Over the past 20 years obesity has emerged as the

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2005;3:992 996 The Association of Gastrointestinal Symptoms With Weight, Diet, and Exercise in Weight-Loss Program Participants RONA L. LEVY,* JENNIFER A. LINDE, KAYLA A. FELD, MICHAEL D. CROWELL, and ROBERT W. JEFFERY *University of Washington, Seattle, Washington; Division of Epidemiology and Community Health, University of Minnesota School of Public Health, Minneapolis, Minnesota; Harvard University, Cambridge, Massachusetts; and Mayo Clinic, Scottsdale, Arizona Background & Aims: Studies on the relationship between gastrointestinal (GI) symptoms and obesity are limited. Research on the relationship between GI symptoms (including irritable bowel syndrome [IBS]), weight, and weight-related behaviors are rare. This study assessed rates of GI symptoms in a sample of obese patients in a weight-loss program and explored relationships among GI symptoms and obesity, binge eating, dieting (fat and fruit/fiber consumption), and physical activity. Methods: A total of 983 participants (70% women) had a mean body mass index (BMI) of 33.2 5.7 kg/m 2 (range, 25.1 60.8 kg/m 2 ) and a mean age of 52.7 12.4 years (range, 20.4 89.8 y). Participants completed a questionnaire about diet and physical activity and a standardized self-report Rome II questionnaire assessing IBS status and GI symptoms. Results: In bivariate analyses BMI was associated positively with abdominal pain and diarrhea whereas healthier diet (lower fat and higher fruit/fiber intake) and higher physical activity were associated with fewer GI symptoms. In multivariate models BMI was not associated with GI symptoms; physical activity remained a protective factor. Conclusions: Although physiologic mechanisms still need to be explored, associations between GI symptoms and diet and exercise behaviors may have implications for the treatment of both obesity and GI symptoms. Over the past 20 years obesity has emerged as the most important nutrition problem in the United States. 1 The average body weight of Americans has increased by approximately 10% during this time. More than half of the adult population is overweight; nearly 1 in 3 adults is clinically obese. Because obesity clearly is related to dietary variables (ie, fat intake is related positively to obesity and fruit/ vegetable intake and physical activity are related inversely to obesity; 2 and Linde, Utter, Jeffery, Sherwood, Pronk, and Boyle, unpublished data), it seems plausible that obesity also might be related to gastrointestinal (GI) symptoms, and anecdotal reports indicate that common GI disorders such as irritable bowel syndrome (IBS) are observed commonly among individuals seeking treatment for obesity as well. To date, however, there are few research studies on the relationship between GI symptoms, weight, and weight-related behaviors (diet and physical activity). Because GI symptoms could affect or be affected by weight itself and by weight-related behaviors, a better understanding of the relationship between them potentially is important. Recent population studies published by Delgado-Aros et al 3 and Talley et al 4 have examined the relationship between body mass index (BMI) and GI symptoms in community samples. These studies found significant positive associations between BMI and diarrhea, 3,4 abdominal pain with nausea or vomiting, 3 and vomiting, upper abdominal pain, and bloating. 