemp S, Morley S, Anderson E. Coping with ep-ilepsy: do illness representations play a role? Br JClin Psychol 1999;38:43e58.

41. Llewellyn CD, McGurk M, Weinman J. Illnessand treatment beliefs in head and neck cancer: isLeventhal’s common sense model a useful frame-work for determining changes in outcomes overtime? J Psychosom Res 2007;63:17e26.

42. Rutter CL, Rutter DR. Longitudinal analysis ofthe illness representation model in patients with ir-ritable bowel syndrome (IBS). J Health Psychol2007;12:141e148.

43. Scharloo M, Kaptein AA, Weinman J, et al.Illness perceptions, coping and functioning in pa-tients with rheumatoid arthritis, chronic obstructivepulmonary disease and psoriasis. J Psychosom Res1998;44:573e585.

44. Gray SE, Rutter DR. Illness representations inyoung people with chronic fatigue syndrome. Psy-chol Health 2007;22:159e174.

45. Knibb RC, Horton SL. Can illness perceptionsand coping predict psychological distress amongst


allergy sufferers? Br J Health Psychol 2008;13:103e119.

46. Rutter CL, Rutter DR. Illness representation,coping and outcome in irritable bowel syndrome(IBS). Br J Health Psychol 2002;7:377e391.

47. Evans D, Norman P. Illness representations,coping and psychological adjustment to Parkinson’sdisease. Psychol Health 2009;24:1181e1196.

48. Moss-Morris R, Weinman J, Petrie KJ. The re-vised illness perception questionnaire (IPQ-R). Psy-chol Health 2002;17:1e16.

49. Williams A-L. Recruitment challenges for end-of-life research. J Hospice Palliat Nurs 2007;9:79e85.

50. Jordhoy MS, Kaasa S, Fayers P, et al. Challengesin palliative care research; recruitment, attritionand compliance: experience from a randomizedcontrolled trial. Palliat Med 1999;13:299e310.

51. Thompson GN, Chochinov HM, Wilson KG,et al. Prognostic acceptance and the well-being ofpatients receiving palliative care for cancer. J ClinOncol 2009;27:5757e5762.

Documents

Illness Perceptions, Adjustment to Illness, and Depression in a Palliative Care Population