- Email: [email protected]
Pathways connecting cognitive behavioral therapy and change in bowel symptoms of IBS
Journal of Psychosomatic Research 70 (2011) 278 – 285
Pathways connecting cognitive behavioral therapy and change in bowel symptoms of IBS Michael Jones a,⁎, Natasha Koloski b , Philip Boyce c , Nicholas J. Talley d a Psychology Department, Macquarie University, North Ryde, NSW, Australia School of Psychology, University of Queensland, St. Lucia, Queensland, Australia c Discipline of Psychiatry, Sydney Medical School-Westmead, Westmead Hospital, Wentworthville, NSW, Australia d Faculty of Health, University of Newcastle, Callaghan, NSW, Australia b
Received 13 January 2010; received in revised form 12 August 2010; accepted 2 October 2010
Abstract Objective: A single previous paper on this topic found a direct pathway between cognitive behavioral therapy (CBT) and an irritable bowel syndrome (IBS) global symptom score. This is controversial since under the biopsychosocial model, the expectation is that CBT's effect would be mediated by mood. Using more sensitive bowel symptom scales and measurements at additional time points, we aimed to compare the relative strengths of direct pathways between CBT and change in IBS symptoms and indirect pathways that operate via mood state using structural equation modeling. Methods: Our data set included 105 people with Rome I IBS randomized to individual CBT (n=34), relaxation therapy (n=36), and usual medical care (n=35). The primary outcome was defined as adequate relief of IBS symptoms in terms of the distress, frequency, and impairment according to the Bowel Symptom
Severity Scale. Outcomes in functional status (according to the 36item Short-Form Health Survey) and psychological status (Hospital Anxiety and Depression Scale) were secondary outcomes. Results: Our data suggest indirect pathways that operate via mood, most clearly anxiety but to a lesser extent depression. Statistically significant pathways were identified that lead from CBT to change in mood state thence to change in bowel symptoms, followed by further changes in mood then changes in bowel symptoms. Our data provide no evidence of direct effect of CBT on bowel symptoms. Conclusions: The present study suggests that CBT may operate via changes in mood state while not ruling out the possibility of direct effects. Our findings do not directly support, but are consistent with, a biopsychosocial model. © 2011 Published by Elsevier Inc.
Keywords: Cognitive behavior therapy; CBT; Iritable bowel syndrome; IBS; Mood; Bowel symptoms
Introduction Despite a prolonged period of study, few very effective treatments for the irritable bowel syndrome (IBS) exist and management often remains elusive [1]. While a number of pharmacological agents show promise [2], none can yet be considered adequate treatment for the full spectrum of IBS presentations to physicians [3]. Psychological therapies [cognitive behavioral therapy (CBT), relaxation therapy, behavioral therapy, psychotherapy, and hypnotherapy] have been studied based on evidence that a considerable ⁎ Corresponding author. Psychology Department, C3A 516, Macquarie University, North Ryde, NSW 2109, Australia. Tel.: +61 2 9850 8601; fax: +61 2 9850 8062. E-mail address: [email protected] (M. Jones). 0022-3999/10/$ – see front matter © 2011 Published by Elsevier Inc. doi:10.1016/j.jpsychores.2010.10.004
proportion of patients with IBS experience comorbid psychological disturbances such as anxiety and depression [4], but with inconsistent results [5]. The need to identify effective treatments for IBS is crucial, with estimates that IBS costs the United States at least $8 billion dollars annually [6] and is associated with significant impairments in QOL for its sufferers [7]. Of all the psychological therapies, CBT has received the most attention as a psychological treatment for IBS [5]. CBT consists of four important components (current health status, thoughts, behaviors, and emotions) and their interaction with one another to explain a person's pain or anxiety [8]. For example, John is a 66-year-old retired, married man who has weekend plans to finish painting his wife's bookcases (behavior). Unfortunately, he wakes up feeling ill on Saturday morning (health) and is unable to complete the
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285
project (behavior). As a result, he feels angry and anxious about not getting to his work (emotions), believing that he is disappointing his wife (thought). He thinks, “My wife will think that I do not care about helping her decorate the study.” This belief raises his anxiety (emotion) as well as his frustration about not feeling up to par. This makes it even harder for John to figure out how to face the day, and consequently he stays in bed (behavior), which in turn only serves to raise his anxiety and strengthens his negative thoughts about his wife's reaction. This is shown in Fig. 1. The empirical evidence generally favors the efficacy of CBT in relieving symptoms and improving the quality of life (QoL) of patients with IBS [9–17]. Although many of the earlier studies reporting a positive benefit of CBT in IBS have been criticized on methodological grounds [9–15], the more recent methodologically rigorous studies are more convincing [16–17]. Blanchard et al. [16] found a modest although clear benefit attributable to CBT in global IBS symptoms. Boyce et al. [17] showed that CBT improved IBS symptoms, anxiety levels, and physical functioning post treatment, although they found no discernible difference in symptom response to CBT compared with relaxation therapy or “usual care.” Interestingly, little effort appears to have been expended in understanding how CBT might relieve symptoms in IBS. The biopsychosocial model provides a theoretical framework for explaining how CBT might affect IBS symptoms (Fig. 2) [18]. According to this model, early life factors (genetic predisposition and environmental factors) influence and interact with the later development of psychosocial risk factors (e.g., psychological disorders, coping styles) and physiological dysfunction (e.g., motility, visceral sensitivity) and their interaction via the brain–gut axis, increasing susceptibility to developing IBS [18]. Hence, it is postulated that a primary disturbance in gut sensitivity or motility, or both, interacts with disturbed processing in the central nervous system (CNS)—in particular, emotional regulation that is itself modulated by psychosocial stressors. Thus, it is suspected that psychological treatments such as CBT work by reducing the impact of CNS activity on gut function in patients with
Fig. 1. Cognitive/behavioral model.
279
Fig. 2. Biopsychosocial model of FGIDs and emotion.
IBS. For example, CBT interventions systematically teach patients the cognitive and behavioral skills needed to better manage stress, anxiety, and/or IBS symptoms and its sequelae. These skills can reduce the fear associated with stress, anxiety, or IBS symptoms, helping patients approach normal activity and assisting with management of the emotional consequences of living with IBS. By such intervention, a more positive cycle of behavioral and cognitive change is possible, resulting in improvements in psychological and physical functioning and reductions in IBS symptoms. Despite the usefulness of this model, however, it is difficult to validate comprehensively. Lackner et al. [19] were the first to attempt to explain the relationship between a group-centered CBT intervention and IBS symptom response. They adopted a path model approach to understanding the relationship between undertaking CBT and response in symptom reports; these authors concluded that CBT works directly on IBS symptoms. We found this result very surprising since it runs counter to the biopsychosocial model, which argues for a connection between symptom perception and emotion within the CNS based on a feedback loop between perception and emotions. The aim of the current work is to provide further empirical evidence in the debate over whether CBT is associated with improvements in bowel symptoms through direct effects or because it affects mood and therefore operates indirectly on bowel symptoms. Our work, while superficially replicating that of Lackner et al. [19], also provides a fresh look at a biopsychosocial model view of the role of CBT in IBS using an independent trial of CBT compared with active and control groups and more time points than that used by Lackner et al. [19]. We have taken an approach that prespecifies a path model, which includes both direct and indirect paths between CBT and subsequent changes in both mood and bowel symptoms.