4 The present study sought to expand work in this area by studying the relationships among GI symptoms, weight, and weight-related behaviors in a large sample of overweight and obese patients who had participated recently in a weight-loss program. Relationships were examined between GI symptoms (including IBS status) and BMI, and the relationship between GI symptoms and specific weight-loss behaviors of physical activity, fat intake, and fruit/vegetable intake. Binge eating disorder (BED), a condition characterized by frequent consumption of large amounts of food in short periods of time accompanied by feelings of loss of control, also was included in the analyses. Our working hypotheses were that GI symptoms and BMI would be related positively and that GI symptoms would be related inversely to behaviors that reduce obesity risk (ie, high exercise, low fat intake, and high fruit/vegetable intake). Abbreviations used in this paper: BED, binge eating disorder; BMI, body mass index; GI, gastrointestinal; IBS, irritable bowel syndrome; MCO, managed care organization. 2005 by the American Gastroenterological Association 1542-3565/05/$30.00 PII: 10.1053/S1542-3565(05)00696-8

October 2005 GI SYMPTOMS, WEIGHT, DIET, AND EXERCISE 993 Table 1. Demographics and Symptom Reporting Frequencies in the Study Sample Sex Male 294 (29.9%) Female 689 (70.1%) Mean age (SD) 54.5 (11.8) Education High school or less 194 (19.7%) Some college or more 789 (80.3%) Ethnicity Non-White 65 (6.6%) White 917 (93.4%) Marital status Not married 260 (26.4%) Married or living with partner 723 (73.5%) Current smoking status Nonsmoker 915 (93.3%) Smoker 66 (6.7%) Mean BMI (SD) 33.4 (5.6) Overweight (25.0 29.9 kg/m 2 ) 313 (31.8%) Obese class I (30.0 34.9 kg/m 2 ) 372 (37.9%) Obese class II (35.0 39.9 kg/m 2 ) 179 (18.2%) Obese class III ( 40.0 kg/m 2 ) 119 (12.1%) BED (yes) 56/967 (5.8%) IBS positive 128/963 (13.3%) Abdominal pain (yes) 187/963 (19.4%) Decreased stools (yes) 57/973 (5.9%) Increased stools (yes) 136/973 (14.0%) Constipation (yes) 165/973 (17.0%) Diarrhea (yes) 243/973 (25.0%) Bloating (yes) 203/973 (20.9%) Recent exercise (kcal/wk, mean/sd) 1202.1 (1048.8) Fat intake (servings/mo, mean/sd) 31.2 (6.1) Fruit/vegetable/fiber intake (servings/mo, mean/sd) 26.1 (5.0) NOTE. N 983. Methods Study Description Data used in this report were collected from 983 men and women participating in a 24-month randomized trial evaluating telephone- and mail-based interventions for weight loss. 5 Participants were members of a managed care organization (MCO) recruited by mail, clinic flyers, postings to the MCO web site, and by physician referral. Eligibility criteria for study inclusion were broad: age 18 years or older; not currently pregnant, lactating, or planning a pregnancy within the next 6 months; and BMI greater than 27 kg/m 2 based on self-reported height and weight. Inclusion in the present analyses required completion of weight measurement and questionnaires about weight-related behaviors and GI symptoms at 24 months. The University of Minnesota and MCO institutional review board committees approved the study protocol. Study participants were randomized to 1 of 3 conditions (a mail-based weight intervention, a telephone-based weight intervention, or a usual care group composed of access to telephone- and clinic-based weight-loss counseling at a modest cost). Details of these procedures and their effects on behaviors and weight are described elsewhere. 5 Measures The following measures were included in the analyses. Demographic characteristics. Age, sex, education, ethnicity, and marital status were assessed at baseline. Smoking. Smoking status was defined as current use of cigarettes (yes or no). Height, weight, and body mass index. During the baseline clinic visit, trained research staff measured height and weight with a calibrated electronic scale and a wall-mounted ruler. Weight measurements were repeated at 24 months. BMI (weight [kg]/height [m 2 ]) was computed. Irritable bowel syndrome status and gastrointestinal symptoms. At 24 months, participants completed a standardized Rome II questionnaire to assess IBS status. 