The Walkerton outbreak revisited at year 8: predictors, prevalence, and prognosis of postinfectious irritable bowel syndrome.
Hank Wang,Lin Chang
DOI: https://doi.org/10.1053/j.gastro.2010.12.008
IF: 29.4
2011-01-01
Gastroenterology
Abstract:Marshall JK, Thabane M, Garg AX, et al; Walkerton Health Study Investigators. Eight year prognosis of postinfectious irritable bowel syndrome following waterborne bacterial dysentery. Gut 2010;59:605–611.Postinfectious irritable bowel syndrome (PI-IBS), defined as the acute onset of symptoms meeting diagnostic criteria for IBS with ≥2 of the following signs or symptoms: fever, vomiting, diarrhea, or positive stool culture (Gastroenterology 2003;124:1662–1671), is a well-recognized but poorly understood complication of acute gastroenteritis because its natural history and outcomes remain largely unknown. Existing studies are limited by either a small study population or a short interval of follow-up. To this end, the Walkerton Health Study (WHS) Investigators have published an 8-year follow-up of 1166 patients since a tragic contamination of the water supply with Escherichia coli 0:157:H7, Campylobacter jejuni, and other pathogens in 2000 (Gastroenterology 2006;131:445–450). This study was initiated 2 years after the outbreak.Patients were eligible for the study cohort if they were adults (age ≥16 years), lived in Walkerton at the time of the outbreak, and had no history of IBS, celiac disease, or inflammatory bowel disease based on self-report and a review of available medical records. Only subjects who enrolled within 2 years of the formation of the WHS were included to minimize recall bias. Patients in the PI-IBS cohort were divided into 3 exposure groups: (1) those who reported no acute illness during the outbreak (controls); (2) those who self-reported gastroenteritis unsubstantiated by a review of available health records; and (3) those who reported acute illness substantiated by review of records, by a local health survey, or by a positive stool culture. IBS with diarrhea (IBS-D) was diagnosed in those reporting >3 bowel movements per day and watery stools or urgency ≥25% of the time. IBS with constipation (IBS-C) was defined in those reporting <3 bowel movements per week and hard/lumpy stools or straining to evacuate ≥25% of the time. Those who did not meet criteria for either IBS-D or IBS-C were diagnosed with unclassified IBS (IBS-U). In addition, the authors also used the Rome III subclassification based on stool form to determine bowel habit subtype (Gastroenterology 2006;130:1480–1491). Risk factors for PI-IBS, including age, gender, exposure status, illness features during the outbreak, self-reported history of anxiety or depression, antibiotic use, and direct contact with livestock, were tested at 4, 6, and 8 years after exposure using univariable and multivariable logistic regression analysis to adjust for confounding factors.By 2003, a total of 4315 individuals were enrolled in the WHS, and ultimately 2246 (52.1%) were excluded by the above criteria. Returning participants in 2008 were older and more likely to be female compared with terminates; however, no differences between exposure to gastroenteritis existed between participants and terminates. By 2008, of the 2069 initially enrolled in the PI-IBS study cohort, only 1166 eligible patients remained—742 subjects exposed to acute gastroenteritis (either self-reported or clinically suspected gastroenteritis) and 424 unexposed controls.The overall prevalence of IBS among the 742 subjects exposed to acute gastroenteritis decreased from 28.3% at 2–3 years to 21.4% at 4 years and to 14.3% at 6 years. At 8 years, the prevalence of IBS was significantly different between cases and controls (15.4% vs 5.4%; P < .0001). In a subanalysis of the 210 patients with PI-IBS in 2002/2003, Rome I criteria for IBS were still fulfilled in 50.0% at 4 years, in 34.4% at 6 years, and in 34.8% at 8 years. A numerically lower but similar proportion of subjects without acute gastroenteritis but with “sporadic” IBS defined by Rome I criteria in 2002 or 2003 continued to meet criteria for IBS over the same period of time (10/45 subjects [22.2%]). In 2008, 1.8% of cases with a history of acute gastroenteritis were diagnosed with IBS for the first time compared with 1.9% of controls (P = .869). Additional post hoc analysis was performed assuming conservatively that all patients who failed to return were healthy and did not