280
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285
Methods Study subjects Subjects were recruited through a tertiary-referral hospital outpatient clinic and newspaper advertisements. To be eligible for the study, participants needed to have a diagnosis of IBS made by a gastroenterologist according to the Rome I criteria [20]. The Rome I criteria were considered to be the most valid standardized criteria at the time of the study. This diagnosis was also confirmed by an independent assessor using a Structured Interview for Bowel Symptoms. Using a structured format, each symptom of the Rome criteria is probed and then rated as present or absent. If the symptom is present, an enquiry is made about the duration of the symptoms to ensure that they meet the duration criteria specified in the Rome criteria. Other inclusion criteria included no structural bowel pathology that would account for symptoms, a sufficient understanding of English to be able to understand the therapy, and an age of 18 years or older. Subjects were excluded if they answered positively to an interviewer's questions regarding the presence of any major current medical or psychotic illness, a history of alcoholism, current psychological treatment and current use of antidepressants or antipsychotic medications, or current use of medications that could affect bowel function (e.g., antispasmodics, prokinetics). Interventions Subjects were randomized into one of three therapy conditions: usual medical care alone, relaxation training, and CBT. Individuals randomized to usual medical care alone (n=34) received three sessions with a gastroenterologist that included discussion of symptoms and dietary advice and written information on the same was handed out. Individuals randomized to relaxation training (n=36) received routine clinical care but also saw a clinical psychologist who provided 8 weeks of relaxation strategies in 30-min face-toface sessions. Individuals randomized to receive CBT (n=35) also received routine clinical care and relaxation training. The CBT intervention consisted of 8 weeks of 60-min faceto-face sessions with a clinical psychologist. The CBT approach was a manual-based program that incorporated realistic symptom appraisal, enhanced coping strategies, cognitive restructuring, and problem solving. For the purpose of this analysis, interest lay in comparing symptom response among those receiving CBT with those not receiving CBT, and hence, the routine clinical care and relaxation training groups have been combined into a single non-CBT group.
adequate before randomization, and to assess the effects of treatment on baseline symptoms. Randomization was conducted blinded to investigators of the study. A series of sealed opaque envelopes were prepared containing a card with the treatment conditions. They were randomly allocated participant identification numbers (using a random number generator) before the start of the study, with a check to ensure that equal numbers were allocated to each treatment group. After randomization, eight sessions of treatment took place weekly over an 8-week period for all subjects in the CBT and relaxation arms. The same clinical psychologist treated patients in both the relaxation and CBT groups. Data were collected at randomization (baseline), mid-treatment (4 weeks after baseline), end of treatment (at 8 weeks), and at 26 and 52 weeks of follow-up. To avoid therapist bias, all of the self-report scales were completed by the patients with the therapist blind to the results. Data were coded and scored at the conclusion of treatments. None of the subjects received additional counseling or psychological treatment during the course of the study. Assessment All subjects completed a range of self-report measures without any assistance from the investigators. GI symptom severity was assessed using the Bowel Symptom Severity Scale (BSSS) [14]. The BSSS measures frequency, disability, and distress for each of eight GI symptoms associated with IBS (loose stools, hard stools, abdominal pain, more than three bowel motions daily, bloating, urgency to defecate, inability to have a bowel motion in the past week, and abdominal discomfort) over the past week. Each item is rated on a six-point scale. Scores are summed to give a measure of bowel symptom frequency, disability and distress. The BSSS was developed by Boyce et al. (2000) for a pilot study as no other measure of bowel symptom severity was available. The results from the pilot study showed that the BSSS was reliable in terms of internal consistency (α=.88), responsive to change and had good face validity (Boyce et al., 2000) and thus was chosen for the main study reported here. Psychological distress was assessed with the reliable and valid Hospital Anxiety and Depression Scale (HAD) [21]. The questionnaire contains 14 items that assess anxiety (seven items) and depression (seven items) (Zigmond & Snaith, 1983). Each item is rated on a four-point scale ranging from 0 to 3, such that each subscale has a maximum score of 21. Higher scores on the HAD indicate increased severity on the depression or anxiety subscales, with a score of 7 or less on either subscale indicating a “non case,” 8–10 as a “doubtful case,” and a score of 11 or more as a “case” of anxiety or depression (Zigmond & Snaith, 1983).
Procedure Outcomes Participants completed a 2-week washout and review period before randomization. A baseline washout period is useful to ensure that symptom frequency and severity are
The primary outcome was bowel symptom severity in terms of frequency, distress, and disability.
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285
281
Models
Table 1 Baseline and change scores for bowel symptoms and emotional state
Our prespecified path model postulates both direct effects of CBT on change in bowel symptoms and a bioemotional feedback mechanism in which change in emotional state leads to improved symptom perception, which feeds back to improved emotional state in an ongoing cycle. This model is presented in more technical detail in Fig. 2. Only one of the three treatment groups was CBT; however, relaxation therapy is an element of CBT. In our primary analysis, relaxation therapy has been grouped with usual care but we have undertaken a sensitivity analysis in which relaxation therapy is grouped with CBT.
Score
Statistical approach The aim of the analysis is to determine whether the relationship between undergoing CBT and change in IBS symptoms is attributable to a direct connection between therapy and physical symptom change, operates indirectly through emotional state, or both. These questions have been addressed through path analysis [22] in which the strength of path coefficients representing direct and indirect relationships is compared and the fit of prespecified models is assessed. To answer the research question posed in this study, path analysis is superior to multiple regression for several reasons. An important difference is that path models allow the a priori specification of a causal pathway which, if supported, would strengthen confidence that the hypothesized pathway represents reality and is not just a proxy for the actual pathway. Where multiple regression models only allow specification of a particular component of the hypothesized pathway, a path model allows the entire path to be specified simultaneously.
Results Demographics Participants were of average age 42 years (S.D.=11, range 19–66) and 81% were female. Observed effects A total of N=105 subjects were randomized to routine clinical care (n=34, 32%), relaxation training (n=36, 34%), or cognitive behavior therapy (CBT, n=35, 33%). In the current analysis, n=6 subjects (routine care n=0, relaxation n=4, and CBT n=2) were omitted due to missing data on vital items. The original trial [17] found no statistically significant or substantive differences between treatment arms with respect to change in bowel symptom scores with overall changes in symptom scores, and these findings are replicated in Table 1 when comparing subjects randomized to CBT compared with those randomized to either relaxation
CBT mean (S.D.), n
BSSS baseline Frequency 21.5 (4.3) n=33 Distress 17.3 (5.3) n=33 interference 15.9 (6.9) n=33 HAD baseline Anxiety 9.1 (3.7) n=33 Depression 5.3 (4.0) n=33 BSSS baseline–midpoint Frequency −1.4 (4.6) n=26 Distress −0.1 (4.5) n=26 interference −1.7 (6.7) n=26 HAD baseline–midpoint Anxiety 0.1 (3.2) n=24 depression −0.1 (2.7) n=24 BSSS midpoint–endpoint Frequency −2.0 (3.6) n=22 Distress −2.5 (4.3) n=22 interference −0.9 (3.1) n=22 HAD midpoint–endpoint Anxiety −0.9 (2.2) n=21 Depression −0.7 (1.7) n=21 a
Non-CBT Mean (S.D.), n
Pa
20.8 (4.5) n=66 17.0 (5.1) n=66 15.5 (5.7) n=66
.4 .8 N.9
8.6 (3.5) n=68 5.5 (3.7) n=68
.7 .8
−1.8 (4.4) n=49 −2.0 (4.3) n=49 −2.0 (4.5) n=49
.7 .04 0.2
−1.6 (3.2) n=50 −0.9 (2.4) n=50
.05 .2
−0.8 (4.4) n=43 −0.9 (3.5) n=43 −0.3 (3.8) n=43
.1 .07 .6
−0.0 (3.0) n=41 −0.5 (2.3) n=41
.3 .8
Mann–Whitney test.
or control groups (combined). Modest differences in change in distress score between baseline and midpoint are reversed in the midpoint to end-point scores. Path model In the postulated path model (Fig. 3) CBT starts a process of sequential changes in mood state that lead to changes in symptom score that invokes a feedback in emotional state that further alters bowel symptom scores. While the model does not provide an adequate fit to the observed covariance structure (χ2=285.9, 29 df, Pb.0005), it does lend support to the indirect effect hypothesis. The path from CBT to change in HAD anxiety from baseline to midpoint reaches statistical significance (P=.02), and the change in anxiety from baseline to midpoint is associated with changes in the bowel symptom interference and disturbance domains over the same period (P=.02 and P=.005, respectively). Change in bowel symptom disturbance from baseline to midpoint is associated with change in both anxiety and depression (Pb.0005 and P=.01) and change in frequency score was associated with change in depression score (P=.004). Finally, change in anxiety from mid- to end-point is associated with change in both bowel symptom frequency and disturbance over the same period (P=.02 and P=.05, respectively). The one unexpected finding in these data is a positive path coefficient between CBT and initial change in anxiety. This coefficient is however consistent with the results reported in Table 1 where the mean change in HAD anxiety from baseline to midpoint is 0.1 for CBT subjects but −1.6 for non-CBT subjects.
