Society: AGA

Introduction:
Fecal incontinence is associated with substantial impacts on quality of life. Prevalence estimates range from 2% to as high as 20% based on the methodology used, and is likely underreported by patients. Prior studies have demonstrated inconsistent effects of body-mass index on fecal incontinence. This study sought to further clarify the impact of modifiable measured physical indices of body mass on fecal incontinence in a nationally representative cross-sectional cohort of participants.
Methods:
This study analyzed the National Health and Nutrition Examination Survey (NHANES), an annual cross-sectional, nationally representative health survey. Surveys collected between 2005-2010 included survey components assessing bowel health. Fecal incontinence was defined by accidental bowel leakage of solid stool, liquid, or mucus at least once in the past month. Sample weight adjustment was performed to account for complex survey design. Stepwise multivariate logistic regression models were constructed to assess risk factors for fecal incontinence.
Results:
14,345 participants were identified with responses to the bowel health survey. After stepwise adjustment for covariates (including demographics, comorbidities, dietary factors, socioeconomic factors, depression scores, diarrhea, marijuana use, Figure 1), 7,530 participants were included in the final multivariate model (Figure 2). Waist circumference-height ratio was a significant predictor of fecal incontinence (OR 1.73, 95% CI 1.04-2.88, p=0.036) comparing the 4^th to the 1^st quartile of values. Quartiles of BMI, conversely, did not significantly predict fecal incontinence.
Conclusions:
Waist circumference-height ratio is a significant risk factor for fecal incontinence. While waist-circumference-height ratio predicted fecal incontinence, body-mass index did not. This suggests bowel continence depends on how body mass is distributed. Further study of the role of central deposition of body mass in neuromuscular control of bowel continence is warranted.

BACKGROUND. Some women with urge-predominant fecal incontinence (FI) have a stiffer rectum and rectal hypersensitivity. In a phase 2 study, the a₂ adrenergic agonist clonidine, which increases rectal compliance and reduces rectal sensation, did not reduce FI in all women with urge FI, but had borderline significant effects in women with diarrhea and urge FI. Up to 25% of patients with idiopathic diarrhea have bile acid malabsorption, which may respond to bile acid sequestrants. HYPOTHESES. Clonidine and colesevelam (CC) will improve fecal continence in women with urge FI. METHODS. Symptoms were recorded for 2-4 weeks before and 4 weeks during randomization to placebo or oral clonidine 0.1mg bid and colesevelam 625 mg tid in 59 women with urge-predominant FI who did not have an organic GI disorder. Bowel symptoms (daily diaries and other questionnaires), anal pressures, and rectal sensation were assessed. The last 2 wks of the baseline period and weeks 2-4 of treatment phase were analyzed. Fecal Incontinence Symptom Severity (FISS) score, fecal incontinence severity index (FISI) and FISI-QoL were computed pre- and during treatment. For the primary outcome (number of FI episodes), missing data were conservatively imputed using the average for all participants. RESULTS. At baseline, placebo and CC groups were not different (Table 1); on average patients had moderately severe FI. After the baseline phase, 3, 1 and 51 patients respectively completed 1, 2, and 4 wks of therapy; 1 eligible patient did not begin therapy and 3 patients were not eligible to receive therapy. Several measures of FI (number of episodes of FI, satisfaction with treatment) improved with placebo (Table 2). The primary outcome (≥50% reduction in FI episodes) was not different between CC (13/24 [54%]) and placebo (17/32 [53%]). Compared to placebo, CC improved the FISS but not other bowel symptoms. Among 21 and 30 patients who completed 4 wks of CC and placebo, the change (treatment – baseline) in FI episodes during therapy was directly correlated with the rectal urge sensory threshold (r_s= 0.50, P =.03) in the CC but not in the placebo group (r_s= 0.17, P =.34). CONCLUSION. In this controlled study, 53% of patients treated with placebo reported a >50% reduction in the number of FI episodes; for this outcome, effects of CC were not significant vs placebo. But, CC did improve the FISS score, which incorporates frequency, type, and amount of leakage and rectal urgency vs placebo. CC but not placebo was more effective in pts with a higher rectal sensory threshold volume for urgency, which suggests that is more likely to be effective in pts without rectal hypersensitivity. Future controlled trials should evaluate the utility of CC in patients who have documented bile acid malabsorption.

