Does resilience ‘buffer’ against depression in prostatecancer patients? A multi-site replication study

C.F. SHARPLEY, PHD, PROFESSOR, Brain-Behaviour Research Group, University of New England, Armidale, NSW,V. BITSIKA, PROFESSOR, Brain-Behaviour Research Group, Bond University, Robina, Qld, A.C. WOOTTEN, RESEARCH

DIRECTOR, Australian Prostate Cancer Research Centre, Epworth Hospital, Richmond, Vic., and Brain-BehaviourResearch Group, University of New England, Armidale, NSW, & D.R.H. CHRISTIE, PROFESSOR, Premion, Qld,

Australia, and Brain-Behaviour Research Group, University of New England, Armidale, NSW, Australia

SHARPLEY C.F., BITSIKA V., WOOTTEN A.C. & CHRISTIE D.R.H. (2014) European Journal of Cancer Care23, 545–552Does resilience ‘buffer’ against depression in prostate cancer patients? A multi-site replication study

Although psychological resilience has been shown to ‘buffer’ against depression following major stressors, nostudies have reported on this relationship within the prostate cancer (PCa) population, many of whom are atelevated risk of depression, health problems and suicide. To investigate the effects of resilience upon anxiety anddepression in the PCa population, postal surveys of 425 PCa patients were collected from two sites: 189 PCapatients at site 1 and 236 at site 2. Background data plus responses to depression and resilience scales werecollected. Results indicated that total resilience score was a significant buffer against depression across bothsites. Resilience had different underlying component factor structures across sites, but only one (common) factorsignificantly (inversely) predicted depression. Within that factor, only some specific items significantly pre-dicted depression scores, suggesting a focused relationship between resilience and depression. It may be con-cluded that measures of resilience may be used to screen depression at-risk PCa patients. These patients mightbenefit from resilience training to enhance their ability to cope effectively with the stress of their diagnosis andtreatment. A focus upon specific aspects of overall resilience may be of further benefit in both these processes.

Keywords: oncology, cancer, depression, prostate, resilience.

INTRODUCTION

Prostate cancer (PCa) patients experience depression atgreater prevalence than their non-PCa peers (Kunkel et al.2000; Kronenwetter et al. 2005; Couper et al. 2006;Sharpley et al. 2008). These depressed men are more likelyto be admitted to hospital for emergency treatment, haveoutpatient visits, suffer death from multiple causes (VanGastel et al. 1997), and are almost twice as likely tocommit suicide within the first year after diagnosis as are

men of comparable age who do not have PCa (Chorbovet al. 2007). Ways of understanding which patients aremost ‘at risk’ of developing depression and these unwel-come side-effects of depression are an important aspect ofthe overall treatment regime for PCa. Although someresearch has suggested that there are genetic factors[e.g. the short form of the serotonin transporter gene,5-HTTLPR (Karg et al. 2011)] that increase vulnerabilityto depression following major stress (such as receiving adiagnosis of PCa), other research has focused upon thosefactors which protect the individual from depression aftermajor stress. That protective effect has sometimes beentermed ‘resilience’ as it describes the ability of somefactors to apparently equip people to resist the depressiveeffects of major stress.

For example, some of the factors which have beenshown to reduce the risk of developing depression in a

Correspondence address: Christopher Francis Sharpley, Brain-BehaviourResearch Group, School of Science & Technology, University of NewEngland, Armidale, NSW 2351, Australia (e-mail: csharpley@OntheNet.com.au).

Accepted 17 November 2013DOI: 10.1111/ecc.12170

European Journal of Cancer Care, 2014, 23, 545–552

bs_bs_banner

Original article

© 2014 John Wiley & Sons Ltd

wide range of people who undergo the stress of cancerinclude personality and demographic characteristics suchas spirituality (Wenzel et al. 2002), family relationships(Orbuch et al. 2005) and hopefulness (Ho et al. 2010).These characteristics has been tested as predictors of theincidence and severity of anxiety and depression in cancerpatients who are children (Orbuch et al. 2005), adoles-cents (Woodgate 1999) and across the lifespan (Rowland &Baker 2005), as well as in particular groups of cancerpatients such as those experiencing ovarian (Wenzel et al.2002) and breast cancer (Deshields et al. 2006). That lineof research has defined resilience as an outcome variablethat describes the effects of selected demographic andpersonality factors upon mental and physical health. Thatis, a person with (say) high levels of spirituality or soundfamily relationships may be found to be more resistant todeveloping depression following major stress than aperson who has low levels of those characteristics.

