A randomized controlled trial of psychotherapy in patients with chronic functional dyspepsia

GASTROENTEROLOGY 2000;119:661– 669

A Randomized Controlled Trial of Psychotherapy in Patients With Chronic Functional Dyspepsia JANE HAMILTON,* ELSPETH GUTHRIE,* FRANCIS CREED,* DAVID THOMPSON,‡ BARBARA TOMENSON,* RAYMOND BENNETT,‡ KIERAN MORIARTY,§ WILLIAM STEPHENS,储 and RICHARD LISTON§ *School of Psychiatry and Behavioural Sciences, University of Manchester, Manchester Royal Infirmary, Manchester, England; ‡G.I. Science Group, University of Manchester, Hope Hospital, Salford, England; §Department of Gastroenterology, Bolton General Hospital, Bolton, Lancashire, England; and 储Department of Gastroenterology, Trafford General Hospital, Trafford, Greater Manchester, England

See editorial on page 869. Background & Aims: This study aimed to determine whether brief psychodynamic-interpersonal (PI) psychotherapy is more efficacious than a psychological control for patients with chronic, intractable functional dyspepsia (FD), and whether patients with abnormal gastric function respond differently than those with normal gastric function. Methods: Ninety-five consecutive patients with chronic symptoms of FD who had failed to respond to conventional pharmacologic treatments were approached. Thirty-seven received PI psychotherapy, and 36 the control condition (supportive therapy). Patients completed self-report questionnaires before and after treatment and 12 months later. The patients’ gastroenterologists, who were blind to the study groups, conducted independent ratings before and after treatment. Forty-nine patients also underwent a radioisotope gastric emptying study. An intention-to-treat analysis was used with baseline scores as covariates. Results: At the end of treatment, there were significant advantages for PI psychotherapy compared with controls for the gastroenterologists’ (P ⴝ 0.002) and patients’ total symptom score (P ⴝ 0 .015). One year after treatment, the symptomatic scores were similar. However, a subanalysis showed that PI therapy was superior to the control condition at 1 year, when patients with severe heartburn were excluded. There was no difference in outcome between patients with normal and abnormal gastric emptying. Conclusions: PI therapy may have both shortand long-term effects in patients with dyspepsia, but further evaluation is required. Its cost-effectiveness needs to be determined.

yspepsia is a common complaint in Western countries, and many individuals seek medical treatment from primary care or hospital outpatient services.1 Only a minority of patients with dyspepsia actually have peptic ulcer disease; most have shown no structural or bio-

D

chemical pathology and are classified as having nonulcer dyspepsia or functional dyspepsia (FD).2 The cause of the condition is unclear, and there is no recognized treatment for FD.3 Possible pathophysiologic mechanisms include hypersensitivity to gastric distention,4 Helicobacter pylori–induced gastritis,5 and abnormal gastric function.6 Patients with FD report poorer health status, mental health, and social functioning than patients with structural gastrointestinal pathology.7 Overinvestigation and high use of secondary health care services is common,8 and costs of treating the condition with pharmacologic agents are high and usually unsuccessful.9 Psychological factors have been identified as important contributors to symptoms in FD,10 suggesting that psychological therapy could be beneficial to patients with this condition. Evidence from previous studies support this suggestion.11–13 The most recent study13 showed that cognitive therapy is superior to routine treatment provided by gastroenterologists. However, because the study did not control for the additional time and attention the patients in the cognitive therapy group received, it could be argued that the benefit was a result of these nonspecific factors of psychological treatment rather than the particular treatment itself. We recently showed that psychodynamic-interpersonal (PI) therapy provides better symptom improvement for patients with chronic, intractable symptoms of the irritable bowel syndrome than a psychological placebo (i.e., time spent with a sympathetic individual).14 Abbreviations used in this paper: ANCOVA, analysis of covariance; FD, functional dyspepsia; GERD, gastroesophageal reflux disease; PI, psychodynamic-interpersonal; SPRS, Sheffield Psychotherapies Rating Schedule. © 2000 by the American Gastroenterological Association 0016-5085/00/$10.00 doi:10.1053/gast.2000.16493

662

HAMILTON ET AL.

This suggested that the therapy had an effect independent of the time and attention spent with the patient. In the present study, we compared the efficacy of PI psychotherapy with a psychological placebo control that we have termed “supportive therapy” in patients with chronic symptoms of FD. Our intention was to control for the exact amount of time and attention that patients in the PI psychotherapy group receive. Because the treatment consists of 7 sessions, and the first session is at least 3 hours long, we had to design a psychological placebo that was credible to patients. In other words, we had to design an intervention in which patients would feel genuinely helped and supported, but would not be the recipient of specific psychological techniques, linking psychological factors to dyspeptic symptom production. We therefore designed a low-grade, psychological intervention based on the basic principles of counseling. A proportion of patients with severe FD have an abnormal pattern of distribution of food within the stomach; a smaller proportion of the meal is located in the proximal stomach after ingestion.6 The explanation for this abnormal intragastric food distribution is unknown, but one suggestion is that it represents vagal neural damage.6 We hypothesized that if symptoms in patients with evidence of abnormal gastric function are caused by neural damage, then they would respond less well to psychological treatment than those with normal gastric function in whom the symptoms might have a more psychological explanation.

