w w w . e l s e v i e r . c o m / l o c a t e / p a i n

PAIN� 145 (2009) 196–203

Psychodynamic interpersonal therapy and improvement in interpersonal difficultiesin people with severe irritable bowel syndrome

Thomas Hyphantis a,b, Else Guthrie a, Barbara Tomenson a, Francis Creed a,*

a Psychiatry Research Group, Medical School, University of Manchester, Manchester, UKb Department of Psychiatry, Medical School, University of Ioannina, Ioannina, Greece

a r t i c l e i n f o

Article history:Received 24 July 2008Received in revised form 22 June 2009Accepted 6 July 2009

Keywords:PainCircumplexPsychodynamic interpersonal psychotherapyIBS

0304-3959/$36.00 � 2009 International Associationdoi:10.1016/j.pain.2009.07.005

* Corresponding author. Address: Psychiatry ReseRawnsley Building, Manchester Royal Infirmary, Ox9WL, United Kingdom. Fax: +44 161 273 2135.

E-mail addresses: francis.creed@man.ac.uk, hyphan

a b s t r a c t

The aim of the present study was to assess the relationship between change in interpersonal difficultieswith change in chronic pain, health status and psychological state in 257 Irritable Bowel Syndrome (IBS)patients in a randomized control trial comparing psychotherapy, antidepressant and usual care. Weassessed at three time points interpersonal problems (IIP-32), abdominal pain and bowel symptoms, psy-chological distress (SCL-90), and health status (SF-36). Analysis included repeated measures (ANOVA) toassess change over time and multiple regressions to identify whether change in IIP was associated withoutcome after controlling for psychological status. The main findings were: (1) difficulties with socialinhibition and dependency were associated with longer disease duration; (2) change in mean IIP-32 over15 months was significantly correlated with changes in pain, but these relationships were mediated bychange in psychological distress; (3) change in IIP-32 was an independent predictor of improved healthstatus at 15 months only in the psychotherapy group. These results indicate that improvement in inter-personal problems in IBS patients appear to be primarily associated with reduced psychological distressbut, in addition, the association with improved health status following psychotherapy suggests that spe-cific help with interpersonal problems may play a role in improving health status of patients with chronicpainful IBS.

� 2009 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction

Irritable bowel syndrome (IBS) is a common chronic pain disor-der which forms a majority of patients in gastroenterology clinicsand often leads to high healthcare use and much time missed fromwork [15]. In common with other chronic pain patients, those withIBS have difficulties in interpersonal relationships [26,31], whichrelate to pain coping and outcome of treatment [37]. Several stud-ies have found an association between insecure attachment and re-lated interpersonal difficulties with poor pain self efficacy, anxietyand poor coping [4,29,30]. These have all been cross-sectionalstudies, however, and all these authors suggested examiningwhether improved personal relationships are associated with re-duced pain and reduced disability [4,24], which is what we havetested in this study.

Patients with IBS are said to have difficulties with being asser-tive [26]. Such a submissive interpersonal style has been relatedto pain catastrophising [25], which is associated with soliciting

for the Study of Pain. Published by

arch Group, Medical School,ford Road, Manchester M13

tis@ioa.forthnet.gr (F. Creed).

support or empathy from others [4,24] and may relate to increasedpain and disability [25]. One study suggested that difficulty withbeing assertive, which was associated with persistent and diar-rhoea-predominant IBS, arose because the illness has a deleteriouseffect on interpersonal relationships but this study was cross-sec-tional and could not assess causality [26]. Furthermore the studydid not control for psychological distress, which is correlated withchronic pain and interpersonal relations [27].

In our trial of patients with severe IBS we found that both anti-depressants and psychotherapy led to improved heath status in thelong term but there was no apparent difference between the treat-ments. The first aim of the present study was to assess whetherchanges in interpersonal difficulties, symptoms of Irritable bowelsyndrome (IBS), health status and psychological state showed con-gruent changes over time.

Our second aim was a preliminary examination of whether theassociation between change in interpersonal relationships and out-come was different in different treatment groups. Both psycholog-ical treatments and antidepressants may help IBS patients[10,11,25,35]. Psychodynamic interpersonal therapy is designedto help people with their interpersonal difficulties, which may ex-plain how it helps some people with IBS in addition to relievingdepression and anxiety [16–19,21,33]. Antidepressants, on the

Elsevier B.V. All rights reserved.

T. Hyphantis et al. / PAIN� 145 (2009) 196–203 197

other hand, may help IBS patients by relieving pain in addition toanxiety and depression [10,11]. No previous study has examinedchange in IIP in relation to outcome in IBS or chronic pain.

We tested the following hypotheses in patients with severe IBS:

(1) that improvement in interpersonal relationships over15 months is associated with improvement in pain and dis-ability, but these relationships are mediated by psychologi-cal distress.

(2) that improved health status, the outcome measure whichshowed greatest long-term change in our trial, is associatedwith improvement in interpersonal difficulties followingboth psychotherapy and antidepressant treatments.

Prior to testing these hypotheses we assessed (a) whether thefactor structure of the brief IIP was similar in this population toprevious studies and (b) the baseline relationships between thevariables we tested in the longitudinal study.