An alternative way of considering resilience lies in theconstruct of ‘Psychological resilience’ (Fredrickson et al.2003) defined as a set of specific behavioural or attitudinalskills which help an individual cope effectively withstress and avoid becoming depressed (von AmmonCavanagh et al. 2001; Bitsika et al. 2011; Sharpley et al.2011, 2012). Under this definition, psychological resil-ience is an intervention or buffer variable between stressand depression (Luthar & Cicchetti 2000), possiblyworking by an active physiological process that reducesautonomic responses to stressors (Charney 2004). Variousaspects of this construct of psychological resilience havebeen identified, including the ability to rebound from dis-appointments (Brooks 2005), positive adjustment behav-iours in adverse circumstances (Tedeschi & Kilmer 2005),or simply successful adaptation to challenging lifestressors (Alvord & Grados 2005). Psychological resiliencehas been shown to intervene between the experience oftraumatic events and the individual’s later return to opti-mism in the face of such stressors as old age (Jopp & Rott2006), chronic pain (Karoly & Reuhlman 2006) and terror-ist attack (Bonanno et al. 2007), and can reduce depressionthat is induced by stressful events (Andreescu et al. 2007).It has been suggested that psychological resilience has abiological basis that relies upon plasticity of the rewardand fear circuits in the brain (Bergstrom et al. 2007), thatthere may be possible neurological mediators of the resil-ient response to extreme stress (Charney 2004) leading tothe suggestion that resilience may be analysed at variouslevels (Cicchetti & Blender 2006), and that preventative aswell as treatment modalities should be considered(Haglund et al. 2007). This definition of psychologicalresilience may be measured via standardised scales and

training can be developed to build a behavioural repertoirethat assists individuals to become more resilient in theface of stressors (Vergun 2012).

However, at the time of writing, an electronic literaturesearch using the terms ‘prostate cancer’, ‘depression’ and‘resilience’ failed to identify any published papers thatinvestigated the influence of this definition of psychologi-cal resilience as a set of measureable behaviours that mod-erated the effects of depression experienced by PCapatients, although several studies had reported testing itseffect in other cancer patients such as those receivingradiation therapy (Strauss et al. 2007) and in samples ofunspecified cancer types (Min et al. 2013). Therefore, thisstudy aimed to explore the potential protective or ‘buffer-ing’ effect of psychological resilience upon depression inPCa patients by determining if PCa patients who pos-sessed high levels of psychological resilience were lesslikely to experience depression than PCa patients whopossessed low level of psychological resilience. Toenhance generalisability of results, data were collectedfrom PCa patients at two sites which provided diagnosticand treatment services to PCa patients.

METHODS

Sites and participants

Site 1 was a private provider of radiotherapy treatmentsfor a wide range of cancers at five centres, with consistentstaff and protocols across centres (Premion, Qld, Aus-tralia). Site 2 was a private provider of surgical treatmentfor PCa. A total of 350 PCa patients from site 1 wereinvited by letter to participate and 189 (54.0%) completedusable questionnaires; 400 patients were invited from site2, with 236 (59.0%) providing usable data. All participantshad PCa limited to the primary site and regional draininglymph nodes using conventional staging investigations.Treatments included radiotherapy and/or surgery andhormone therapy (HT) when required. Other inclusioncriteria were: (1) the diagnosis of PCa was proven histo-logically; (2) all of the treatment options were properlyconsidered by patients via discussion with their GP (asreported to their oncologist), a radiation oncologist and/ora urologist; and (3) patients were included regardless of thetype of HT they had been prescribed. Unwillingness toparticipate in the study was the only exclusion criterion.

Scales

Background questionnaire

Age (in years), living situation (with wife/partner,widowed, separated/divorced, never married), month and

SHARPLEY ET AL.

© 2014 John Wiley & Sons Ltd546

year of first diagnosis, treatments received and continuing(radiotherapy, surgery, HT, none), present status of theircancer [cancer undergoing initial treatment, no obvioussign of cancer (in remission), cancer re-occurring afterprevious treatment].