Materials and Methods The trial was designed as a randomized controlled trial. The same therapist delivered both interventions, to exclude potential for differences in treatment caused by different therapist effects. Only the therapist knew which type of psychological intervention the patients were receiving. Both patients and the assessors of the patients’ symptoms were blind to the trial limbs. Patients were offered entry to the study if they fulfilled the following criteria for FD: prominent symptoms of upper abdominal pain with or without any of the following symptoms: early satiety, upper abdominal distention and bloating, acid reflux, loss of appetite, and nausea, in the presence of routine hematologic and biochemical screening, ultrasonography, and gastroscopy.1 In addition, patients were required to have had continuous symptoms for at least 6 months that had been unresponsive to at least 2 medical treatments, including antacid treatment, proton pump inhibitors, prokinetics, and Helicobacter eradication therapy. Not all patients received all of the earlier-mentioned treatments. Helicobacter pylori serology status was determined in most patients, but recruitment of patients was independent of the result because of the uncertainty surrounding its role in FD.15

GASTROENTEROLOGY Vol. 119, No. 3

Because at the time the study was planned there was no uniform agreement on the subgroup of patients with dyspepsia and reflux, patients with reflux were included in the study. However, because now patients with prominent reflux are considered to have gastroesophageal reflux disease (GERD) unless proved otherwise,16 a subanalysis was conducted, excluding patients who reported severe symptoms of heartburn. All patients were recruited from the gastroenterology clinics in 3 hospitals in the Manchester area, 1 university teaching hospital and 2 district general hospitals, using the following procedure. The case notes of all patients attending the clinics were screened by a researcher associated with the study to determine eligibility for recruitment. If a patient was considered eligible, the gastroenterologist was asked to assess the patient in more detail and offer the opportunity to participate in the study. Patients were then interviewed by a member of the research team within 1 week of identification. This method of recruitment ensured that all eligible patients were offered the opportunity to participate in the study. Patients were invited to participate in the study to assess 1 of 2 potentially helpful treatments. They were told the treatments were unproven and may or may not be beneficial. If a patient agreed to participate, he/she was then allocated randomly to either PI psychotherapy or the psychological control condition. Patients were allowed to continue on medication if they so wished. Randomization was performed using a computer-generated series of random numbers supplied by the trial statistician (B.T.).

Assessments The following assessments were performed at entry, at the end of the 12-week intervention period, and at follow-up 12 months later. Symptom rating by patients. The main outcome measure of the study was the patients’ self-rating of their dyspeptic symptoms. All patients rated each of the following symptoms on a 5-point scale (0, none; 1, mild; 2, moderate; 3, severe; and 4, very severe): upper abdominal pain or discomfort, fullness after eating, abdominal bloating or swelling, belching or flatulence, vomiting, regurgitation of food, heartburn or acid reflux, and poor appetite. The sum of these scores was taken as an overall measure of upper gastrointestinal symptom severity, and termed the “patients’ total symptom score.” Symptom rating by physicians. The patients’ gastroenterologists completed a rating similar to that conducted by the patients, at the beginning and end of treatment. The gastroenterologists remained blind to the treatment groups. Their responses to the 8-item scale were summed in the same way as with the patients’ responses to produce a “gastroenterologists total symptom score.” The gastroenterologists did not conduct assessments at the 12-month follow-up because most patients had stopped attending clinics by this time. Instead, the follow-up assessments were conducted either by mail or by visiting the patients at their homes.

September 2000

1-year assessments. In addition to repeating the main outcome measure (patients’ total symptom score), patients were also asked to rate their overall physical health and well-being in comparison with their pretrial status, 15 months previously. They were asked to rate whether they felt much worse, worse, the same, better, or much better. Psychological status. Psychological status was rated using the SCL-90-R.17 The Global Symptom Index, which is calculated by the total score divided by 90, represents the best single indicator of the current severity of a psychological disorder and is used in most instances in which a single summary measure is required. Patients completed this measure at entry, at the end of treatment, and 1 year later. Psychological outcome was considered a secondary outcome. Health care use. The number of gastrointestinal clinic visits, medications, in-patient stays, and procedures were recorded for both patient groups for 12 months before and 12 months after treatment. These data were verified using hospital notes and hospital databases.

