And Consensus-Based Practice Guidelines for the Diagnosis of Irritable Bowel Syndrome

REVIEW ARTICLE Evidence- and Consensus-Based Practice Guidelines for the Diagnosis of Irritable Bowel Syndrome

Ronnie Fass, MD; George F. Longstreth, MD; Mark Pimentel, MD; Steven Fullerton, MPH; Simcha M. Russak, MA; Chiun-Fang Chiou, PhD; Eileen Reyes, BS; Paul Crane, MD; Glenn Eisen, MD; Bill McCarberg, MD; Joshua Ofman, MD, MSHS

Background: Irritable bowel syndrome (IBS) presents nostic tests, and the internal validity of current diagnostic a significant diagnostic and management challenge for symptom criteria. Few studies met accepted methodologi- primary care practitioners. Improving the accuracy and cal criteria. While symptom criteria have been validated, timeliness of diagnosis may result in improved quality the utility of endoscopic and other diagnostic interven- and efficiency of care. tions remains unknown. An analysis of the literature, combined with consensus from experienced clinicians, re- Objective: To systematically appraise the existing diag- sulted in the development of a diagnostic algorithm relevant nostic criteria and combine the evidence with expert opin- to primary care that emphasizes a symptom-based diag- ion to derive evidence- and consensus-based guidelines for nostic approach, refers patients with alarm symptoms to a diagnostic approach to patients with suspected IBS. subspecialists, and reserves radiographic, endoscopic, and other tests for referral cases. The resulting algorithm high- Methods: We performed a systematic literature review lights the reliance on symptom criteria and comprises a pri- (January 1966–April 2000) of computerized biblio- mary module, 3 submodules based on the predominant graphic databases. Articles meeting explicit inclusion cri- symptom pattern (constipation, diarrhea, and pain) and se- teria for diagnostic studies in IBS were subjected to criti- verity level, and a subspecialist referral module. cal appraisal, which formed the basis of guideline statements presented to an expert panel. To develop a Conclusions: The dearth of available evidence high- diagnostic algorithm, an expert panel of specialists and lights the need for more rigorous scientific validation to primary care physicians was used to fill in gaps in the identify the most accurate methods of diagnosing IBS. Un- literature. Consensus was developed using a modified Del- til such time, the diagnostic algorithm presented herein phi technique. could inform decision making for a range of providers caring for primary care patients with abdominal discom- Results: The systematic literature review identified only fort or pain and altered bowel function suggestive of IBS. 13 published studies regarding the effectiveness of com- peting diagnostic approaches for IBS, the accuracy of diag- Arch Intern Med. 2001;161:2081-2088

RRITABLE BOWEL syndrome (IBS) clude the Rome I criteria,6 the Manning is a common disorder character- criteria,7 and the recently developed Rome ized by abdominal pain, bloat- II consensus criteria.8 These criteria have ing, and disturbed defecation. Ir- been used in research protocols to facili- ritable bowel syndrome remains tate study inclusion. However, they have un- the most common disorder encountered dergone limited validation, particularly in I 1 by gastroenterologists. The incidence of primary care settings. IBS is reported to be 15% to 20% in the In addition to symptom criteria, sev- general population,2,3 with prevalence rates eral diagnostic algorithms, such as that dependent upon the symptom criteria4 proposed by Schmulson and Chang,9 have used to define the condition. Further- been developed to facilitate the diagnosis more, functional bowel complaints such and management of IBS. However, most as IBS are responsible for nearly 50% of guidelines were developed for use in the visits to gastroenterologists.5 specialty care setting and targeted for pa- Presently there are no known bio- tients with more severe symptoms. While chemical or structural markers for identi- many algorithms use structured and vali- fying patients with IBS. In most cases, a di- dated expert panel methods, most panels agnosis of IBS is based on typical symptoms consisted of academic center subspecial- and negative results of a limited diagnostic ists, who may not reflect the understand- The affiliations of the authors evaluation. Consequently, symptom crite- ing and concerns of provider organiza- appear in the acknowledgment ria for diagnosis have been proposed. Cur- tions focused on the provision of care in section at the end of the article. rently, the most widely accepted criteria in- the primary care setting.

(REPRINTED) ARCH INTERN MED/ VOL 161, SEP 24, 2001 WWW.ARCHINTERNMED.COM 2081

Downloaded From: https://jamanetwork.com/ on 09/26/2021 Table 1. Comparison of Diagnostic Criteria*

Manning Criteria† Kruis Questionnaire‡ Rome I Criteria§ Rome II Criteriaሻ Abdominal distension Patient fills out: Ͼ3 mo of continuous or recurrent Ն12 wk of continuous or recurrent Pain relief with bowel 1. Did you come because of abdominal symptoms: symptoms: movement pain? Abdominal pain: Abdominal pain or discomfort More frequent stools Do you suffer from flatulence? Relieved with defecation Plus with pain Do you suffer from irregularities of and/or Ն2 of the following: Looser stools with pain bowel movements? Associated with change in Relieved with defecation Mucus per rectum¶ 2. Have you suffered from your stool consistency and/or Feeling of incomplete complaints for more than 2 years? Plus Associated with change in evacuation¶ 3. How can your abdominal pain be Ն2 of the following on at least frequency of stool described: burning, cutting, very 25% of days: and/or strong, terrible, feeling of pressure, Altered stool frequency Associated with change in dull, boring, not so bad? (Ͼ3/d or Ͻ3/wk) form (appearance) of 4. Have you ever noticed alternating Altered stool form (lumpy/ stool constipation and diarrhea? hard or loose/watery) Physician fills out: Altered stool passage 1. Abnormal physical findings and/or (straining, urgency, or history pathognomonic for any feeling of incomplete diagnosis other than IBS evacuation) 2. ESR Ͼ20 mm/2 h Passage of mucus 3. Leukocytosis (Ͼ10 000/cm)3 Bloating or feeling of 4. Hemoglobin: female, Ͻ12 g/dL; abdominal distension male, Ͻ14 g/dL 5. History of blood in stool

