Mechanical bowe l preparation for elec tive colorectal surgery (Review) Guenaga KKFG, Atallah ÁN, Castro AA, Matos D, Wille-Jørgensen P This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library 2008, Issue 4 http://www.thecochranelibrary.com Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaborat ion. Published by John Wi ley & Sons, Ltd. T A B L E O F C O N T E N T S 1HEADER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3PLAIN LANGUAGE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9AUTHORS’ CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11CHARACTERISTICS OF STUDIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19DATA AND ANALYSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20Analysis 1.1. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 1 Anastomosis l eakage stratified for colonic or rectal surgery. . . . . . . . . . . . . . . . . . . . . . . . . . . 22Analysis 1.2. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 2 Overall anastomotic leakage for colorectal surgery. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23Analysis 1.3. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 3 Mortality. . . . . 23Analysis 1.4. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 4 Peritonitis. . . . . 24Analysis 1.5. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 5 Reoperation. . . . 24Analysis 1.6. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 6 Wound infection. . 25Analysis 1.7. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 7 Infectious extra- abdominal complications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26Analysis 1.8. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 8 Non-infectious extra- abdominal complications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26Analysis 1.9. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 9 Surgical site infections. 27Analysis 1.10. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 10 Sensitivity analysis 1 - Studies with dubious randomisation procedure excluded. . . . . . . . . . . . . . . . . . . . 29Analysis 1.11. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 11 Sensitivity analysis 2 - Studies published as abstract only e xcluded. . . . . . . . . . . . . . . . . . . . . . . . 31Analysis 1.12. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 12 Sensitivity analysis 3 - Studies including children excluded. . . . . . . . . . . . . . . . . . . . . . . . . . . 33Analysis 1.13. Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 13 Sensitivity analysis 4 - Studies including patients without anastomosis excluded. . . . . . . . . . . . . . . . . . . . 35WHAT’ S NEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36HISTORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36CONTRIBUTIONS OF AUTHORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37DECLARATIONS OF INTEREST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37SOURCES OF SUPPORT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37NOTES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37INDEX TERMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iMechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaborat ion. Published by John Wi ley & Sons, Ltd. [Intervention review] Mechanical bowel preparation for elective colorectal surge ry Katia KFG Guenaga 1 , Álvaro N Atallah 2 , Aldemar A Castro 3 , Delcio Matos 4 , Peer Wille-Jørgensen 5 1 Surgical Gastroenterology Department, Ferderal University of São Paulo, Guarujá, Brazil. 2 Brazilian Cochrane Centre, Universidade Feder al de São Paulo / Escola Paulista de Medicina, São Paulo, Brazil. 3 Department of Public Health, State University of Heath Science, Maceió, Brazil. 4 Brazilian Cochrane Centre, Universidade Federal de São Paulo, São Paulo, Brazil. 