PI Therapy PI therapy has been developed in England over the past 20 years. It is derived from the conversational model of psychotherapy developed by Hobson.18 It is manualized19 and has been used in 3 previous, randomized controlled trials, 2 of depression,20,21 and our previous study of psychological treatment in irritable bowel syndrome.14 It has many similarities to the Interpersonal Therapy developed by Klerman and Weissman,22 but PI therapy focuses much more on the therapist-patient relationship. The method emphasizes the formation of a strong collaborative working alliance with the patient, in which the patient’s interpersonal difficulties can be explored, revealed, and modified. A 3-hour first session of therapy is used to engage patients, followed by 6 sessions of 50 minutes.

Psychological Control Condition (Supportive Therapy) This intervention was used specifically to control for the time, sympathy, support, and attention received by the patients in the PI therapy group. Patients were encouraged to express their feelings and distress concerning their symptoms. Because many patients with dyspepsia spontaneously raise concerns about their diet and sensitivity to particular foods, the therapist specifically focused on diet and the relationship between eating and dyspepsia (manual available on request from Dr. Guthrie). A history of the patient’s symptoms was taken in the first session, and issues related to eating, family attitudes to eating, and food were explored, but without offering specific dietary advice. The therapist was allowed to respond to patient concerns about the dyspepsia by providing simple explanations about its cause. The therapist did not use any of the specific elements of PI therapy. Patients received a long first session of therapy (3 hours) plus 6 sessions (50 minutes) thereafter, thus matching exactly and controlling for the contact time used in the PI therapy.

PSYCHOLOGICAL TREATMENT OF FUNCTIONAL DYSPEPSIA

663

Adherence to Trial Interventions The PI therapy sessions and the supportive therapy sessions were audiotaped and randomly selected for ratings of adherence using the Sheffield Psychotherapies Rating Schedule (SPRS).23 Treatment integrity was maintained by weekly individual supervision from an experienced therapist (E.G.) with regular monitoring of audiotaped material. One session per therapy was taken at random and rated by 2 raters according to the SPRS rating scale. Ninety-five percent of sessions were correctly assigned to the appropriate treatment modality by using the PI subscale of the SPRS.

Gastric Emptying Studies All patients undergoing gastric emptying studies had the same test meal consisting of 415 mL (425 g) of a commercially available semiliquid soup to which 12 MBq of technetium 99m–labeled diethylenetriaminepentaacetic acid was added. Images of both anterior and posterior views of the abdomen were acquired for 1 minute each at 5, 10, 15, 20, 30, 40, 50, 60, 75, and 90 minutes after beginning ingestion of the meal by standing the patient in front of the low-energy collimator of a gamma camera. In the intervals between image acquisition, the patient sat comfortably in a chair near the camera. All images were stored on magnetic tape for analysis after completion of acquisition. A region of interest for the whole stomach was drawn by using the initial anterior and posterior images. From this, regions of interest for the proximal and distal parts of the stomach were derived by dividing the whole region of interest into 2 equal-sized areas, as described previously.6 The total counts for the whole stomach, proximal stomach, and distal stomach were determined for all 20 images. Counts were corrected for radioactive decay and room background. In addition, to compensate for the movement anteriorally of the radioactivity as emptying proceeds, the geometric mean of the anterior and posterior corrected counts for each region was calculated at each time point. The counts at the start of the study (time 0), when theoretically all the ingested radioactivity is contained in the stomach, was obtained for each region by extrapolation from the corrected counts at the first 2 time points (at 5 and 10 minutes after beginning meal ingestion). Gastric function was assessed by determining the gastric emptying time for the whole stomach (the time for the corrected counts to fall to 50% of the value at time 0) and the initial proximal proportion (the percentage of the initial whole stomach counts contained in the proximal stomach). We have established norms for gastric emptying time and initial proximal proportion from studies conducted on a large series of patients over several years. Gastric emptying times for the whole stomach are considered abnormal if they are longer than 66 minutes, and initial proximal proportions are abnormal if they fall below 54%.

664

HAMILTON ET AL.

GASTROENTEROLOGY Vol. 119, No. 3

Figure 1. Course of patients through the study. FU, follow-up.

Statistical Analysis The study used an intention-to-treat analysis. Data are presented as the mean and SD of the mean. Data that were not normally distributed are presented in the form of medians and interquartile ranges. Comparisons between continuous data were made using the Student t test followed by an analysis of covariance (ANCOVA), with the baseline value of the variable under test as covariate. No attempts were made to statistically account for missing values. Data are presented in the form of 95% confidence intervals (CIs) and P values. Categorical data were compared using ␹2 tests. Before and after change in patients was analyzed using paired t tests. The relationship between change in dyspeptic symptoms and change in psychological symptoms was examined using Pearson correlation coefficients. The main analyses were repeated, excluding patients with severe symptoms of heartburn, because such patients would now be considered to have GERD as a primary diagnosis. The study was submitted to and approved by the Salford Ethical Committee. All patients provided written consent after delivery of both a verbal explanation and written information.