6 The instrument included specific questions about the frequency of the following GI symptoms over the past 3 months: decreased stools, increased stools, constipation, diarrhea, abdominal pain, and bloating. Dietary intake. The Block Screening Questionnaire for Fat and the Block Screening Questionnaire for Fruit/ Vegetable/Fiber Intake 7 were used to assess usual dietary intake for these foods. The frequency of consumption of 15 high-fat foods and 9 foods high in fiber (fruit and vegetables) over the past 6 months was assessed with response categories of less than once a month, 2 3 times per month, 1 2 times per week, 3 4 times per week, 5 or more times per week. Scores reported here are servings per month of fat and fruit/vegetable/ fiber foods summed across items. The reliability and validity of this instrument has been shown previously. 8,9 Binge eating status. BED was assessed using 3 items from the Questionnaire on Eating and Weight Patterns, which has been used in community and weight-treatment seeking populations to determine the presence of BED and to validate diagnostic criteria. 10,11 Participants are asked the frequency of overeating episodes and feelings of loss of control over their eating during the past 6 months. Participants were classified as binge eaters if they reported binge episodes and feelings of loss of control on at least 2 days per week for the past 6 months. This BED questionnaire was developed by the group who field-tested and validated the diagnostic criteria for BED in the Diagnostic and Statistical Manual of Mental Disorders, and during instrument development and validation. The measure has a score of.60 for clinician (eating disorder specialist) vs self-reported symptom status, which is comparable with agreement rates for other psychiatric diagnoses. 11,12 Physical activity. Recent exercise, in kilocalories expended per week, was assessed using the Paffenbarger Activity Questionnaire. 13 The Paffenbarger Activity Questionnaire assesses city blocks walked and stairs climbed per day and provides spaces for participants to indicate leisure activities pursued during the past week. The measure yields an estimate of energy expenditure per week in leisure time physical activity and has well-established reliability and validity in dieting,

994 LEVY ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 10 Table 2. Bivariate Odds Ratios for Associations of BMI, Binge Eating Status, Physical Activity, and Dietary Intake With IBS and GI Symptom Status IBS status Abdominal pain Decreased stools Increased stools Constipation Diarrhea Bloating BMI (continuous) 1.04 (1.01 1.07) 1.03 (1.01 1.06) 1.01 (.96 1.06) 1.03 (1.00 1.06).99 (.96 1.02) 1.05 (1.02 1.07) 1.02 (.99 1.04) P value.02.01.70.09.58.0003.26 BED status (yes) 1.91 (.98 3.73) 3.28 (1.87 5.75) 1.71 (.65 4.47) 1.05 (.49 2.27) 1.93 (1.04 3.58) 1.13 (.61 2.08) 3.76 (2.16 6.54) P value.06.0001.28.90.04.70.0001 Recent exercise (kcal/wk).93 (.86 1.00).88 (.82.94).88 (.80.95).90 (.84.96).97 (.90 1.05).91 (.85.97).92 (.87.99) P value.05.0001.002.002.46.003.02 Fat intake (servings/mo) 1.02 (.99 1.06) 1.02 (.99 1.05).97 (.93 1.02) 1.03 (1.00 1.07) 1.00 (.97 1.03) 1.03 (1.01 1.06) 1.00 (.97 1.03) P value.16.17.26.04.80.02.93 Fruit/vegetable/fiber (servings/mo).97 (.93 1.01).98 (.95 1.02).96 (.90 1.01).97 (.93 1.00).97 (.94 1.01).98 (.95 1.01).97 (.94 1.00) P value.11.27.11.08.15.21.04 NOTE. N 895 973 because of missing values. Values in parentheses represent 95% confidence intervals. obese populations. 14 This variable was log transformed for analyses because of a skewed distribution. Data Analysis Statistical analyses were conducted using SAS version 8.2 (SAS, Cary, NC). 15 Logistic regression models were used to calculate odds ratios with 95% confidence intervals for associations of IBS status and GI symptoms by BMI, binge eating status category, and with dietary intake and physical activity. The analysis was conducted in 3 steps. In step 1 we examined bivariate associations; step 2 added age, sex, and smoking as covariates; step 3 used a multivariate model that included all predictor variables and demographic variables and smoking, if they were significant in analysis step 2. Results Demographics and Body Mass Index Of 1801 participants enrolling in the trial, 1155 (64%) completed a questionnaire on IBS status at 24 months, 