282
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285
Fig. 3. Longitudinal model of CBT indirect effects on bowel symptoms with standardized path coefficients shown.
The sensitivity analysis, in which relaxation therapy is grouped with CBT, generally supported the primary analysis. There were no statistically significant direct paths between CBT/relaxation and change in bowel symptoms (all PN.05). Two notable differences are (i) no statistically significant pathways from CBT/relaxation to change in anxiety or depression and (ii) stronger evidence of indirect pathways between change in anxiety and depression and change in bowel symptoms (Table 2).
Discussion Understanding the therapeutic processes responsible for IBS symptom change is an important area of research. Apart from assisting with the development of more effective and patient-friendly treatments, it also may give clues to the underlying etiology of IBS. This study aimed to shed some light on the potential mechanisms responsible for symptom improvement in patients with IBS undergoing CBT by testing
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285 Table 2 Unstandardized path coefficients (S.E.) and P values for path model Primary model a
Sensitivity analysis b
Path
Estimate S.E.
P
Estimate C.R.
P
hadadbm←cbt/cbtplus hadddbm←cbt/cbtplus bsssfdbm←cbt/cbtplus bsssddbm←cbt/cbtplus bsssidbm←cbt/cbtplus bsssidbm←hadddbm bsssddbm←hadddbm bsssfdbm←hadddbm bsssidbm←hadadbm bsssddbm←hadadbm bsssfdbm←hadadbm hadadme←bsssfdbm hadadme←bsssddbm hadddme←bsssddbm hadddme←bsssidbm hadddme←bsssfdbm hadadme←bsssidbm bsssfdme←hadadme bsssddme←hadadme bsssidme←hadadme bsssfdme←hadddme bsssddme←hadddme bsssidme←hadddme bsssfdme←cbt/cbtplus bsssddme←cbt/cbtplus bsssidme←cbt/cbtplus
1.743 0.759 −0.109 0.963 −0.973 0.639 0.155 −0.073 0.415 0.426 0.259 0.136 −0.225 −0.166 0.088 0.163 0.095 0.419 0.318 0.269 −0.072 −0.050 0.154 −0.806 −1.315 −0.301
.024 .216 .924 .351 .418 .004 .421 .732 .019 .005 .124 .075 .005 .005 .084 .004 .165 .017 .052 .081 .751 .814 .440 .437 .174 .740
0.923 −0.403 0.284 0.024 −1.541 0.579 0.151 −0.054 0.426 0.463 0.239 0.138 −0.232 −0.162 0.085 0.162 0.099 0.450 0.360 0.329 −0.085 −0.060 0.099 0.298 0.184 1.260
.244 .517 .798 .981 .185 .009 .434 .800 .013 .002 .146 .070 .003 .007 .096 .005 .146 .010 .030 .029 .710 .783 .617 .776 .852 .161
0.770 0.613 1.136 1.034 1.202 0.223 0.193 0.212 0.177 0.153 0.168 0.077 0.080 0.059 0.051 0.057 0.068 0.176 0.164 0.154 0.228 .213 .200 1.037 .968 .907
1.165 −0.647 0.256 0.024 −1.326 2.631 0.782 −0.253 2.479 3.083 1.453 1.809 −2.920 −2.715 1.667 2.837 1.455 2.562 2.173 2.177 −0.371 −0.276 0.501 0.285 0.187 1.401
a
Primary analysis compares CBT group vs. relaxation and usual care combined. b Sensitivity analysis compares CBT and relaxation groups vs. usual care.
both the direct effects of CBT on symptoms and the indirect effects of CBT on symptoms operating via effects in emotional state, conceptualized within a biopsychosocial framework. Our data provide no evidence of a direct effect of CBT on the frequency, disturbance, or interference of bowel symptoms. This is in contrast to the finding of Lackner et al. [19] that CBT works directly on IBS symptoms. This discrepancy, however, is most likely due to some key differences between the two studies investigated. Most noticeable is that the primary endpoint in the trial that Lackner et al. investigated was global improvement compared with continuous bowel symptom indices of frequency, disturbance, and interference used in the trial we studied [17]. There is evidence that a patient's appraisal of symptom improvement and severity of symptoms is only modestly correlated [23]. Thus, it is difficult to compare the two trials, although we would argue a focus on key individual symptoms is more relevant to interpreting the outcome of a clinical trial. Moreover, the BSSS has been shown to be reliable and sensitive to change [14] and is a more relevant measure of bowel severity than other measures such as the Irritable Bowel Syndrome Severity Scale [24], which rely on trichotimizing severity into mild, moderate, and severe categories. Other differences between the two
283
studies is the CBT approach that Lackner et al. [16] investigated was group-centered vs. individual CBT sessions assessed in the current study [17]. The control groups between the two trials were also different with a patient education/support group (that involved education, sharing of experiences, support, etc.) and a waitlist control used in the CBT trial that Lackner et al. assessed [16]. This is compared with a relaxation group and usual care control conditions in the trial we investigated [17]. Our CBT group also received usual care on top of CBT intervention. Moreover, it is important to note that in the trial we investigated we did not find CBT to be more effective than the control groups, with all three groups reporting symptom improvement. IBS patients in the trial Lackner et al. [16] investigated were also likely to be more severely affected with IBS as they needed to score of greater than or equal to 2 on a global rating of IBS severity scale before entering the trial compared with just meeting IBS Rome I criteria in ours [17]. In this study we recruited subjects with IBS from a range of sources including tertiary referral centers and newspaper advertisements, and thus the results of this study should be generalizable to a large proportion of the IBS population. Hence, the difference between the present findings and those of Lackner et al. [19] need to be considered in the context of the differences between the studies. However, based on the current data, we conclude that the hypothesis that CBT works directly on IBS symptoms does not seem to be supported. We did find some evidence to support the hypothesis of indirect effects of CBT on symptoms operating via emotion, although we would not claim it to be definitive. In Model 2 we found that change in both anxiety and depression had a direct effect on bowel symptoms score changes. These findings are consistent with a biopsychosocial model of IBS that argues that abdominal pain and bowel symptoms experienced by persons with IBS is not only affected by underlying biological factors (e.g., gut motility and visceral sensitivity) but also by psychological factors (e.g., depression and anxiety, coping skills) and social factors (e.g., a supportive home environment) [18]. The brain–gut axis is believed to be crucial in the biopsychosocial model in understanding how psychological factors may be causal in IBS [25]. The brain–gut axis incorporates a bidirectional neural pathway that links cognitive and emotional centers in the brain with neuroendocrine centers, the enteric nervous system, and the immune system [25]. It is postulated that a primary disturbance in gut sensitivity or motility, or both, interacts with disturbed processing in the CNS—in particular, emotional regulation that is itself modulated by psychosocial stressors. Psychological treatments such as CBT are believed to work by reducing the impact of CNS activity on gut function in patients with IBS. Our findings are consistent with other lines of research supporting a biopsychosocial conceptualization for IBS including studies showing a increased psychological dysfunction in
284
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285
people with IBS [4], with longitudinal studies showing a temporal relationship between psychological factors and IBS symptoms [26,27] and other psychophysiological research [28,29]. In our final model evaluating a potential bioemotional feedback mechanism, improvement in emotional state leads to improved symptom perception that feeds back to improved emotional state in an ongoing cycle lending some support to the indirect effect hypothesis, although the model itself overall was not statistically significant. For example, we found that the path from CBT to change in HAD anxiety from baseline to midpoint reached statistical significance, while the change in anxiety from baseline to midpoint was also associated with change in bowel symptom interference and disturbance domains over the same period. Change in disturbance from baseline to midpoint was associated with change in both anxiety and depression. Similarly, the change in anxiety from midpoint to endpoint was also associated with change in both bowel frequency and disturbance over the same period. Thereby, our observations support a positive bioemotional feedback mechanism in which improvement in emotional state, particularly anxiety, leads to improved symptom perception which feeds back to improved emotional state in an ongoing cycle. This study has several strengths and weaknesses. In particular, the results of the study must be interpreted with the constraints of any uncontrolled study in mind. Path modeling has an advantage over linear models in that a hypothesized causal pathway is specified a priori but remains, nonetheless, an assessment of associations and the model specified is potentially not the only model that could be found to fit the observed covariance structure in the data. If the object of the model is to test putative mechanisms of action, then this point is very important. It is possible that there exist superior model specifications but that does not detract from the comparisons made in the study that has focused on direct vs. indirect pathways between CBT and bowel symptoms in IBS. It is very possible that other pathways may explain the symptom improvement seen from this CBT trial. For example, it could be that CBT helps change negative cognitions, or improves social relationships and that these could be responsible for symptom change. Moreover, it is possible that the heterogeneity of IBS may impact on the outcome of CBT. In the present study, we do not know whether IBS patients in either trial also met criteria for other psychiatric disorders, in particular somatization or abuse. Creed et al. [30] have shown that patients with IBS and high levels of somatization respond well to CBT. Thus, future studies may also need to determine the mechanism of action of CBT in “pure” IBS patients and those with comorbid psychiatric conditions. Also, the therapist–patient relationship itself may also effect symptom change. However, understanding the mechanism through which CBT affects bowel symptoms was not our objective but rather we sought to determine whether the effects were direct, indirect, or both.
