Biofeedback and Self-Regulation, VoL 13, No. 2, 1988
Notes and Observations
Bowel Sound Biofeedback as a Treatment for Irritable Bowel Syndrome Cynthia L. Radnitz and Edward B. Blanchard 1 State University of New York at Albany
Using an electronic stethoscope placed on subjects" abdomens, bowel sound biofeedback was administered to five subjects suffering from irritable bowel syndrome (functional diarrhea). They were instructed to alternately increase and decrease colonic sounds in an attempt to gain control over bowel activity. Using daily ratings o f diarrhea as the primary dependent measure, three of five subjects reduced mean ratings enough at posttreatment to meet our 50% criterion for success (100%, 94 %, and 54 %). A t 1-year follow-up, two o f the three short-term successes had maintained their level o f improvementeach had ratings 75% below those o f pretreatment. Descriptor Key Words: irritable bowel syndrome; bowel sound biofeedback; electronic stethoscope; functional diarrhea.
Irritable bowel syndrome (IBS) is the diagnosis typically given to gastroenterological patients who present with symptoms including abdominal pain or tenderness and a change in bowel habit (constipation and/or diarrhea) and for which no organic or anatomical cause can be found (Latimer, 1983). Traditionally, IBS has served as a residual category for a wide range of functional gastrointestinal complaints. Approximately 8 (Whitehead, Winget, Fedoravicius, Wooley, & Blackwell, 1982) to 17°70 (Drossman, Sandier, McKee, & Lovitz, 1982) of Americans suffer from IBS each year. No universally accepted treatment has been found for the disorder.
~Address all correspondence to Edward B. Blanchard, Center for Stress and Anxiety Disorders, 1535 Western Avenue, Albany, New York 12203. 169 0363-3586/88/0600-0169506.00/0 © 1988PlenumPublishingCorporation
170
Radnitz and Blanchard
A number of studies of the psychological treatment of IBS have been conducted (see the review by Blanchard, Schwarz, & Radnitz, 1987), yet few of these have distinguished between the diarrhea-predominant and constipation-predominant forms of the disorders (Whitehead et al., 1982), and almost all have used a very global, stress management approach to treatment (e.g., Neff & Blanchard, 1987; Whitehead & Schuster, 1985; Whorwell, Prior, & Faragher, 1984; Svedlund, Sjodin, Ottosson, & Dotevall, 1983). Furman (1973) took an alternative approach in treating five patients who complained of diarrhea and abdominal cramping. Furman taught his patients to increase and decrease bowel sounds using biofeedback. Operating under the assumption that the occurrence of bowel sounds represented colonic activity, he believed that learning to control these sounds would represent colonic control. He placed an electronic stethoscope on subjects' abdomens and instructed them to alternate increasing and decreasing bowel sounds. Verbal positive reinforcement was given when subjects achieved the desired objective. In a series of case studies, Furman reported that subjects became asymptomatic once they gained control of their bowel sounds. Although Furman reported successfully treating his patients and continues to do so (S. Furman, personal communication, October 25, 1985), 2 other attempts to replicate this procedure have not been successful (e.g., Weinstock, 1976; O'Connell & Russ, 1978). The present study sought to evaluate Furman's treatment using a multiple baseline across subjects methodology.
METHOD
Subjects Subjects participated in the study as a result of either media publicity or physician's referral. Subjects were selected for study if they had been diagnosed as having IBS and they reported diarrhea as their most debilitating symptom. Of the 15 subjects screened, 2 declined to participate, and 7 were excluded because their most problematic symptom was not diarrhea. Demographic characteristics are listed in Table I. The sample was composed of three females and three males. The average age was 31 years, and the average duration of symptomatology was 9.7 years. IBS was independently diagnosed by both a physician and an advanced doctoral-level student in clinical psychology (CLR). One subject ($6) dropped out after four treatment sessions because he could not fit appointments into his schedule. 2Theauthors'appreciationis extendedto Dr. Furmanfor his helpfuladviceon the proper way to conduct bowel sound biofeedback.