2. Methods

For this study we used data that were collected during a ran-domised controlled trial of patients with severe, chronic IBS to as-sess the cost effectiveness of psychotherapy and antidepressants incomparison to treatment as usual [6]. We recruited from sevengastroenterology clinics in the UK all patients who fulfilled bothROME I criteria for IBS and the criteria for ‘‘severe” IBS. Rome I cri-teria require 3 months of continuous or recurring symptoms of: (1)abdominal pain, accompanied by pain relieved by defecation andassociated with change in frequency or consistency of stool; (2)at least two of the following: irregular pattern of defecation, al-tered stool consistency, incomplete rectal evacuation and/or ur-gency or straining; abdominal bloating or distension, and/ormucus in stools [36]. ‘‘Severe” IBS refers to patients with durationof symptoms >6 months, failure to respond to ‘‘usual” medicaltreatment, including antispasmodics and laxatives or antidiarrhealmedication administered for a minimum of 3 months and severeabdominal pain, defined as >59 on a visual analogue scale [13].

The trial involved random allocation of the patients to eight ses-sions of psychodynamic interpersonal therapy [16,18], or 3 monthsof treatment with 20 mg daily of the SSRI antidepressant, paroxe-tine, or routine care by gastroenterologist and general practitioner[6]. Patients allocated to psychotherapy received one long (approx-imately 2 h) and 7 shorter (45 min) individual sessions over3 months. They were encouraged to discuss their symptoms indepth; emotional factors were explored, and links between symp-toms and emotional factors were identified. Therapists weretrained by a member of the study team (E.G.) using a manualand a videotaped training package; continued conformity by thetherapist to the model was ensured by weekly supervision withE.G. [6]. After 3 months of treatment, all patients receiving psycho-therapy or paroxetine returned to their general practitioner, whodecided what further management was required over the nextyear. Patients were excluded from the trial if they had a psychoticdisorder, severe personality disorder, active suicidal ideation orconsumed more than 50 units of alcohol per week, but patientswith other psychiatric disorders were included.

The assessments we quote in this study were made at baseline(entry to the trial), after 3 months of treatment and at 12 monthsafter treatment was completed (i.e. 15 months after baseline). Fulldetails of the trial have been reported previously [6], including theCONSORT details, and will not be repeated here.

The following self-administered questionnaires were completedby each participant at each time point. Severity of current abdom-inal pain was assessed using visual analogue scales taken from theMcGill Pain Questionnaire, relating to the severity of ‘‘usual”

abdominal pain and its severity ‘‘today” [13]. In addition, each par-ticipant completed a daily diary recording the severity of their bo-wel symptoms for 14 days prior to each assessment. Psychologicaldistress was measured using the Global Severity Index (GSI) ofthe SCL-90 [12]. Health status was measured using the ShortForm-36 (SF-36) [39], which corresponds closely to patients’ ratingof the disruption their daily lives [20,40]. We used the physicalcomponent summary (SF-36-PCS) score as the main outcome var-iable; a low score indicates poor health status [38]. This is a com-posite score of the scales: physical function, role limitationphysical, bodily pain and health perception.

For the assessment of interpersonal problems, the Inventory ofInterpersonal Problems-32 (IIP-32) [2] was used. The IIP-32 is aself-report measure developed as a shortened version of the origi-nal 127-item Inventory of Interpersonal problems [22], aiming toassess the difficulties people experience in their interpersonal rela-tionships and comprises eight subscales which have shown highinternal consistency and confirmatory factor analysis has repli-cated the eight-factor structure [2]. We present results for baseline,3 and 15 months later.

At the initial assessment only, a trained psychiatrist, whoworked independent of treating clinicians and was blind to treat-ment group, assessed psychiatric diagnosis using the Schedules forClinical Assessment in Neuropsychiatry (SCAN) [41]. The detailsof the IBS symptom pattern (diarrhoea- or constipation-predomi-nant) and IBS duration were ascertained using the questionnaireof Drossman [36]. Diarrhoea-predominant IBS refers to patientswho had more than three bowel movements a day or watery stoolsor urgency or having to rush to have a bowel movement whereasconstipation-predominant IBS refers to patients who had fewerthan three bowel movements a week or lumpy stools or strainingduring a bowel movement.

A history of sexual abuse was documented using the Sexual andPhysical Abuse Questionnaire [14,28]. In this report sexual abuserefers to either forced touching or forced penetration (rape),against one’s will either as a child or adult.

2.1. Statistical analysis

All the statistical analyses were performed using the StatisticalPackage for the Social Sciences (SPSS) 15.0 (SPSS Inc., Chicago, IL,USA) for Windows and Stata Statistical Software: Release 9 (Col-lege Station, TX: Statacorp LP. 2005). Summary statistics for allvariables were calculated. Normality was tested by the Kolmogo-rov–Smirnov test [1].

Since we used the 32-item version of the IIP for the first time inIBS patients, a confirmatory principal component factor analysiswas performed to confirm that the factorial structure of this ver-sion in IBS patients is comparable to that of the original versionof the IIP-32 [2].

Univariate analyses to asses the independent associations be-tween demographic, clinical or psychopathology variables and IIPscores used one-way analyses of variance, two-tailed t-tests, andPearson’s or Spearman’s correlations as appropriate [1].

Repeated measures (ANOVA) for IIP mean score adjusted forbaseline score was performed to test for a significant change ofIIP over time (baseline, 3 months, and 15 months), and the rela-tionship of this with treatment.

