Systematic Review on the Management of Chronic Constipation in North America

Abstract

2.1 Methods Evidence-based guidelines should have: a transparent link between the evidence and the recommendations; explicit criteria for inclusion of studies to serve as the evidence; comprehensive searching of the literature for relevant studies; a standardized and explicit system for grading the quality of study design; a standardized and explicit system for grading recommendations; and, recommendations that acknowledge the magnitude of the treatment benefit, the adverse events associated with the treatment, and individual patient preferences that may guide the application of guideline recommendations (1–3). In order to fulfill these requirements, Task Force members followed the techniques used to produce a previous American College of Gastroenterology evidence-based monograph (4). Each section of the systematic review has been numbered to provide a link between the evidence and the recommendations. Standard techniques for literature searching and study selection were utilized for each section of this document (5, 6). Data about study methodology and study results were extracted onto standard forms and summaries of data are presented in tables and graphs to insure that the quality of study design was defined and that the magnitude of treatment was quantified. A commonly used system for grading recommendations in evidence-based guidelines (7) was adapted for this document (see Section 2.2) and insured that an explicit and transparent process was used to make recommendations based upon the strength of the evidence. Adverse events and individual patient preferences may affect the application of these recommendations. Therefore, adverse events have been assessed, relative contraindications to specific treatments have been described, and Task Force members contributed their expert advice based upon their clinical experience. Literature Search In order to identify Chronic Constipation (CC) Therapy Trials, the following literature search techniques were employed. Separate PUBMED and MEDLINE searches of English language articles from 1966–2003 were performed with different combinations of the following search terms: “constipation,” “laxatives, stimulant,” “laxatives, osmotic,” “laxatives, irritant,” “laxatives, bulk,” “fecal softeners,” “sorbitol,” “lactulose,” “milk of magnesia,” “magnesium sulphate,” “bisacodyl,” “calcium polycarbophil,” “polyethylene glycol,” “danthron,” “cascara,” “ispaghula, “bran,” “celandin,” “docusate,” “poloxalkol,” “mineral oil,” “glycerine,” “psyllium,” “methylcellulose,” “senna,” and “tegaserod.” Exploded terms were reviewed, and where appropriate, the search was expanded to include them. Manual searches of reference lists from relevant articles also were performed to identify additional studies that may have been missed during the computer-assisted search. In order to identify relevant studies about the Epidemiology of CC and the Diagnostic Approach to Patients with CC Symptoms, the following literature search techniques were employed. For Epidemiology of CC, only English language articles published in full manuscript form were considered. A search of the MEDLINE database from 1966–2003 was performed using the exploded (exp) MESH terms: exp constipation AND (exp incidence OR exp prevalence OR exp prognosis OR exp natural history OR exp epidemiology OR exp quality of life). For EMBASE, a search from 1998–2002 was performed and modified to include only articles that mentioned constipation in the title: constipation.ti AND (exp incidence OR exp prevalence OR exp prognosis OR exp natural history OR exp epidemiology OR exp quality of life). The Current Contents database from Week 1, 2002 to Week 6, 2003 was searched with the terms constipation.mp AND (incidence.mp OR prevalence.mp OR prognosis.mp OR natural history.mp OR epidemiology.mp OR quality of life.mp). Manual searches of reference lists from potentially relevant articles also were performed to identify any additional studies that might have been missed during the computer-assisted search. For Diagnostic Approach to the Patient with CC Symptoms, a bibliographic database search of MEDLINE from 1966 was performed. The search terms used included “constipation,” “colonic diseases,” “Rome criteria,” “colonoscopy,” “barium enema,” “defecation,” and various systemic or local disorders that cause constipation such as diabetes or colon cancer. Only human studies were considered. Bibliographies from all potentially relevant articles were searched manually. A medical education and research company (EBMed, LLC, Anaheim Hills, CA) assisted the Task Force with literature searches, application of study selection criteria, data extraction and analysis, and assessment of methodologic quality of individual trials. Study Selection Criteria The titles and abstracts of all citations identified by the literature searches were reviewed. Potentially relevant studies were retrieved, and the selection criteria were applied. Since a North American perspective was used, only treatments available in the United States were examined and only epidemiologic studies from North American populations were reviewed. For CC therapy trials, the selection criteria were: (a) RCT; (b) population of adult patients with CC; (c) comparison of CC therapy versus placebo or control therapy; (d) evaluation of relief of CC symptoms; (e) results published in English in full manuscript form (or adequate data available after written communication with investigators); and, (f) therapy available in the United States. For Epidemiology of CC studies, the selection criteria were: (a) studies of population-based samples of CC patients in North America; (b) population of adult patients (inclusion of pediatric patients within an adult study population was allowed); (c) results reported on prevalence, incidence, quality of life, or natural history of CC; (d) results published in full manuscript form; and (e) English language only. Case definition for constipation in these epidemiologic trials included patient self-report, physician or ICD-9 diagnosis, consensus criteria from Rome I or Rome II. For trials about the Diagnostic Approach to the Patient with CC Symptoms, study inclusion criteria were: (1) population of patients with CC symptoms (note: the definition of constipation is extremely variable or inadequately described in most studies and no single symptom was used to define this condition; therefore, we included any study that stated patients had CC symptoms.); (2) performance of diagnostic tests to diagnose organic disorders responsible for CC symptoms; including complete blood cell count, biochemical laboratory tests, colonoscopy, flexible sigmoidoscopy, barium enema, and plain abdominal radiography; (3) prevalence of organic disorders responsible for CC symptoms or number of patients with abnormalities on physiologic testing. Data Extraction and Analysis For CC therapy