have IBS. The prevalence of IBS decreased from 30.4% in 2002/2003 to 8.3% in 2008 among cases exposed to acute gastroenteritis and from 10.1% to 3.3% in controls over that same period.Risk factors for PI-IBS were evaluated at each time period by multivariable logistic regression. Consistent predictors of PI-IBS included female gender and weight loss for all time points, young age at all points except 2004, prolonged diarrhea at all time points except 2008, and abdominal pain during the acute illness in 2002/2003 and in 2004. Other variables predictive of PI-IBS at a single time point include preexisting psychological disorders (2008), peak stool frequency (2006), fever (2008), and blood per rectum during the acute enteric illness (2002/2003).Of the 2069 enrolled in 2002/2003 eligible for the PI-IBS study cohort, 488 total patients (cases and controls) met Rome I criteria for IBS at enrollment (23.6%). Of these, 41.8% were subclassified as IBS-D, 10.7% as IBS-C, and 47.5% as IBS-U. When bowel habit was subclassified by Rome III criteria in these patients, 42.4% had IBS-D, 20.1% had IBS-C, 15.4% had mixed IBS, and 22.3% had IBS-U. Over the subsequent 6-year follow-up period, a significant number of IBS-D patients were reclassified by their original bowel habit criteria as IBS-C, IBS-U, and non-IBS; similar phenotypic instability was seen in those initially classified as IBS-C and IBS-U.The authors point out several weaknesses of the study. First, only a minority (exact number not revealed) of patients had culture-proven infection. Second, of the original 4315 WHS enrollees, only 53.5% returned for reassessment in 2008, although no differences in exposure to gastroenteritis was reported between returnees and terminates. Third, misclassification of PI-IBS could not be completely excluded because the authors were unable to assess the presence of gastrointestinal symptoms and psychological profile before the outbreak.CommentThe current WHS is the largest study published to date following patients exposed and unexposed to acute gastroenteritis for >2 years. The size and duration of this study clearly add significant value to our understanding of PI-IBS. Furthermore, the authors have shown more definitively that the severity of the initial acute bacterial gastroenteritis (including abdominal cramps, fever, weight loss, blood in stool, and duration of diarrhea >7 days) all predict presence of PI-IBS at various time points between 2 and 8 years after infection. Previously, a long duration of diarrhea had been shown in some studies to predict PI-IBS (BMJ 1997;314:779–782; Am J Gastroenterol 2009;104:2267–2274), but not in others (Aliment Pharmacol Ther 2007;26:535–544). The follow-up WHS analysis confirmed findings from prior studies demonstrating that young age, female gender, and a history of prior anxiety or depression increase risk for PI-IBS (Clin Gastroenterol Hepatol 2007;5:465–469; Gut 2007;56:1066–1071). The authors have also shown that the prevalence of PI-IBS at 6 and 8 years are similar (14.3% vs 15.4%), whereas previously it was reported that 57% of patients in a small cohort of 14 patients had PI-IBS at 6 years (Gut 2002;51:410–413).In this study, the overall prevalence of IBS in the returning 742 cohort participants at 8 years was significantly greater in the cases exposed to acute bacterial gastroenteritis in 2000 as compared with controls representing sporadic cases (15.4% vs 5.4%; P < .0001). In the 210 participants who were diagnosed with IBS in 2002/2003 (either sporadic or PI-IBS), persistence of illness was seen in both groups, although the percentage was greater in the PI-IBS group (34.8% vs 22.2% at the 8-year follow-up). These values differ markedly from those reported by El-Serag et al in a systematic review based on 5 studies that reported that only 12%–38% of symptom-free patients experienced complete resolution of symptoms at a median follow-up duration of 2 years (Aliment Pharmacol Ther 2004;19:861–870). In a 6th study, 29% of patients were found to have IBS-related visits during a median follow-up of 12 years, a number more consistent with these findings (Ann Intern Med 1995;122:107–112.) Reasons that may account for differences in prevalence of IBS persistence between the WHS study and the studies included in El Serag et al's systematic review is that the latter included both retrospective and prospective studies, diagnostic criteria varied, and the prevalence of women ranged