Resilience

Resilience was assessed by the Connor–Davidson Resil-ience Scale (CD-RISC) (Connor & Davidson 2003). TheCD-RISC includes 25 items such as ‘I like a challenge’ and‘I bounce back after illness or hardship’ and was reportedby the scale authors to have five factors (in their originalnorming samples) that measure ‘Personal competence,high standards and tenacity’, ‘Trust in one’s instincts,tolerance of negative affect, strengthening effects ofstress’, ‘Positive acceptance of change and secure relation-ships with others’, ‘Control’ and ‘Spiritual influences’(Connor & Davidson 2003). Total scores on the CD-RISCare significantly correlated (0.83) with total scores on theKobasa Hardiness Measure and negatively correlated withtotal scores on the Perceived Stress Scale (−0.76) indicat-ing high concurrent validity. The CD-RISC has acceptablereliability, ranging from 0.89 (Cronbach’s alpha) to 0.87(test–retest reliability) (Connor & Davidson 2003).

Depression

The Patient Health Questionnaire Depression Scale(PHQ9) was used to assess depression. The PHQ9 wasdeveloped from Diagnostic and Statistical Manual ofMental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) (APA 2000) criteria for Major Depressive Disorder andhas been shown to possess excellent validity for patientswith severe, moderate and mild depression (Kroenke et al.2001).

Procedures

Patients who met the inclusion criteria for the study wereidentified from the databases at each site and were posteda package including the Participant Information State-ment, Background Questionnaire, CD-RISC and PHQ9,plus a stamped and addressed envelope for return of thequestionnaire package. All procedures were approved bythe Human Research Ethics Committees at each site.

Statistical analyses

Data were analysed with SPSS 20. Following tests fornormality, patients from each site were tested for differ-

ences in age and time since diagnosis via MANOVA, andother background variables were tested for site differencesvia Chi-square procedures. MANOVA was also used totest for the presence of significant differences in PHQ9and CD-RISC scores across sites. Pearson correlationswere used to test for initial significant associationsbetween PHQ9 and CD-RISC scores at each site. Becausethe major aim of this study was to explore the relationshipbetween resilience and depression by testing for the pres-ence of significant differences in depression in PCapatients who exhibited ‘high’ versus ‘low’ levels of psy-chological resilience, ANOVA of PHQ9 scores was con-ducted using the median split for CD-RISC scores to form‘high’ versus ‘low’ resilience subgroups within each sitesample. The factor structure of an instrument can varyacross samples (Tabachnick & Fidell 2007), and thereforePrincipal Components analysis was used to identify thefactor structure of the CD-RISC in these two site samples,and then hierarchical regression used to determine whichof the CD-RISC factors most powerfully predicted PHQ9scores within each site.

RESULTS

Site effects: MANOVA on background variables of age andtime since diagnosis across site produced a significantmain effect [F(2, 406) = 97.890, P < 0.001], with significantunivariate effects for both of these variables, indicatingthat site 1 patients were older than site 2 patients but thatsite 2 patients had received their diagnoses a significantlylonger time before the survey than site 1 participants(Table 1). Appendix 1 presents the other background datafor participants at sites 1 and 2. A series of Chi-squareanalyses was conducted on: living situation (ns);present disease status (Χ(2) = 67.029, Cramer’s V = 0.400,P < 0.000) with 28.6% of patients in site 1 reportingcurrent cancer being treated but only 1.3% of site 2patients falling into this category; past treatment(Χ(2) = 292.799, Cramer’s V = 0.852, P < 0.000), where thepercentages of patients at each site who had receivedradiotherapy, surgery and HT were quite different (combi-nations of these three treatments also varied across sitesin similar directions); and current treatment (Χ(2) = 36.914,

Table 1. Means (SD) for age (years) and time since diagnosis(months) for both sites

Variable Site 1 Site 2 F

Age 70.193 (6.35) 66.571 (6.72) 10.855*Time since diagnosis 24.967 (31.49) 67.833 (33.37) 61.723*

*P < 0.001.

Resilience buffers depression in prostate cancer

© 2014 John Wiley & Sons Ltd 547

Cramer’s V = 0.486, P < 0.000), where many patients werereceiving radiotherapy at site 1 but none at site 2.

Table 2 shows the scale data for both sites. UsingMANOVA, there was a significant main effect for the twopsychological variables of PHQ9 and CD-RISC accordingto site [F(2,403) = 3.159, P < 0.05, Wilks Lambda, partialeta square = 0.015] and a single significant univariateeffect for PHQ9 at the Bonferroni-corrected Type I errorrate of 0.0167 [F(1,405) = 6.312, P = 0.012], where site 1patients had significantly higher depression scores thansite 2 patients. Further analyses were conducted sepa-rately for each site.