5 Department of Surgical Gastroenterology K, Bispebjerg Hospital, Copenhagen NV, Denmark Contact address: Katia KFG Guenaga, Surgical Gastroenterology Department, Ferderal University of São Paulo, Marivaldo Fernandes, 152 apto. 13, Guarujá, São Paulo, 11 440-050, Brazil. kaci@uol.com.br. (Editorial group: Cochrane Colorectal Cancer Group.) Cochrane Database of Systematic Reviews, Issue 4, 2008 (Status in this issue: Edited) Copyright © 2008 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. DOI: 10.1002/14651858.CD001544.pub2 This version first published online: 24 January 2005 in Issue 1, 2005. Re-published online with edits: 8 October 2008 in Issue 4, 2008. Last assessed as up-to-date: 20 October 2004. ( Dates and statuses?) This record should be cited as: Guenaga KKFG, Atallah ÁN, Castro AA, Matos D, Wille-Jørgensen P. Mechanical bowel preparation for elective colorectal surgery. Cochrane Database of Systematic Reviews 2005, Issue 1. Art. No.: CD001544. DOI: 10.1002/14651858.CD001544.pub2. A B S T R A C T Background For over a century the presence of bowel content during surgery has been linked to anastomotic leakage. Mechanical bowel preparation has been considered an efficient agent against leakage and infectous complications. This dogma is not based on solid evidence, but on observational data and expert’s opinions. Objectives To dete rmine the effectiveness and safety of prophylactic mechanical bowel preparation for morbidity and mortality rates in elective colorectal surgery. Search strategy We searched MEDLINE, EMBASE, LILACS, and the Cochrane Central Register of Controlled Trials. We also searched relevant medical journals, and conference proceedings from major gastroenterological congresses and contacted experts in the field. We used the search strategy described by the Colorectal Cancer Review Group, without limitations for date of publication and language. I Selection criteria Randomised, cl inical trials that compared any strategy in mechanical bowel preparation with no mechanical bowel preparation. Data collection and analysis Data were independently extracted by the reviewers and cross-checked. The same reviewers assessed the methodological quality of each trial. Details of the randomisation (generation and concealment), blinding, whether an intention-to-treat analysis was done, and the number of patients lost to follow-up was recorded. For analysis the Peto odds ratio (OR) was used as defaults. Main results 1Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaborat ion. Published by John Wi ley & Sons, Ltd. Of the 1592 patients (9 trials), 789 were allocated to mechanical bowel preparation (Group A) and 803 to no preparation (Group B) before elective colorectal surgery. For anastomotic le akage (main outcome) the results were: - Low anterior resection: 9.8% (11 of 112 patients in Group A) compared with 7.5% (9 of 119 patients in Group B); Peto OR 1.45, 95% confidence interval (CI): 0.57 to 3.67 (non-significant); - Colonic surgery: 2.9% (Group A) compared with 1.6% (Group B) ; Peto OR 1.80, 95% CI: 0.68 to 4.75 (non-significant); Overall anastomotic leakage: 6.2% (Group A) compared with 3.2% (Group B); Peto OR 2.03, 95% CI: 1.276 to 3.26 (p=0.003). For the secondary outcome of wound infection the result was: 7.4% (Group A) compared with 5.4% (Group B); Peto OR 1.46, 95% CI: 0.97 - to 2.18 (p=0.07); Sensitivity analyses excluding studies with dubious randomisation, studies published as abstracts only, and studies involving children did not change the overall conclusions Authors’ conclusions There is no convincing evidence that mechanical bowel preparation is associated with reduced rates of anastomotic leakage after elective colorectal surgery. On th e contrary, there is evidence that this intervention may be associated with an increased rate of anastomotic leakage and wound complications. It is not possible to be conclusion on the latter issue because of the clinical hete r ogeneity of trial inclusion criteria, methodological inadequacies in tr ial (in particular, poor reporting of concealment and allocation), potential performance biases, and failure of intention-to-treat analyses. Nevertheless, the dogma that mechanical bowel preparation is necessary before elective colorectal surgery should be reconsidered. 2Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaborat ion. Published by John Wi ley & Sons, Ltd. P L A I N L A N G U A G E S U M M A R Y Key findings: Preoperative mechanical bowel preparation before colorectal sur g ery does not reduce anastomotic leakage. Preoperative mechanical bowel preparation before colorectal surgery is a widely-practised treatment, but its efficacy has never been proven outside observational studies and animal experiments. This systematic review of nine trials (1592 patients) found that there is no convincing evidence that mechanical bowel preparation is associated with reduced rates of anastomotic leakage after elective colorectal surgery, but on the contrary, there is evidence that this intervention may be associated with an increased rate of anastomotic leakage and wound complications. There was no difference in other outcomes, such as mortality, peritonitis, re-operation, infectious extra-abdominal complication, non- infectious extra-abdominal complication, and surgical site infection. Mechanical bowel preparation before colorectal surgery cannot be recommended as routine. B A C K G R O U N D The impor tance of efficient mechanical bowel preparation in pre- venting infectious complications and anastomotic dehiscence after colorectal surger y has been adogma among surgeonsfor more than a century ( Halsted 1887; Thornton 1997). Clinical experiences and observational studies have shown that mechanical removal of gross faeces from the colon has been associated with decreased morbidity and mortality in patients undergoing operations of the colon ( Nichols 1971). One author (Chung 1979) was categorical: “One of the most important factors within the control of the sur- geon, that affect the outcome of a col onic operation, is the degree of emptiness of the bowels”. An early randomised clinical trial questioned this view and con- cluded that vigorous mechanical bowel preparation is not neces- sary ( Hughes 1972). Omission of enemas and bowel washes from the preoperative procedures will be welcomed by both patients and nursing staff. One trial ( Irving 1987) questioned the necessity of preoperative or intra operative mechanical bowel preparation of the colon, before primary anastomosis. The authors argue that preoperative bowel preparation is time-consuming, expensive, and unpleasant for pa- tients - even dangerous on occasion - and completely unnecessary. Traditionally, “bowel preparation” has been used to reduce faecal mass and also bacterial counts. Most surgeons consider mechan- ical bowel preparation to be essential, and the systematic admin- istration of appropriate antibiotics has been shown effective in re- ducing infectious complications in numerous randomised trials. Furthermore, mechanical bowel preparation is recommended by many guidelines from surgical associations and scientific societies ( ASCGBI 2001; Moore 1999; SIGN 1997). Different methods of mechanical bowel preparation have bee n tested and approved and the potential danger of having faeces in contact with a newly performed anastomosis when the colon was not prepared has been discussed ( Grabham 1995 ; Mealy 1992 ). Both experimental studies (Smith 1983; O’Dwyer 1989; Schein 1995), and clinical trials inemergency surgery (Baker 1990; Dorudi 1990; Duthie 1990) have been published in order to sup- port this theory. Two randomised trials from Ireland and Brazil concluded that the role of bowel preparation in colorectal surgery requires re-evalua- tion ( Burke 1994; Santos 1994). If bowel preparation is shown to be needless, it could mean a shorter hospital stay for the patient and avoidance of the potential complications associated with the cleansing procedure such as gastric intolerance, low serum potas- sium level, bowel explosion, mucosal lesions, electrolyte distur- bance and fluid overload. Analysed in isolation, the results of published trials have not shown any significant difference in outcomes between patients who un- derwent mechanical bowel preparation and those who did not, but as the individual studies contain a high risk of a statistical type II er ror it seemed justified to perform a meta-analysis. O B J E C T I V E S To determine the necessity of prophylactic mechanical bowel preparation in patients undergoing elective colorectal surgery. The incidence of anastomotic dehiscence is increasing as more anal the anastomosis is performed ( Goligher 1970). Because bowel preparation might have different effect in colon and rectum, we will