from 36% to 100%. An additional finding in the WHS was that there was no greater likelihood of developing de novo IBS at 8 years in cases exposed to acute bacterial gastroenteritis versus controls.Some limitations of the study are worth mentioning. First, even though the authors excluded enrollees beyond 2002/2003, recall bias of symptoms even at 2 years can cause misclassification of both cases and controls. Patients exposed to a severe cause of acute gastroenteritis, for instance, may be less apt to remember any prior milder IBS symptoms before the outbreak. Furthermore, Parry et al demonstrated that IBS is more frequently seen before diagnosis in patients with acute bacterial gastroenteritis (Am J Gastroenterol 2003;98:327–331), making exclusion of IBS especially important. The WHS also did not report in this study about other functional somatic syndromes such as fibromyalgia, chronic fatigue syndrome, and interstitial cystitis, which frequently coexist with IBS. It is conceivable that participants did not have preinfection IBS but had another functional syndrome that could have predisposed them to developing IBS. Second, in evaluating differences between participants and terminates in Table 2, the authors characterized all patients from the WHS cohort including those excluded by the study criteria. The addition of these patients to this table make it difficult to truly assess for differences between participants included in the study and terminates from the study, a potential concern if it is believed that those with more severe illness are disproportionately more likely to seek health care (thereby increasing the prevalence of PI-IBS). The terminates in 2008 had some features that are associated with greater risk for PI-IBS (fever, weight loss, more with a longer duration of diarrhea) but other features that are not (fewer women, lower stool frequency). Third, it is unclear if there was a misdiagnosis of PI-IBS or if these patients had developed other diseases at the time of follow-up because health records were not reviewed throughout the duration of the study. For instance, a small increased risk of inflammatory bowel disease has been reported after acute infectious gastroenteritis (1.2% in cases vs 0.5% in controls; Gastroenterology 2009;137:415–418). Fourth, no assessment of the severity of PI-IBS at various time points was performed. It would have been interesting to determine whether certain clinical factors predict a more severe disease course in PI-IBS. Recent studies have identified clinical features that predict IBS severity in a non–PI-IBS population (Am J Gastroenterol 2008;103:2536–2543) and that specific features of pain drive IBS illness severity more than other features including intensity, constancy, frequency, and predictability of pain (Aliment Pharmacol Ther 2010;32:1192–1202). Lastly, patients were diagnosed by Rome I criteria as opposed to Rome II criteria, which are more restrictive and have been shown to produce lower estimates of prevalence (Am J Gastroenterol 2000;95:3176–183). In a meta-analysis, the incidence of PI-IBS ranged from 3% to 14% in studies that used Rome II criteria versus 16% to 32% in studies that used Rome I criteria (Aliment Pharmacol Ther 2007;26:535–544).Recently, the WHS investigators have also identified several genes that encode proteins involved in epithelial cell barrier function and the innate immune response to enteric bacteria that are associated with development of PI-IBS (Gastroenterology 2010;138:1502–1513). This finding suggests that certain individuals may be more predisposed to developing PI-IBS when exposed to a gastroenteritis. However, alterations in gene expression need to be confirmed with microarray analysis because gene expression can be altered by environmental influences (eg, chronic stress which is a predictor of PI-IBS (Gut 1999;44:400-6). Future studies should evaluate whether Toll-like receptor-9, interleukin-6, and CDH1 genetic variants are also associated with non–PI-IBS.The WHS cohort continues to provide invaluable data regarding the long-term prognosis of PI-IBS. As these patients are followed over the course of their lives, it will be interesting to see how and if their symptoms continue to evolve. Will they be more likely to develop inflammatory bowel disease or other gastrointestinal or nongastrointestinal illnesses? Will they continue to demonstrate subtype phenotypic instability? Will their symptoms be