1166 (65%) completed a questionnaire on GI symptoms, 1000 (56%) provided measured weights for calculation of 24-month BMI, and 983 (55%) completed all study measures. As compared with those who were present at baseline only, those completing all study measures were older (mean age, 52 vs 49 y, t[1684] 6.90; P.0001); weighed less (mean BMI, 33.8 vs 34.7 kg/m 2, t[1671] 3.38; P.001); ate more fruits, vegetables, and fiber (mean, 25.3 vs 23.9 servings per month, t[1762] 5.73; P.0001); and were more likely to be married (73% vs 67%, 2 (1) 8.41; P.01), white (93% vs 88%, 2 (1) 14.35; P.001), and college-educated (80% vs 74%, 2 (1) 9.13; P.01). A descriptive summary of all variables used in this report is shown in Table 1. The mean BMI of the group was 33.2 5.7 kg/m 2 (range, 25.1 60.8 kg/m 2 )at24 months. The mean age was 52.7 12.4 years (range, 20.4 89.8 y). Seventy percent of the sample were women, 93% were white, 74% were married or living with a partner, and 80% had attended at least some college. Ninety-three percent of the sample was nonsmoking. Table 3. Demographic-Adjusted Odds Ratios for Associations of BMI, Binge Eating Status, Physical Activity, and Dietary Intake With IBS and GI Symptom Status IBS status a Abdominal pain a Decreased stools b Increased stools c Constipation c Diarrhea c Bloating c BMI (continuous) 1.02 (.99 1.06) 1.03 (1.00 1.05) 1.00 (.96 1.05) 1.02 (.99 1.06).98 (.95 1.02) 1.04 (1.02 1.07) 1.00 (.97 1.03) P value.16.07.90.13.30.001.99 BED status (yes) 1.93 (.97 3.85) 3.45 (1.94 6.14) 1.76 (.67 4.63) 1.06 (.49 2.31) 1.93 (1.04 3.61) 1.11 (.60 2.05) 4.00 (2.24 7.15) P value.06.0001.25.88.04.75.0001 Recent exercise (kcal/wk).95 (.88 1.02).89 (.83.95).88 (.81.97).90 (.84.97).98 (.91 1.06).91 (.85.97).93 (.87 1.00) P value.17.001.01.004.59.002.04 Fat intake (servings/mo) 1.03 (.99 1.06) 1.03 (1.00 1.05).98 (.94 1.03) 1.04 (1.01 1.07) 1.00 (.97 1.03) 1.03 (1.00 1.05) 1.00 (.97 1.03) P value.13.09.45.02.98.03.94 Fruit/vegetable/fiber (servings/mo) 1.00 (.95 1.04) 1.00 (.96 1.03).96 (.91 1.02).97 (.93 1.01).98 (.95 1.02).99 (.96 1.02).99 (.96 1.03) P value.80.83.17.11.41.51.71 NOTE. N 893 973 because of missing values. Values in parentheses represent 95% confidence intervals. Values in boldface were statistically significant in bivariate models. a Age, sex, smoking status entered as covariates. b Sex entered as a covariate. c Age and sex entered as covariates.

October 2005 GI SYMPTOMS, WEIGHT, DIET, AND EXERCISE 995 Table 4. Multivariate Odds Ratios for Associations of BMI, Binge Eating Status, Physical Activity, and Dietary Intake With IBS and GI Symptom Status IBS status a Abdominal pain a Decreased stools b Increased stools c Constipation c Diarrhea c Bloating c Age (continuous).98 (.96 1.00).99 (.98 1.01).98 (.96 1.01).99 (.97 1.01).99 (.97 1.00).97 (.95.98) P value.07.52.24.33.10.001 Sex (female) 2.55 (1.42 4.60) 2.04 (1.29 3.24) 2.00 (.91 4.43) 1.41 (.87 2.28) 1.67 (1.06 2.66) 1.14 (.79 1.66) 2.16 (1.37 3.41) P value.002.002.09.16.03.48.001 Smoking status (yes) 1.97 (1.04 3.73) 1.74 (.96 3.15) P value.04.07 BMI (continuous) 1.02 (.98 1.05) 1.01 (.98 1.05).98 (.93 1.03) 1.01 (.97 1.05).97 (.94 1.01) 1.03 (1.00 1.06).98 (.95 1.01) P value.42.38.37.60.12.08.16 BED status (yes) 1.68 (.78 3.64) 2.85 (1.50 5.44) 2.02 (.72 5.65) 1.01 (.43 2.38) 1.60 (.76 3.36).97 (.49 1.92) 4.00 (2.10 7.62) P value.19.001.18.99.22.93.001 Recent exercise (kcal/wk).96 (.87 1.06).89 (.82.97).84 (.76.93).93 (.85 1.03).97 (.88 1.08).91 (.84.99).93 (.85 1.01) P value.44.01.001.15.58.03.10 Fat intake (servings/mo) 1.03 (.99 1.07) 1.02 (.99 1.06).99 (.94 1.04) 1.04 (1.01 1.08) 1.01 (.98 1.04) 1.03 (1.00 1.06) 1.01 (.98 1.04) P value.10.14.66.02.61.06.39 Fruit/vegetable/fiber (servings/mo).98 (.94 1.03) 1.00 (.96 1.04).97 (.91 1.03).98 (.94 1.02).98 (.94 1.02) 1.03 (1.00 1.06) 1.01 (.98 1.04) P value.46.99.34.37.25.75.74 NOTE. N 842 850 because of missing values. P values in boldface were statistically significant in bivariate models. P