To test hypothesis 1, we assessed change in IIP with change inabdominal pain severity, bowel symptoms, psychological distressand health status between baseline and 15 months, adjusting forbaseline scores with ANCOVA.

Multiple regression analyses were performed to assess whetherSCL-90 global severity index score mediated the association ofchange in pain and bowel symptoms with change in interpersonaldifficulties between trial entry and follow-up. These analyses used

198 T. Hyphantis et al. / PAIN� 145 (2009) 196–203

the method of Baron and Kenny [3]. To fulfil the requirements ofmediation according to this method, the following three associa-tions should be significant in multiple regression equations: (1)change in pain and change in IIP score, (2) change in pain andmediating variable (change in SCL-90 global severity index), and(3) mediating variable and change in IIP score. Addition ofmediating variable to multiple regression leads to relationship 1becoming non-significant. The first multiple regression analysis in-cluded change in IIP score between trial entry and follow-up asdependent variable with the following as independent variables:age, sex, years of education, widowed/separated or divorced, base-line scores of IIP, SF-36 Pain, diary diarrhoea and diary constipa-tion, treatment group (as two dummy variables, psychotherapyand antidepressants) and changes between baseline and15 months in SF-36 Pain, diary diarrhoea and diary constipationscores. In the second analysis, we added the SCL-90 global severityscore at baseline as well as its change between baseline and15 month and noted whether this rendered the previous associa-tions between change in pain and/or bowel symptoms and changein IIP score nonsignificant. The STATA impute command was usedto replace missing values in all the independent variables. Sobeltests were performed to gauge whether the mediator (change inSCL-90 global severity index) significantly carries the influence ofeach one significant independent variable to the dependent vari-able (i.e. change in interpersonal relationships). These analyses in-cluded the 214 patients on whom we had IIP data at baseline andfollow-up; the drop-outs were younger than the remainder but didnot differ on baseline IIP mean score, distress scores or IBS painvariables.

To test hypothesis 2, i.e. to assess the significant predictors ofthe improvement in the SF-36 physical component of health-re-lated quality of life in each treatment group, separate multipleregression analyses were performed for each treatment group.The dependent variable was the change of SF-36 physical compo-nent score between baseline and follow-up. All variables whichwere associated with change of SF-36 physical component scoresat p < 0.01 in the preceding univariate analyses in the entire sam-ple (data not shown) were entered into the multiple regressionanalysis as independent variables. Therefore, independent vari-ables were severe sexual abuse, age, unemployment due to poorhealth, baseline SF-36 physical component score, baseline IIPmean score, change of IIP mean score between baseline and15 months, change of SCL-90-R global severity index betweenbaseline and 15 months and change of VAS pain today score be-tween baseline and 15 months. Data were complete for all theseindependent variables for the patients with data on the dependentvariable, except for four patients with missing data on change inIIP score. The Stata impute command was used to replace thesefour missing values.

3. Results

3.1. Patient characteristics

The demographic and clinical characteristics of the participantsin our trial have been reported previously [6]. Briefly, a total of 257subjects (81% of eligible patients) were recruited to the study.There were no differences in demographic and diagnostic variablesbetween those patients who agreed and those who declined to en-ter the trial (N = 60). Ages ranged from 19 to 65 years, with a med-ian of 39 years. The majority of the patients were women (79.8%),married (65.8%), and 54.5% had 12 or more years of education.

The IBS was chronic (median duration, 8 years; interquartilerange, 9 years) and led to restricted activities on a mean of 12.1(SD, 11.8) days per month before baseline assessment, while sev-

enty patients (27.2%) were unemployed through illness. Mean typ-ical pain score was 67.4 of 100. Fifty-nine patients (22.9%) reportedthat they had experienced sexual abuse. Seventy four patients(28.8%) had diarrhoea-predominant IBS, 59 (23.0%) had constipa-tion-predominant IBS, and 124 (48.2%) had the general form ofthe disorder.

Of the 85 patients randomized to psychotherapy, 59 (69.4%)completed all eight sessions, and 43 of the 86 patients (50%) ran-domized to paroxetine completed the 12-week course (x2, 5.91;df, 1; p = 0.013).

The main findings of the study were that patients treated withpsychotherapy or antidepressants compared with those receivingusual care showed a significant improvement in health status12 months post treatment [6]. In the present study we examinewhether reduction in IIP score is associated with such improve-ment in health status.

3.2. Preliminary (factor) analysis of IIP-32

Of the 257 IBS patients, 225 (87.5%) completed all 32 items ofIIP and these were used in analysis. Principal component analysisshowed a Kaiser-Mayer-Olkin statistic of 0.86, an average commu-nality of 0.679 and a significant Bartlett’s test for sphericity (x2,4290; df, 496; p < 0.0005), all supporting the factorability of thecorrelation matrix. An inspection of the scree plot revealed a largefirst component and a number of ‘‘elbows” resulting in a very smalleigenvalue after eight components. An eight component extractionwas rotated with both Varimax and Oblimax rotations. Both ro-tated solutions revealed the presence of a coherent multidimen-sional structure with eight components. Table 1 presents theresults of the Oblimax rotation, since this rotation showed theclearer loadings. The eight components explain a cumulative70.75% of the variance. As shown in Table 1, all items have theirstrongest loadings exactly as suggested by Barckam et al. [2] withonly four of the possible 224 non-design loadings being above ofthe criterion of 0.3 and all but one are below 0.4, the remainingbeing 0.45. These results indicate that IIP-32 responses of the IBSpatients conform very well to the design of the instrument.