comparable in severity with those with de novo IBS? Finally, do PI-IBS patients respond to different therapies (eg, anti-inflammatory agents) compared with non–PI-IBS patients? The WHS Investigators have followed a truly unique cohort of patients unmatched in their size, their duration of follow-up, and their availability of DNA samples. From a national tragedy affecting thousands and resulting in 6 deaths, the WHS Investigators have published a detailed, 8-year follow-up of survivors of this cohort, contributing immensely to our knowledge of the natural history of PI-IBS. Marshall JK, Thabane M, Garg AX, et al; Walkerton Health Study Investigators. Eight year prognosis of postinfectious irritable bowel syndrome following waterborne bacterial dysentery. Gut 2010;59:605–611. Postinfectious irritable bowel syndrome (PI-IBS), defined as the acute onset of symptoms meeting diagnostic criteria for IBS with ≥2 of the following signs or symptoms: fever, vomiting, diarrhea, or positive stool culture (Gastroenterology 2003;124:1662–1671), is a well-recognized but poorly understood complication of acute gastroenteritis because its natural history and outcomes remain largely unknown. Existing studies are limited by either a small study population or a short interval of follow-up. To this end, the Walkerton Health Study (WHS) Investigators have published an 8-year follow-up of 1166 patients since a tragic contamination of the water supply with Escherichia coli 0:157:H7, Campylobacter jejuni, and other pathogens in 2000 (Gastroenterology 2006;131:445–450). This study was initiated 2 years after the outbreak. Patients were eligible for the study cohort if they were adults (age ≥16 years), lived in Walkerton at the time of the outbreak, and had no history of IBS, celiac disease, or inflammatory bowel disease based on self-report and a review of available medical records. Only subjects who enrolled within 2 years of the formation of the WHS were included to minimize recall bias. Patients in the PI-IBS cohort were divided into 3 exposure groups: (1) those who reported no acute illness during the outbreak (controls); (2) those who self-reported gastroenteritis unsubstantiated by a review of available health records; and (3) those who reported acute illness substantiated by review of records, by a local health survey, or by a positive stool culture. IBS with diarrhea (IBS-D) was diagnosed in those reporting >3 bowel movements per day and watery stools or urgency ≥25% of the time. IBS with constipation (IBS-C) was defined in those reporting <3 bowel movements per week and hard/lumpy stools or straining to evacuate ≥25% of the time. Those who did not meet criteria for either IBS-D or IBS-C were diagnosed with unclassified IBS (IBS-U). In addition, the authors also used the Rome III subclassification based on stool form to determine bowel habit subtype (Gastroenterology 2006;130:1480–1491). Risk factors for PI-IBS, including age, gender, exposure status, illness features during the outbreak, self-reported history of anxiety or depression, antibiotic use, and direct contact with livestock, were tested at 4, 6, and 8 years after exposure using univariable and multivariable logistic regression analysis to adjust for confounding factors. By 2003, a total of 4315 individuals were enrolled in the WHS, and ultimately 2246 (52.1%) were excluded by the above criteria. Returning participants in 2008 were older and more likely to be female compared with terminates; however, no differences between exposure to gastroenteritis existed between participants and terminates. By 2008, of the 2069 initially enrolled in the PI-IBS study cohort, only 1166 eligible patients remained—742 subjects exposed to acute gastroenteritis (either self-reported or clinically suspected gastroenteritis) and 424 unexposed controls. The overall prevalence of IBS among the 742 subjects exposed to acute gastroenteritis decreased from 28.3% at 2–3 years to 21.4% at 4 years and to 14.3% at 6 years. At 8 years, the prevalence of IBS was significantly different between cases and controls (15.4% vs 5.4%; P < .0001). In a subanalysis of the 210 patients with PI-IBS in 2002/2003, Rome I criteria for IBS were still fulfilled in 50.0% at 4 years, in 34.4% at 6 years, and in 34.8% at 8 years. A numerically lower but similar proportion of subjects without acute gastroenteritis but with “sporadic” IBS defined by Rome I criteria in 2002 or 2003 continued to meet criteria for