While we took bowel symptom severity as a primary endpoint on the basis that symptom reduction is the therapeutic goal in a functional disorder such as IBS, from a patient perspective however, it could be argued that ultimately it is QoL that matters and it would be useful for future research to replicate our work and that of Lackner et al. [19] including QoL as an endpoint. In conclusion, the counterintuitive hypothesis that CBT would directly affect bowel symptoms was not supported. Instead this study suggests that CBT may operate via changes in mood state, consistent with the biopsychosocial model. We acknowledge however that neither this nor previous studies definitively show how CBT affects bowel symptoms and this must remain an active area of research. References [1] Camilleri M, Mayer EA, Drossman DA, Heath A, Dukes GE, Mc-Sorley D, et al. Improvement in pain and bowel function in female irritable bowel patients with alosetron, a 5-HT3 receptor antagonist. Aliment Pharmacol Ther 1999;13:1149–59. [2] Bradesi S, Tillisch K, Mayer E. Emerging drugs for irritable bowel syndrome. Expert Opin Emerg Drugs 2006;11:293–313. [3] Andersen V, Camilleri M. Irritable bowel syndrome: recent and novel therapeutic approaches. Drugs 2006;66:1073–88. [4] Walker EA, Roy-Byrne PP, Katon WJ. Irritable bowel syndrome and psychiatric illness. Am J Psychiatry 1990;147:565–72. [5] Talley N, Owen B, Boyce P, Paterson K. Psychological treatments for irritable bowel syndrome: a critique of controlled treatment trials. Am J Gastroenterol 1996;91:277–86. [6] Talley NJ, Gabriel SE, Harmsen WS, Zinsmeister AR, Evans RW. Medical costs in community subjects with irritable bowel syndrome. Gastroenterol 1995;109:1736–41. [7] Koloski NA, Talley NJ, Boyce PM. The impact of functional gastrointestinal disorders (FGIDs) on quality of life. Am J Gastroenterol 2000;95:67–71. [8] Nezu AM, Nezu CM, Friedman SH, Haynes SN. Case formulation in behavior therapy: problem-solving and functional analytic strategies. In: Eells TD, editor. Handbook of psychotherapy case formulation. New York: Guilford Press, 1997. p. 368–401. [9] Blanchard EB, Schwarz SP, Suls JM, Gerardi MA, Scharff L, Greene B, et al. Two controlled evaluations of multicomponent psychological treatment of irritable bowel syndrome. Behav Res Ther 1992;30: 175–89. [10] Greene B, Blanchard EB. Cognitive therapy for irritable bowel syndrome. J Consult Clin Psychol 1994;62:576–82. [11] Payne A, Blanchard EB. A controlled comparison of cognitive therapy and self-help support groups in the treatment of irritable bowel syndrome. J Consult Clin Psychol 1995;63:779–86. [12] Van Dulmen AM, Fennis JFM, Bleijenberg G. Cognitive behavioral group therapy for irritable bowel syndrome: effects and long-term follow-up. Psychosom Med 1996;58:508–14. [13] Toner BB, Segal ZV, Emmott S, Myran D, Ali A, DiGasbarro I, et al. Cognitive-behavioral group therapy for patients with irritable bowel syndrome. Int J Group Psychother 1998;48:215–43. [14] Boyce P, Gilchrist J, Talley NJ, Rose D. Cognitive-behavior therapy as a treatment for irritable bowel syndrome: a pilot study. Aust N Z J Psychiatry 2000;34:300–9. [15] Heymann-Monnikes I, Arnold R, Florin I, Herda C, Melfsen S, Mönnikes H, et al. The combination of medical treatment plus multicomponent behavioural therapy is superior to medical treatment alone in the therapy for irritable bowel syndrome. Am J Gastroenterol 2000;95:981–94.
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285 [16] Blanchard EB, Lackner JM, Sanders K, Krasner SS, Keefer L, Payne A, et al. A controlled evaluation of group cognitive therapy in the treatment of irritable bowel syndrome. Behav Res Ther 2007;45: 633–48. [17] Boyce PM, Talley NJ, Balaam B, Koloski NA, Truman G. Randomized controlled trial of cognitive behavior therapy, relaxation training, and routine clinical care for the irritable bowel syndrome. Am J Gastro 2003;98:2209–18. [18] Drossman DA. Gastrointestinal illness and the biopsychosocial model. Psychosom Med 1998;60:258–67. [19] Lackner JM, Jaccard J, Krasner SS, Katz LA, Gudleski GD, Blanchard EB. How does cognitive behavior therapy for irritable bowel syndrome work? A mediational analysis of a randomized clinical trial. Gastro 2007;133:433–44. [20] Thompson WG, Doteval G, Drossman DA. Irritable bowel syndrome. Guidelines for the diagnosis. Gastroenterol Int 1989;2:92–5. [21] Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983;67:361–70. [22] Keith T. Multiple regression and beyond. Boston: Pearson, 2006. p. 254–88. [23] Meissner JS, Blanchard EB, Malamood HS. Comparison of treatment outcome measures for irritable bowel syndrome. Appl Psychophysiol Biofeedback 1997;22:55–62. [24] Francis CY, Morris J, Whorwell PJ. The irritable bowel severity scoring system: a simple method of monitoring irritable bowel
[25]
[26]
[27]
[28]
[29]
[30]
285
syndrome and its progress. Aliment Pharmacol Ther 1997;11: 395–402. Drossman DA, Corazziari E, Talley NJ, Thompson WG, Whitehead WE, editors. Rome II. The functional gastrointestinal disorders— diagnosis, pathophysiology and treatment: a multinational consensus. 2nd ed. McLean: Degnon Associates, 2000. p. 355–75. Koloski NA, Talley NJ, Boyce PM. Does psychological distress modulate functional gastrointestinal symptoms and health care seeking? A prospective community cohort study. Am J Gastroenterol 2003;98:789–97. Gwee KA, Graham JC, McKendrick MW, Collins SM, Marshall JS, Walters SJ, et al. Psychometric scores and persistence of irritable bowel syndrome after infectious diarrhoea. Lancet 1996;347: 150–3. Dorn SD, Palsson OS, Thiwan SI, Kanazawa M, Clark WC, van Tilburg MA, et al. Increased colonic sensitivity in irritable bowel syndrome strongly influenced by a psychological tendency to report pain and urge. Gut 2007;56:1202–9. Silverman DH, Munakata JA, Ennes H, Mandelkern MA, Hoh CK, Mayer EA. Regional cerebral activity in normal and pathological perception of visceral pain. Gastroenterol 1997;112:64–72. Creed F, Tomenson B, Guthrie E, Ratcliffe J, Fernandes L, Read N, et al. The relationship between somatisation and outcome in patients with severe irritable bowel syndrome. J Psychosom Res 2008;64: 613–20.