*IBS indicates irritable bowel syndrome; ESR, erythrocyte sedimentation rate. †From Manning et al.7 ‡From Kruis et al.10 Questions that are not associated with a score are excluded. §From Drossman et al.6 ࿣From Thompson et al.11 ¶Differences between IBS patients and controls were not statistically significant (P Ͼ.05).

The objective of the present racy of diagnostic tests, and conven- ment) and met one of the following study was to arrive at evidence- and ing an expert panel to synthesize this criteria: compared 2 diagnostic mo- consensus-based guidelines for a di- information and develop consensus- dalities, distinguished IBS from an- agnostic approach to patients with based recommendations about the other condition, appraised indi- suspected IBS with specific rel- diagnosis of IBS. vidual symptoms for their diagnostic evance to primary care providers. Our association, or provided sensitivity/ evidence-based approach to guide- SYSTEMATIC REVIEW specificity data on a diagnostic mo- line development relied upon a sys- OF THE LITERATURE dality. tematic review of the published medical literature and consensus from an We searched 4 computerized bib- EVALUATION expert panel of experienced clini- liographic databases (MEDLINE, OF PUBLISHED STUDIES cians in a variety of health care set- HEALTHSTAR, Evidence-Based tings. A modified Delphi technique Medicine, and the Cochrane Data- The symptom criteria were assessed was used for instances in which the base) to identify English-language for their test characteristics and per- published literature could not in- articles published between January formance (Table 1) based on ab- form the decision-making process. 1966 and April 2000. The focus of straction of the following informa- The purpose of this effort is to pro- the search was on articles that evalu- tion from each study: study design1 vide primary care and nonacademic ated the performance of diagnostic (ie, gold standard, diagnostic perfor- clinicians with guidance to improve tests and procedures for IBS. Search mance, and disease prevalence) and the evaluation and diagnosis of pa- terms and strategies were devel- diagnostic accuracy2 (ie, sensitivity, tients with IBS. oped in cooperation with an expert specificity, and diagnostic odds ratio). librarian experienced in advanced The quality of each study was METHODS search strategies of health-related assessed by summing the weights of computerized databases. In addi- the study characteristics met. These A 3-phase approach was used to con- tion, the search included the bibli- weights were obtained from a mul- struct the evidence-based guide- ographies of key reviews and of all tivariate regression analysis re- lines for diagnosing IBS. The phases articles that met the search criteria. ported in a recent publication by Lij- include a systematic literature re- Articles were accepted for re- mer and colleagues12 that evaluated view of diagnostic studies in IBS, a view if they used an objective gold design-related bias in assessments of comprehensive appraisal of prior standard (ie, Manning, Kruis, or diagnostic tests. The potential range studies and estimates of the accu- Rome I criteria or clinical assess- of each study’s total score was from