IBS over the same period of time (10/45 subjects [22.2%]). In 2008, 1.8% of cases with a history of acute gastroenteritis were diagnosed with IBS for the first time compared with 1.9% of controls (P = .869). Additional post hoc analysis was performed assuming conservatively that all patients who failed to return were healthy and did not have IBS. The prevalence of IBS decreased from 30.4% in 2002/2003 to 8.3% in 2008 among cases exposed to acute gastroenteritis and from 10.1% to 3.3% in controls over that same period. Risk factors for PI-IBS were evaluated at each time period by multivariable logistic regression. Consistent predictors of PI-IBS included female gender and weight loss for all time points, young age at all points except 2004, prolonged diarrhea at all time points except 2008, and abdominal pain during the acute illness in 2002/2003 and in 2004. Other variables predictive of PI-IBS at a single time point include preexisting psychological disorders (2008), peak stool frequency (2006), fever (2008), and blood per rectum during the acute enteric illness (2002/2003). Of the 2069 enrolled in 2002/2003 eligible for the PI-IBS study cohort, 488 total patients (cases and controls) met Rome I criteria for IBS at enrollment (23.6%). Of these, 41.8% were subclassified as IBS-D, 10.7% as IBS-C, and 47.5% as IBS-U. When bowel habit was subclassified by Rome III criteria in these patients, 42.4% had IBS-D, 20.1% had IBS-C, 15.4% had mixed IBS, and 22.3% had IBS-U. Over the subsequent 6-year follow-up period, a significant number of IBS-D patients were reclassified by their original bowel habit criteria as IBS-C, IBS-U, and non-IBS; similar phenotypic instability was seen in those initially classified as IBS-C and IBS-U. The authors point out several weaknesses of the study. First, only a minority (exact number not revealed) of patients had culture-proven infection. Second, of the original 4315 WHS enrollees, only 53.5% returned for reassessment in 2008, although no differences in exposure to gastroenteritis was reported between returnees and terminates. Third, misclassification of PI-IBS could not be completely excluded because the authors were unable to assess the presence of gastrointestinal symptoms and psychological profile before the outbreak. CommentThe current WHS is the largest study published to date following patients exposed and unexposed to acute gastroenteritis for >2 years. The size and duration of this study clearly add significant value to our understanding of PI-IBS. Furthermore, the authors have shown more definitively that the severity of the initial acute bacterial gastroenteritis (including abdominal cramps, fever, weight loss, blood in stool, and duration of diarrhea >7 days) all predict presence of PI-IBS at various time points between 2 and 8 years after infection. Previously, a long duration of diarrhea had been shown in some studies to predict PI-IBS (BMJ 1997;314:779–782; Am J Gastroenterol 2009;104:2267–2274), but not in others (Aliment Pharmacol Ther 2007;26:535–544). The follow-up WHS analysis confirmed findings from prior studies demonstrating that young age, female gender, and a history of prior anxiety or depression increase risk for PI-IBS (Clin Gastroenterol Hepatol 2007;5:465–469; Gut 2007;56:1066–1071). The authors have also shown that the prevalence of PI-IBS at 6 and 8 years are similar (14.3% vs 15.4%), whereas previously it was reported that 57% of patients in a small cohort of 14 patients had PI-IBS at 6 years (Gut 2002;51:410–413).In this study, the overall prevalence of IBS in the returning 742 cohort participants at 8 years was significantly greater in the cases exposed to acute bacterial gastroenteritis in 2000 as compared with controls representing sporadic cases (15.4% vs 5.4%; P < .0001). In the 210 participants who were diagnosed with IBS in 2002/2003 (either sporadic or PI-IBS), persistence of illness was seen in both groups, although the percentage was greater in the PI-IBS group (34.8% vs 22.2% at the 8-year follow-up). These values differ markedly from those reported by El-Serag et al in a systematic review based on 5 studies that reported that only 12%–38% of symptom-free patients experienced complete resolution of symptoms at a median follow-up duration of 2 years (Aliment Pharmacol Ther 2004;19:861–870). In a 6th study, 29% of patients were found to have IBS-related visits during a median follow-up of 12 years, a number more consistent with