3.3. Interpersonal difficulties, demographic, clinical andpsychopathology variables at baseline

Table 2 shows that socio-demographic variables and symptompattern (diarrhoea-predominant, constipation-predominant andgeneral) were not significantly associated with IIP mean scorebut depressive, generalized anxiety and panic disorders were sig-nificantly associated. Further analysis (not shown) found that theseresults held when each of the eight IIP subscales were examinedindividually.

Table 3 shows that IIP mean score was associated with SCL-90global severity index but not with bodily pain, bowel symptomsor duration of IBS. Further analysis with the eight IIP subscalesshowed that no subscale score was associated with bowel symp-tom pattern (data not shown). The two subscales concerning diffi-culties with social inhibition and dependency (‘‘hard to besociable” and ‘‘too dependent”) were associated with the durationof IBS after adjustment for age, sex, marital status and psycholog-ical distress (partial correlation coefficients r = 0.166, p < 0.009 andr = 0.125, p < 0.05, respectively).

3.4. Change in interpersonal difficulties and change in pain, bowelsymptoms, psychological distress and health status

Repeated measures (ANOVA) analysis showed that IIP meanscore reduced with time (F = 11.84, p < 0.001) but there was no sig-nificant treatment by time interaction (F = 0.79, p = 0.45). This

Table 1Factor loadings of the IIP-32 in the IBS sample.

Factors

Items Hard to be sociable Hard to be supportive Too aggressive Too open Too caring Hard to be assertive Hard to be involved Too dependent

7 .9013 .8671 .7379 .71914 .87915 .87513 .78616 .55528 .92330 .89220 .82721 .79124 .83929 .79810 �.463 .32717 �.455 .39125 .86526 .79918 .65632 .5474 .84211 .8326 .7742 .73319 .7585 .63312 .458 .4578 .364 .35223 .71731 .60322 .60027 .444a 0.89 0.86 0.87 0.63 0.78 0.85 0.80 0.75

Note.Jbliminrotation with Kaisernormalization; loadings censored at 0.3; item loadings that correspond to the loadings of the original version are highlighted with bold characters.

Table 2IIP-32 total mean score and demographic, clinical and psychiatric categoricalvariables.

N Mean SD p

SexMale 50 1.16 0.04Female 197 1.06 0.09 t = �1.01 0.311

EducationLess than GCSE 112 1.04 0.63GCSE or more 135 1.11 0.65 t = �0.86 0.391

Unemployment due to ill healthNo 179 1.07 0.63Yes 68 1.11 0.68 t = �0.40 0.688

Marital statusMarried 159 1.15 0.74Singles 50 1.11 0.62 F2,244 = 2.78 0.064Div/Sep 38 0.86 0.57

Sexual abuseNo 188 1.05 0.64Yes 59 1.18 0.63 t = �1.36 0.173

Rome diagnosis (symptom pattern)General 124 1.07 0.63Diarrhoea predominant 74 1.17 0.66 F2,244 = 1.29 0.277Constipation predominant 59 0.99 0.63

Depressive disorderNo 174 0.93 0.58Yes 73 1.43 0.64 t = �5.96 <0.0005

Generalized anxiety disorderNo 212 1.04 0.64Yes 35 1.33 0.61 t = �2.48 0.014

Panic disorderNo 218 1.01 0.61Yes 29 1.61 0.61 t = �4.97 <0.0005

T. Hyphantis et al. / PAIN� 145 (2009) 196–203 199

reduction of IIP mean score between baseline and 15 months fol-low-up was associated with the reduction of all measurements ofpain and with improvement of disability (Table 4).

Fig. 1a and b show the temporal sequence of changes. In thepsychotherapy and antidepressant groups improvement in painand psychological distress occur during the 3 months of treatmentbefore the reduction in IIP; the latter occurs predominantly duringthe follow-up year. This means that change in IIP should not be re-garded as the primary change with treatment as the other param-eters (pain, psychological distress and health status) show changebefore change in IIP.

A much greater change in IIP score is apparent at follow-up andmultiple regression analysis was used to identify the variablesmost closely associated with this change; they are shown in

Table 3Correlations of baseline IIP-32 total mean score with demographic, baseline clinicaland baseline psychiatric continuous variables.

ra p

Age �0.031 0.623Age of onset �0.099 0.122Duration of IBS 0.077 0.229VAS typical pain �0.099 0.122VAS pain today 0.057 0.375Number of days with pain 0.016b 0.826SF-36 physical component 0.022 0.731SF-36 pain �0.037 0.558Diarrhoea 0.002 0.998Constipation 0.034 0.780SCL-90 Global severity index 0.676 <0.0005

a Pearson’s correlation coefficients except:b Spearman’s correlation coefficient VAS = visual analogue scale.

Table 4Associations of change in IIP mean score with change in psychological distress, pain,diary measure diarrhoea and diary measure constipation scores between baseline andfollow-up, adjusted for baseline scores.