(REPRINTED) ARCH INTERN MED/ VOL 161, SEP 24, 2001 WWW.ARCHINTERNMED.COM 2082

Downloaded From: https://jamanetwork.com/ on 09/26/2021 0 to 8 unweighted and from 0 to 13.2 DIAGNOSTIC ALGORITHM weighted. The study characteristics Table 2. Criteria for Grading included spectrum (clinical popula- Guidelines in the form of a diagnos- the Quality of Diagnostic Studies tion vs case-control), verification tic algorithm were developed incor- (complete vs different reference porating the best available evi- Score tests vs partial), interpretation of test dence and expert opinion regarding Study Characteristic Raw Weighted* results (blinded vs not blinded), the diagnosis of IBS. Expert opin- Spectrum patient selection (consecutive vs ion was used when evidence in the Clinical population 1 3.0 nonconsecutive), data collection (pro- literature did not exist to inform the Case-control 0 0 spective vs retrospective vs un- decision. Guideline statements were Verification known), details test (sufficient vs in- incorporated into the algorithm Complete 1 3.2 Different reference 0 1.0 sufficient), details reference test when they met 1 of 2 criteria: if there tests (sufficient vs insufficient), and de- was strong literature-based evi- Partial 0 0 tails population (sufficient vs dence or if the expert panel voted Interpretation of test insufficient). Studies were classified that the guideline statement was ap- results as low quality if they scored in the propriate (ie, a median score Ͼ5). Blinded 1 1.3 lowest tertile. Conversely, studies The algorithm details whether there Not blinded 0 0 Patient selection were classified as being of medium- was agreement or disagreement Consecutive 1 0.9 high quality if their scores were in the among the expert panel. The meth- Nonconsecutive 0 0 middle or high tertile (Table 2). odology of the algorithm flow- Data collection charts is consistent with the format Prospective 1 1.0 EXPERT PANEL previously adopted by the Agency for Unknown 0 0.5 Health Care Policy Research.14 Retrospective 0 0 Details test Because many areas were not ad- Sufficient 1 1.7 dressed in the published literature, in Insufficient 0 0 order to complete a diagnostic algo- RESULTS Details reference test rithm, an expert panel was as- Sufficient 1 0.7 sembled consisting of physicians from SYSTEMATIC REVIEW Insufficient 0 0 3 medical settings: an academic medi- Details population Sufficient 1 1.4 cal center, a Veterans Affairs medi- The initial search strategy identified Insufficient 0 0 cal center, and a large group-model 291 titles. One hundred twenty- health maintenance organization. eight abstracts remained after ex- *The values used to develop the weights for Guideline statements were devel- plicit title rejection criteria were ap- the weighted score are based on results oped from the systematic review, ex- plied. Further evaluation of these 128 reported by Lijmer et al.12 Data that were not reported in the article by Lijmer et al are scored isting guidelines identified in the abstracts resulted in 88 articles for fi- as 0. supplemental literature review, and nal review. Of the 88 articles identi- expert opinion. The panel was asked fied, 28 (32%) met the criteria for in- to vote on the relative appropriate- clusion in the systematic review. more symptoms were present with all ness of each guideline statement by 6 symptoms included, the positive considering both the expected costs PERFORMANCE AND predictive value was 63%. Groups and health benefits. Response op- QUALITY EVALUATION have continued to determine sensi- tions were based on a scale from 1 to OF DIAGNOSTIC CRITERIA tivity and specificity data for 2 of 4 9, ranging from extremely inappro- and 2 or 3 of 6 symptoms (Table 3). priate1 to extremely appropriate.9 A Manning Criteria When 2 of 4 symptoms were score of 5 was considered equivocal. used, the Manning criteria yielded The RAND appropriateness method- Of the 3 sets of diagnostic criteria a sensitivity and specificity of 91% ology for scoring responses13 was used identified in this review, the Man- and 70%, respectively.7,16,17 In addi- as the basis for determining consen- ning criteria appeared to be the most tion, when 2 or more of 6 criteria sus. Results were evaluated with re- extensively studied. Manning and were used, sensitivity ranged from spect to tertile1-9 after discarding the colleagues7 originally identified only 84% to 94% and specificity ranged lowest and highest scores. Agree- 4 symptoms that were significantly from 55% to 76%.7,16-18,21 In articles ment was reached if the remaining 4 more prevalent in IBS patients than in which 3 or more symptoms were scores fell within any 3-point range. in organic controls. Two further assessed (irrespective of whether it Disagreement occurred if at least 1 of symptoms approached statistical sig- was out of 4 or 6), sensitivity ranged the remaining 4 ratings fell within the nificance (mucus per rectum and sen- from 63% to 90% and specificity lowest tertile and at least 1 score fell sation of incomplete evacuation). Us- ranged from 70% to 93%.16-20 within the highest tertile. Unclear ing the 4 significant symptoms, The diagnostic ability of the opinion was defined as all of the votes subjects having less than 2 symp- Manning criteria also depended on falling within adjoining tertiles. Ex- toms had a positive predictive value the control group used. All but 1 perts were allowed to modify their for IBS of 12%. If 2 or more symp- study compared the ability to dis- votes after independently reviewing toms were present, the positive pre- tinguish IBS from organic gastroin- the results of the group’s ratings. dictive value was 74%. Finally, if 2 or testinal (GI) disease.15 However, in-

(REPRINTED) ARCH INTERN MED/ VOL 161, SEP 24, 2001 WWW.ARCHINTERNMED.COM 2083

Downloaded From: https://jamanetwork.com/ on 09/26/2021 Table 3. Summary of Studies Validating Standard Diagnostic Criteria*

IBS Diagnostic Score† Gold Type of Controls, Patients, Sensitivity, Specificity, Odds Criteria Source, y Scoring Method Standard Control No. No. % % Ratio Raw Weighted Manning Manning et al,7 Ն2 of 4 criteria Clinical Organic GI 33 32 91 70 22.2 6 10.9 1978 Ն2 of 6 criteria Clinical Organic GI 33 32 94 55 12.5 Talley et al,15 1989 No. of warnings Clinical Healthy 154 82 65 86 11.4 Organic GI 101 82 58 74 3.9 67 NUD-OGD 134 82 42 85 4.1 Talley et al,16 1991; Ն2 of 4 criteria 91 70 NC Talley,17 1992 Ն2 of 6 criteria 94 55 NC Clinical Organic GI 33 32 6 9.1 Ն3 of 4 criteria 63 85 NC Ն3 of 6 criteria 84 76 NC Jeong et al,18 1993 Ն2 of 6 criteria Non-IBS 84 54 6.5 Clinical 114 58 7 12.3 Ն3 of 6 criteria referrals 67 70 4.8 Rao et al,19 1993 Ն3 of 6 criteria Clinical Healthy 45 65 66 93 27.4 Organic GI 23 65 66 61 3.0 8 13.2 NUD 35 65 66 91 20.9 Dogan and Unal,20 Ն3 of 6 criteria Clinical Organic GI 182 165 90 86 57.4 7 12.3 1996 Thompson,21 1997 Ն2 of 6 criteria Clinical None NC 156 NC NC NC 1 3 Kruis Kruis et al,10 1984 Kruis Ն44 Clinical Organic GI 209 108 64 99 183.1 8 13.2 Frigerio et al,22 Kruis (male) Ն44 93 15 47 94 12.7 1992 Kruis (female) Ն44 108 37 59 95 30.2 Kruis (male 201 52 56 95 21.8 + female) Ն44 8 13.2 Modified Kruis Clinical Organic GI 93 15 67 90 18.7 (male) Ն44 Modified Kruis 108 37 68 92 23.8 (female) Ն44 Modified Kruis (male 201 52 68 91 21.5 + female) Ն44 Dogan and Unal,20 Kruis Ն44 Clinical Organic GI 182 165 81 91 44.8 7 12.3 1996 Rome Thompson,21 1997 Rome guidelines Clinical None NC 156 NC NC NC 1 3 Vanner et al,23 Rome guidelines Clinical Non-IBS 52 46 65 100 18.8 6 9 1999 referrals