these findings (Ann Intern Med 1995;122:107–112.) Reasons that may account for differences in prevalence of IBS persistence between the WHS study and the studies included in El Serag et al's systematic review is that the latter included both retrospective and prospective studies, diagnostic criteria varied, and the prevalence of women ranged from 36% to 100%. An additional finding in the WHS was that there was no greater likelihood of developing de novo IBS at 8 years in cases exposed to acute bacterial gastroenteritis versus controls.Some limitations of the study are worth mentioning. First, even though the authors excluded enrollees beyond 2002/2003, recall bias of symptoms even at 2 years can cause misclassification of both cases and controls. Patients exposed to a severe cause of acute gastroenteritis, for instance, may be less apt to remember any prior milder IBS symptoms before the outbreak. Furthermore, Parry et al demonstrated that IBS is more frequently seen before diagnosis in patients with acute bacterial gastroenteritis (Am J Gastroenterol 2003;98:327–331), making exclusion of IBS especially important. The WHS also did not report in this study about other functional somatic syndromes such as fibromyalgia, chronic fatigue syndrome, and interstitial cystitis, which frequently coexist with IBS. It is conceivable that participants did not have preinfection IBS but had another functional syndrome that could have predisposed them to developing IBS. Second, in evaluating differences between participants and terminates in Table 2, the authors characterized all patients from the WHS cohort including those excluded by the study criteria. The addition of these patients to this table make it difficult to truly assess for differences between participants included in the study and terminates from the study, a potential concern if it is believed that those with more severe illness are disproportionately more likely to seek health care (thereby increasing the prevalence of PI-IBS). The terminates in 2008 had some features that are associated with greater risk for PI-IBS (fever, weight loss, more with a longer duration of diarrhea) but other features that are not (fewer women, lower stool frequency). Third, it is unclear if there was a misdiagnosis of PI-IBS or if these patients had developed other diseases at the time of follow-up because health records were not reviewed throughout the duration of the study. For instance, a small increased risk of inflammatory bowel disease has been reported after acute infectious gastroenteritis (1.2% in cases vs 0.5% in controls; Gastroenterology 2009;137:415–418). Fourth, no assessment of the severity of PI-IBS at various time points was performed. It would have been interesting to determine whether certain clinical factors predict a more severe disease course in PI-IBS. Recent studies have identified clinical features that predict IBS severity in a non–PI-IBS population (Am J Gastroenterol 2008;103:2536–2543) and that specific features of pain drive IBS illness severity more than other features including intensity, constancy, frequency, and predictability of pain (Aliment Pharmacol Ther 2010;32:1192–1202). Lastly, patients were diagnosed by Rome I criteria as opposed to Rome II criteria, which are more restrictive and have been shown to produce lower estimates of prevalence (Am J Gastroenterol 2000;95:3176–183). In a meta-analysis, the incidence of PI-IBS ranged from 3% to 14% in studies that used Rome II criteria versus 16% to 32% in studies that used Rome I criteria (Aliment Pharmacol Ther 2007;26:535–544).Recently, the WHS investigators have also identified several genes that encode proteins involved in epithelial cell barrier function and the innate immune response to enteric bacteria that are associated with development of PI-IBS (Gastroenterology 2010;138:1502–1513). This finding suggests that certain individuals may be more predisposed to developing PI-IBS when exposed to a gastroenteritis. However, alterations in gene expression need to be confirmed with microarray analysis because gene expression can be altered by environmental influences (eg, chronic stress which is a predictor of PI-IBS (Gut 1999;44:400-6). Future studies should evaluate whether Toll-like receptor-9, interleukin-6, and CDH1 genetic variants are also associated with non–PI-IBS.The WHS cohort continues to provide invaluable data regarding the long-term prognosis of PI-IBS. As these patients