171
Bowel Sound Biofeedback
Table I. Demographic Characteristics Sex S1 $2 S3 $4 $5 $6 (dropped) Mean
M M F F F M
Age 25 31 31 26 37 38
Marital status Single Married Single Married Married Married
31.3
Duration of symptoms (years) 17 23 4 4 3 7 9.7
Procedure
Subjects were matched into pairs according to age, sex, and duration of symptoms. Since $6 dropped out of the study during treatment, results of $5 (to whom $6 was matched) are included singularly. One member of each pair monitored symptoms for 2 weeks, while the other monitored for 5 weeks, in the context of a multiple baseline across subjects design (Hersen & Barlow, 1976). Treatment began immediately after each subject completed baseline monitoring. We chose a multiple baseline design because we wanted to compare symptom ratings between subjects in different stages of treatment (e.g., one would expect subjects during a baseline phase to have higher symptom ratings than subjects in treatment). Subjects used diaries to record medication intake, change in diet, and gastrointestinal symptoms associated with IBS: abdominal pain and tenderness, constipation, diarrhea, nausea, belching, flatulence, and vomiting. Symptoms were rated once per day on a scale of 0 (not a problem) to 4 (debilitating problem) (see Neff & Blanchard, 1987, for reliability data). Symptom monitoring continued through treatment, for 2 weeks following treatment, and for 2 weeks at 1-month, 3-month, 6-month, and 1-year follow-ups. After subjects completed baseline monitoring, they received 10 treatment sessions held twice a week for 5 weeks. During the first session, subjects were given a treatment rationale: that by learning to control bowel sounds through both increasing and decreasing the amount of activity, they could learn to regulate bowel activity and thus obtain symptom relief. They were asked to place the head of the electronic stethoscope on their abdomens at a location where they reported regularly experiencing the greatest pain and/or spasms. The stethoscope was connected to a speaker system and a polygraph, so that bowel sounds were both broadcast to subjects and recorded on paper. The sounds were recorded on a 7P3 wide-band AC preamplifier ~nd integrated with a 7P10 preamplifier, set on ramp mode, of a Grass model 7 polygraph. This provided a cumulative depiction of activity in mil-
Radnitz and Blanchard
172
limeters/minute of cumulative pen excursion, which could easily be converted to numerical scores. Recordings were used as a measure of success in learning the technique. All 10 sessions used identical protocols. In each session, subjects spent the first 6 minutes listening to their bowel sounds. Then, the sound was turned off, and subjects attempted two 3-minute self-control phases in which they sought both to increase and then to decrease bowel sound activity without the feedback. Which direction came first depended on subjects' baseline levels (e.g., if baseline bowel sounds occurred only sporadically, increasing would be done first, and vice versa). Next, the sound was turned on again, and eight 4-minute biofeedback phases were conducted, alternating in each direction. During ari individual phase, subjects were asked to increase or decrease bowel sound activity for the entire phase. These sounds were amplified and fed back to them. During this procedure verbal positive reinforcement was given to subjects when they succeeded in changing in the desired direction. For home practice, subjects received regular stethoscopes, and instructions to practice daily. They were told to try increasing and decreasing bowel sounds without listening for 1 minute each and then to try two 4-minute phases of increasing and two of decreasing with the feedback (listening directly through the earpieces of the stethoscope).