Correlation coefficient p-value

Change in VAS pain today 0.201 0.003Change in VAS pain typical 0.208 0.003Change in diarrhoea 0.301 <0.005Change in constipation 0.200 0.008Change in SF-36 physical component 0.236 0.001Change in SF-36 health perceptions 0.355 <0.0005Change in SF-36 pain 0.281 <0.0005Change in psychological distress (SCL-90 GSI) 0.567 <0.0005

200 T. Hyphantis et al. / PAIN� 145 (2009) 196–203

Table 5. Model 1 shows that the improvement of pain was a signif-icant predictor of improvement in interpersonal difficulties, butthis is not so in model 2, when the improvement in SCL-90 globalseverity index was added to the linear regression equation.

Sobel tests showed that the indirect effect of each one signifi-cant independent variable (i.e. change in pain, change in diary diar-rhoea and change in diary constipation) on the dependent variable(i.e. change in interpersonal relationships) through the mediator(i.e. change in SCL-90 global severity index) was significant (Sobeltest statistics (Z): 4.59, p < 0.0005; 3.62, p < 0.0004 and 3.09,p < 0.002, respectively). Sobel tests for the indirect effects of base-line values of the independent variables on the dependent variablethrough change in SCL-90 global severity index values were not

Psychotherapy group

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

z-sc

ore

IIP

SF-36 pcs

SCL-90 global severity index

VAS pain today

Paroxetine group

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

3 months 15 monthsenilesab

3 months 15 monthsenilesab

z-sc

ore IIP

SF-36 pcs

SCL-90 global severity index

Vas pain today

a

b

Fig. 1. Effect size for changes from baseline for IIP-32, SCL-90 global severity indexSF-36 physical component summary and VAS-pain today in psychotherapy (a) andantidepressant group (b).

significant. This indicates that change in psychological distressmediates the relationships between improvement of pain andimprovement in interpersonal difficulties.

3.5. Health status improvement and treatment modality (hypothesis 2)

Table 6 shows the results of the separate multiple regressionanalyses performed for each treatment group with SF-36 physicalcomponent score as the dependent variable. It can be seen thatfor both, psychotherapy and antidepressant groups, the baselinevalue of this variable was a significant predictor. However, in thepsychotherapy group, the other independent predictors of changein SF-36 physical component score were change in IIP mean scoresbetween baseline and 15 months, reported severe sexual abuseand unemployment, but not change in SCL-90 global severity in-dex. By contrast, in the antidepressant group, age and change inSCL-90 global severity index were the only additional independentpredictors of change in SF-36 physical component score.

4. Discussion

There are several new findings in this study. First, in patientswith severe IBS interpersonal problems concerning difficultieswith social inhibition and dependency, were associated with long-er disease duration, after controlling for psychological distress. Wedid not find that the overall mean score of the IIP was associatedwith duration or with diarrhoea-predominant IBS after adjustmentfor psychological distress. Thus we only partially confirmed theprevious findings of Lackner and Gurtman [26] regarding durationof IBS and interpersonal difficulties.

Second, we found that change in IIP score over time was clearlyassociated with change in pain, bowel symptom pattern and healthstatus and this association was mediated by psychological distress.Reduction of psychological distress, and improvement of pain, oc-curred mostly during the 3 months of treatment whereas change inIIP score occurred mostly during the subsequent 1 year, indicatingthat the improvement of psychological distress precedes theimprovement of interpersonal difficulties. This suggests that theimprovement in IIP score may be secondary to reduction of distressrather than a direct result of the therapy.

Third, IIP mean score (baseline and change values) showed bor-derline significance as predictors of our main trial outcome,improvement in health status (SF-36 physical component score)only in the psychotherapy group. This contrasts with the antide-pressant group where the reduction of psychological distress wasassociated with improved health status. This raises the possibilitythat for those receiving psychotherapy, reduction of distress maynot be the only driver for improved health status. It is possible thatthe improvement in health status over the follow-up year is linkedin some more direct way to improved interpersonal relationships.A small accompanying qualitative study suggested that peoplereceiving psychotherapy were twice as likely as those receivingantidepressants to appreciate that stress affected their bowelsymptoms and had made changes in their lives to reduce stressduring the follow-up year (C. Rigby, personal communication). Thismay have included improved interpersonal relationships and mayhave been related to improvement in health status. This conceptneeds to be tested in future research.

It has been suggested that incorporating strategies to addressinterpersonal concerns may increase the efficacy of treatment ofchronic pain as chronic pain patients are said to be overly nurtur-ant, exploitable, non-assertive and socially avoidant according tothe circumplex version of IIP [29,31,37]. Our results suggest thatthere may be benefit in doing so. On the one hand we have shownthat interpersonal relationships are related to the level of psycho-logical distress and that distress mediates the relationship between

Table 5Multiple regression analyses to predict change in IIP mean score between baseline and follow-up (N = 214).