are followed over the course of their lives, it will be interesting to see how and if their symptoms continue to evolve. Will they be more likely to develop inflammatory bowel disease or other gastrointestinal or nongastrointestinal illnesses? Will they continue to demonstrate subtype phenotypic instability? Will their symptoms be comparable in severity with those with de novo IBS? Finally, do PI-IBS patients respond to different therapies (eg, anti-inflammatory agents) compared with non–PI-IBS patients? The WHS Investigators have followed a truly unique cohort of patients unmatched in their size, their duration of follow-up, and their availability of DNA samples. From a national tragedy affecting thousands and resulting in 6 deaths, the WHS Investigators have published a detailed, 8-year follow-up of survivors of this cohort, contributing immensely to our knowledge of the natural history of PI-IBS. The current WHS is the largest study published to date following patients exposed and unexposed to acute gastroenteritis for >2 years. The size and duration of this study clearly add significant value to our understanding of PI-IBS. Furthermore, the authors have shown more definitively that the severity of the initial acute bacterial gastroenteritis (including abdominal cramps, fever, weight loss, blood in stool, and duration of diarrhea >7 days) all predict presence of PI-IBS at various time points between 2 and 8 years after infection. Previously, a long duration of diarrhea had been shown in some studies to predict PI-IBS (BMJ 1997;314:779–782; Am J Gastroenterol 2009;104:2267–2274), but not in others (Aliment Pharmacol Ther 2007;26:535–544). The follow-up WHS analysis confirmed findings from prior studies demonstrating that young age, female gender, and a history of prior anxiety or depression increase risk for PI-IBS (Clin Gastroenterol Hepatol 2007;5:465–469; Gut 2007;56:1066–1071). The authors have also shown that the prevalence of PI-IBS at 6 and 8 years are similar (14.3% vs 15.4%), whereas previously it was reported that 57% of patients in a small cohort of 14 patients had PI-IBS at 6 years (Gut 2002;51:410–413). In this study, the overall prevalence of IBS in the returning 742 cohort participants at 8 years was significantly greater in the cases exposed to acute bacterial gastroenteritis in 2000 as compared with controls representing sporadic cases (15.4% vs 5.4%; P < .0001). In the 210 participants who were diagnosed with IBS in 2002/2003 (either sporadic or PI-IBS), persistence of illness was seen in both groups, although the percentage was greater in the PI-IBS group (34.8% vs 22.2% at the 8-year follow-up). These values differ markedly from those reported by El-Serag et al in a systematic review based on 5 studies that reported that only 12%–38% of symptom-free patients experienced complete resolution of symptoms at a median follow-up duration of 2 years (Aliment Pharmacol Ther 2004;19:861–870). In a 6th study, 29% of patients were found to have IBS-related visits during a median follow-up of 12 years, a number more consistent with these findings (Ann Intern Med 1995;122:107–112.) Reasons that may account for differences in prevalence of IBS persistence between the WHS study and the studies included in El Serag et al's systematic review is that the latter included both retrospective and prospective studies, diagnostic criteria varied, and the prevalence of women ranged from 36% to 100%. An additional finding in the WHS was that there was no greater likelihood of developing de novo IBS at 8 years in cases exposed to acute bacterial gastroenteritis versus controls. Some limitations of the study are worth mentioning. First, even though the authors excluded enrollees beyond 2002/2003, recall bias of symptoms even at 2 years can cause misclassification of both cases and controls. Patients exposed to a severe cause of acute gastroenteritis, for instance, may be less apt to remember any prior milder IBS symptoms before the outbreak. Furthermore, Parry et al demonstrated that IBS is more frequently seen before diagnosis in patients with acute bacterial gastroenteritis (Am J Gastroenterol 2003;98:327–331), making exclusion of IBS especially important. The WHS also did not report in this study about other functional somatic syndromes such as fibromyalgia, chronic fatigue syndrome, and interstitial cystitis, which frequently coexist with IBS. It is conceivable that participants did not have