Pathways connecting cognitive behavioral therapy and change in bowel symptoms of IBS Michael Jones a,⁎, Natasha Koloski b , Philip Boyce c , Nicholas J. Talley d a Psychology Department, Macquarie University, North Ryde, NSW, Australia School of Psychology, University of Queensland, St. Lucia, Queensland, Australia c Discipline of Psychiatry, Sydney Medical School-Westmead, Westmead Hospital, Wentworthville, NSW, Australia d Faculty of Health, University of Newcastle, Callaghan, NSW, Australia b
Received 13 January 2010; received in revised form 12 August 2010; accepted 2 October 2010
Abstract Objective: A single previous paper on this topic found a direct pathway between cognitive behavioral therapy (CBT) and an irritable bowel syndrome (IBS) global symptom score. This is controversial since under the biopsychosocial model, the expectation is that CBT's effect would be mediated by mood. Using more sensitive bowel symptom scales and measurements at additional time points, we aimed to compare the relative strengths of direct pathways between CBT and change in IBS symptoms and indirect pathways that operate via mood state using structural equation modeling. Methods: Our data set included 105 people with Rome I IBS randomized to individual CBT (n=34), relaxation therapy (n=36), and usual medical care (n=35). The primary outcome was defined as adequate relief of IBS symptoms in terms of the distress, frequency, and impairment according to the Bowel Symptom
Severity Scale. Outcomes in functional status (according to the 36item Short-Form Health Survey) and psychological status (Hospital Anxiety and Depression Scale) were secondary outcomes. Results: Our data suggest indirect pathways that operate via mood, most clearly anxiety but to a lesser extent depression. Statistically significant pathways were identified that lead from CBT to change in mood state thence to change in bowel symptoms, followed by further changes in mood then changes in bowel symptoms. Our data provide no evidence of direct effect of CBT on bowel symptoms. Conclusions: The present study suggests that CBT may operate via changes in mood state while not ruling out the possibility of direct effects. Our findings do not directly support, but are consistent with, a biopsychosocial model. © 2011 Published by Elsevier Inc.
Keywords: Cognitive behavior therapy; CBT; Iritable bowel syndrome; IBS; Mood; Bowel symptoms
Introduction Despite a prolonged period of study, few very effective treatments for the irritable bowel syndrome (IBS) exist and management often remains elusive [1]. While a number of pharmacological agents show promise [2], none can yet be considered adequate treatment for the full spectrum of IBS presentations to physicians [3]. Psychological therapies [cognitive behavioral therapy (CBT), relaxation therapy, behavioral therapy, psychotherapy, and hypnotherapy] have been studied based on evidence that a considerable ⁎ Corresponding author. Psychology Department, C3A 516, Macquarie University, North Ryde, NSW 2109, Australia. Tel.: +61 2 9850 8601; fax: +61 2 9850 8062. E-mail address: [email protected] (M. Jones). 0022-3999/10/$ – see front matter © 2011 Published by Elsevier Inc. doi:10.1016/j.jpsychores.2010.10.004
proportion of patients with IBS experience comorbid psychological disturbances such as anxiety and depression [4], but with inconsistent results [5]. The need to identify effective treatments for IBS is crucial, with estimates that IBS costs the United States at least $8 billion dollars annually [6] and is associated with significant impairments in QOL for its sufferers [7]. Of all the psychological therapies, CBT has received the most attention as a psychological treatment for IBS [5]. CBT consists of four important components (current health status, thoughts, behaviors, and emotions) and their interaction with one another to explain a person's pain or anxiety [8]. For example, John is a 66-year-old retired, married man who has weekend plans to finish painting his wife's bookcases (behavior). Unfortunately, he wakes up feeling ill on Saturday morning (health) and is unable to complete the
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285
project (behavior). As a result, he feels angry and anxious about not getting to his work (emotions), believing that he is disappointing his wife (thought). He thinks, “My wife will think that I do not care about helping her decorate the study.” This belief raises his anxiety (emotion) as well as his frustration about not feeling up to par. This makes it even harder for John to figure out how to face the day, and consequently he stays in bed (behavior), which in turn only serves to raise his anxiety and strengthens his negative thoughts about his wife's reaction. This is shown in Fig. 1. The empirical evidence generally favors the efficacy of CBT in relieving symptoms and improving the quality of life (QoL) of patients with IBS [9–17]. Although many of the earlier studies reporting a positive benefit of CBT in IBS have been criticized on methodological grounds [9–15], the more recent methodologically rigorous studies are more convincing [16–17]. Blanchard et al. [16] found a modest although clear benefit attributable to CBT in global IBS symptoms. Boyce et al. [17] showed that CBT improved IBS symptoms, anxiety levels, and physical functioning post treatment, although they found no discernible difference in symptom response to CBT compared with relaxation therapy or “usual care.” Interestingly, little effort appears to have been expended in understanding how CBT might relieve symptoms in IBS. The biopsychosocial model provides a theoretical framework for explaining how CBT might affect IBS symptoms (Fig. 2) [18]. According to this model, early life factors (genetic predisposition and environmental factors) influence and interact with the later development of psychosocial risk factors (e.g., psychological disorders, coping styles) and physiological dysfunction (e.g., motility, visceral sensitivity) and their interaction via the brain–gut axis, increasing susceptibility to developing IBS [18]. Hence, it is postulated that a primary disturbance in gut sensitivity or motility, or both, interacts with disturbed processing in the central nervous system (CNS)—in particular, emotional regulation that is itself modulated by psychosocial stressors. Thus, it is suspected that psychological treatments such as CBT work by reducing the impact of CNS activity on gut function in patients with
Fig. 1. Cognitive/behavioral model.
279
Fig. 2. Biopsychosocial model of FGIDs and emotion.
IBS. For example, CBT interventions systematically teach patients the cognitive and behavioral skills needed to better manage stress, anxiety, and/or IBS symptoms and its sequelae. These skills can reduce the fear associated with stress, anxiety, or IBS symptoms, helping patients approach normal activity and assisting with management of the emotional consequences of living with IBS. By such intervention, a more positive cycle of behavioral and cognitive change is possible, resulting in improvements in psychological and physical functioning and reductions in IBS symptoms. Despite the usefulness of this model, however, it is difficult to validate comprehensively. Lackner et al. [19] were the first to attempt to explain the relationship between a group-centered CBT intervention and IBS symptom response. They adopted a path model approach to understanding the relationship between undertaking CBT and response in symptom reports; these authors concluded that CBT works directly on IBS symptoms. We found this result very surprising since it runs counter to the biopsychosocial model, which argues for a connection between symptom perception and emotion within the CNS based on a feedback loop between perception and emotions. The aim of the current work is to provide further empirical evidence in the debate over whether CBT is associated with improvements in bowel symptoms through direct effects or because it affects mood and therefore operates indirectly on bowel symptoms. Our work, while superficially replicating that of Lackner et al. [19], also provides a fresh look at a biopsychosocial model view of the role of CBT in IBS using an independent trial of CBT compared with active and control groups and more time points than that used by Lackner et al. [19]. We have taken an approach that prespecifies a path model, which includes both direct and indirect paths between CBT and subsequent changes in both mood and bowel symptoms.