stratify the analyses for colon and rectum separately whenever possible. M E T H O D S Criteria for consideri ng stu dies for this review Types of studies 3Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaborat ion. Published by John Wi ley & Sons, Ltd. (i) Randomised clinical trials comparing preoperative mechanical bowel preparation versus no preparation (or placebo) in (ii) patients undergoing elective colorectal surgery and in which (iii) the primary outcome (anastomotic leakage) is clearly stated in both treatment arms. To be included in this review, trials had to meet all three criteria. EXCLUSION CRITERIA: Studies evaluating two or more different cleansing methods; stud- ies including patients undergoing emergency surgery. Types of participants Patients undergoing elective colorectal surgery. Types of interventions Any strategy in mechanical bowel preparation for patients under- goingelective colorectal surgery compared to no mechanical bowel preparation. Types of outcome measures PRIMARY OUTCOME MEASURES: 1) A nastomotic leakage, defined as discharge of faeces from the anastomosis site, externalising through the drainage opening or the wound incision; or just th e existence of an abscess adjacent to the anastomosis site. The anastomotic leakage was confirmed by either clinical or radiological investigation. The type of surgery and anastomosis site were stratified in: A: Low anterior resection, extra-peritoneal anastomosis (rectum considered extra-peritoneal); B: Colonic surgery, intra-peritoneal anastomosis. 2) Overall anastomotic leakage: total number of anastomotic de- hiscence in all of colon and rectum. SECONDARY OUTCOME MEASURES: 3) Mortality: number of postoperative deaths related to the surgery. 4) Per itonitis: presence of postoperative infections at the abdomi- nal cavity, localized (abscess) or not. 5) Re-operation: surgical re-intervention for anastomotic compli- cation. 6) Wound infection: defined as a discharge of pus from th e ab- dominal wound. 7) Infectious extra-abdominal complication: postoperative infec- tious complication at extra-abdominal site. 8) Non-infectious extra-abdominal complications (e.g. deep ve- nous thrombosis, cardiac complications, wound rupture). 9) Overall infections in surgical sites. SENSITIVITY AND SUBGROUP ANALYSES 10) Anastomotic leakage and wound infection in studies with adequate randomisation. 11) Anastomotic leakage and wound infection in studies published as full articles. 12) Anastomotic leakage and wound infection in studies only deal- ing with adult patients. 13) Anastomotic leakage and wound infection in studies in which bowel continuity was restored. Search methods for identification of studies See: Collaborative Colorectal Cancer Review Group search strat- egy ( Wille-Jørgensen 1999). The studies were identified from the following sources: MED- LINE, EMBASE, CINAHL, LILACS, SCISEARCH, Controlled Clinical Trials Database, Trials Register of the Cochrane Colorec- tal Cancer Group, and the Cochrane Central Register of Con- trolled Trials (CENTRAL). Reference lists were checked, hand- searching was carried out, and through letters sent to study au- thors. Conference proceedings from major gastrointestinal confer- ences (World Congress of Gastroenterology, Annual Meetings of American Sociery of Colon and Rectal Surgery, Annual meetings of Association of Coloproctology of Great Britain and Ireland, Tripartites meetings) were scrutinised back to 1994 (last possible retrieval of abstract-material). There were no limits regarding lan- guage, date, or other restrictions in the searches. Al l searches were performed up to July 2004. Search strategy: #1 Cochrane Collaboration search strategy for randomised con- trolled trials ( Handbook 2004) #2 Tw INTESTIN* or Tw BOWE L #3 Tw LARGE or Tw GROSSO or Tw GRUES O #4 #2 and #3 #5 Tw COLO* or Tw CECO #6 Tw RECT* or Tw RET* #7 #4 or #5 or #6 #8 #3 and #7 #9 Tw PREPARA* #10 Tw SURGERY or SURGICAL #11 #8 and #9 and #10 #12 #1 and #11 Data collection and an alysis LOCATING AND SELECTING STUDIES The reviewers (KFG and PWJ) independently selected the trials to be included in this review. Disagreement on selection was solved in a consensus