preinfection IBS but had another functional syndrome that could have predisposed them to developing IBS. Second, in evaluating differences between participants and terminates in Table 2, the authors characterized all patients from the WHS cohort including those excluded by the study criteria. The addition of these patients to this table make it difficult to truly assess for differences between participants included in the study and terminates from the study, a potential concern if it is believed that those with more severe illness are disproportionately more likely to seek health care (thereby increasing the prevalence of PI-IBS). The terminates in 2008 had some features that are associated with greater risk for PI-IBS (fever, weight loss, more with a longer duration of diarrhea) but other features that are not (fewer women, lower stool frequency). Third, it is unclear if there was a misdiagnosis of PI-IBS or if these patients had developed other diseases at the time of follow-up because health records were not reviewed throughout the duration of the study. For instance, a small increased risk of inflammatory bowel disease has been reported after acute infectious gastroenteritis (1.2% in cases vs 0.5% in controls; Gastroenterology 2009;137:415–418). Fourth, no assessment of the severity of PI-IBS at various time points was performed. It would have been interesting to determine whether certain clinical factors predict a more severe disease course in PI-IBS. Recent studies have identified clinical features that predict IBS severity in a non–PI-IBS population (Am J Gastroenterol 2008;103:2536–2543) and that specific features of pain drive IBS illness severity more than other features including intensity, constancy, frequency, and predictability of pain (Aliment Pharmacol Ther 2010;32:1192–1202). Lastly, patients were diagnosed by Rome I criteria as opposed to Rome II criteria, which are more restrictive and have been shown to produce lower estimates of prevalence (Am J Gastroenterol 2000;95:3176–183). In a meta-analysis, the incidence of PI-IBS ranged from 3% to 14% in studies that used Rome II criteria versus 16% to 32% in studies that used Rome I criteria (Aliment Pharmacol Ther 2007;26:535–544). Recently, the WHS investigators have also identified several genes that encode proteins involved in epithelial cell barrier function and the innate immune response to enteric bacteria that are associated with development of PI-IBS (Gastroenterology 2010;138:1502–1513). This finding suggests that certain individuals may be more predisposed to developing PI-IBS when exposed to a gastroenteritis. However, alterations in gene expression need to be confirmed with microarray analysis because gene expression can be altered by environmental influences (eg, chronic stress which is a predictor of PI-IBS (Gut 1999;44:400-6). Future studies should evaluate whether Toll-like receptor-9, interleukin-6, and CDH1 genetic variants are also associated with non–PI-IBS. The WHS cohort continues to provide invaluable data regarding the long-term prognosis of PI-IBS. As these patients are followed over the course of their lives, it will be interesting to see how and if their symptoms continue to evolve. Will they be more likely to develop inflammatory bowel disease or other gastrointestinal or nongastrointestinal illnesses? Will they continue to demonstrate subtype phenotypic instability? Will their symptoms be comparable in severity with those with de novo IBS? Finally, do PI-IBS patients respond to different therapies (eg, anti-inflammatory agents) compared with non–PI-IBS patients? The WHS Investigators have followed a truly unique cohort of patients unmatched in their size, their duration of follow-up, and their availability of DNA samples. From a national tragedy affecting thousands and resulting in 6 deaths, the WHS Investigators have published a detailed, 8-year follow-up of survivors of this cohort, contributing immensely to our knowledge of the natural history of PI-IBS. ReplyGastroenterologyVol. 140Issue 2PreviewWe are grateful to Drs Wang and Chang for their thoughtful review of our study of the 8-year prognosis of postinfectious irritable bowel syndrome (IBS) in the Walkerton Health Study cohort. The outbreak of acute gastroenteritis in this community in 2000 was a public health tragedy, but has provided valuable insight into the natural history and long-term health consequences of acute gastroenteritis. Full-Text PDF