Results The recruitment and progress of patients through the study is shown in Figure 1 according to the CONSORT statement,24 which recommends a standardized format for presenting outcome data from a randomized controlled trial. Twenty patients (21.1%) of the 95 patients who were eligible declined to participate. Thirty-

seven patients were randomized to receive PI therapy, and 36 were randomized to the psychological control condition. At 1-year follow-up, 58 patients (79.5%) were reassessed. Fourteen patients refused further contact despite repeated attempts to collect follow-up data, and 1 patient died of a myocardial infarction. There was no significant difference in baseline total symptom scores for patients who were reassessed at 1 year compared with those for whom assessments could not be completed (P ⫽ 0.47; 95% CI for difference, ⫺1.77 to 3.76). Ten patients rated their symptoms of heartburn or acid reflux as severe. Entry Measures The mean age of the patients in both trial groups was similar: PI psychotherapy, 40 years (SD, 12); psychological control, 40 years (SD, 14) (P ⫽ 0.97; CI for difference, – 6.31 to 6.08). There was no difference between the study groups for demographic variables, symptom duration, and Helicobacter status (Table 1). Symptom Rating by Patients The patients’ total symptom ratings of upper gastrointestinal symptoms are shown in Table 2. The 2 trial groups had no significant differences at baseline. However, at the end of treatment, patients in PI therapy reported a significantly greater improvement in their symptoms than those receiving the control condition

September 2000

PSYCHOLOGICAL TREATMENT OF FUNCTIONAL DYSPEPSIA

665

Table 1. Demographic Characteristics and Use of Health Care by Patients Recruited to the Study According to the 2 Trial Groups Demographic characteristics and indicators of health care use Men Marital status Single Married/cohabitating Divorced/separated Widow Social class Professional Low-grade professional Technical Semiskilled Unskilled Unemployed Receiving disability allowance Helicobacter status Negative Positive Admitted to hospital in past 2 yr Invasive tests in past 12 mo

PI (n ⫽ 37)

Supportive psychotherapy (n ⫽ 36)

␹2

df

P

15 (40.5%)

14 (38.9%)

0

1

1.00

6 (16.2%) 26 (70.3%) 5 (13.5%) —

9 (25%) 24 (66.6%) 1 (2.7%) 2 (5.5%)

5.33

3

0.15

4 (10.8%) 6 (16.2%) 13 (35.1%) 8 (21.6%) 5 (13.5%) 10 (27%) 9 (24.3%)

1 (2.7%) 5 (13.8%) 6 (16.6%) 13 (36.1%) 11 (30.5%) 9 (25%) 7 (19.4%)

7.91 0 0.05

4 1 1

0.095 1.0 0.83

10 (27%) 25 (67.6%) 21/35 (60%) 32 (86.5%)

15 (41.6%) 17 (47.2%) 21/33 (63.6%) 35 (97.2%)

1.68 0 1.55

1 1 1

0.20 0.95 0.21

Mann–Whitney U test

Duration of symptoms GP visits past 6 mo GI OPD visits past year

Median (IQR)

Median (IQR)

z

P

24 (12–36) 4 (2–8) 2 (1–4)

24 (12–48) 4 (3–7) 3 (2–4)

⫺0.26 ⫺0.30 ⫺1.11

0.81 0.76 0.27

GP, general practitioner; GI, gastrointestinal; OPD, outpatient department; IQR, interquartile range.

(P ⫽ 0.015). The mean change in the total symptom score in the PI group was 3.92 (SD, 5.07) compared with 0.68 (SD, 4.30) for controls. The raw scores for the 3 main individual symptom variables are shown in Table 2. Symptom Rating by Physicians The total symptom score, as determined by the gastroenterologists, was similar in both groups at the start of the trial (Table 2). At the end of the treatment, however, the gastroenterologists rated the patients who received PI therapy as having improved significantly more than patients receiving the psychological control condition (supportive therapy) (Table 2). 1-Year Assessments One year after treatment, both groups reported a significant reduction in total symptoms compared with the start of the trial (PI group, P ⬍ 0.001; control condition, P ⫽ 0.002; paired t tests), but there was no difference between the 2 groups in the degree of improvement (Table 2). When patients with severe symptoms of heartburn were excluded from the analysis, however, a significant advantage for PI therapy was found at 1 year (Table 2).

More patients in the PI group (22/37, 59%) than the control group (11/36, 31%) rated their overall wellbeing as much better than their pretrial status (P ⫽ 0.0187, Fisher exact test; odds ratio [OR], 3.33; 95% CI, 1.27– 8.76). Significantly more patients in the PI group (20/37, 54.1%) also rated their physical health as much better compared with patients in the control condition (10/36, 27.7%) (P ⫽ 0.032, Fisher exact test; OR, 3.05; 95% CI, 1.15– 8.11). Psychological Status The results of the psychological status analysis are shown in Table 3. At entry, there was no difference in baseline scores between the 2 groups, with mean SCL90-R global symptom index score for all patients of 0.81 (SD, 0.56). No difference as found between the 2 groups at the end of treatment or at 1-year follow-up; patients in both groups showed a similar reduction in psychological symptoms. When patients with severe heartburn were excluded from the analysis, the pattern remained similar. Relationship Between Symptom Ratings and Psychological Status Improvement in dyspeptic symptoms was significantly correlated with improvement in psychological

666

HAMILTON ET AL.