Predictor variables Model 1; change in IIP as dependent variable Model 2 (final); change in IIP as dependent variable

Beta p Beta p

Age 0.006 0.925 0.002 0.968Sex �0.001 0.992 0.024 0.579Education �0.034 0.549 �0.001 0.974Widowed, divorced, separated �0.022 0.705 0.003 0.951Baseline IIP-32 mean score 0.521 <0.0005 0.775 <0.0005Baseline SF-36 pain 0.158 0.014 �0.037 0.469Baseline diary diarrhoea �0.188 0.004 �0.089 0.073Baseline diary constipation �0.026 0.684 0.037 0.447SF-36 pain improvement between baseline and 15 months 0.149 0.033 �0.028 0.607Diary diarrhoea improvement between baseline and

15 months0.201 0.003 0.100 0.059

Diary constipation improvement between baseline and15 months

0.133 0.041 �0.038 0.453

Psychotherapy 0.014 0.829 �0.008 0.868Antidepressant 0.004 0.953 �0.005 0.915Baseline SCL-90-R GSI – – �0.626 <0.0005SCL-90-R GSI improvement between baseline and 15 months – – 0.603 <0.0005Regression statisticsR square adjusted 0.343 0.632F-values F(13,200) = 9.53, p < 0.0005 F(15,198) = 25.4, p < 0.0005

T. Hyphantis et al. / PAIN� 145 (2009) 196–203 201

pain and IIP, so reducing distress should be an important aspect ofmanaging chronic pain. On the other hand our results are compat-ible with the suggestion that improved health status one year aftertreatment with psychotherapy may be associated with improvedinterpersonal relationships.

Our results are not identical to those of Lackner as we found so-cial inhibition and difficulties with dependency were associatedonly with the duration of IBS. The difference probably reflects thefact that we adjusted for psychological distress and all our patientshad severe pain as a selection criterion of our trial. This means thatpain scores, and others, showed little variation at baseline whereasthe greater spread of scores after 15 months could explain why our

Table 6Multiple regression analyses to predict change in SF-36 physical component summary sco

Variables Unstandardized regressioncoefficient (B)

Psychotherapy groupSevere sexual abuse 10.317Age �0.152Baseline SF-36 physical component score �0.512Unemployed due to poor health �5.427Baseline IIP-32 mean score �3.299IIP-32 mean improvement between

baseline and 15 months4.59

SCL-90-R GSI improvement betweenbaseline and 15 months

3.992

VAS pain today improvement betweenbaseline and 15 months

0.003

Adjusted R square 0.428 (F = 6.43,p < 0.0005)

Antidepressant groupSevere sexual abuse 4.510Age �0.172Baseline SF-36 physical component score �0.375Unemployed due to poor health �2.845Baseline IIP-32 mean score �1.913IIP-32 mean improvement between

baseline and 15 months�0.554

SCL-90-R GSI improvement betweenbaseline and 15 months

4.689

VAS pain today improvement betweenbaseline and 15 months

0.040

Adjusted R square 0.301 (F = 4.82, p < 0.0005)

change scores showed clear association between pain measuresand IIP when baseline scores had not.

The association between certain interpersonal difficulties andduration of IBS, even after adjustment for psychological distress,is compatible with the suggestion that interpersonal difficultiesare a consequence of a chronic painful condition [34]. Pain patientswith a very long history have been said to perceive their spouses asless supportive and more punitive than patients with shorter dura-tion pain [5]. On the other hand, people with difficulties withassertiveness or social inhibition may develop IBS early in lifeand/or cope poorly with the symptoms of this relapsing and remit-ting illness [23].

re between baseline and follow-up for psychotherapy and antidepressant groups.

Std. Error Standardized regressioncoefficient (beta)

p

3.502 0.324 0.0050.092 �0.192 0.1040.114 �0.495 <0.00052.262 �0.267 0.021.761 �0.244 0.0672.34 0.276 0.05

2.376 0.198 0.099

0.034 0.010 0.933

2.451 0.185 0.0700.082 �0.221 0.0410.094 �0.488 <0.00052.411 �0.146 0.2421.656 �0.136 0.2522.209 �0.041 0.803

2.238 0.328 0.040

0.032 0.136 0.214

202 T. Hyphantis et al. / PAIN� 145 (2009) 196–203

There are some limitations of this study, which need to be rec-ognized. First, this is a secondary analysis of a dataset that was col-lected for another purpose. Another limitation lies in the absenceof a healthy control group, which could allow us clearer conclu-sions with regard to IBS patients’ specific interpersonal difficulties.In addition, our results pertain to patients with severe IBS and maynot hold for people with less severe IBS. We used only a self-reportquestionnaire to assess interpersonal difficulties and it is not clearwhether such an instrument measures the discomfort in relation-ships that is associated with eliciting sympathetic response fromothers [37]. Similarly we did not measure catastrophising whichmeans we cannot compare our results with those which foundan association between interpersonal problems and pain catastro-phizing [4,24–26].

This paper does overcome some of the limitations of the previ-ous studies, however, notably the prospective design. Strengths ofour study include also the representative nature of our sample(81% of the eligible patients were recruited), the detailed measures,our ability to demonstrate the factorial structure of the maininstrument used (IIP-32) and the reasonably large sample size, atleast as compared to other studies using IIP.

The prospective design and response to treatment are the majorassets of this study. We found that interpersonal problems im-proved over time and this is, we believe, the first study that showsthis in IBS patients. We found a clear association between improve-ment in IIP score and improvement in all our measures of pain anddisability and we also found that this was mediated by change inpsychological distress. This occurred irrespective of treatmentgroup and it is most plausible that improved mood leads to feelingbetter about interpersonal relationships. The additional finding isalso new-improvement in IIP was associated directly with im-proved health status in the psychotherapy group but not in theantidepressant group. This finding suggests, perhaps, that themode of action of antidepressants and psychotherapy may not beidentical. We interpret this finding as suggesting that a componentof the change following psychotherapy was some aspect ofimprovement of interpersonal difficulties (sexual abuse was alsoa predictor of improved health status [7]) whereas following anti-depressant treatment a reduction in the level of distress was theprimary change [6].