meeting. Only studies designed and stated as ran- domised controlled trials were considered for inclusion. CRITICAL APPRAISAL OF STUDIES The re viewers assessedthe methodological quality of each trial. We recorded details of the randomisation method, blinding, whether anintention-to-treatanalysiswas done, and the number ofpatients lost to follow-up to evaluate the risk of bias in the individual studies ( Handbook 2004). We assessed the external validity of the studies in an analysis of the characteristics of the participants and the interventions as collected below. COLLECTING DATA 4Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaborat ion. Published by John Wi ley & Sons, Ltd. We included studies in which allocationconcealment was regarded adequate were included. A few studies with unclear allocation concealment were included as well . The reviewers independently extracted and cross-checked the data. The result of each trial was summarised in 2 x 2 tables for each outcome. We evaluated study validity according to participants and inter- ventions: PARTICIPANTS: Category of disease (colorectal cancer, inflam- matory disease, megacolon, polyposis, diverticular disease), gen- der, age, topography, operative procedure, antibiotic therapy, sur- geon experience. The calculation of the sample size and the sample representativeness was observed. INTERVENTIONS: Types of mechanical bowel preparation: an- terograde(oral) or retrograde (enemas)versus no mechanical bowel preparation. Information data from the studies published more than once, was only included once. Data were entered into Review Manager 4.2 by single data-entry by KFG and controlled by PWJ. ANALYSING AND PRESENTING RESU LTS If appropriate we stratified the studies for different meta-analy- sis (Review Manager 4,2) according to the analysis of the defined outcomes. We used various techniques: in the dichotomous out- come measures, the combined logarithm of the Peto Odds Ratio (fixed effect model) was used as default. We performed a te st for statistical heterogeneity in each case. If we detected heterogeneity, results were reported as Odds Ratio using random-effects mod- elling. For the analysis, we reviewed only patients who underwent elective colorectal surgery; according to type of interventions, type of participants, to assess whether there were important differences between them. A ll inclusion criteria had to be met. We assessed statistical heterogeneity and potential publication bias in th e results of the meta-analysis both by inspection of graphical presentations (“funnel plot”: plotting the study weight or sample size (on the “Y” axis) against the Odds Ratio (on the “x” axis) and by calculating a test of heterogeneity (standard chi-squared test on N degrees of freedom where N equals the number of tri- als contributing data minus one). The funnel plot is possible for outcomes described in five or more studies. Three possible reasons for heterogeneity were pre-specified: (i) that responses differ ac- cording to difference in the quality of the trial; (ii) that response differ according to sample size; (iii) that response differ according to clinical heterogeneity. If we detected heterogeneity, sensitivity analyses were performed in subgroups. SENSITIVITY ANALYSIS. We used a fixed sample model with Peto Odds ratio was used as default. If heterogeneity was apparent, a random effects model was applied. IMPROVING AND UPDATING THIS REVIEW As a minimum, updates will be considered on an biannual basis. This is the first update performed two years after the first appear- ance in The Cochrane Library. Three additional studies have been included, and one study previously included as an abstract is now included as a full paper version. R E S U L T S Description of studies See: Characteristicsof included studies; Characteristics of excluded studies . We identified fourteen studies of which nine were included, and five trials were excluded. The reason for exclusion was absence of a control group ( Irving 1987, Dorudi 1990, Duthie 1990), elemental diet in the control group ( Matheson 1978), or lack of description of the primary outcome and insufficient description of the secondary