GASTROENTEROLOGY Vol. 119, No. 3

Table 2. Comparison of Patients’ and Gastroenterologists’ Ratings of Dyspeptic Symptoms During the Trial Period: PI Therapy Versus Supportive Psychotherapy PI therapy Symptom ratings Patients’ total score at entry Upper abdominal pain Fullness Bloating Total score at entrya Patients’ total score at end Upper abdominal pain Fullness Bloating Total score at enda Patients’ total score at FU Upper abdominal pain Fullness Bloating Total score at FUa Gastroenterologists’ total score at entry Gastroenterologists’ total score at end

Student t test

Supportive therapy

ANCOVA P

Mean

SD

n

Mean

SD

n

P

14.8 2.8 2.2 2.1 14.7 10.9 1.7 1.8 1.5 10.5 8.7 1.3 1.4 1.2 7.8

(5.2) (0.9) (1.3) (1.4) (5.5) (6.4) (0.9) (1.2) (1.2) (6.4) (5.8) (1.2) (1.2) (1.2) 5

37

(4.2) (1.3) (1.1) (1.2) (4.6) (5.5) (1.3) (1.2) (1.2) (5.7) (6.2) (1.3) (1.5) (1.5) 6.1

36

0.13

⫺0.5 to 3.9

—

30 31

0.21 0.31

⫺0.9 to 4.2 ⫺4.4 to 1.4

— 0.015b

28 27

0.20 0.70

⫺5.2 to 1.1 ⫺3.8 to 2.5

0.004 0.37

28

13.1 2.1 2.2 2 13.1 12.4 1.9 2 2.1 12.6 9.3 1.4 1.6 1.3 9.9

24

0.18

⫺5.2 to 1.0

0.037

10.2

(4.2)

35

10.2

(3.3)

35

0.98

⫺1.8 to 1.8

—

6.8

(5.1)

30

10.1

(3.2)

28

0.005

⫺5.5 to ⫺1.0

33 37

32 31

95% CI

0.002c

FU, follow-up. aExcluding patients with heartburn. bANCOVA with baseline patient total symptom score as covariate. cANCOVA with baseline gastroenterologists’ score as covariate.

symptoms in the PI therapy group, both at the end of treatment (r ⫽ 0.51, P ⫽ 0.001) and 1 year later (r ⫽ 0.55, P ⫽ 0.001). For the controls, there was no relationship between improvement in symptoms of FD and improvement in psychological status, at either the end of treatment (r ⫽ 0.16, P ⫽ 0.40) or at 1 year (r ⫽ 0.24, P ⫽ 0.25). Health Use No differences were found between the 2 groups in the use of services in the year after end of therapy: number of hospital admissions (P ⫽ 0.32), number of surgeries (P ⫽ 0.82), number of visits to gastrointestinal clinic (P ⫽ 0.53), and number of medications (P ⫽ 0.628). Because we used an intention-to-treat analysis, all data on patients who dropped out were included. In

both groups, however, there was a reduction of service use. In relation to hospital visits for the PI therapy group, the mean number of visits in the year before the trial was 3 (SD, 2.62) compared with a mean of 1.19 (SD, 1.40) in the year after study (P ⫽ 0.0014). For the control condition, the corresponding figures were 3.28 visits (SD, 1.97) for the year before the study and 1.72 visits (SD, 2.32) for the year after trial (P ⫽ 0.002). Similar reductions in both groups were also witnessed in the number of hospital admissions, surgeries, and medications prescribed in the year after the study ended. When patients with severe heartburn were excluded from the analysis, the pattern remained similar; there were reductions in service use in both groups but no difference between the groups.

Table 3. Comparison of Psychological Symptom Scores During the Trial Period: PI Versus Supportive Psychotherapy Group difference in mean PI therapy

Student t test

Supportive psychotherapy

Psychological symptom scores (SCL-90-R GSI scores)

Mean

SD

n

Mean

SD

n

P

95% CI

ANCOVA P

Entry End of trial 1-Year follow-up

0.90 0.67 0.59

(0.59) (0.48) (0.4)

37 37 31

0.71 0.67 0.56

(0.52) (0.52) (0.54)

36 31 27

0.14 0.99 0.77

⫺0.07 to 0.45 ⫺0.24 to 0.24 ⫺0.21 to 0.28

— 0.25 0.66

NOTE. ANCOVA with baseline GSI score as covariate.