280
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285
Methods Study subjects Subjects were recruited through a tertiary-referral hospital outpatient clinic and newspaper advertisements. To be eligible for the study, participants needed to have a diagnosis of IBS made by a gastroenterologist according to the Rome I criteria [20]. The Rome I criteria were considered to be the most valid standardized criteria at the time of the study. This diagnosis was also confirmed by an independent assessor using a Structured Interview for Bowel Symptoms. Using a structured format, each symptom of the Rome criteria is probed and then rated as present or absent. If the symptom is present, an enquiry is made about the duration of the symptoms to ensure that they meet the duration criteria specified in the Rome criteria. Other inclusion criteria included no structural bowel pathology that would account for symptoms, a sufficient understanding of English to be able to understand the therapy, and an age of 18 years or older. Subjects were excluded if they answered positively to an interviewer's questions regarding the presence of any major current medical or psychotic illness, a history of alcoholism, current psychological treatment and current use of antidepressants or antipsychotic medications, or current use of medications that could affect bowel function (e.g., antispasmodics, prokinetics). Interventions Subjects were randomized into one of three therapy conditions: usual medical care alone, relaxation training, and CBT. Individuals randomized to usual medical care alone (n=34) received three sessions with a gastroenterologist that included discussion of symptoms and dietary advice and written information on the same was handed out. Individuals randomized to relaxation training (n=36) received routine clinical care but also saw a clinical psychologist who provided 8 weeks of relaxation strategies in 30-min face-toface sessions. Individuals randomized to receive CBT (n=35) also received routine clinical care and relaxation training. The CBT intervention consisted of 8 weeks of 60-min faceto-face sessions with a clinical psychologist. The CBT approach was a manual-based program that incorporated realistic symptom appraisal, enhanced coping strategies, cognitive restructuring, and problem solving. For the purpose of this analysis, interest lay in comparing symptom response among those receiving CBT with those not receiving CBT, and hence, the routine clinical care and relaxation training groups have been combined into a single non-CBT group.
adequate before randomization, and to assess the effects of treatment on baseline symptoms. Randomization was conducted blinded to investigators of the study. A series of sealed opaque envelopes were prepared containing a card with the treatment conditions. They were randomly allocated participant identification numbers (using a random number generator) before the start of the study, with a check to ensure that equal numbers were allocated to each treatment group. After randomization, eight sessions of treatment took place weekly over an 8-week period for all subjects in the CBT and relaxation arms. The same clinical psychologist treated patients in both the relaxation and CBT groups. Data were collected at randomization (baseline), mid-treatment (4 weeks after baseline), end of treatment (at 8 weeks), and at 26 and 52 weeks of follow-up. To avoid therapist bias, all of the self-report scales were completed by the patients with the therapist blind to the results. Data were coded and scored at the conclusion of treatments. None of the subjects received additional counseling or psychological treatment during the course of the study. Assessment All subjects completed a range of self-report measures without any assistance from the investigators. GI symptom severity was assessed using the Bowel Symptom Severity Scale (BSSS) [14]. The BSSS measures frequency, disability, and distress for each of eight GI symptoms associated with IBS (loose stools, hard stools, abdominal pain, more than three bowel motions daily, bloating, urgency to defecate, inability to have a bowel motion in the past week, and abdominal discomfort) over the past week. Each item is rated on a six-point scale. Scores are summed to give a measure of bowel symptom frequency, disability and distress. The BSSS was developed by Boyce et al. (2000) for a pilot study as no other measure of bowel symptom severity was available. The results from the pilot study showed that the BSSS was reliable in terms of internal consistency (α=.88), responsive to change and had good face validity (Boyce et al., 2000) and thus was chosen for the main study reported here. Psychological distress was assessed with the reliable and valid Hospital Anxiety and Depression Scale (HAD) [21]. The questionnaire contains 14 items that assess anxiety (seven items) and depression (seven items) (Zigmond & Snaith, 1983). Each item is rated on a four-point scale ranging from 0 to 3, such that each subscale has a maximum score of 21. Higher scores on the HAD indicate increased severity on the depression or anxiety subscales, with a score of 7 or less on either subscale indicating a “non case,” 8–10 as a “doubtful case,” and a score of 11 or more as a “case” of anxiety or depression (Zigmond & Snaith, 1983).
Procedure Outcomes Participants completed a 2-week washout and review period before randomization. A baseline washout period is useful to ensure that symptom frequency and severity are
The primary outcome was bowel symptom severity in terms of frequency, distress, and disability.
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285
281
Models
Table 1 Baseline and change scores for bowel symptoms and emotional state
Our prespecified path model postulates both direct effects of CBT on change in bowel symptoms and a bioemotional feedback mechanism in which change in emotional state leads to improved symptom perception, which feeds back to improved emotional state in an ongoing cycle. This model is presented in more technical detail in Fig. 2. Only one of the three treatment groups was CBT; however, relaxation therapy is an element of CBT. In our primary analysis, relaxation therapy has been grouped with usual care but we have undertaken a sensitivity analysis in which relaxation therapy is grouped with CBT.
Score
Statistical approach The aim of the analysis is to determine whether the relationship between undergoing CBT and change in IBS symptoms is attributable to a direct connection between therapy and physical symptom change, operates indirectly through emotional state, or both. These questions have been addressed through path analysis [22] in which the strength of path coefficients representing direct and indirect relationships is compared and the fit of prespecified models is assessed. To answer the research question posed in this study, path analysis is superior to multiple regression for several reasons. An important difference is that path models allow the a priori specification of a causal pathway which, if supported, would strengthen confidence that the hypothesized pathway represents reality and is not just a proxy for the actual pathway. Where multiple regression models only allow specification of a particular component of the hypothesized pathway, a path model allows the entire path to be specified simultaneously.
Results Demographics Participants were of average age 42 years (S.D.=11, range 19–66) and 81% were female. Observed effects A total of N=105 subjects were randomized to routine clinical care (n=34, 32%), relaxation training (n=36, 34%), or cognitive behavior therapy (CBT, n=35, 33%). In the current analysis, n=6 subjects (routine care n=0, relaxation n=4, and CBT n=2) were omitted due to missing data on vital items. The original trial [17] found no statistically significant or substantive differences between treatment arms with respect to change in bowel symptom scores with overall changes in symptom scores, and these findings are replicated in Table 1 when comparing subjects randomized to CBT compared with those randomized to either relaxation
CBT mean (S.D.), n
BSSS baseline Frequency 21.5 (4.3) n=33 Distress 17.3 (5.3) n=33 interference 15.9 (6.9) n=33 HAD baseline Anxiety 9.1 (3.7) n=33 Depression 5.3 (4.0) n=33 BSSS baseline–midpoint Frequency −1.4 (4.6) n=26 Distress −0.1 (4.5) n=26 interference −1.7 (6.7) n=26 HAD baseline–midpoint Anxiety 0.1 (3.2) n=24 depression −0.1 (2.7) n=24 BSSS midpoint–endpoint Frequency −2.0 (3.6) n=22 Distress −2.5 (4.3) n=22 interference −0.9 (3.1) n=22 HAD midpoint–endpoint Anxiety −0.9 (2.2) n=21 Depression −0.7 (1.7) n=21 a
Non-CBT Mean (S.D.), n
Pa
20.8 (4.5) n=66 17.0 (5.1) n=66 15.5 (5.7) n=66
.4 .8 N.9
8.6 (3.5) n=68 5.5 (3.7) n=68
.7 .8
−1.8 (4.4) n=49 −2.0 (4.3) n=49 −2.0 (4.5) n=49
.7 .04 0.2
−1.6 (3.2) n=50 −0.9 (2.4) n=50
.05 .2
−0.8 (4.4) n=43 −0.9 (3.5) n=43 −0.3 (3.8) n=43
.1 .07 .6
−0.0 (3.0) n=41 −0.5 (2.3) n=41
.3 .8
Mann–Whitney test.
or control groups (combined). Modest differences in change in distress score between baseline and midpoint are reversed in the midpoint to end-point scores. Path model In the postulated path model (Fig. 3) CBT starts a process of sequential changes in mood state that lead to changes in symptom score that invokes a feedback in emotional state that further alters bowel symptom scores. While the model does not provide an adequate fit to the observed covariance structure (χ2=285.9, 29 df, Pb.0005), it does lend support to the indirect effect hypothesis. The path from CBT to change in HAD anxiety from baseline to midpoint reaches statistical significance (P=.02), and the change in anxiety from baseline to midpoint is associated with changes in the bowel symptom interference and disturbance domains over the same period (P=.02 and P=.005, respectively). Change in bowel symptom disturbance from baseline to midpoint is associated with change in both anxiety and depression (Pb.0005 and P=.01) and change in frequency score was associated with change in depression score (P=.004). Finally, change in anxiety from mid- to end-point is associated with change in both bowel symptom frequency and disturbance over the same period (P=.02 and P=.05, respectively). The one unexpected finding in these data is a positive path coefficient between CBT and initial change in anxiety. This coefficient is however consistent with the results reported in Table 1 where the mean change in HAD anxiety from baseline to midpoint is 0.1 for CBT subjects but −1.6 for non-CBT subjects.