*IBS indicates irritable bowel syndrome; GI, gastrointestinal; NUD-OGD, nonulcer dyspepsia–oesophagogastroduodenoscopy; and NC, not calculable. †Sums of scores from Table 1.

terpretation of the validation studies greater, to identify IBS (or Ն3 symp- tween 69% and 100% in these 2 pa- is problematic since many of the toms from the list in Table 1), the tient groups. However, the study had control group patients experienced sensitivity was reported as 64% and a relatively small sample size and or had upper GI symptoms rather the specificity as 99%.10 combined the absence of red flag fea- than the organic lower GI disor- tures with symptom criteria. ders from which IBS more gener- Rome I Criteria ally needs to be distinguished. The Comparisons of Criteria Manning criteria fared better when Rome I criteria were developed used to distinguish patients with through expert consensus as the first Although the Rome I criteria have not IBS from healthy controls (sensi- of an ongoing series of criteria for the been well tested in a controlled fash- tivity, 65%-66%; specificity, 86%- standardization of diagnostic crite- ion, studies have tried to compare re- 93%)15,19 than when used to distin- ria for IBS.5,8 The 3 elements of the sults between the various criteria. Two guish IBS from organic GI disease Manning criteria that were eluci- articles compared the agreement (sensitivity, 58%-94%; specificity, dated in factor analysis constitute the among various diagnostic criteria for 55%-93%).7,15-17,19,20 first part of the Rome I criteria. De- IBS. There was good agreement be- spite this, the published validation tween the Manning and Rome I di- Kruis Score of these criteria is minimal. Table 3 agnosis of IBS in 1 study (␬=0.72).24 summarizes data from 2 studies, only Additionally, in a large population- Kruis and colleagues10 used a point one of which provided an evalua- based study, 98% of subjects who system whereby functional symp- tion of the sensitivity and specific- tested positive for the Rome I crite- toms received positive values and ity of the criteria.23 The Rome I cri- ria also met the Manning criteria.25 “red flag” symptoms received nega- teria demonstrated a sensitivity of However, of the subjects who tested tive values (Table 3). Based on this 65% and specificity of 100%. The positive for 2 or more of the Man- point system, using a score of 44 or positive predictive value ranged be- ning criteria, only 37% were positive

(REPRINTED) ARCH INTERN MED/ VOL 161, SEP 24, 2001 WWW.ARCHINTERNMED.COM 2084

Downloaded From: https://jamanetwork.com/ on 09/26/2021 using the Rome I criteria. The lower LR Literature Review New Patient Primary Alarm Symptoms (LR) prevalence rate could be due to the Recommendation Care Office Visit a. Weight Loss inclusion of pain as a necessary pre- EA Expert Consensus Agreement b. GI Bleeding condition in the Rome I criteria. ED Expert Consensus Disagreement c. Anemia d. Fever e. Frequent Nocturnal Symptoms Quality of the Criteria Alarm Symptoms? Record Yes (LR, EA) in Chart Each of the studies used to validate Consider Gastroenterologist standard diagnostic criteria was No scored based on quality criteria. The Supspecialist Referral ∗ Reevaluate for raw scores ranged from 1 through Positive for Rome II Criteria ? Go to No Persistent Symptoms (LR, EA) 8, and weighted scores ranged from at a Later Date 3 to 13.2. Of the 7 validation stud- Yes ies for the Manning criteria, all but Module 2 1 were of medium to high quality. The 3 validation studies on the Kruis Age >50 y? criteria all received a medium to high (LR, ED) quality score. Only 1 of the 2 validation studies for the Rome I crite- No ria obtained a medium to high qual- Predominant Symptom? ity score. Two additional studies (LR, EA) were identified that compared diagnostic criteria.24,25 However, these studies did not compare the crite- Constipation (LR, EA) Diarrhea (LR, EA) Abdominal Pain (LR, EA) ria with a diagnostic gold standard. CBC (LR, EA) CBC (LR, EA) CBC (LR, EA) TSH (LR) TSH (LR, EA) DIAGNOSTIC ALGORITHM Flexible Sigmoidoscopy (LR, ED) ESR (LR, EA) Electrolytes (LR) Electrolytes (LR) The diagnostic algorithm was de- Flexible Sigmoidoscopy (LR, ED) veloped based on consensus of the guideline statements that were de- Go to Go to Go to rived from the systematic review, supplemental review, and expert opinion. The algorithm consisted of Module 1 Module 1 Module 1 a primary module, a primary care workup module that comprised 3 Figure 1. Irritable bowel syndrome diagnostic algorithm for a new primary care patient. Asterisk indicates predominant symptom patterns that experts agree that chronic abdominal pain plus 2 or more Manning criteria is an acceptable (constipation, diarrhea, and pain), alternative to the Rome I criteria. GI indicates gastrointestinal; CBC, complete blood cell count; TSH, thyrotropin (thyroid-stimulating hormone) level; and ESR, erythrocyte sedimentation rate. and a subspecialist referral module (Figures 1, 2, and 3). While the evidence suggests that 3 categories: mild (can be ignored if disorders, although evidence is forth- the Manning criteria have the great- the patient does not think about it), coming. Conservative empirical tri- est number of validation studies, the moderate (cannot be ignored but does als may include the use of antidiar- expert panel reached consensus and not affect patient’s lifestyle), and se- rheal agents for predominant diarrhea selected the Rome II criteria as the pri- vere/very severe (affects patient’s symptoms or antispasmodics for pre- mary diagnostic symptom criteria. lifestyle).26 Predominant symptom dominant pain symptoms. Trials The Rome II criteria incorporate the patterns were chosen by the expert could also include psychosocial coun- most valid elements of the Manning panel based on the categorization of seling, stress reduction, or biofeed- criteria while broadening inclusion the Rome Working Group.11 back based on needs assessment. Fur- with the addition of abdominal dis- After patients are categorized, a ther diagnostic consideration may be comfort or pain and potentially thorough history should be taken to undertaken depending upon re- greater discrimination between IBS identify previous interventions, thera- sponses to this trial. Upon referral to and other functional disorders. In ad- pies, and medications used. In some a subspecialist, traditional invasive dition, the subcategorization of IBS on cases, a psychosocial assessment is and noninvasive testing is recom- the validated Rome I platform facili- recommended. The expert panel mended, if necessary, to establish the tates management in a clinical algo- achieved consensus that an empiri- diagnosis and to arrive at a therapeu- rithm. Given the validity of the Man- cal trial of therapy based on the pre- tic approach targeted at the predomi- ning criteria, the panel alternatively dominant symptom complex does aid nant symptom complex. accepted chronic abdominal pain plus in the treatment of patients with sus- In the development of the al- 2 or more Manning criteria as an ac- pected IBS. Failure to respond to em- gorithm, there were areas of agree- ceptable criterion for the algorithm. pirical trials may have diagnostic im- ment and disagreement that im- Severity of illness was classified into plications as in other functional GI pacted the specific diagnostic