It has not been possible to identify a single major predictor ofoutcome in this group of patients with severe IBS, unlike the studyof temporomandibular disorder patients reported by Rudy [32].This probably results from our selection of a relatively homoge-nous group of patients with severe IBS. We have shown previouslythat depressive and panic disorders, neurasthenia and markedsomatisation predict a poor outcome [8,9] and that a history of sex-ual abuse predicts a good response [7]. In this paper we aimed toidentify whether interpersonal difficulties also predict outcomeand it seems that this is the case in the psychotherapy group,though the effect was of borderline significance, possibly reflectingsmall sample size. Since this analysis showed that a history of sex-ual abuse and interpersonal difficulties are independent predictorsof outcome in the psychotherapy, even after controlling for thereduction of distress, we conclude that specific interpersonalchange following our brief psychotherapy is an important aspectof the long-term outcome following this type of treatment. This ef-fect was not found in the antidepressant group suggesting that thetwo treatments act differently in this respect. Both aspects of treat-ment could be utilised in the treatment of chronic pain patients, inaddition to cognitive behavioural treatment, which acts principallyon different aspects of the person’s response to pain.

The main clinical implication of this study is that IBS patients’interpersonal difficulties improved over time and that thisimprovement was associated with improved health status in bothtreatment groups. Therefore, the choice of treatment might be

determined by patient preference or availability of psychotherapy.Future research should aim to clarify the specific mechanism of ac-tion of psychotherapy in IBS patients and further study of the rela-tionship between patients’ interpersonal difficulties and IBS.

Acknowledgments

This work supported by the Medical Research Council of theUnited Kingdom and the North Western Region Health Authority(U.K.) R&D Directorate. There was not any financial interest whichcould create a potential conflict of interest or the appearance of aconflict of interest with regard to the present submitted work.SmithKlineBeecham provided the paroxetine but was not involvedin the design, conduct, or analysis of the trial.

References

[1] Altman DG. Practical statistics for medical research. London: Chapman andHall; 1991. p. 285.

[2] Barkham M, Hardy GE, Startup M. The IIP-32: a short version of the inventoryof interpersonal problems. Br J Clin Psychol 1996;35:21–35.

[3] Baron RM, Kenny DA. The moderator–mediator variable distinction in socialpsychological research: conceptual, strategic, and statistical considerations. JPers Soc Psychol 1986;51:1173–82.

[4] Buenaver LF, Edwards RR, Haythornthwaite JA. Pain-related catastrophizingand perceived social responses: inter-relationships in the context of chronicpain. Pain 2007;127:234–42.

[5] Cano A. Pain catastrophizing and social support in married individuals withchronic pain: the moderating role of pain duration. Pain 2004;110:656–64.

[6] Creed F, Fernandes L, Guthrie E, Palmer S, Ratcliffe J, Read N, Rigby C,Thompson D, Tomenson B. North of England IBS Research Group. The cost-effectiveness of psychotherapy and paroxetine for severe irritable bowelsyndrome. Gastroenterology 2003;124:303–17.

[7] Creed F, Guthrie E, Ratcliffe J, Fernandes L, Rigby C, Tomenson B, Read N,Thompson DG. Reported sexual abuse predicts impaired functioning but agood response to psychological treatments in patients with severe irritablebowel syndrome. Psychosom Med 2005;67:490–9.

[8] Creed F, Ratcliffe J, Fernandes L, Palmer S, Rigby C, Tomenson B, Guthrie E, ReadN, Thompson DG. North of England IBS Research Group. Outcome in severeirritable bowel syndrome with and without accompanying depressive, panicand neurasthenic disorders. Br J Psychiatry 2005;186:507–15.

[9] Creed F, Tomenson B, Guthrie E, Ratcliffe J, Fernandes L, Read N, Palmer S,Thompson DG. The relationship between somatisation and outcome inpatients with severe irritable bowel syndrome. J Psychosom Res2008;64:613–20.

[10] Creed F. Antidepressants and psychological interventions for functionalgastrointestinal disorders. Gastroenterol Hepatol 2006;1:62–6.

[11] Creed F. How do SSRIs help patients with irritable bowel syndrome? Gut2006;55:1065–7.

[12] Derogatis LR. The SCL-90-R Manual-II: scoring* administration and proceduresfor the SCL-90-R. 2nd ed. Towson, MD: Clinical Psychometric Research; 1992.

[13] Drossman DA, Li Z, Toner BB, Diamant NE, Creed FH, Thompson D, Read NW,Babbs C, Barreiro M, Bank L, Whitehead WE, Schuster MM, Guthrie EA.Functional bowel disorders: a multicenter comparison of health status, anddevelopment of illness severity index. Digest Dis Sci 1995;40:986–95.

[14] Drossman DA, Talley N, Leserman J, Olden KW, Barreiro MA. Sexual andphysical abuse and gastrointestinal illness: a review and recommendations.Ann Intern Med 1995;123:782–94.