September 2000

Gastric Emptying Forty-nine patients agreed to undergo gastric scintigraphy at entry into the trial: 28 in the PI group and 21 in the control group. There were no significant differences between the 2 groups on any measure of gastric function. The mean whole stomach emptying times for the PI group and controls, respectively, were 48.1 minutes (SD, 15.9) and 42 minutes (SD, 12.0) (P ⫽ 0.17; 95% CI for difference, –2.6 to 14.9). Only 6 patients had a delayed gastric emptying time (⬎66 minutes), but 19 patients (38.8%) had lower than normal initial proximal food residence (⬍54%), 11 in the PI therapy group and 8 in the psychological control group. To assess the relationship between gastric dysfunction and treatment response, patients who underwent gastric scintigraphy were divided into 2 groups. Of the 49 patients studied, 31 rated themselves as having improved (better or much better) and 18 as having not improved (determined by assessment on the physical health rating at 1-year follow-up). There was no evidence that patients with low initial proximal food residence had a different response to treatment than patients with normal gastric function. Twelve (63%) with abnormal function improved after treatment compared with 18 (67%) who were normal (x2 ⫽ 0, P ⫽ 1). Twenty-seven patients in PI therapy group underwent gastric scintigraphy; 20 were improvers and 6 nonimprovers. The mean whole-stomach emptying time for the improvers was 48.7 minutes (SD, 16.6) compared with 49.3 minutes (SD, 13.8) for controls (P ⫽ 0.93; 95% CI for difference, ⫺14.9 to 13.7). No difference on any of the other measures of gastric function was found between improvers and nonimprovers. Data for patients in the control condition showed a similar pattern.

Discussion This study shows that although PI therapy improves gastrointestinal symptoms in patients with chronic FD in the short term, its long-term advantage over less-specific forms of therapy is less clear-cut. When patients with GERD were included in the analysis, no difference between psychotherapy and controls was found at 12-month follow-up, but when the patients with GERD were excluded, the psychotherapy group had a significantly greater reduction in gastrointestinal symptoms at follow-up. The rapid improvement of patients in the PI psychotherapy group can be clearly attributable to the specific effects of this form of psychotherapy because (1) at the end of treatment, the PI group had fared significantly better than the controls, and (2) improvements in gastrointestinal symptoms were significantly

PSYCHOLOGICAL TREATMENT OF FUNCTIONAL DYSPEPSIA

667

correlated with improvements in psychological symptoms in the PI group, but not the control group. Additional advantages of PI therapy vs. the supportive placebo were observed 1 year after treatment on the secondary measures of well-being and physical health, suggesting that PI therapy may specifically help patients to manage and cope with their symptoms better than a supportive approach. We have used the term “supportive therapy” to describe the psychological control. It should not be assumed that this intervention was inert (i.e., akin to a drug placebo). It would have been impossible to produce a psychological control condition that was credible to patients without it having therapeutic value; we therefore expected the control condition to help patients. Our main concern in designing the study was to determine whether PI therapy had specific effects, over and above those obtained from, and attributable to, the nonspecific effects of psychotherapy. The results suggest that PI therapy produces a much more rapid response but the long-term differential effects are not as marked, unless patients with heartburn are excluded. Although there was a strong rationale for excluding patients with GERD, this analysis was performed post hoc, so the findings must be treated with caution. Our results need to be confirmed by studies that exclude patients with GERD at the outset. Patients who received psychotherapy showed a mean improvement of nearly 4 points on the dyspeptic symptom measure, compared with a change of less than 1 for the control patients. If patients with GERD were excluded, the mean symptom scores for the psychotherapy group decreased by nearly 50% over the trial period, compared with a decrease of 25% for the control group. Coupled with the reports of greater well-being in the psychotherapy group at follow-up, these results suggest that many patients in the psychotherapy group experienced clinical improvement. The patients who were recruited into the study, however, had chronic and intractable symptoms, and complete resolution of symptoms would not be expected in such cases. In our previous trial of psychotherapy in patients with chronic irritable bowel syndrome,14 we found that symptomatic improvement in gut symptoms was closely related to an improvement in psychological symptoms. We now found a similar relationship between changes in dyspeptic symptoms and psychological status after PI therapy. However, although we found improvement in psychological status in the control group, it was unrelated to changes in gastrointestinal symptoms. A key aspect of PI therapy involves helping patients understand

668

HAMILTON ET AL.