Table 1. Baseline Characteristics

Table 2. Primary, Secondary and Additional End Points

Table 2. Primary, Secondary and Additional End Points

Background: Controversy exists regarding the optimum diagnostic criteria for a defecatory disorder. The commonly employed Rome definition utilises symptoms (functional constipation (FC) or constipation predominant irritable bowel syndrome (IBS-C)) plus disordered defecatory physiology to diagnose a functional defecation disorder (FDD). Biofeedback is the gold standard treatment for FDD and response to biofeedback could therefore be used to assess the validity of the current Rome IV diagnostic criteria.
Aim: To compare the outcome of biofeedback therapy in patients with a formal Rome IV FDD diagnosis to a group of patients presenting with defecatory symptoms who, for various reasons, do not meet Rome criteria.
Method: 380 consecutive patients (mean age 50yrs, 348F) presenting to a tertiary referral clinic with defecatory symptoms who underwent biofeedback therapy between 2003-2019 were included. Prospectively collected data included Rome and Knowles constipation questionnaires, structured clinical interview, anorectal manometry, sensation testing, and balloon expulsion. Two or more point improvement in patient-reported bowel satisfaction visual analog score was used as the main outcome. Six symptoms (stool consistency, time spent in toilet, sense of blockage during defecation, sense of incomplete emptying, digitation and abdominal pain preceding altered bowel movements) were evaluated for their utility in identifying successful biofeedback therapy using backward stepwise logistic regression.
Results: 76% of all patients reported two or more point improvement in bowel satisfaction on completion of biofeedback. There was no significant difference in biofeedback outcome between patients who met Rome FDD compared to those who did not (p=0.33). There was also no difference between those who met Rome FC or IBS-C and those that did not in terms of biofeedback outcome (p=0.48). However, having two or more abnormal physiology measures showed a trend towards better prediction of biofeedback outcome compared to those with more normal physiology (p=0.076; Figure 1). Digitation during defecation and abdominal pain preceding abnormal bowel movements were the only variables independently associated with biofeedback outcome on regression analysis (p=0.014, p=0.041 respectively; Table 1).
Conclusion: Despite current recommendations, a Rome IV diagnosis of FDD was no better at predicting better biofeedback outcome than patients referred with defecatory symptoms that did not meet Rome criteria. Physiology criteria appeared better than FC or IBS-C criteria in predicting biofeedback outcome. Other defecatory symptoms, especially digitation and abdominal pain with altered bowel movements, show promise to be combined with abnormal physiology to predict outcome of biofeedback. These results should be considered in further refinement of Rome diagnostic criteria for FDD.

Introduction: High resolution anorectal manometry (HRAM) is an important tool in assessing anorectal dysfunction. The recent International Anorectal Physiology Working Group protocol and London classification (LC) proposed standardized criteria for performing and interpreting HRAM. However, comparative data on LC versus conventional models in patient-reported outcome measures remain limited. We aimed to compare LC diagnoses of HRAM findings with conventional measures in predicting patient-reported symptom outcomes.
Methods: Consecutive adults referred for HRAM at a tertiary center were enrolled. Validated instruments were prospectively collected at time of study, including the lower GI domains of the Gastrointestinal Patient-Reported Outcomes Measurement Information System (GI-PROMIS), global symptom severity (GSS, 100-mm visual analog scale), and 12-item Short-Form Health Survey (SF-12) for health-related quality of life. Statistical analyses were performed using Chi-squared or student t-test and Spearman correlation (univariate); and generalized linear regression (multivariate).
Results: 384 patients (80.5% female, 53.1 yrs, BMI 27.0) were enrolled. Older age was associated with worse incontinence (R=0.192, p=0.0001) but lower constipation (R=-0.172, p=0.0006) scores. These two hindgut domains also correlated with worse GSS and physical component score of SF-12. On HRAM, anal resting tone correlated positively with constipation (R=0.22, p<0.0001), but negatively with incontinence (R=-0.126, p=0.014). Amongst conventional measures, prolonged balloon expulsion test (BET) (p=0.040) and manometric patterns of dyssynergia (p=0.017), particularly types I and II, correlated with worse constipation scores, while low squeeze pressure increment correlated with worse incontinence scores (p=0.04). When categorized per LC, major (but not minor) findings for disorders of anal tone and contractility (DATC) predicted incontinence scores, while minor (but not inconclusive) findings for disorders of anorectal coordination (DARC) predicted constipation scores. On multivariate models, conventional measures and LC diagnoses both independently predicted worse constipation and incontinence (Table 1). However, LC models demonstrated improved model fit versus conventional measure models in predicting both constipation and incontinence symptoms (Table 2).
Conclusions: LC and conventional HRAM parameters independently predicted severity of patient-reported outcomes for constipation and incontinence. These results support the current major/minor and minor/inconclusive LC designations in DATC and DARC, respectively. Models constructed using LC demonstrated improved model fit compared to conventional diagnostics in predicting symptom severity. LC provides standardized measures on HRAM that correlate with patient-reported outcomes that may help guide symptom management.

INTRODUCTION: Chronic constipation is one of the most common reasons for gastroenterology referral. Many patients will achieve long-lasting remission from pelvic floor physical therapy (PT), but care is often delayed or never offered. We performed an economic analysis comparing a variety of management strategies for chronic constipation.

