Biofeedback treatment of constipation: a critical review

This review was designed to 1) critically examine the research design used in investigations of biofeedback for pelvic floor dyssynergia, 2) compare the various biofeedback treatment protocols for pelvic floor dyssynergia-type constipation used in this research, 3) identify factors that influence treatment outcome, and 4) identify goals for future biofeedback research for pelvic floor dyssynergia.

A comprehensive review of both the pediatric and adult research from 1970 to 2002 on "biofeedback for constipation" was conducted using a Medline search in all languages. Only prospective studies including five or more subjects that described the treatment protocol were included. In addition, a meta-analysis of these studies was performed to compare the outcome of different biofeedback protocols for treating constipation.

Thirty-eight studies were reviewed, and sample size, treatment protocol, outcome rates, number of sessions, and etiology are shown in a table. Ten studies using a parallel treatment design were reviewed in detail, including seven that randomized subjects to treatment groups. A meta-analysis (weighted by subjects) was performed to compare the results of two treatment protocols prevalent in the literature. The mean success rate of studies using pressure biofeedback (78 percent) was superior (P = 0.018) to the mean success rate for studies using electromyography biofeedback (70 percent). However, the mean success rates comparing studies using intra-anal electromyography sensors to studies using perianal electromyography sensors were 69 and 72 percent, respectively, indicating no advantages for one type of electromyography protocol over the other (P = 0.428). In addition to the varied protocols and instrumentation used, there also are inconsistencies in the literature regarding the severity and etiology of symptoms, patient selection criteria, and the definition of a successful outcome. Finally, no anatomic, physiologic, or demographic variables were identified that would assist in predicting successful outcome. Having significant psychological symptoms was identified as a factor that may influence treatment outcome, but this requires further study.

Although most studies report positive results using biofeedback to treat constipation, quality research is lacking. Specific recommendations are made for future investigations to 1) improve experimental design, 2) clearly define outcome measures, 3) identify the etiology and severity of symptoms, 4) determine which treatment protocol and which component of treatment is most effective for different types of subjects, 5) systematically explore the role of psychopathology in this population, 6) use an adequate sample size that allows for meaningful analysis, and 7) include long-term follow-up data.