ways in which psychological factors may exacerbate their symptoms. This linking process may be an important therapeutic mechanism. One of the traditional problems of trials of treatment in functional bowel disorders is that patients are recruited in a nonsystematic way, or by advertisement, which reduces the clinical use of the study. Our method of recruitment ensured that the patients who entered the study were representative of clinic patients with chronic symptoms. In addition, patients were not selected on the basis of their suitability for psychotherapy, or for their motivation to participate in treatment, which is an important confounding factor if patients are recruited by advertisement. Another problem in previous trials has been the comparison of an intensive psychological treatment, with a treatment-as-usual group. In such studies, a psychological control condition has not been included, which means that patients, and often assessors, are able to identify the trial groups (because one group is obviously having more treatment than the other). Patients who receive the treatment-as-usual condition may feel disappointed; this treatment expectation may have a detrimental effect on their outcome, thus enhancing the apparent effect of the psychological treatment. The rigor of our trial design meant that both patients and their assessors (the physicians) remained blind to the study groups. Because we did not include a no-treatment group, it could be argued that the improvement reported by the patients in the year after therapy was spontaneous and unrelated to any intervention. The patients in the study were selected, however, for chronicity and stability of symptoms so that a substantial, spontaneous improvement in gastrointestinal symptoms would have been unlikely. Had we included a no-treatment group, it would have been difficult to maintain blindness, and confounding factors such as patient expectation would have been introduced. Any trial of the efficacy of psychological treatment involves certain compromises. In this study, we chose to evaluate the specific effect of PI therapy, over and above the effect obtained by regular contact with a therapist using supportive techniques. Had the control group not received this intervention, but continued with treatment as usual, the difference in outcome between the treatment and control groups may have been greater. This difference, however, would have been a result of both the specific and nonspecific effects of the therapy and patient expectation effects. Only 3 other studies of psychological treatment in patients with functional bowel disorders have used a

GASTROENTEROLOGY Vol. 119, No. 3

psychological control condition.14,25,26 In our previous study,14 patients in the control condition received support and attention, but not the same amount of time as patients in the treatment condition. In a study of cognitive therapy,25 patients in the control condition performed symptom monitoring over a 10-week period, but did not receive any active support and were only seen once during the study. In another study of cognitive therapy, control patients entered a patient support group.26 The patients spent the same amount of time in treatment as those who received the cognitive therapy, but they did not receive one-on-one attention. The present study is the first trial of psychological treatment in functional bowel disorders to control exactly for both the time and attention given to patients in the psychotherapy condition. We found no relationship between gastric dysfunction and treatment outcome. Patients with abnormal intragastric function did as well as patients with normal function. Thus, underlying intragastric dysfunction should not be considered a contraindication for psychological treatment in patients with FD, nor can it be concluded that gastric dysfunction is itself responsible for symptoms. The role of gastric dysfunction in the development and persistence of symptoms of FD is unclear and needs further clarification. However, gastric function was only measured on one occasion, so we do not know the variability of the abnormalities detected by this test. It is possible that underlying gastric dysfunction conveys vulnerability to developing dyspeptic symptoms in some individuals, but that other factors, such as psychological status, determine how such individuals cope with the symptoms. Use of health service decreased for both groups, which may have been related to the improvement of symptoms or to natural fluctuations of contact with services. There was no advantage for PI therapy over the psychological placebo control, although evidence from a study of psychiatric disorders shows that PI therapy reduces health care costs in comparison with usual treatment.27 We are unable to comment on the relative economic advantages of PI therapy in the present study because the 2 treatment interventions actually cost the same to deliver because they were conducted by the same therapist for the same time. The economic impact of psychological treatment in functional gastrointestinal disease, therefore, warrants further investigation. Two potential biases in our results need to be addressed. The first is the number of patients interviewed at 1 year; we were only able to gain assessments on 58 (80%) of the 73 patients who started the study. In such

September 2000

a relatively small study, the absence of follow-up data for these patients could have affected the results in either direction. However, no evidence from the baseline measures showed that the patients who did not return questionnaires differed in their symptomatology from those who completed assessments. The second bias is that the therapist in this study ( J.H.) saw both groups of patients. The possibility that she favored one treatment approach over another cannot be excluded. However, the integrity of each psychological approach was closely monitored via ratings of audiotapes of the sessions and weekly supervision, so that the magnitude of this effect is likely to have been small. In addition, both groups showed similar improvements in psychological status during the trial period, which negates the likelihood of therapist favorability. Because the patients in this study represent the extreme end of the spectrum of this disorder, the results should not be generalized to less-severely disabled populations. Not all patients with severe and chronic symptoms will have psychological treatment (20 declined to enter the study and approximately 15% dropped out during treatment). The results of this study suggest that psychological treatments for chronic FD are promising. Further evaluation, particularly health economic study, is indicated before widespread implementation can be recommended.