282
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285
Fig. 3. Longitudinal model of CBT indirect effects on bowel symptoms with standardized path coefficients shown.
The sensitivity analysis, in which relaxation therapy is grouped with CBT, generally supported the primary analysis. There were no statistically significant direct paths between CBT/relaxation and change in bowel symptoms (all PN.05). Two notable differences are (i) no statistically significant pathways from CBT/relaxation to change in anxiety or depression and (ii) stronger evidence of indirect pathways between change in anxiety and depression and change in bowel symptoms (Table 2).
Discussion Understanding the therapeutic processes responsible for IBS symptom change is an important area of research. Apart from assisting with the development of more effective and patient-friendly treatments, it also may give clues to the underlying etiology of IBS. This study aimed to shed some light on the potential mechanisms responsible for symptom improvement in patients with IBS undergoing CBT by testing
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285 Table 2 Unstandardized path coefficients (S.E.) and P values for path model Primary model a
Sensitivity analysis b
Path
Estimate S.E.
P
Estimate C.R.
P
hadadbm←cbt/cbtplus hadddbm←cbt/cbtplus bsssfdbm←cbt/cbtplus bsssddbm←cbt/cbtplus bsssidbm←cbt/cbtplus bsssidbm←hadddbm bsssddbm←hadddbm bsssfdbm←hadddbm bsssidbm←hadadbm bsssddbm←hadadbm bsssfdbm←hadadbm hadadme←bsssfdbm hadadme←bsssddbm hadddme←bsssddbm hadddme←bsssidbm hadddme←bsssfdbm hadadme←bsssidbm bsssfdme←hadadme bsssddme←hadadme bsssidme←hadadme bsssfdme←hadddme bsssddme←hadddme bsssidme←hadddme bsssfdme←cbt/cbtplus bsssddme←cbt/cbtplus bsssidme←cbt/cbtplus
1.743 0.759 −0.109 0.963 −0.973 0.639 0.155 −0.073 0.415 0.426 0.259 0.136 −0.225 −0.166 0.088 0.163 0.095 0.419 0.318 0.269 −0.072 −0.050 0.154 −0.806 −1.315 −0.301
.024 .216 .924 .351 .418 .004 .421 .732 .019 .005 .124 .075 .005 .005 .084 .004 .165 .017 .052 .081 .751 .814 .440 .437 .174 .740
0.923 −0.403 0.284 0.024 −1.541 0.579 0.151 −0.054 0.426 0.463 0.239 0.138 −0.232 −0.162 0.085 0.162 0.099 0.450 0.360 0.329 −0.085 −0.060 0.099 0.298 0.184 1.260
.244 .517 .798 .981 .185 .009 .434 .800 .013 .002 .146 .070 .003 .007 .096 .005 .146 .010 .030 .029 .710 .783 .617 .776 .852 .161
0.770 0.613 1.136 1.034 1.202 0.223 0.193 0.212 0.177 0.153 0.168 0.077 0.080 0.059 0.051 0.057 0.068 0.176 0.164 0.154 0.228 .213 .200 1.037 .968 .907
1.165 −0.647 0.256 0.024 −1.326 2.631 0.782 −0.253 2.479 3.083 1.453 1.809 −2.920 −2.715 1.667 2.837 1.455 2.562 2.173 2.177 −0.371 −0.276 0.501 0.285 0.187 1.401
a
Primary analysis compares CBT group vs. relaxation and usual care combined. b Sensitivity analysis compares CBT and relaxation groups vs. usual care.
both the direct effects of CBT on symptoms and the indirect effects of CBT on symptoms operating via effects in emotional state, conceptualized within a biopsychosocial framework. Our data provide no evidence of a direct effect of CBT on the frequency, disturbance, or interference of bowel symptoms. This is in contrast to the finding of Lackner et al. [19] that CBT works directly on IBS symptoms. This discrepancy, however, is most likely due to some key differences between the two studies investigated. Most noticeable is that the primary endpoint in the trial that Lackner et al. investigated was global improvement compared with continuous bowel symptom indices of frequency, disturbance, and interference used in the trial we studied [17]. There is evidence that a patient's appraisal of symptom improvement and severity of symptoms is only modestly correlated [23]. Thus, it is difficult to compare the two trials, although we would argue a focus on key individual symptoms is more relevant to interpreting the outcome of a clinical trial. Moreover, the BSSS has been shown to be reliable and sensitive to change [14] and is a more relevant measure of bowel severity than other measures such as the Irritable Bowel Syndrome Severity Scale [24], which rely on trichotimizing severity into mild, moderate, and severe categories. Other differences between the two
283
studies is the CBT approach that Lackner et al. [16] investigated was group-centered vs. individual CBT sessions assessed in the current study [17]. The control groups between the two trials were also different with a patient education/support group (that involved education, sharing of experiences, support, etc.) and a waitlist control used in the CBT trial that Lackner et al. assessed [16]. This is compared with a relaxation group and usual care control conditions in the trial we investigated [17]. Our CBT group also received usual care on top of CBT intervention. Moreover, it is important to note that in the trial we investigated we did not find CBT to be more effective than the control groups, with all three groups reporting symptom improvement. IBS patients in the trial Lackner et al. [16] investigated were also likely to be more severely affected with IBS as they needed to score of greater than or equal to 2 on a global rating of IBS severity scale before entering the trial compared with just meeting IBS Rome I criteria in ours [17]. In this study we recruited subjects with IBS from a range of sources including tertiary referral centers and newspaper advertisements, and thus the results of this study should be generalizable to a large proportion of the IBS population. Hence, the difference between the present findings and those of Lackner et al. [19] need to be considered in the context of the differences between the studies. However, based on the current data, we conclude that the hypothesis that CBT works directly on IBS symptoms does not seem to be supported. We did find some evidence to support the hypothesis of indirect effects of CBT on symptoms operating via emotion, although we would not claim it to be definitive. In Model 2 we found that change in both anxiety and depression had a direct effect on bowel symptoms score changes. These findings are consistent with a biopsychosocial model of IBS that argues that abdominal pain and bowel symptoms experienced by persons with IBS is not only affected by underlying biological factors (e.g., gut motility and visceral sensitivity) but also by psychological factors (e.g., depression and anxiety, coping skills) and social factors (e.g., a supportive home environment) [18]. The brain–gut axis is believed to be crucial in the biopsychosocial model in understanding how psychological factors may be causal in IBS [25]. The brain–gut axis incorporates a bidirectional neural pathway that links cognitive and emotional centers in the brain with neuroendocrine centers, the enteric nervous system, and the immune system [25]. It is postulated that a primary disturbance in gut sensitivity or motility, or both, interacts with disturbed processing in the CNS—in particular, emotional regulation that is itself modulated by psychosocial stressors. Psychological treatments such as CBT are believed to work by reducing the impact of CNS activity on gut function in patients with IBS. Our findings are consistent with other lines of research supporting a biopsychosocial conceptualization for IBS including studies showing a increased psychological dysfunction in
284
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285
people with IBS [4], with longitudinal studies showing a temporal relationship between psychological factors and IBS symptoms [26,27] and other psychophysiological research [28,29]. In our final model evaluating a potential bioemotional feedback mechanism, improvement in emotional state leads to improved symptom perception that feeds back to improved emotional state in an ongoing cycle lending some support to the indirect effect hypothesis, although the model itself overall was not statistically significant. For example, we found that the path from CBT to change in HAD anxiety from baseline to midpoint reached statistical significance, while the change in anxiety from baseline to midpoint was also associated with change in bowel symptom interference and disturbance domains over the same period. Change in disturbance from baseline to midpoint was associated with change in both anxiety and depression. Similarly, the change in anxiety from midpoint to endpoint was also associated with change in both bowel frequency and disturbance over the same period. Thereby, our observations support a positive bioemotional feedback mechanism in which improvement in emotional state, particularly anxiety, leads to improved symptom perception which feeds back to improved emotional state in an ongoing cycle. This study has several strengths and weaknesses. In particular, the results of the study must be interpreted with the constraints of any uncontrolled study in mind. Path modeling has an advantage over linear models in that a hypothesized causal pathway is specified a priori but remains, nonetheless, an assessment of associations and the model specified is potentially not the only model that could be found to fit the observed covariance structure in the data. If the object of the model is to test putative mechanisms of action, then this point is very important. It is possible that there exist superior model specifications but that does not detract from the comparisons made in the study that has focused on direct vs. indirect pathways between CBT and bowel symptoms in IBS. It is very possible that other pathways may explain the symptom improvement seen from this CBT trial. For example, it could be that CBT helps change negative cognitions, or improves social relationships and that these could be responsible for symptom change. Moreover, it is possible that the heterogeneity of IBS may impact on the outcome of CBT. In the present study, we do not know whether IBS patients in either trial also met criteria for other psychiatric disorders, in particular somatization or abuse. Creed et al. [30] have shown that patients with IBS and high levels of somatization respond well to CBT. Thus, future studies may also need to determine the mechanism of action of CBT in “pure” IBS patients and those with comorbid psychiatric conditions. Also, the therapist–patient relationship itself may also effect symptom change. However, understanding the mechanism through which CBT affects bowel symptoms was not our objective but rather we sought to determine whether the effects were direct, indirect, or both.