RESULTS Subjects" Reaction to Treatment Patients indicated that bowel sound biofeedback seemed to be a logical therapy for IBS. On a scale of 0 (not logical) to 9 (very logical), the mean rating for the plausibility of this therapy was 7.5 (range = 7-8). Subjects did not find therapy sessions aversive, and all reported practicing at least once a day. All were satisfied with the results of therapy at posttreatment assessment (even the nonsuccessful subjects), indicating that satisfaction was not related to treatment outcome. When asked to rate the level of confidence they had that training would be successful in reducing future problems, subjects gave a mean rating of 7 on a scale of 0 to 9. Control of Bowel Sounds To ascertain whether subjects were able to learn to control bowel sounds, polygraph recordings of cumulative activity during increasing and decreasing phases were compared. For each subject, we decided a priori to
Bowel Sound Biofeedback
173
reduce only the data from sessions 1, 3, 5, 7, and 9. Only biofeedback phases were analyzed, so that 20 phases each of increasing and decreasing over five sessions were compared. Phases were scored by measuring the cumulative depiction of activity in millimeters. Only 1-minute segments that were not broken by a phase change (approximately 76% of the recorded minutes) were used. Scores for each phase were computed by the following formula: Total number of millimeters Number of minutes For each subject, means were computed for all increasing and decreasing phases that occurred during a session (see Table II). One-tailed, correlated t tests were conducted between increasing and decreasing activity level scores in order to see if subjects did, in fact, learn to control their bowel sounds. Two subjects, $2 and $3, were able to control bowel sound activity to a significant degree, t(19) = 1.80, p < .05, and t(19) = 1.93, p < .05, respectively, while one subject, $4, showed a slight trend, t(19) = 1.10, p < .20. In contrast, S1 did not demonstrate control, t(19) = .45, while $5 exhibited change contrary to expectations, t(19) = - 2 . 6 7 . Her decreasing trials were higher than her increasing trials. Clinical Outcome Diarrhea ratings from symptom diaries were the primary dependent variable since-this was the symptom targeted in treatment. We summed daily ratings to obtain a score for each week. Plots of average weekly ratings for each patient are presented in Figure 1. To calculate percent improvement, we used the following formula:
Percent Reduction = 100 Score
x
Symptom Symptom Baseline s c o r e - E n d of Treatment Score Baseline Symptom Score
A 50% reduction was chosen as our criterion for success, in keeping with the criterion introduced by Neff and Blanchard (1987). Results will be discussed in terms of both percentage improvement and pairwise comparisons of subjects. Means and percentage improvement scores for diarrhea ratings at posttreatment are listed in Table III. We used a multiple baseline across subjects design so that we would be able to compare subjects who were simultaneously at different stages of treatment. S1 did not improve according to our criterion and, in fact, worsened by 60%. $2
174
Radnitz and Blanchard
7
O
o
-
Z
Z ~3
I
~t3
O
c~ c~
~3 ¢D
I
o
z O
~3
c~
.=_
-i
z < ~3 0
~
0
"~, m
0
z
I
E c~
< o
Q
..=
Bowel Sound Biofeedback
175
did improve, however, by 94%, and was deemed a treatment success. Examination of Figure 1 reveals that while S1 was in treatment and $2 was in baseline, S1 had lower diarrhea ratings than $2. S1 continued to do better than $2 until $2 finished treatment. At posttreatment S2's ratings were lower than S l's. Both $3 and $4 were successful in treatment. $3 improved by 54°70 while $4 improved by 100%. Figure 1 shows that during weeks 4 and 5, when $3 was in treatment and $4 was still in baseline, $3 had lower diarrhea ratings than $4. $3 continued to do better than $4 until posttreatment, when $4 became symptom-free, whereas $3 still had a mild diarrhea problem. The remaining subject, $5, improved only by 20%, not enough to be counted as a treatment success. As described earlier, two of the three subjects who were clinical successes, $2 and $3, showed statistically significant control of bowel sounds, while the other successful subject, $4, showed a slight trend toward establishing control, p < .20. On the other hand, one of the failures, S 1, showed essentially no control, while the other failure, S5, showed change in the op-
Base ~ 10
Line i
TreaLmenL
Post-Treatment
ol
8
/\', i
6 b.O
"~
4
~
2
% t~
0
sa
\
i
i
i
i
i
i
1
2
3
4
5
i
",~
i
I
i
1
~
t
6
?
8
9
10
11
12
WEEKS Fig. 1, Mean weekly diarrhea ratings for subjects
in both the 2-week and 5-week baseline conditions during baseline, treatment, and posttreatment phases.