References 1. Talley NJ, Zinsmeister AR, Schleck CD, Melton LJ. Dyspepsia and dyspepsia subgroups: a population-based study. Gastroenterology 1992;102:1259 –1268. 2. Talley NJ, Piper DW. The association between non-ulcer dyspepsia and other gastrointestinal disorders. Scand J Gastroenterol 1985;20:896 –900. 3. Holtmann G, Talley NJ. Functional dyspepsia. Current treatment recommendations. Drugs 1993;45:918 –930. 4. Schmulson MJ, Mayer EA. Gastrointestinal sensory abnormalities in functional dyspepsia. Baillieres Clin Gastroenterol 1998;12: 545–556. 5. McColl KE. Helicobacter pylori 1998 –1998. Eur J Gastroenterol Hepatol 1999;11:13–16. 6. Troncon LEA, Bennett RJM, Ahluwalia NK, Thompson DG. Abnormal intragastric distribution of food during gastric emptying in functional dyspepsia patients. Gut 1994;35:327–332. 7. Talley NJ, Weaver AL, Zinsmeister AR. Impact of dyspepsia on quality of life. Dig Dis Sci 1995;40:584 –589. 8. Drossman DA, Li Z, Andruzzi E, Temple RD, Talley NJ, Thompson WG, Whitehead WE, Janssens J, Funch-Jensen P, Corazziari E, Richter JE, Koch GG. U.S. householder survey of functional gastrointestinal disorders. Prevalence sociodemography and health impact. Dig Dis Sci 1993;38:1569 –1580. 9. Tebaldi M, Heading RC. Clinical economics review: functional (non-ulcer) dyspepsia. Aliment Pharmacol Ther 1998;12:11–19.

PSYCHOLOGICAL TREATMENT OF FUNCTIONAL DYSPEPSIA

669

10. Bennett EJ, Piesse C, Palmer K, Badcock CA, Tennant CC, Kellow JE. Functional gastrointestinal disorders: psychological, social and somatic features. Gut 1998;42:414 – 420. 11. Giles SL. Separate and combined effects of biofeedback training and brief individual psychotherapy in the treatment of gastrointestinal disorders. Dissertation Abstr Int 1976;39:2495B. 12. Uhes MJ, Sweet AA, Cowles S. A behaviourally orientated stress reduction program and its effects upon self-reported gastric complaints. Stress 1989;5:57– 61. 13. Haug TT, Svebak S, Wilhelmsen I, Berstad A, Ursin H. Psychotherapy in functional dyspepsia. J Psychosom Res 1994;38: 281–291. 14. Guthrie E, Creed F, Dawson D, Tomenson B. A controlled trial of psychological treatment for the irritable bowel syndrome. Gastroenterology 1991;100:450 – 457. 15. Marshall BJ, McGechie DB, Roger PA, Glancy RJ. Pyloric campylobacter infection and gastroduodenal disease. Med J Aust 1985;142:439 – 444. 16. Talley NJ, Stanghellini V, Heading RC, Koch KL, Malagelada JR, Tytgat GN. Functional gastroduodenal disorders. Gut 1999; 45(suppl II):II37–II42. 17. Derogatis LR. SCL-90-R. Administration, scoring and procedures manual–II for the r(evised) version. Towson, MD: Clinical Psychometric Research, Inc., 1992. 18. Hobson RF. Forms of Feeling. London: Tavistock, 1987. 19. Shapiro DA, Firth JA. Exploratory therapy manual for the Sheffield Psychotherapy Project (memo no.733). (Available from University of Leeds, Leeds, England.) 20. Shapiro D, Firth J. Prescriptive v exploratory psychotherapy: outcomes of the Sheffield Psychotherapy project. Br J Psychiatry 1987;151:790 –799. 21. Shapiro DA, Barkham M, Rees A, Hardy GE, Reynolds S, Startup M. Effects of treatment duration and severity of depression on the effectiveness of cognitive-behavioural and psychodynamic– interpersonal psychotherapy. J Consult Clin Psychol 1994;62: 522–534. 22. Klerman GL, Weissman MM, Rounsaville BJ, Chevron ES. Interpersonal psychotherapy of depression. New York: Basic Books, 1984. 23. Shapiro DA, Startup MJ. Raters’ manual for the Sheffield Psychotherapy Rating Scale (memo no. 1154). (Available from University of Leeds, Leeds, England: Psychological Therapies Research Centre.) 24. Begg C, Eastwood S, Horton R, Moher D, Olkin I, Pitkin R, Rennie D, Schulz KF, Simel D, Stroup DF. Improving the quality of reporting of randomized controlled trials. The CONSORT statement. JAMA 1996;276:637– 639. 25. Greene B, Blanchard EB. Cognitive therapy for irritable bowel syndrome. J Consult Clin Psychol 1994;62:576 –582. 26. 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 –786. 27. Guthrie E, Moorey J, Margison F, Barker H, Palmer S, McGrath G, Tomenson B, Creed F. Cost-effectiveness of brief psychodynamicinterpersonal therapy in high utilizers of psychiatric services. Arch Gen Psychiatry 1999;56:519 –526. Received June 17, 1999. Accepted May 3, 2000. Address requests for reprints to: Elspeth Guthrie, M.D., Rawnsley Building, Manchester Royal Infirmary, Oxford Road, Manchester, M13 9BX, England. e-mail: [email protected]; fax: (44) 161-2732135. Supported by the Sir Jules Thorn Trust.