trials, data about study methodology and study results were abstracted onto standard forms for the performance of systematic reviews. Data were extracted about: (a) study population; (b) intervention: dosage and schedule of treatment versus placebo or control therapy; (c) study duration; and (d) proportion of patients achieving improvement in global CC symptoms, stool frequency, stool consistency, and adverse events. The Rome committee recommended study design techniques to minimize bias in trials of functional GI disorders (8) (Table 2.1.1). Data about the use of these techniques were extracted. We also used a single grading system that has been validated for appropriate study design of therapy trials examined in a systematic review or meta-analysis (9) (Table 2.1.2). Data about the use of proper randomization, blinding and follow-up were extracted from each study and summarized in tabular form. Because of wide variation in study design, study endpoints, and dosages of study medication, however, no attempt was made to combine results into meta-analyses. Based on Rome Committee recommendations (8), global improvement in symptoms is the ideal endpoint for these trials because this endpoint encompasses improvement in the multiple symptoms (e.g., straining, sense of incomplete evacuation, bloating, etc.) of CC. Because few trials evaluated this endpoint, however, Task Force members agreed to expand the endpoints to include stool frequency and stool consistency. Results of individual RCTs are presented in tabular form. When individual RCTs reported numerical results, these results were included in the tables. When individual RCTs provided graphs or declarative statements without numerical results, then declarative statements were included in the tables. Due to the lack of RCTs, no data about the management of pelvic floor disorders are included.Table 2.1: Table 2.1.1. Qualitative Assessment of Study Methodology Scale*#Table 2.1: Table 2.1.2. Jadad Criteria-Validated Scale to Assess Quality of Study Design for Treatment Trials in a Systematic Review or Meta-AnalysisFor Epidemiology of CC studies, data about study methodology and study results also were abstracted onto standard forms. Data were abstracted about: (a) symptom-based definition of CC (e.g., patient self-report or Rome Criteria); (b) sample size and case ascertainment technique to identify CC patients; (c) prevalence, incidence, prevalence of CC sub-groups, gender distribution, race distribution, socioeconomic distribution, diseases associated with CC, and mean age of onset of symptoms; and (d) disease activity (e.g., prevalence of CC over time and frequency of CC flares within a specified period of time) and/or quality of life data. For trials about the Diagnostic Approach to the Patient with CC Symptoms, data were abstracted about: (a) symptom-based criteria used to identify CC patients; (b) diagnostic evaluation performed; (c) prevalence of confirmed organic GI disease, resulting in an alternative diagnosis to explain CC symptoms or prevalence of abnormal results on diagnostic tests. 2.2 Levels of Evidence and Grading of Recommendations Quantitative Assessment of Study Methodology Previous reviews and epidemiologic studies (10–12) have established criteria that minimize bias in trials about therapy, including the use of randomization, concealed allocation, double-blinding, and complete patient follow-up. A single grading system has been validated for appropriate study design of therapy trials examined in a systematic review or meta-analysis (9) (Table 2.1.2). In order to assess the strength of individual studies about CC therapies, data about the use of these techniques were extracted, summarized in a quantitative scale, and presented in tabular form. This scale estimates the rigor of an individual trial: a trial with a low quality score may be more likely to produce inaccurate or biased results and a trial with a high quality score may be more likely to produce accurate and unbiased results (8–10). Additionally, the Rome committees described the preferred design of treatment trials for functional gastrointestinal disorders (8) (Table 2.1.1). Data about use of these additional Rome committee criteria for appropriate design of treatment trials also were extracted and are presented in descriptive form. No standard criteria are available to rate the quality of study design of epidemiologic studies in a systematic review. Population-based studies may be preferable to studies of referral populations because referral populations may provide inflated estimates of the prevalence and incidence of a disorder. Therefore, epidemiologic studies in this review are limited to population-based studies. Levels of Evidence Levels of Evidence have been defined previously (see Table 1.2). Level I evidence represents high quality RCTs. These RCTs have few limitations in their study design, which should minimize Type I errors. Thus, if a RCT shows a significant difference between treatment and placebo, then this finding probably did not result from biased study design. These RCTs also have adequate power to minimize Type II errors. Thus, if a RCT does not show a significant difference between treatment and placebo, then this finding probably did not occur because of an inadequate sample size of patients. Level II evidence represents intermediate quality RCTs. These RCTs have important limitations in their study design, which could produce a Type I error. Intermediate quality RCTs also may be susceptible to Type II errors because of inadequate sample size. Level III–V evidence comes from non-randomized trials or case series. These are observational studies that are prone to multiple biases that produce Type I errors. For this review, Level III–V evidence was not used to make recommendations about CC therapies. Level III–V evidence was used only to make recommendations about the Diagnostic Approach to the Patient with CC Symptoms or about Epidemiology of CC because data on these topics are available only from observational studies. Grading of Recommendations Recommendations are listed as Grade A, Grade B, or Grade C (see Table 1.2). Grade A recommendations are supported by the strongest (Level I) evidence. Task Force members strongly believe that these recommendations are accurate based upon the evidence. Grade B recommendations are supported by intermediate quality (Level II) evidence. Task Force members believe that Grade B recommendations may have important limitations because of the intermediate quality of the evidence. These recommendations may change in the future if high quality (Level I) evidence becomes available. Grade C recommendations are supported by observational studies (Level III–V evidence). The strength of evidence behind these recommendations is limited, and Grade C recommendations are provided only because they represent the best evidence about the Epidemiology of CC and the Diagnostic Approach to Patients with CC Symptoms.