Site 1

Pearson correlations between current cancer status,past treatment and current treatment showed no signifi-cant relationships with PHQ9. There was a signifi-cant inverse correlation between PHQ9 and CD-RISC(r = −0.491, P < 0.001). Using a median split on CD-RISC(83), two subgroups were created: low resilience (n = 92)and high resilience (n = 80), and there was a significantdifference in PHQ9 scores across these two groups [lowresilience = 14.96 (5.27), high resilience = 12.16 (3.56):F(1,171) = 16.084, P < 0.001], indicating that resiliencewas significantly inversely associated with depressionamong these PCa patients from site 1.

However, these analyses are based upon scores from theCD-RISC as a unidimensional construct. Although theoriginal development of the CD-RISC reported fivefactors, such factor structures are sometimes restricted tothe original sample from which they were derived andmay vary according to the demographic and other aspectsof different samples (Tabachnick & Fidell 2007). Thecurrent sample was quite different from the ‘communitysample, primary care outpatients, general psychiatric out-patients, clinical trial of generalized anxiety disorder, andtwo clinical trials of PTSD’ (Connor & Davidson 2003,p. 76) samples that were used by the developers of theCD-RISC, and so the current CD-RISC data from these

PCa patients were investigated to identify their underly-ing factor structure. The Kaiser–Meyer–Olkin measure ofsampling adequacy was 0.920, above the recommendedlevel of 0.6, Bartlett’s Test of Sphericity was significantand there were many correlations above 0.3 in the corre-lation matrix, thus supporting the use of factor analysiswith these data. Five components had eigenvalues of 1.0or more but the component matrix showed only threeitems that loaded on the final two factors. Inspection ofthe screeplot and parallel analysis indicated that a three-factor solution was most appropriate. After rotation, thethree-factor solution explained 57.16% of the variance,with only minor inter-factor correlations. The patternmatrix listed 15 CD-RISC items loading on factor 1(43.20% of the variance), which measured ‘Confidence tocope with change’; factor 2 (six items, 7.61% of the vari-ance) measured ‘Being able to take difficult actions’, andfactor 3 (four items, 6.35% of the variance) measured‘Trusting in a higher power’.

Hierarchical regression on PHQ9 score, with the threeCD-RISC factors entered in order, showed that factor 1produced an R square of 0.280 [F(1,170) = 66.098,P < 0.001], but that factors 2 and 3 produced R squarechanges of 0.002 and 0.000, respectively (ns), indicatingthat only the ‘Confidence to cope with change’ factor wasa significant buffer against depression in site 1. Linearregression on PHQ9 score using the 15 CD-RISC itemscomprising factor 1 as predictor variables produced an Rsquare of 0.373 [F(14,178) = 6.976, P < 0.001] but onlyCD-RISC items 14 (‘Under pressure, I stay focused andthink clearly’: Beta = −0.257, t = −2.533, P = 0.012), 13(‘During times of stress/crisis, I know where to turn forhelp’: Beta = −0.196, t = −0.288, P = 0.023) and 6 (‘I try tosee the humorous side of things when I am faced withproblems’: Beta = −0.175, t = −2.101, P = 0.037) were sig-nificant (inverse) predictors of PHQ9 scores, suggestingthat these behaviours/attributes were the key aspects indefining the relationship between resilience and depres-sion among site 1 participants.

Site 2

Although current cancer status and past treatment werenot significantly correlated with PHQ9, current treatmentwas significantly inversely associated with PHQ9(r = −0.400, P < 0.001). Inspection of Appendix 1 andANOVA indicated that this was due to the significantlyhigher PHQ9 scores by patients from site 2 who werereceiving HT (M = 15.304, SD = 5.076) than patients whowere not receiving any treatment at all [M = 11.683,SD = 2.461: F(1,85) = 19.697, P < 0.001].

Table 2. Scale data from both sites

Scale Mean SD 5% mean Min. Max.Cronbach’salpha

Site 1PHQ9 13.64 4.75 13.03 10 39 0.89CD-RISC 78.65 15.46 79.86 28 100 0.93

Site 2PHQ9 12.57 3.93 11.99 10 32 0.89CD-RISC 80.71 13.36 81.83 23 100 0.92

PHQ9, Patient Health Questionnaire Depression Scale;CD-RISC, Connor–Davidson Resilience Scale.

SHARPLEY ET AL.