[15] Drossman DA. The functional gastrointestinal disorders and the Rome IIIprocess. Gastroenterology 2006;130:1377–90.

[16] Guthrie E, Creed F, Dawson D, Tomenson B. A controlled trial of psychologicaltreatment for the irritable bowel syndrome. Gastroenterology1991;100:450–7.

[17] Guthrie E, Kapur N, Mackway-Jones K, Chew-Graham C, Moorey J, Mendel E,Francis FM, Sanderson S, Turpin C, Boddy G. Predictors of outcome followingbrief psychodynamic-interpersonal therapy for deliberate self-poisoning. AustNZ J Psychiatry 2003;37:532–6.

[18] Guthrie E. Brief psychotherapy with patients with refractory irritable bowelsyndrome. Br J Psychother 1991;8:175–88.

[19] Guthrie E. Psychodynamic interpersonal therapy. Adv Psychiat Treat1999;5:135–45.

[20] Hahn BA, Kirchdoerfer LJ, Fullerton S, Mayer E. Patient-perceived severity ofirritable bowel syndrome in relation to symptoms, health resource utilizationand quality of life. Aliment Pharmacol Therap 1997;11:553–9.

[21] Hamilton J, Guthrie E, Creed F, Thompson D, Tomenson B, Bennett R, MoriartyK, Stephens W, Liston R. A randomized controlled trial of psychotherapy inpatients with chronic functional dyspepsia. Gastroenterology2000;119:661–9.

[22] Horowitz LM, Rosenberg SE, Baer BA, Ureño G, Villaseñor VS. Inventory ofinterpersonal problems: psychometric properties and clinical applications. JConsult Clin Psychol 1988;56:885–92.

T. Hyphantis et al. / PAIN� 145 (2009) 196–203 203

[23] Jones MP, Wessinger S, Crowell MD. Coping strategies and interpersonalsupport in patients with irritable bowel syndrome and inflammatory boweldisease. Clin Gastroenterol Hepatol 2006;4:474–81.

[24] Keefe FJ, Lipkus I, Lefebvre JC, Hurwitz H, Clipp E, Smith J, Porter L. The socialcontext of gastrointestinal cancer pain: a preliminary study examining therelation of patient pain catastrophizing to patient perceptions of social supportand caregiver stress and negative responses. Pain 2003;103:151–6.

[25] Lackner JM, Gurtman MB. Pain catastrophizing and interpersonal problems: acircumplex analysis of the communal coping model. Pain 2004;110:597–604.

[26] Lackner JM, Gurtman MB. Patterns of interpersonal problems in irritable bowelsyndrome patients: a circumplex analysis. J Psychosom Res 2005;58:523–32.

[27] Leonard MT, Cano A, Johansen AB. Chronic pain in a couples context: a reviewand integration of theoretical models and empirical evidence. J Pain2006;7:377–90.

[28] Leserman J, Drossman DA, Li Z, Toomey TC, Nachman G. The reliability andvalidity of a sexual and physical abuse history questionnaire in female patientswith gastrointestinal disorders. Behav Med 1995;21:141–50.

[29] McWilliams LA, Asmundson GJ. The relationship of adult attachmentdimensions to pain-related fear, hypervigilance, and catastrophizing. Pain2007;127:27–34.

[30] Meredith P, Strong J, Feeney JA. Adult attachment, anxiety, and pain self-efficacy as predictors of pain intensity and disability. Pain 2006;123:146–54.

[31] Monsen K, Havik OE. Psychological functioning and bodily conditions inpatients with pain disorder associated with psychological factors. Br J MedPsychol 2001;74 Part 2:183–95.

[32] Rudy TE, Turk DC, Kubinski JA, Zaki HS. Differential treatment responses ofTMD patients as a function of psychological characteristics. Pain 1995;61:103–12.

[33] Shapiro DA, Firth JA. Exploratory therapy manual for the SheffieldPsychotherapy Project (Memo No.733) [Available from University of Leeds,Leeds, England].

[34] Silk DB. Impact of irritable bowel syndrome on personal relationships andworking practices. Eur J Gastroenterol Hepatol 2001;13:1327–32.

[35] Spanier JA, Howden CW, Jones MP. A systematic review of alternativetherapies in the irritable bowel syndrome. Arch Intern Med 2003;163:265–74.

[36] Thompson WG, Creed F, Drossman DA, Heaton KW, Mazzacca G. Functionalbowel disease and functional abdominal pain. Gastroenterol Int 1992;5:75–91.

[37] Thorn BE, Keefe FJ, Anderson T. The communal coping model and interpersonalcontext: problems or process? Pain 2004;110:505–7.

[38] Ware JE, Kosinski M, Keller SD. SF-36 physical and mental health summaryscales: a user’s manual. Boston: The Health Institute, New Engl. Med. Centre;1994.

[39] Ware JE, Sherbourne CD. The MOS 36-item short-form health survey(SF-36): 1. Conceptual framework and item selection. Med Care 1992;30:473–83.

[40] Whitehead WE, Burnett CK, Cook III E, Taub E. Impact of irritable bowelsyndrome on quality of life. Digest Dis Sci 1996;41:2248–53.

[41] World Health Organization. Schedules for clinical assessment inneuropsychiatry. Geneva, Switzerland: Division of Mental Health; 1994.