(REPRINTED) ARCH INTERN MED/ VOL 161, SEP 24, 2001 WWW.ARCHINTERNMED.COM 2085

Downloaded From: https://jamanetwork.com/ on 09/26/2021 jective, a panel was assembled that LR Literature Review Primary Care Follow-up Visit Recommendation reflects the needs and concerns of EA Expert Consensus Agreement primary care providers. ED Expert Consensus Disagreement Preliminary efforts toward the

Symptom Severity? development of diagnostic guide- (LR) lines entailed the systematic review of previous investigations. Of the studies identified in the review, 8 as- Mild: Moderate: Severe/Very Severe: 7,15-21 Can Be Ignored if Patient Doesn’t Can’t Be Ignored. Affects Patient’s sessed the Manning criteria, 3 Think About It Doesn’t Affect Patient’s Lifestyle Lifestyle evaluated the Kruis criteria,10,20,22 and (LR) (LR) (LR) 2 evaluated the Rome I criteria21,23 (Table 3). The latest Rome II crite- History and Physical Examination ria are based on needed improve- ments to the Rome I criteria and use All Primary Symptoms • Education/Reassurance (LR) the most valid elements of the Man- • Lifestyle (LR) ning criteria. However, it is evident • Prescription Medicine (LR) that more research is needed to bet- Constipation Only ter validate both the Rome I and • Diet (LR) • Fiber/Laxatives (LR) Rome II criteria. Still, the potential • Psychosocial (LR) advantages of the Rome II criteria in- • Depression/Anxiety (LR) clude simplicity and improved sen- Diarrhea Only sitivity as a result of the inclusion of • Diet (LR) discomfort and pain as symptoms. • Loperamide (LR) • Psychosocial (LR) Furthermore, the Rome II criteria • Depression/Anxiety (LR) have potentially greater specificity Abdominal Pain Only given that they do not include the • Antispasmodics (LR) second part of the Rome I criteria • Tricyclics (LR) (non–pain-related symptoms), which had poor clustering in factor Empirical Trial of Targeted analysis. Therapy (LR) Over the past decade, research has begun to reveal differences in Treatment Response at No Yes, Record in Chart 4, 8, 12 wk (LR, ED) physiological findings in IBS sub- Confirm Diagnosis jects with different predominant Yes Yes symptom patterns, including dysmo- tility, gut hypersensitivity, and al- Gastroenterologist Diagnostic Uncertainty? Further Therapeutic Module 2 Consider Subspecialist (EA) Management tered brain activation, among oth- Referral ers.11 To subclassify by predominant No symptom, diarrhea vs constipation, new criteria were needed to better Patient Concern? Consider (EA) identify these subgroups. Thus, the Rome II criteria may offer improved Figure 2. Irritable bowel syndrome diagnostic algorithm for constipation, diarrhea, and abdominal pain discriminative ability for diagnosing (module 1). patients with IBS.8 While abstracts as- sessing the validity of the Rome II cri- approach. For example, all of the ex- dence, supplemented by expert opin- teria have been presented,27-31 no full- perts agreed that it is “inappropri- ion, to arrive at evidence- and con- length reports on the validity of the ate” to have all patients who are sensus-based guidelines for a Rome II criteria had been published “ . . . referred to the gastroenterol- diagnostic approach to patients with at the time of this report. ogy subspecialty unit with sus- suspected IBS. Scant data were iden- The Manning and Rome crite- pected IBS be given an anorectal ma- tified in the published literature re- ria have gained much attention. nometry exam.” Conversely, experts garding the effectiveness of compet- However, the Kruis score10 has not disagreed whether it would be ap- ing diagnostic approaches, the been as widely adopted, possibly be- propriate that all patients “ ...re- accuracy of diagnostic tests, and the cause of the inclusion of red flag ferred to the gastroenterology sub- internal validity of current diagnos- symptoms as part of the scoring al- specialty unit with suspected IBS be tic symptom criteria. As a result, it gorithm. Red flag symptoms are given a large bowel exam.” was necessary to rely upon previ- quite common in subjects with IBS, ously published validation studies, and, based on the present review, COMMENT previously developed practice guide- blood in the stool may be seen in up lines, and the consensus opinion of to 31% of IBS subjects, with no ob- The objective of the present study our expert panel when developing jective cause identified on subse- was to use the best available evi- guidelines. To achieve the study’s ob- quent evaluation.15 Blood in the stool