© 2014 John Wiley & Sons Ltd548

As for site 1, there was a significant inverse correlationbetween CD-RISC and PHQ9 scores (r = −0.562,P < 0.001). The median CD-RISC score was 84. When splitinto two subgroups according to this value, and withcurrent treatment covaried out of the analysis, ANOVAindicated that those participants whose CD-RISC scorewas below the median had PHQ9 scores that were signifi-cantly higher (M = 14.00, SD = 4.809) than those partici-pants whose CD-RISC was above the median [M = 11.070,SD = 1.738: F(1,87) = 31.361, P < 0.001].

Factor analysis of the 25 CD-RISC items (Kaiser–Meyer–Olkin = 0.927, Bartlett’s Test of Sphericity was sig-nificant, many correlations above 0.3) again revealed fivecomponents with eigenvalues greater than 1.0. However,inspection of the screeplot and parallel analysis indicatedthat a two-factor solution was more appropriate, and rota-tion was undertaken with this number of stipulatedfactors. This two-factor solution explained 47.18% of thevariance with inter-component correlations less than0.25. The pattern matrix listed 22 CD-RISC items loadingon factor 1 and three items on factor 2. Inspection of theitems suggested that factor 2 in this site was very similarto factor 3 from site 1, comprising three of the four itemsthat made up the ‘Trusting in a higher power’ factor in thesite 1 solution. Therefore, factor 1 from site 2 was named‘Confidence to cope and take difficult actions’ and factor 2from this site was also named as factor 3 was for site 1 (i.e.‘Trusting in a higher power’). Linear regression on PHQ9scores indicated that only factor 1 significantly predictedPHQ9 score (Beta = −0.557, t = −9.931, P < 0.001). Linearregression on PHQ9 score using the 22 CD-RISC itemscomprising factor 1 as predictor variables produced an Rsquare of 0.514 [F(22,233) = 10.157, P < 0.001] but onlyeight CD-RISC items had statistically significant t valuesfrom their standardised beta weights and six of those wereinverse predictors of PHQ9 score. Table 3 shows thetwo CD-RISC items that were direct predictors of PHQ9and the six that were inverse predictors of depression.Although there is some difference between the group ofCD-RISC items that significantly predicted PHQ9 scoresacross the two sites, three of the 6 items which weresignificant inverse predictors of PHQ9 scores in site 2participants were also the three significant inverse predic-tors of PHQ9 scores in site 1 participants, suggesting thatthese may be the most generalisable aspects of resiliencebehaviour across the two sites.

DISCUSSION

Despite significant differences between the two sites inseveral of the background variables (time since diagnosis,

current and past treatments, current disease state anddepression severity), some aspects of psychological resil-ience were significantly associated with lower levels ofdepression in each of the two samples, thus indicatingsome degree of generalisability of this finding and addingto its previously reported robust effect as a buffer againstdepression in other populations. The finding of differentfactor structures for the CD-RISC across the two sites iscongruent with the comments made by Tabachnick andFidell (2007) that different samples require re-analysis offactor structures on selected scales, and enabled morereliable testing of the associations between the underlyingfactors of psychological resilience as it was experiencedbythese samples than would have been likely with theoriginal five-factor structure of the CD-RISC that wasreported by the test authors (Connor & Davidson 2003).

Regression analysis of CD-RISC factors from each sitesample showed that the same aspects of the overallconstruct of resilience might be more powerful than othersin helping PCa patients cope with their diagnosis andtreatment without becoming depressed. That is, stayingfocused when under pressure, knowing where to turn forhelp during a crisis, and maintaining a humorous outlookin the face of problems appear to be the aspects of psycho-logical resilience that are generalisable as potential‘buffers’ against depression arising from the stress of PCaacross these two samples of PCa patients. These data canalso provide some guidance as to which aspects of resil-ience could be most effectively trained within PCa patientsto best equip them to avoid depression after their initialdiagnosis. In addition to those ‘buffering’ aspects of psy-chological resilience which could be valuable in the face ofstress, the negative effects of those CD-RISC items which

Table 3. Site 2, factor 1 Connor–Davidson Resilience Scale (CD-RISC) items that significantly predicted Patient Health Question-naire Depression Scale (PHQ9) scores

CD-RISC item Beta t P

I prefer to take the lead in solvingproblems rather than lettingothers make all the decisions

0.205 2.515 0.013

I take pride in my achievements 0.147 2.038 0.043I feel in control of my life −0.237 −3.340 0.001I try to see the humorous side of

things when I am faced withproblems

−0.183 −2.614 0.010

I am able to adapt when changesoccur

−0.151 −2.469 0.014

I believe I can achieve my goals,even if there are obstacles

−0.169 −2.194 0.029

Under pressure, I stay focused andthink clearly

−0.195 −2.243 0.013

During times of stress/crisis,I know where to turn for help

−0.120 −2.024 0.044

Resilience buffers depression in prostate cancer

© 2014 John Wiley & Sons Ltd 549

were significant direct predictors of depression in site 2patients (i.e. taking pride in achievements, taking the leadin solving problems rather than enlisting others’ help)might also be conceptualised as attitudes to avoid in PCapatients seeking to reduce the depressive effect of stress.