values in italics were statistically significant in bivariate and demographics-adjusted models. Values in parentheses represent 95% confidence intervals. a Age, sex, smoking status entered as covariates. b Sex entered as a covariate. c Age and sex entered as covariates. Irritable Bowel Syndrome, Gastrointestinal, and Binge Eating Disorder Symptom Prevalence Prevalence of IBS status in the sample was 13.3%. A total of 19.4% reported abdominal pain, 17% reported constipation, 25.0% reported diarrhea, and 20.0% reported bloating. A total of 5.8% met the criteria for probable BED. Associations of Irritable Bowel Syndrome and Gastrointestinal Symptoms with Diet, Physical Activity, and Binge Eating Disorder Table 2 shows the results of our first analysis step: bivariate associations between GI symptoms and weightrelated variables. BMI was related positively to IBS, abdominal pain, and diarrhea. BED was related positively to abdominal pain, constipation, and bloating. Physical activity was related inversely to all GI symptoms except constipation. Fat intake was associated positively with increased stools and diarrhea. Fruit/vegetable/fiber intake was associated inversely with bloating. Adding demographic variables and smoking as covariates had relatively little effect on these relationships (see Table 3), although the association between fruit/vegetable/fiber intake and bloating was attenuated considerably, as were the associations between BMI, IBS, and abdominal pain. In our third analysis step, multivariate analyses (see Table 4), the most robust findings were that BED was associated strongly (positively) with abdominal pain and bloating and physical activity generally was protective against GI symptoms. Discussion This study has made a number of contributions to our knowledge about the relationship between IBS, GI symptoms, obesity, and weight-loss behaviors. This was a large study of GI symptoms in obese persons or persons volunteering for weight-loss programs. Our first hypothesis, that GI symptoms significantly would be associated positively with BMI, was supported only partially. We observed positive associations between BMI and both abdominal pain and diarrhea in bivariate analyses. These relationships remained fairly stable with controls for demographic characteristics, but were attenuated severely by adding weight-related behavior variables to the analyses as covariates. This finding suggests that weightrelated behaviors may mediate the relationship between BMI and GI symptoms. The observation that physical activity was the strongest correlate of GI symptoms overall suggests physical inactivity in the obese might be important particularly to their increased GI symptom rates. It is interesting to speculate about possible clinical implications of the strong protective effect that physical activity seemed to have on GI symptomatology. Similar to the results of Chami et al 16 and Crowell et al, 17 we also found significant associations between probable BED status and some GI symptoms, specifically abdominal pain and bloating. Alterations in GI function and gut peptides have been reported and may influence ingestive behaviors and GI symptoms in these populations. Increased gastric capacity and delayed gastric emptying have been reported in patients who binge eat and

996 LEVY ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 10 appear to be associated with abnormalities in mealrelated ghrelin and insulin patterns that may be factors of GI symptoms related to transit time (ie, abdominal pain and bloating). 18 Mechanisms similar to those affecting obesity and BMI in general may be operating in this population, although any suggestions as to these mechanisms would be speculative given the currently available data in this area. There are some limitations of the study that are worth noting. First, participants in this study were overweight and obese individuals seeking treatment for weight loss, which limits the generalizability of the findings to similar groups. Self-report data is another area of study limitation because these data might be subject to memory or other biasing effects. However, in support of the validity of these data, the self-report measures we used for weight-related behaviors were predictive of BMI and sensitive to changes made during the course of weightloss programs. 