outcomes ( Hughes 1972) see “Characteristics of excluded studies”. One of theincluded studies was published in Portuguese ( Fillmann 1995 ), and identified in the Lilacs database. One study was in Spanish ( Tabusso 2002). The others were published in English language. Three studies were published as abstracts only (Brown- son 1992, Bucher 2003, Fa-Si-Oen 2003). Data from the latter study were retrieved from another publication ( Slim 2004). Three new studies were identified and included in this update ( Bucher 2003, Tabusso 2002, Fa-Si-Oen 2003). Two were con- ference proceedings ( Bucher 2003, Fa-Si-Oen 2003). TYPES OF PARTICIPANTS The inclusion criteria was the same for all studies: patients ad- mitted for elective colorectal surgery. O ne trial ( Santos 1994) in- cluded children. Two studies included patients without anasto- mosis ( Fillmann 1995, Santos 1994); one study (Brownson 1992) excluded these patients in only one of the outcomes: anastomosis leakage; two of them ( Burke 1994, Miettinen 2000) excluded pa- tients for whom bowel continuity was not restored. In two of the new trials included in the review ( Bucher 2003 , Tabusso 2002) one of the inclusion criteria was patients undergo- ing elective “left-sided” colorectal surgery. One new trials stated only “elective colorectal surgery” ( Fa-Si-Oen 2003). None of the studies reported the use of preoperative adjuvant chemotherapy or radiation. Seven stated use of prophylactic an- tibiotics, and there was no information on this from two studies ( Zmora 2003, Bucher 2003) . Five trials ( Burke 1994, Fillmann 1995, Miettinen 2000, Santos 1994 , Zmora 2003 ) described the two allocation groups as being equal according to gender, age, types of operation, and diagnosis. Three of them (Brownson 1992, Bucher 2003, Fa-Si-Oen 2003) did not give details. One ( Tabusso 2002 ) described a statistic difference betwee n the two groups regarding age, hemoglobinlevel and serum albumin. The criteria for excl usion of patients were reported in different ways: A) patients who had been taken antibiotics for at least 15 days before surgery, or if there was evidence of infection, or any as- sociated disease requiring antibiotic therapy, and patients in whom 5Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaborat ion. Published by John Wi ley & Sons, Ltd. the mechanical bowel preparation was not fe asible (Santos 1994); B) any patients wh o could not tolerate the preparation; C) patients who had bowel preparation for colon one week before surgery, patients who where unable to drink the solution, patients not re- quiring opening of the bowel, and one patient who refused to be randomised ( Miettinen 2000); D: Two trials excluded patients in whom bowel continuity was not restored (Burke 1994, Miettinen 2000). One trial (Brownson 1992) excluded the patients in whom bowel continuity was not restored in the analysis of the primary outcome: anastomotic leakage. E: One of the trials ( Zmora 2003) included only patients with primary anastomosis. F) no patients were excluded, ( Fillmann 1995) ; G) did not give details on ex- clusion (Brownson 1992, Bucher 2003, Tabusso 2002, Fa-Si-Oen 2003 ). Two studies, ( Fillmann 1995 , Santos 1994 ), included patients undergoing any of the following surgical procedures: abdomi- nal excision of the rectum, Hartman’s procedure, defunctioning colostomy, colonic anastomosis with colostomy; some of the pa- tients without anastomosis. One of the trials ( Tabusso 2002) in- cluded patients without anastomosis. TYPES OF INTERVENTIONS All of the included studies compared mechanical bowel prepara- tion with no preparation of the bowel prior to colorectal surgery : Preparation of the bowel was either polyethy lene glycol ele ctr olyte solution; laxatives (mineral oil, agar and phenolphthalein); man- nitol; enemas (water, 900 ml; glycerin, 100 ml); sodium picosul- phate 10 mg; Bisacodyl (10 mg)+enemas; and diets, low and non- residue. Only two studies mentioned the experience of the surgeon. wo of them ( Burke 1994, Miettinen 2000) described the operations performed by or under the supervision of a consultant surgeon; one ( Santos 1994), described the operations performed by senior