(REPRINTED) ARCH INTERN MED/ VOL 161, SEP 24, 2001 WWW.ARCHINTERNMED.COM 2086

Downloaded From: https://jamanetwork.com/ on 09/26/2021 alone could represent hemorrhoidal EA Expert Consensus Agreement Subspecialist Referral bleeding in IBS patients, yet, in the ED Expert Consensus Disagreement Initial Visit Kruis score, this would incur a pen- EU Expert Consensus Unclear alty of 98 points—enough to fail to LR Literature Review Recommendation meet the criteria for IBS. This may ex- CE Literature Review Conflicting plain the relatively low sensitivity and Evidence Predominant Symptom? high specificity of the Kruis score. (LR, EA) Frigerio and colleagues22 adjusted the score to exclude a diagnosis of organic digestive disease in patients with 44 or more points. Still, their Constipation (LR, EA) Diarrhea (LR, EA) Abdominal Pain (LR, EA) modification was unable to signifi- History and Physical Examination History and Physical Examination History and Physical Examination cantly improve the sensitivity of the criteria. One of the Following Tests: One of the Following Tests: Tests to Consider: Based on the present evalua- Flexible Sigmoidoscopy (LR, EA) CBC (Recheck) (LR, EA) Small Bowel Follow-through Thyroid Function (LR) ESR (Recheck) (LR, EA) (LR, EA) tion, studies that assessed standard di- Colonoscopy (LR, ED) TSH, FOBT (LR, EA) Colonoscopy (LR, ED) agnostic criteria were generally of Barium Enema (LR, ED) Thyroid Function (LR) medium to high quality. However, Ova and Parasites Stool only 1 study of medium to high qual- Examination (LR, EU) Flexible Sigmoidoscopy (LR, EA) ity evaluating the Rome I criteria was or Colonoscopy (LR, ED) identified in the review. Although sur- veys and symptom criteria have been

used as an aid to identify IBS and to Negative Negative No No distinguish IBS from other func- Findings? Findings? tional disorders,26,32-35 procedures have Yes Yes most often been relied upon to rule

out organic disease. While the ex- Other Tests to Consider: Another Test to Consider: pert panel reached agreement regard- Colonic Transit Study (LR, ED) Small Bowel Follow-through ing the use of sigmoidoscopy and Defacography (LR, ED) (LR, EA) colonoscopy in the guidelines, no vali- Anorectal Manometry (LR, ED) dated evidence was found to sup- EMG (LR, ED) port the diagnostic value of these and Directed Treatment Directed Treatment or or other commonly used invasive and Consider Referring Back noninvasive procedures (eg, blood to Primary Care for Treatment, if Appropriate tests or colonic transit studies). More- over, recent evidence suggests that Figure 3. Irritable bowel syndrome diagnostic algorithm for gastrointestinal subspecialty workup (module testing for and treating small intes- 2). CBC indicates complete blood cell count; ESR, erythrocyte sedimentation rate; TSH, thyrotropin tine bacterial overgrowth in IBS may (thyroid-stimulating hormone) level; FOBT, fecal occult blood test; and EMG, electromyography. result in improved outcomes,36 but diagnostic utility in the primary care and the fact that younger patients ity of targeted empirical trials is that practice setting requires further vali- without alarm symptoms should be a treatment response may become a dation. seen initially in the primary care set- diagnostic indicator in itself. There- The diagnostic algorithm pre- ting. There is agreement that symp- fore, the predictive value of empiri- sented represents an accumulation tom severity should play a role in the cal therapy must be assessed in pro- of the best available evidence- and intensity of treatment. However, it is spective trials. consensus-based expert opinion from well known that IBS patients often There are several limitations to a variety of practice settings. We recog- present with extraintestinal symp- the present study. First, the findings nize that there is a lack of expert con- toms, especially psychological comor- of our systematic review were likely sensus in many areas—both among bidity, that may dramatically influ- confounded by publication bias; that experts and between experts and the ence the classification of severity. is, small studies with positive find- published literature. Guideline state- While we advocate empirical trials in ings are selectively published. Thus, ments were incorporated that were our algorithm as an aid to manage- it is possible that studies with nega- deemed appropriate even if there was ment, the diagnostic validity of this tive findings regarding the discrimi- disagreement among experts. This en- approach remains unclear. Further- native capability of symptom crite- hances the flexibility of the algo- more, the utility of newly available ria or poor test characteristics of rithm, allowing providers greater op- medications targeting the pathophysi- standard diagnostic tests may not have portunity to employ their own ological mechanisms of IBS remains been discovered in our review. There judgment. Accepted components of unclear. Still, these medications hold were also many gaps in the litera- the algorithm include the differen- promise for more targeted empirical ture, which meant that expert opin- tial treatment of patients based on pre- trials based on the pathophysiologi- ion was required to develop the al- dominant symptom type (constipa- cal mechanism of the predominant gorithm. Additionally, several areas of tion vs diarrhea vs abdominal pain) symptom complex. The potential util- disagreement remained even after a