While we took bowel symptom severity as a primary endpoint on the basis that symptom reduction is the therapeutic goal in a functional disorder such as IBS, from a patient perspective however, it could be argued that ultimately it is QoL that matters and it would be useful for future research to replicate our work and that of Lackner et al. [19] including QoL as an endpoint. In conclusion, the counterintuitive hypothesis that CBT would directly affect bowel symptoms was not supported. Instead this study suggests that CBT may operate via changes in mood state, consistent with the biopsychosocial model. We acknowledge however that neither this nor previous studies definitively show how CBT affects bowel symptoms and this must remain an active area of research. References [1] Camilleri M, Mayer EA, Drossman DA, Heath A, Dukes GE, Mc-Sorley D, et al. Improvement in pain and bowel function in female irritable bowel patients with alosetron, a 5-HT3 receptor antagonist. Aliment Pharmacol Ther 1999;13:1149–59. [2] Bradesi S, Tillisch K, Mayer E. Emerging drugs for irritable bowel syndrome. Expert Opin Emerg Drugs 2006;11:293–313. [3] Andersen V, Camilleri M. Irritable bowel syndrome: recent and novel therapeutic approaches. Drugs 2006;66:1073–88. [4] Walker EA, Roy-Byrne PP, Katon WJ. Irritable bowel syndrome and psychiatric illness. Am J Psychiatry 1990;147:565–72. [5] Talley N, Owen B, Boyce P, Paterson K. Psychological treatments for irritable bowel syndrome: a critique of controlled treatment trials. Am J Gastroenterol 1996;91:277–86. [6] Talley NJ, Gabriel SE, Harmsen WS, Zinsmeister AR, Evans RW. Medical costs in community subjects with irritable bowel syndrome. Gastroenterol 1995;109:1736–41. [7] Koloski NA, Talley NJ, Boyce PM. The impact of functional gastrointestinal disorders (FGIDs) on quality of life. Am J Gastroenterol 2000;95:67–71. [8] Nezu AM, Nezu CM, Friedman SH, Haynes SN. Case formulation in behavior therapy: problem-solving and functional analytic strategies. In: Eells TD, editor. Handbook of psychotherapy case formulation. New York: Guilford Press, 1997. p. 368–401. [9] Blanchard EB, Schwarz SP, Suls JM, Gerardi MA, Scharff L, Greene B, et al. Two controlled evaluations of multicomponent psychological treatment of irritable bowel syndrome. Behav Res Ther 1992;30: 175–89. [10] Greene B, Blanchard EB. Cognitive therapy for irritable bowel syndrome. J Consult Clin Psychol 1994;62:576–82. [11] Payne A, Blanchard EB. A controlled comparison of cognitive therapy and self-help support groups in the treatment of irritable bowel syndrome. J Consult Clin Psychol 1995;63:779–86. [12] Van Dulmen AM, Fennis JFM, Bleijenberg G. Cognitive behavioral group therapy for irritable bowel syndrome: effects and long-term follow-up. Psychosom Med 1996;58:508–14. [13] Toner BB, Segal ZV, Emmott S, Myran D, Ali A, DiGasbarro I, et al. Cognitive-behavioral group therapy for patients with irritable bowel syndrome. Int J Group Psychother 1998;48:215–43. [14] Boyce P, Gilchrist J, Talley NJ, Rose D. Cognitive-behavior therapy as a treatment for irritable bowel syndrome: a pilot study. Aust N Z J Psychiatry 2000;34:300–9. [15] Heymann-Monnikes I, Arnold R, Florin I, Herda C, Melfsen S, Mönnikes H, et al. The combination of medical treatment plus multicomponent behavioural therapy is superior to medical treatment alone in the therapy for irritable bowel syndrome. Am J Gastroenterol 2000;95:981–94.
M. Jones et al. / Journal of Psychosomatic Research 70 (2011) 278–285 [16] Blanchard EB, Lackner JM, Sanders K, Krasner SS, Keefer L, Payne A, et al. A controlled evaluation of group cognitive therapy in the treatment of irritable bowel syndrome. Behav Res Ther 2007;45: 633–48. [17] Boyce PM, Talley NJ, Balaam B, Koloski NA, Truman G. Randomized controlled trial of cognitive behavior therapy, relaxation training, and routine clinical care for the irritable bowel syndrome. Am J Gastro 2003;98:2209–18. [18] Drossman DA. Gastrointestinal illness and the biopsychosocial model. Psychosom Med 1998;60:258–67. [19] Lackner JM, Jaccard J, Krasner SS, Katz LA, Gudleski GD, Blanchard EB. How does cognitive behavior therapy for irritable bowel syndrome work? A mediational analysis of a randomized clinical trial. Gastro 2007;133:433–44. [20] Thompson WG, Doteval G, Drossman DA. Irritable bowel syndrome. Guidelines for the diagnosis. Gastroenterol Int 1989;2:92–5. [21] Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983;67:361–70. [22] Keith T. Multiple regression and beyond. Boston: Pearson, 2006. p. 254–88. [23] Meissner JS, Blanchard EB, Malamood HS. Comparison of treatment outcome measures for irritable bowel syndrome. Appl Psychophysiol Biofeedback 1997;22:55–62. [24] Francis CY, Morris J, Whorwell PJ. The irritable bowel severity scoring system: a simple method of monitoring irritable bowel
[25]
[26]
[27]
[28]
[29]
[30]
285
syndrome and its progress. Aliment Pharmacol Ther 1997;11: 395–402. Drossman DA, Corazziari E, Talley NJ, Thompson WG, Whitehead WE, editors. Rome II. The functional gastrointestinal disorders— diagnosis, pathophysiology and treatment: a multinational consensus. 2nd ed. McLean: Degnon Associates, 2000. p. 355–75. Koloski NA, Talley NJ, Boyce PM. Does psychological distress modulate functional gastrointestinal symptoms and health care seeking? A prospective community cohort study. Am J Gastroenterol 2003;98:789–97. Gwee KA, Graham JC, McKendrick MW, Collins SM, Marshall JS, Walters SJ, et al. Psychometric scores and persistence of irritable bowel syndrome after infectious diarrhoea. Lancet 1996;347: 150–3. Dorn SD, Palsson OS, Thiwan SI, Kanazawa M, Clark WC, van Tilburg MA, et al. Increased colonic sensitivity in irritable bowel syndrome strongly influenced by a psychological tendency to report pain and urge. Gut 2007;56:1202–9. Silverman DH, Munakata JA, Ennes H, Mandelkern MA, Hoh CK, Mayer EA. Regional cerebral activity in normal and pathological perception of visceral pain. Gastroenterol 1997;112:64–72. Creed F, Tomenson B, Guthrie E, Ratcliffe J, Fernandes L, Read N, et al. The relationship between somatisation and outcome in patients with severe irritable bowel syndrome. J Psychosom Res 2008;64: 613–20.