Although in no way challenging the previously reportedbeneficial effects of other ‘buffers’ against depression incancer patients, such as spirituality, family relationshipsand hopefulness, these findings present a focused report ofthe specific aspects of psychological resilience (as meas-ured by the CD-RISC) that may be of potential assistanceto PCa patients in resisting depression. The results of thisstudy raise two important potential avenues for furtherresearch and clinical applications. First, screening forlevels of resilience (or specific aspects of the CD-RISC) ator soon after the time of diagnosis could be incorporatedinto usual patient intake procedures and provide an indi-cation of which of these patients might benefit most fromearly intervention to equip them with skills and strategiesfor avoiding later depression. The CD-RISC is not arestricted instrument and may be used by a range ofhealthcare workers for this purpose, given as a simplepaper-and-pencil scale during patient intake. Second, suchinterventions for depressed PCa patients could focus uponresilience training as a cognitive-behavioural trainingprocess that is aimed at helping participants to learn howto mentally and emotionally respond to stressors, as it hasbeen successfully applied in other settings such as mili-tary combat (Vergun 2012), coping with chronic pain(Karoly & Reuhlman 2006), and reducing depression (Rileyet al. 2008), although the data from this study would arguefor specific attention to those aspects of resilience thatwere shown to be significant buffers against depression inthese samples of PCa patients.

There are some limitations to this study. First, althoughthe collection of data from two sites and samples withsignificantly different background variables and depres-sion scores enables greater generalisability than from a

single site, generalisability is restricted to similar popula-tions and settings. Similarly, although the PHQ9 has goodvalidity and reliability, it would be of interest to deter-mine if the results reported here are also present whenclinical interviews or observations of actual behaviour areused to determine depressive status and resilience. TheCD-RISC is one of the best instruments for measuringpsychological resilience but there are others, and investi-gation of the generalisability of these findings acrossinstruments would further clarify the underlying nature ofthe associations between resilience and depression.Further investigation of the roles of specific aspects ofresilience (as indicated by the individual CD-RISC itemsfound to be most powerful inverse predictors of depressionin this study) would advance the current findings andcould also contribute to more focused assessment andintervention procedures for PCa patients.

Notwithstanding these caveats, these data hold impor-tant implications for the potential identification of PCapatients who are most likely to develop depression. Inaddition, resilience training with PCa patients who arethus identified as high risk (or who have already developeddepression) may be an effective preventative/treatmentoption with the adverse mood effects often found in thesemen and which can be of potential detriment to theirpsychological and physical health.

ACKNOWLEDGEMENTS

The authors acknowledge the contribution made by theCollaborative Research Network on Mental Health andWell-being in Rural Communities, supported by theDepartment of Industry, Innovation, Science, Researchand Tertiary Education, Commonwealth Government ofAustralia.

CONFLICT OF INTEREST

None.

REFERENCES

Alvord M. & Grados J. (2005) Enhancingresilience in children: a proactiveapproach. Professional Psychology,Research and Practice 36, 238–245.

von Ammon Cavanagh S., Furlanetto L.,Creech S. & Powell L. (2001) Medicalillness, past depression, and presentdepression: a predictive triad for in-hospital mortality. American Journal ofPsychiatry 158, 43–48.

Andreescu C., Lenze E.J., Dew M.A., BegleyA.E., Mulsant B.H., Dombrovski A.Y.,

Pollock B.G., Stack J., Miller M.D. &Reynolds C.F. (2007) Effect of comorbidanxiety on treatment response andrelapse risk in late-life depression: con-trolled study. The British Journal of Psy-chiatry 190, 344–349.

APA (2000) Diagnostic and StatisticalManual of Mental Disorders, 4th edn,Text Revision. American PsychiatricAssociation, Washington, DC, USA.

Bergstrom A., Jayatissa M., Thykjaer T. &Wiborg O. (2007) Molecular pathwaysassociated with stress resilience and drugresistance in the chronic mild stress rat

model of depression – a gene expressionstudy. Journal of Molecular Neuroscience33, 201–215.