METHODS: We conducted a cost-benefit analysis from a patient perspective and cost-minimization analysis from an insurer perspective. A Markov model was constructed to compare four strategies applicable to patients failing initial over-the-counter (OTC) and lifestyle recommendations: (1) “Drug first”: empiric prescription drug followed by pelvic floor PT for drug failure; (2) “Pelvic floor PT first”: empiric pelvic floor PT followed by prescription drug for PT failure; (3) “Up-front anorectal function testing”: care directed by up-front anorectal function testing; (4) “Repeated laxatives”: rotating OTC laxatives and lifestyle advice. Model inputs were sourced from a recent systematic review of drug trials and recent pelvic floor physical therapy trials, national cost databases, and observational studies evaluating health outcomes and costs with chronic constipation. The time horizon was one year with a willingness-to-pay threshold of $100,000/quality-adjusted-life-year gained. Sensitivity analysis was performed on the expected likelihood of response to pelvic floor PT.

RESULTS: From the insurance perspective, repeated laxatives incurred a cost of $2,566/year. The drug first strategy added $4,580/year in costs for a 0.019 QALY gain/year. The pelvic floor PT first strategy resulted in greater QALY gains and lower costs than repeated laxatives or drug first strategies, making the pelvic floor PT first strategy a dominant (preferred) strategy over the drug first strategy for insurers. From the patient perspective, repeated laxatives incurred a cost of $13,761/year driven by out-of-pocket healthcare costs and the impact of symptom severity on work productivity. To patients, both the pelvic floor PT first and drug first strategies resulted in lower costs and greater QALY gains than repeated laxatives. By optimizing patient selection on expected outcomes, up-front anorectal function testing was the least expensive strategy, saving $5,796 to insurers and $4,923 to patients compared to a drug first strategy. Up-front anorectal function testing also yielded the greatest QALY gains of all strategies (+0.043 QALY gain/year). Therefore, up-front anorectal function testing was the dominant (preferred) strategy from insurance and patient perspectives.

DISCUSSION: Up-front anorectal function testing optimizes cost-effective care for patients with chronic constipation that fail OTC/lifestyle recommendations.

Background/Aims: We previously proposed a novel concept of integrated pressurized volume (IPV), which is calculated by multiplying amplitude, distance, and a specified time period, in high resolution anorectal manometry (HRAM) and found that it enables to predict the results of delayed balloon expulsion (BE) tests better than the conventional manometric parameters such as defecation index. However, the previous study could not use the entire raw dataset because of methodological limitations. We hypothesized that an automatic diagnostic tool using artificial intelligence (AI) could more accurately identify delayed BE test using the HRAM raw dataset and time-series IPVs if all pressure data of the push maneuver period was used without any data loss.
Methods: Three hundred patients with constipation (130 males and 170 females) were enrolled and underwent HRAM and BE test from September 2020 to May 2021. Delayed BE was defined as a requirement of more than 1 min. The time-series IPVs were calculated within a particular time interval during the push maneuver. The raw HRAM data and time-series IPVs were split into multiple window frames and data was augmented with 75% overlap to improve performance. Convolutional neural networks (CNNs) and a long short-term memory (LSTM) network were applied to predict BE test results using time-series data of manometric pressure and IPVs. The CNNs extract features of each dataset and LSTM learns chronological changes in the feature during the push maneuver.
Results: Delayed BE test was observed in 69 (53.1%) male and 49 (28.8%) female patients, respectively. From the receiver operating characteristic (ROC) curve analysis, time-series IPV ratio between the upper 1 cm and lower 3 cm of the anal canal (TS-IPV₁₃ratio) was the best parameter for predicting BE test results in both male and female patients. With the TS-IPV₁₃ ratio parameter, the proposed model achieved a 0.96 F1-score, 0.96 accuracy, 1.00 sensitivity, 0.92 specificity, and 0.99 AUC in female patients, and a 0.97 F1-score, 0.97 accuracy, 1.00 sensitivity, 0.93 specificity, and 0.99 AUC in male patients. In contrast, with the conventional anorectal parameters, the defecation index yielded an AUC of 0.543 for male and an AUC of 0.655 for female patients, respectively.
Conclusions: Our AI model was able to accurately classify delayed BE test from raw HRAM data and time-series IPVs in both male and female patients with an AUC of 0.99, which was significantly better than the conventional manometric parameters. Furthermore, the model used time-variant HRAM pressure data and TS-IPVs throughout the entire time of push maneuver without any data loss; thus, it could be used as a more reliable biomarker to classify delayed BE test compared to conventional parameters.

The proposed convolutional neural networks (CNN) - long short-term memory (LSTM) model architecture for identification of delayed balloon expulsion test using high-resolution anorectal manometry and time-series integrated pressurized volume (TS-IPV).

The proposed convolutional neural networks (CNN) - long short-term memory (LSTM) model architecture for identification of delayed balloon expulsion test using high-resolution anorectal manometry and time-series integrated pressurized volume (TS-IPV).

Seung-Jae Myung

Asan Medical Center, University of Ulsan College of Medicine

Appearances