176
Radnitz and Blanchard
posite direction-that is, decrease trials were higher than increase trails. While far from a perfect dose-response relationship, these results indicate that degree of success in controlling bowel sounds appears to be related to clinical outcome.
Follow-Up Data Two weeks of diaries kept by subjects at 1-month, 3-month, 6-month, and 1-year intervals after the end of treatment served as measures of maintenance. Percentage improvement scores of diarrhea ratings were computed for individual subjects from pretreatment to each follow-up period using the symptom reduction score formula. Determination of success or failure was made for each subject at each follow-up period using the 50% criterion established for posttreatment scores. Follow-up means and percentage improvement scores are listed in Table III. During the first three follow-up periods, four of five subjects had improved by 50% or more, qualifying them as successes. These subjects included $2, $3, and $4, who were successes at posttreatment, plus $5, who became asymptomatic at follow-up. Examination of 1-year follow-up data reveals that the three subjects who were successes at posttreatment ($2, $3, and $4) were still classified as successful, while those who were failures (S 1 and $5) were still failures. In the case of one subject, however, results at 1 year should be interpreted with caution. $3 reported that she began to take Alprazolam for both gastrointestinal and anxiety problems a few months before the 1-year follow-up contact. Thus, her asymptomatic status at 1 year could be attributed to the medication. DISCUSSION On the basis of symptom diary scores, three out of the five patients who completed treatment met our criterion for success in terms of reduction of diarrhea. Moreover, in terms of the experimental analysis, the data in Figure 1 are compatible with an interpretation that improvement began only after treatment was introduced. We recorded bowel sound activity on a polygraph to see if learning control of bowel sounds was related to treatment success. Two of the three successful subjects, $2 and $3, learned to control bowel sound activity to a statistically significant extent, while the third successful subject showed a trend. (At this point we do not know what degree of control "normal" asymptomatic individuals might have.) The two end-of-treatment failures did not show control of bowel sounds. There were no significant differences between Sl's increasing and decreasing trials, while for $5, decreasing trials
Bowel Sound Biofeedback
177
were significantly greater than increasing trials. Although S5's experience with bowel sound biofeedback may be construed as a negative "side effect" of treatment, she nonetheless showed some improvement in symptomatology at both posttreatment and follow-up. While the relationship between success in learning the technique and success in treatment was not a perfect dose-response relationship, the trend was in the appropriate direction to support the idea that a relationship existed between the two. Ideally, clinical improvement should be correlated with physiological control in the form of a dose-response relationship. However, such relationships rarely appear in the biofeedback literature, even in such well-researched areas as chronic headache (Blanchard & Andrasik, 1985). That a perfect relationship between physiological changes and clinical improvement was not found in this study should not detract from the merits of the procedures. The percentage improvement scores for diarrhea ratings indicate a high degree of variability in subjects' responses to treatment. Factors that could have contributed to this variability include medication changes, diet changes, and stress. No subjects reported medication or diet modifications on their diaries, but the two subjects who were treatment failures reported that major life stressors occurred during treatment or posttreatment monitoring. S 1 got married immediately after he completed treatment, while $5 reported taking her board certification exams while treatment was ongoing. The occurrence of these life stressors may have worked against these subjects' attainment of the symptom relief experienced by the other three subjects. In attempting to explain the variable results experienced by subjects at posttreatment, we examined initial baseline activity. The two failures had much lower ratings at baseline than the three successful subjects. Perhaps, they experienced a floor effect; alternately, this procedure may be appropriate only for patients with a more severe diarrhea problem. Exactly