9,14,19 Last, the standardized questionnaires we used looked at GI symptoms and associated behaviors over different time intervals (ie, prior 3 months for Rome II, prior 6 months for diet, and past week for physical activity). Nevertheless, despite these limitations, this research provides useful information that may help to understand better the relationships among obesity, obesity-related behaviors, and GI symptoms and the potential causal pathways between them. References 1. Flegal KM, Kuczmarski RJ, Johnson CL. Overweight and obesity in the United States: prevalence and trends, 1960-1994. Int J Obes Relat Metab Disord 1998;22:39 47. 2. Shah M, French SA, Jeffery RW, et al. Correlates of high fat/ calorie food intake in a worksite population: the Healthy Worker Project. Addict Behav 1993;18:583 594. 3. Delgado-Aros S, Locke G, Camilleri M, et al. Obesity is associated with increased risk of gastrointestinal symptoms: a populationbased study. Am J Gastroenterol 2004;99:1801 1806. 4. Talley NJ, Howell S, Poulton R. Obesity and chronic gastrointestinal symptoms in young adults: a birth cohort study. Am J Gastroenterol 2004;99:1807 1814. 5. Jeffery RW, Sherwood NE, Brelje K, et al. Mail and phone interventions for weight loss in a managed-care setting: weigh-to-be one-year outcomes. Int J Obes 2003;27:1584 1592. 6. Drossman DA, Corazziari E, Talley NJ, et al. The functional gastrointestinal disorders. 2nd ed. McLean, VA: Degnon Associates, 2000. 7. Thompson F, Byers T. Dietary assessment resource manual. J Nutr 1994;124:2296S. 8. Block G, Clifford C, Naughton MD, et al. A brief dietary screen for high fat intake. J Nutr Educ 1989;21:199 207. 9. Caan B, Coates A, Schaffer D. Variations in sensitivity, specificity, and predictive value of a dietary fat screener from Block et al. J Am Diet Assoc 1995;95:564 568. 10. Spitzer RL, Devlin MJ, Walsh BT, et al. Binge eating disorder: a multisite field trial of the diagnostic criteria. Int J Eat Disord 1992;11:191 203. 11. Spitzer RL, Yanovski S, Wadden T, et al. Binge eating disorder: its further validation in a multisite study. Int J Eat Disord 1993;13: 137 153. 12. Williams JBW, Gibbon M, First MB, et al. The structured clinical interview for DSM-III-R (SCID). II: multi-site test-retest reliability. Arch Gen Psychiatry 1992;49:630 636. 13. Paffenbarger R, Wing A, Hyde R. Physical activity as a index of heart attack risk in college alumni. Am J Epidemiol 1978;108: 161 175. 14. Harris JK, French SA, Jeffery RW, et al. Dietary and physical activity correlates of long-term weight loss. Obes Res 1994;2: 307 313. 15. SAS Institute. SASOnlineDoc (version 8). 1999. Available at: http://www.epi.umn.edu/sas8docs/. Accessed March 25, 2004. 16. Chami TN, Andersen AE, Crowell MD, et al. Gastrointestinal symptoms in bulimia nervosa: effects of treatment. Am J Gastroenterol 1995;90:88 92. 17. Crowell MD, Cheskin LJ, Musial F. Prevalence of gastrointestinal symptoms in obese and normal weight binge eaters. Am J Gastroenterol 1994;89:38791. 18. Geliebter A, Yahav EK, Gluck ME, et al. Gastric capacity, test meal intake, and appetitive hormones in binge eating disorder. Physiol Behav 2004;81:735 740. 19. Jeffery RW, Wing RR, Thorson C, et al. Strengthening behavioral interventions for weight loss: a randomized trial of food provision and monetary incentives. J Consult Clin Psychol 1993;61:1038 1045. Address requests for reprints to: Rona L. Levy, PhD, University of Washington, 4101 15th Avenue NE, Seattle, Washington 98105. e-mail: rlevy@u.washington.edu; fax: (206) 543-1228. Supported by National Institute of Health grants R01 HD36069 and R01 DK53826.