Bitsika V., Sharpley C., Aroutzidis A. &Smith D. (2011) The impact of students‘internally-’ versus ‘externally-oriented’coping strategies upon anxiety anddepression: implications for counsellingprocesses. Asia-Pacific Journal of Coun-selling and Psychotherapy 2, 71–81.

Bonanno G.A., Galea S., Bucciarelli A. &Vlahoz D. (2007) What predicts psycho-logical resilience after disaster? The roleof demographics, resources, and life

SHARPLEY ET AL.

© 2014 John Wiley & Sons Ltd550

stress. Journal of Consulting and ClinicalPsychology 75, 671–682.

Brooks R. (2005) The power of parenting. In:Handbook of Resilience in Children (edsBrooks R. & Goldstein S.), pp. 443–458.Kluwer Academic/Plenum, New York,USA.

Charney D.S. (2004) Psychobiologicalmechanisms of resilience and vulnerabil-ity: implications for successful adapta-tion to extreme stress. Focus (SanFrancisco, Calif.) 2, 368–391.

Chorbov V., Lobos E., Todorov A., HeathA., Botteron K. & Todd R. (2007) Rela-tionship of 5_HTTLPR genotypes anddepression risk in the presence of traumain a female twin sample. AmericanJournal of Medical Genetics. Part B,Neuropsychiatric Genetics 144B, 830–833.

Cicchetti D. & Blender J.A. (2006) Amultiple-levels-of-analysis perspectiveon resilience. Annals of the New YorkAcademy of Sciences 1094, 248–258.

Connor K. & Davidson J. (2003) Develop-ment of a new resilience scale: theConnor-Davidson Resilience Scale (CD-RISC). Depression and Anxiety 18,76–82.

Couper J.W., Bloch S., Love A., DuchesneG., Macvean M. & Kissane D. (2006) Thepsychosocial impact of prostate cancer onpatients and their partners. MedicalJournal of Australia 185, 428–432.

Deshields T., Tibbs T., Fan M.-Y. & TaylorM. (2006) Differences in patterns ofdepression after treatment for breastcancer. Psycho-Oncology 15, 398–406.

Fredrickson B., Tugade M., Waugh C. &Larkin G. (2003) A prospective study ofresilience and emotions following theterrorist attacks on the United States onSeptember 11th, 2002. Journal of Person-ality and Social Psychology 84, 365–376.

Haglund M.E.M., Nestadt P.S., CooperN.S., Southwick S.M. & Charney D.S.(2007) Psychobiological mechanisms ofresilience: relevance to prevention andtreatment of stress-related psychopathol-ogy. Development and Psychopathology19, 889–920.

Ho S., Ho J., Bonanno G., Chu A. & Chan E.(2010) Hopefulness predicts resilienceafter hereditary colorectal cancer genetic

testing: a prospective outcome trajecto-ries study. BMC Cancer 10, 279.

Jopp D. & Rott C. (2006) Adaptation in veryold age: exploring the role of resources,beliefs, and attitudes for centenarians’happiness. Psychology and Aging 21,266–280.

Karg K., Burmeister M., Shedden K. & SenS. (2011) The serotonin transporter pro-moter variant (5-HTTLPR), stress, anddepression meta-analysis revisited.Archives of General Psychiatry 68, 444–454.

Karoly P. & Reuhlman L. (2006) Psychologi-cal resilience and its correlates in chronicpain. Pain 123, 90–97.

Kroenke K., Spitzer R. & Williams J. (2001)The PHQ-9: validity of a brief depressionseverity measure. Journal of GeneralInternal Medicine 16, 606–613.

Kronenwetter C., Weidner G., Pettengill E.,Marlin R., Crutchfield L., Mccormac P.,Raisin C. & Ornish D. (2005) A qualita-tive analysis of interview of men withearly stage prostate cancer: the prostatecancer lifestyle trial. Nursing 28, 99–107.

Kunkel E.J.S., Bakker J.R., Myers R.E.,Oyesanmi O. & Gomella L.G. (2000)Biopsychosocial aspects of prostatecancer. Psychosomatics 41, 85–94.

Luthar S. & Cicchetti D. (2000) The con-struct of resilience: implications forinterventions and social policies. Devel-opment and Psychopathology 12, 857–885.