(REPRINTED) ARCH INTERN MED/ VOL 161, SEP 24, 2001 WWW.ARCHINTERNMED.COM 2087

Downloaded From: https://jamanetwork.com/ on 09/26/2021 modified Delphi method was used. ment of Medicine, Vanderbilt Univer- pears in Gastroenterology. 1992;102:746]. Gas- troenterology. 1991;101:927-934. The algorithm was based primarily on sity Medical Center, Nashville, Tenn 17. Talley NJ. Sensitivity and specificity of the Man- expert consensus, yet, in some cases, (Dr Eisen). ning criteria [letter]. Gastroenterology. 1992;102: consensus-based recommendations This study was sponsored by an 1828-1829. 18. Jeong H, Lee HR, Yoo BC, Park SM. Manning cri- were not possible, as clearly eluci- educational grant from the Novartis teria in irritable bowel syndrome: its diagnostic sig- dated in the algorithm. As with all Pharmaceuticals Corp, East Ha- nificance. Korean J Intern Med. 1993;8:34-39. guidelines, providers must use their nover, NJ. 19. Rao KP, Gupta S, Jain AK, Agrawal AK, Gupta JP. Evaluation of Manning’s criteria in the diagnosis best judgment in determining which Corresponding author and re- of irritable bowel syndrome. J Assoc Physicians patients are eligible for the guide- prints: Joshua Ofman, MD, MSHS, Zynx India. 1993;41:357-363. 20. Dogan UB, Unal S. Kruis scoring system and Man- lines and in which cases the guide- Health Inc, 9100 Wilshire Blvd, East ning’s criteria in diagnosis of irritable bowel syn- lines should be strictly adhered to. Fi- Tower, Suite 655, Beverly Hills, CA drome: is it better to use combined? Acta Gas- nally, because of the scope of the 90212 (e-mail: [email protected]). troenterol Belg. 1996;59:225-228. 21. Thompson WG. Gender differences in irritable present study, there are no recom- bowel symptoms. Eur J Gastroenterol Hepatol. mendations in the algorithm regard- REFERENCES 1997;9:299-302. ing the possible impact that sex has 22. Frigerio G, Beretta A, Orsenigo G, Tadeo G, Im- periali G, Minoli G. Irritable bowel syndrome: still on IBS symptom reporting and on 1. Camilleri M, Choi MG. Review article: irritable far from a positive diagnosis. Dig Dis Sci. 1992; symptom-based diagnostic criteria. bowel syndrome [review]. Aliment Pharmacol 37:164-167. Indeed, differences between the sexes Ther. 1997;11:3-15. 23. Vanner SJ, Depew WT, Paterson WG, et al. Pre- 2. Drossman DA, Whitehead WE, Camilleri M. Irri- dictive value of the Rome criteria for diagnosing in health-seeking behavior have been table bowel syndrome: a technical review for prac- the irritable bowel syndrome. Am J Gastroen- reported by Hochstrasser and Angst,37 tice guideline development [review]. Gastroen- terol. 1999;94:2912-2917. terology. 1997;112:2120-2137. 24. Kay L, Jorgensen T, Lanng C. Irritable bowel syn- who found that women sought care 3. Thompson WG, Heaton KW. Functional bowel dis- drome: which definitions are consistent? J In- for GI problems significantly more of- orders in apparently healthy people. Gastroenter- tern Med. 1998;244:489-494. ten than did men. ology. 1980;79:283-288. 25. Hahn BA, Saunders WB, Maier WC. Differences 4. Saito YA, Locke GR, Talley NJ, Zinsmeister AR, Fett between individuals with self-reported irritable Further research is necessary to SL, Melton LJ III. A comparison of the Rome and bowel syndrome (IBS) and IBS-like symptoms. Dig define the most accurate methods of Manning criteria for case identification in epidemio- Dis Sci. 1997;42:2585-2590. patient identification and diagnosis. logical investigations of irritable bowel syndrome. 26. Talley NJ, Phillips SF, Wiltgen CM, Zinsmeister AR, Am J Gastroenterol. 2000;95:2816-2824. Melton LJ III. Assessment of functional gastro- Studies should be performed using es- 5. Thompson WG. The functional gastrointestinal intestinal disease: the bowel disease question- tablished methodological standards bowel disorders. In: Drossman D, ed. The Func- naire. Mayo Clin Proc. 1990;65:1456-1479. tional Gastrointestinal Disorders. McLean, Va: 27. Levy RL, Whitehead WE, Feld AD, et al. Agree- for diagnostic test evaluation and Degnon Associates; 1994:117-134. ment between clinician recorded diagnoses and should compare the most com- 6. Drossman DA, Richter JE, Talley NJ. The Func- Rome I and Rome II diagnostic criteria for irri- monly used criteria and diagnostic tional Gastrointestinal Disorders, Pathophysiol- table bowel syndrome [abstract]. Gastroenterol- ogy, and Treatment: A Multinational Consensus. ogy. 2000;118:A2068. tests. Finally, a prospective evalua- Boston, Mass: Little Brown; 1994. 