which variables (e.g., life stress or initial baseline levels) predict success and failure is an important question for further study. Throughout the first three follow-up periods, four subjects reported diary-based ratings of improvement that were at least 50% below baseline levels. These included the three subjects who were successful at posttreatment plus $5. $5 became asymptomatic between posttreatment and 1-month follow-up. We believe this result was related to the exam she took (and failed) during treatment. Throughout treatment, she seemed quite distressed about it; however, at follow-up she indicated that she had accepted her failure and that she wasn't going to let it upset her any longer. Perhaps this change in attitude toward the event was responsible for her being asymptomatic during the first three follow-up periods. From the experiences of $5 and S1, it seems likely that stressful events militate against treatment success. Results at 1 year seem to support this state-
178
Radnitz and Blanehard
ment. $3 reported undergoing very stressful psychotherapy between 6-month and 1-year follow-ups. She reported symptoms of anxiety that induced gastrointestinal distress to the extent that she began taking Alprazolam. Thus, despite her being asymptomatic at the follow-up contact, we cannot attribute her success at 1 year to the biofeedback treatment. Our variable degree of success is in contrast to the uniformly positive results reported initially by Furman. However, we achieved a higher degree of success than other attempts to replicate this procedure (e.g., Weinstock, 1976). When we contacted Furman in 1985, he was aware of only one other attempt to replicate his procedure. (His comment pointed to important differences in the replication procedure from the original procedure.) Our success relative to other replication attempts could be due to the explicit information we received from Furman regarding how he conducted his procedure. The purpose of this study was to further explore a treatment reported by Furman (1973) to be clinically effective. Follow-up data supported the idea that treatment effects were maintained for 1 year in two of three initially successful subjects. We observed that in three instances, major stressors may have militated against a successful outcome. This leads us to wonder if the addition of a stress management component might improve the clinical utility of bowel sound biofeedback. Future research is needed to assess the effect of stressful events on treatment outcome and the potential incremental utility of each therapy.
REFERENCES
Blanchard, E. B., & Andrasik, F. (1985). Management of chronic headache. New York: Pergamon Press. Blanchard, E. B., Schwarz, S. P., & Radnitz, C. (1987). Psychologicalassessment and treatment of irritable bowel syndrome. Behavior Modification, 11(3), 348-372. Drossman, D. A., Sandier, R. S., McKee, D. C., & Lovitz, A. J. (1982). Bowelpatterns among subjects not seeking health care. Gastroenterology, 83, 529-534. Furman, S. (1973). Intestinal biofeedback in functional diarrhea: A preliminary report. Journal of Behavior Therapy and Experimental Psychiatry, 4, 317-321. Hersen, M., & Barlow, D. H. (1976). Single case experimental designs: Strategies for studying behavior change. New York: Pergamon Press. Latimer, P. R. (1983). Functional gastrointestinal disorders: A behavioral medicine approach. New York: Springer. Neff, D. F., & Blanchard, E. B. (1987). A multi-componenttreatment for irritable bowel syndrome. Behavior Therapy, 18, 70-83. O'Connell, M. F., & Russ, K. L. (1978). A case report comparing two types of biofeedback in the treatment of irritable bowel syndrome. Paper presented at the Ninth Annual Biofeedback Society of America meeting, Albuquerque, NM. Svedlund, J., Sjodin, I., Ottosson, J.-O., & DotevaU,G. (1983). Controlledstudyof psychotherapy in irritable bowel syndrome. Lancet, 2, 581-592. Weinstock, S. A. (1976). The reestablishment of intestinal control in functionalcolitis. Biofeedback and Self-Regulation, 1, 324.
Bowel Sound Biofeedback
179
Whitehead, W. E., & Shuster, M. M. (1985). Gastrointestinal disorders: Behavioral andphysiological basis for treatment. New York: Academic Press. Whitehead, W. E., Winget, C., Fedoravicius, A. S., Wooley, S., & Blackwell, B. (1982). Learned illness behavior in patients with irritable bowel syndrome and peptic ulcer. Digestive Diseases and Sciences, 27, 202-208. WhorweU, P. J., Prior, A., & Faragher, E. B. (1984). Controlled trial of hypnotherapy in the treatment of severe refractory irritable-bowel syndrome. Lancet, 2, 1232-1234.