Min J.-A., Yoon S., Lee C.-U., Chae J.-H.,Lee C., Song K.-Y. & Kim T.-S. (2013)Psychological resilience contributes tolow emotional distress in cancer patients.Supportive Care in Cancer 21, 2469–2476.

Orbuch T., Parry C., Chesler M., Fritz M. &Repetto P. (2005) Parent-child relation-ships and quality of life: resilience amongchildhood cancer survivors. Family Rela-tions 54, 171–183.

Riley A., Valdez C., Barrueco S., Mills C.,Beardslee W., Sandler I. & Rawal P. (2008)Development of a family-based programto reduce risk and promote resilienceamong families affected by maternaldepression: theoretical basis and programdescription. Clinical Child and FamilyPsychology Review 11, 12–29.

Rowland J. & Baker F. (2005) Introduction:resilience of cancer survivors across thelifespan. Cancer 104, 2543–2548.

Sharpley C., Bitsika V. & Christie D. (2008)Psychological distress among prostatecancer patients: fact or fiction? ClinicalMedicine: Oncology 2, 563–572.

Sharpley C., Bitsika V. & Christie D. (2011)How prostate cancer patients cope withthe effects of diagnosis and treatment:development of the effects of prostatecancer coping strategies scale. Journal ofMen’s Health 8, 56–65.

Sharpley C., Bitsika V. & Christie D. (2012)How prostate cancer patients cope: evalu-ation and refinement of the ProstateCancer Patients’ Coping Strategies Scale.Journal of Men’s Health 9, 70–78.

Strauss B., Brix C., Fischer S., Leppert K.,Fuller J., Roehrig B., Schleussner C. &Wendt T. (2007) The influence of resil-ience on fatigue in cancer patients under-going radiation therapy (RT). Journal ofCancer Research and Clinical Oncology133, 511–518.

Tabachnick B. & Fidell L. (2007) UsingMultivariate Statistics, 5th edn. Pearson,New York, USA.

Tedeschi R. & Kilmer R. (2005) Assessingstrengths, resilience, and growth to guideclinical interventions. Professional Psy-chology, Research and Practice 36, 230–237.

Van Gastel A., Schotte C. & Maes M. (1997)The prediction of suicide intent indepressed patients. Acta PsychiatricaScandinavia 96, 254–259.

Vergun D. 2012. Study concludes MasterResilience Training effective. [Online].http://www.army.mil. Available at: http://www.army.mil/article/72431/Study_concludes_Master_Resilience_Training_effective/ (accessed 27 August 2012).

Wenzel L., Donnelly J., Fowler J., Habbal R.,Taylor T., Aziz N. & Cella D. (2002)Resilience, reflection, and residual stressin ovarian cancer survivorship: agynecologic oncology group. Psycho-Oncology 11, 142–153.

Woodgate R. (1999) Conceptual under-standing of resilience in the adolescentwith cancer: part 1. Journal of PediatricOncology Nursing 16, 35–43.

Resilience buffers depression in prostate cancer

© 2014 John Wiley & Sons Ltd 551

APPENDIX 1

Demographic data, sites 1 and 2

Variable

Site 1 Site 2

Frequency Per cent Frequency Per cent

Living situation (n = 187) (n = 236)With wife/partner 170 90.9% 212 89.8%Widowed 11 5.9% 8 3.4%Separated/divorced 3 1.6% 8 3.4%Never married 3 1.6% 8 3.4%

Present status (n = 182) (n = 235)Cancer undergoing initial treatment 52 28.6% 3 1.3%In remission (no signs of recurrent disease) 114 62.6% 204 86.8%Cancer recurring after previous treatment 16 8.8% 28 11.9%

Past treatment (n = 168) (n = 234)Radiotherapy 42 25.0% 1 0.4%Surgery (radical prostatectomy) 1 0.6% 182 77.8%Hormone therapy 6 3.6% 3 1.3%Radiotherapy + surgery 41 24.4% 21 8.9%Surgery + hormone 1 0.6% 12 5.2%Radiotherapy + hormone 60 35.7% 1 0.4%Radiotherapy + surgery + hormone 16 9.5% 9 3.9%No treatment 1 0.6% 5 2.1%

Current treatment (n = 67) (n = 89)Radiotherapy 10 14.9% 0 0.0%Surgery 3 4.5% 2 2.2%Hormone therapy 32 47.7% 23 25.9%Combinations 3 4.5% 1 1.1%No treatment 19 28.4% 63 70.8%

SHARPLEY ET AL.

© 2014 John Wiley & Sons Ltd552