28. Yuri A, Locke G, Prather C, et al. Evaluation of the tion of the impact of systematic ap- 7. Manning AP, Thompson WG, Heaton KW, Mor- Rome II criteria for irritable bowel syndrome in a proaches to care for patients with IBS ris AF. Towards positive diagnosis of the irritable tertiary care outpatient population [abstract]. Gas- bowel. BMJ. 1978;2:653-654. troenterology. 2000;118:A2071. should be performed to document the 8. Thompson WG, Longstreth GF, Drossman DA, 29. Thompson GW, Irvine JE, Pare P. Comparing impact of guidelines on the cost- Heaton KW, Irvine EJ, Muller-Lissner SA. Func- Rome I and Rome II criteria for irritable bowel syn- tional bowel disorders and functional abdominal drome (IBS) in a prospective survey of the Cana- effectiveness and outcomes of care. pain. Gut. 1999;45(suppl II):II43-II47. dian population [abstract]. Am J Gastroenterol. We hope that, until results from com- 9. Schmulson MW, Chang L. Diagnostic approach 2000;96:A2553. parative prospective studies are avail- to the patient with irritable bowel syndrome [re- 30. Saito YA, Locke GR, Talley NJ, Zinsmeister AR, view]. Am J Med. 1999;107(5A):20S-26S. Fett SL, Melton J. The effect of new diagnostic cri- able, the algorithm will inform the de- 10. Kruis W, Thieme C, Weinzierl M, Schussler P, Holl teria for irritable bowel syndrome on community cision-making process for a wide J, Paulus W. A diagnostic score for the irritable prevalence estimates [abstract]. Gastroenterol- range of providers caring for pri- bowel syndrome: its value in the exclusion of or- ogy. 2000;118:A396. ganic disease. Gastroenterology. 1984;87:1-7. 31. Sperber A, Safieh Y, Jaffer A, et al. A comparison mary care patients with abdominal 11. Thompson WG, Longstreth GF, Drossman DA, of the prevalence of IBS using Rome I and Rome discomfort or pain and altered bowel Heaton KW, Irvine EJ, Muller-Lissner SA. C. Func- II criteria in an epidemiology study. [abstract] Gas- tional bowel disorders and fuctional abdominal troenterology. 2000;118:A397. function suggestive of IBS. pain. In: Drossman AD, Talley NJ, Corazziari ETNJ, 32. Kapoor KK, Nigam P, Rastogi CK, Kumar A, Gupta Thompson WG, Whitehead WE, eds. Rome II: The AK. Clinical profile of irritable bowel syndrome. Accepted for publication March 29, Functional Gastrointestinal Disorders. McLean, Va: Indian J Gastroenterol. 1985;4:15-16. Degnon Associates; 2000:351-432. 33. Talley NJ, Phillips SF, Bruce B, Twomey CK, Zins- 2001. 12. Lijmer JG, Mol BW, Heisterkamp S, et al. Empirical meister AR, Melton LJ III. Relation among per- From the Department of Medi- evidence of design-related bias in studies of diag- sonality and symptoms in nonulcer dyspepsia and nostic tests [published erratum appears in the irritable bowel syndrome. Gastroenterology. cine, Southern Arizona Veterans Af- JAMA. 2000;283:1963]. JAMA. 1999;282:1061- 1990;99:327-333. fairs Medical Center, Tucson (Dr 1066. 34. O’Keefe EA, Talley NJ, Tangalos EG, Zinsmeister Fass); Department of Medicine, Kai- 13. Kahn KL, Roth CP, Kosecoff J, et al. Indications for AR. A bowel symptom questionnaire for the el- Selected Medical and Surgical Procedures—A Lit- derly. J Gerontol. 1992;47:M116-M121. ser Permanente Medical Care Pro- erature Review and Ratings of Appropriateness. 35. Agreus L, Talley NJ, Svardsudd K, Tibblin G, Jones gram, San Diego, Calif (Drs Long- Santa Monica, Calif: RAND Corp; 1986. MP. Identifying dyspepsia and irritable bowel syn- streth, Crane, and McCarberg); 14. Using Clinical Practice Guidelines to Evaluate Qual- drome: the value of pain or discomfort, and bowel ity of Care. Vol 1. Bethesda, Md: US Dept of Health habit descriptors. Scand J Gastroenterol. 2000; Departments of Medicine (Drs Pimen- and Human Services; 1995. Publication AHCPR 35:142-151. tel and Ofman) and Health Services 95-0045. 36. Pimentel M, Chow EJ, Lin HC. Eradication of small 15. Talley NJ, Phillips SF, Melton LJ, Mulvihill C, Wilt- intestinal bacterial overgrowth reduces symp- Reseach (Dr Ofman), Cedars-Sinai gen C, Zinsmeister AR. Diagnostic value of the toms of irritable bowel syndrome. Am J Gastro- Health System, Los Angeles, Calif; Manning criteria in irritable bowel syndrome. Gut. enterol. 2000;95:3503-3506. Zynx Health Inc, Los Angeles (Messrs 1990;31:77-81. 37. Hochstrasser B, Angst J. The Zurich Study: XXII, 16. Talley NJ, Zinsmeister AR, Van Dyke C, Melton LJ epidemiology of gastrointestinal complaints and Fullerton and Russak, Drs Chiou and III. Epidemiology of colonic symptoms and the ir- comorbidity with anxiety and depression. Eur Arch Ofman, and Ms Reyes); and Depart- ritable bowel syndrome [published erratum ap- Psychiatry Clin Neurosci. 1996;246:261-272.

(REPRINTED) ARCH INTERN MED/ VOL 161, SEP 24, 2001 WWW.ARCHINTERNMED.COM 2088

Downloaded From: https://jamanetwork.com/ on 09/26/2021