residents. The duration of follow-up was described as follows: A ) 30 days or until h ospital discharge ( Santos 1994); B) 30 days after surgery (Fillmann 1995 , Zmora 2003 ) ; C) 1-2 months after surgery ( Miettinen 2000); D) le ss clearly (Burke 1994, Tabusso 2002): 7 days after surgery ; E) not described (Brownson 1992, Bucher 2003 , Fa-Si-Oen 2003). TYPES OF OUTCOMES MEASUREMENTS PRIMARY OUTCOMES 1) Anastomotic leakage: two of the studies ( Burke 1994; Miettinen 2000 ) stratified the anastomosis between rectal and colonic. Data on stratification were obtained by personal contact with two au- thors ( Zmora 2003, Santos 1994). The others (Brownson 1992, Fillmann 1995, Tabusso 2002, Fa-Si-Oen 2003) did not refer to the site of the anastomosis. Two studies described all anastomosis to be left-sided ( Bucher 2003, Zmora 2003). 2) Overall anastomotic leakage: All the included studies described this outcome. SECONDARY OUTCOMES 3) Mortality: five of the studies described this outcome ( Burke 1994, Fillmann 1995, Miettinen 2000, Santos1994, Zmora2003). 4) Peritonitis: four of the studies (Brownson 1992, Fillmann 1995, Miettinen 2000,Tabusso 2002) included this. 5) Re operation: four trials described this outcome ( Burke 1994, Fillmann 1995, Miettinen 2000, Santos 1994). 6) Wound infection: all of the included studies described it (Brownson 1992, Bucher 2003, Burke 1994 , Fa-Si-Oen 2003, Fillmann 1995, Miettinen 2000 , Santos 1994, Tabusso 2002 , Zmora 2003). 7) Infectious extra-abdominal complication: two studies ( Fillmann 1995; Miettinen 2000) described this outcome. 8) Non-infectious extra-abdominal complication: four studies ( Burke 1994; Fillmann 1995; Miettinen 2000, Zmora 2003) de- scribed this. 9) Surgical site infection: two studies ( Miettinen 2000 , Zmora 2003 ). OTHER CHARACTERISTICS None of the studies contained an indication of how the sample size was calculated. One author ( Fillmann 1995) replied to our enquiry that the sample size was calculated, but didn’t give more details. Two of the studies( Burke 1994, Miettinen 2000) described the sampling as consecutive. SENSITIVITY ANALYSIS AND SUBGROUP ANALYSES ON ANASTOMOTIC LEAKAGE AND WOUND INFECTION In five of the studies (Brownson 1992, Bucher 2003, Burke 1994, Tabusso 2002, Fa-Si-Oen 2003) the allocation method was not well-described . A sensitivity analysis was performed leaving out these studies (outcome 10). As three of the studies (Brownson 1992, Bucher 2003; Fa-Si-Oen 2003) were published as abstracts, an analysis was performed, leaving out these studies (outcome 11). In one study ( Santos 1994) children were included. This study was excluded in the third sensitivity analysis (outcome 12). In three of the trials ( Fillmann 1995, Santos 1994, Tabusso 2002) patients without anastomosis were included and an analysis was carried out without these studies(outcome 13). Risk of bias in included studies None of the studies used an intention to treat analysis. SELECTION BIAS (Sy ste matic differences in comparison groups) In two trials ( Santos 1994, Miettinen 2000), the allocation process was described as randomised cards. One author ( Fillmann 1995) replied to our enquiries and described the process using a random number table. In one study a computer generated list was used ( Zmora 2003). In these studies the allocation process was con- sidered sufficient. In the others (Brownson 1992, Bucher 2003, Burke 1994, Tabusso 2002, Fa-Si-Oen 2003), the allocation pro- cess was not clearly specified and thus considered unclear, leading to a sensitivity analysis. In general, the allocation concealment in all studies was not described. This is known to create biases ( Juni 2002 ). 6Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaborat ion. Published by John Wi ley & Sons, Ltd. PERFORMANCE BIAS (Systematic differences in care pr ovided apart from the intervention being evaluated) None of the studies reported the use of preoperative adjuvant chemotherapy or radiation. They all used prophylactic antibiotics, and all but two (Brownson 1992, Bucher 2003 ) de scribed the two allocation groups as being equal according to gender, age, types of operation, and diagnosis. Another one ( Tabusso 2002) indicated a difference between the all ocation groups with the age, haemoglobin level and serum albumin. No relevant performance bias was thus detected BLINDING One trial ( Fillmann 1995) described as a double-blind, in which orange juice was used as placebo, must be considered only to the surgeon due to the differences in taste for the patient between the intervention and the control. One study ( Burke 1994 ) was described as a single-blind study, as the surgeons were aware of allocation of patients to bowel preparation. The rest of the studies contained no mention of blinding methods. ATTRITION BIAS (Systematic differences in withdrawals from the trial) Brownson 1992, Bucher 2003, Tabusso 2002, Fa-Si-Oen 2003did not describe withdrawals or dropouts. Burke 1994 had 9.1% (17/ 186 patients) withdrawal, and no dropout; Santos 1994 , with 5% (8/157 patients) withdrawal, and no dropout, and Zmora 2003 8.6% (35/415). Two trials (Fillmann 1995, Miettinen 2000) described that all patients completed the study. The author of Fillmann 1995 supplied this information on written request. DETECTION BIAS (Systematic differences in outcomes assess- ment) No studies described any kind of concealment of assessment was described, except for the blinding procedure in the Fillmann-study ( Fillmann 1995). ( Burke 1994) measured the incidence of anastomotic leakage in the first half of the study by performing water soluble contrast enemas in all patients. In the second half of the study, enema was used on clinical suspicion of leakage due to the experience that two of the six l eaks on day 7 after surgery occurred immediately after administration of the routine water-soluble contrast enema. The contrast enema was used on clinical suspicion in four trials ( Burke 1994; Miettinen 2000; Santos 1994; Fillmann 1995). For the diagnostics of the various outcomes, the trials: a) did not describe the methodology (Brownson 1992, Bucher 2003 , Tabusso 2002, Fa-Si-Oen 2003); b) used clinical symptoms and laboratory results ( Burke 1994 , Santos 1994); c) used laboratory results in patients inwh om the clinical diagnosis was unclear ( Fillmann 1995); d) described all of the meth ods used for diagnosing the complica- tions ( Miettinen 2000, Zmora 2003). As stated in the beginning of this section, none of the studies used an intention to treat analysis. Eight patients were excluded after randomisation from one study ( Santos 1994), 17 from another (Burke 1994), and 35 from a third (Zmora 2003). In two studies no patients were excluded ( Fillmann 1995, Miettinen 2000). The others, gave no information on exclusion (Brownson 1992, Bucher 2003 , Tabusso 2002, Fa-Si-Oen 2003). Effects of interventions Nine randomised controlled trials including a total of 1592 pa- tients, of whom 789 were allocated for mechanical bowel prepa- ration (Group A), and 803 for no bowel preparation (Group B) prior to elective colorectal surgery were included. The results of each outcome were: PRIMARY OUTCOMES: 1) Anastomotic leakage - stratified: A) Low anterior resection: 9.8% (11 of 112 patients in Group A) compared to 7.5% (9 of 119 patients in Group B); Peto OR 1.45, 95% CI: 0.57 to 3.67 (non-significant) - no statistical heterogene- ity ( Burke 1994, Miettinen 2000, Santos 1994, Zmora 2003); B) Colonic surgery: 2.9% (11 of 367 patients in Group A) com- pared to 1.6% (6 of 367 patients in Group B) ; Peto OR 1.80, 95% CI: 0.68 to 4.75 (non-significant) - no statistical heterogene- ity ( Burke 1994, Miettinen 2000, Santos 1994, Zmora 2003); 2) Overall anastomotic leakage: Overall anastomotic leakage: 6.2% (48 of 772 patients in Group A) compared to 3.2% (25 of 777 patients in Group B); Peto OR 2.03, 95% CI: 1.276 to 3.26 (p=0.003) - no statistical hetero- geneity. (Brownson 1992, Bucher 2003, Burke 1994, Fillmann 1995 , Santos 1994 , Miettinen 2000 , Tabusso 2002 , Zmora 2003 ,Fa-Si-Oen 2003); SECONDARY OUTCOMES: 3) Mortality: 1% (5of 509patients in Group A) compared to 0.6% (3 of 516 patients in Group B); Peto OR 1.72, 95% CI: 0.43 to 6.95 (non-significant) - no statistical heterogeneity ( Burke 1994, Fillmann 1995, Miettinen 2000, Santos 1994, Zmora 2003); 4) Peritonitis: 5.7% ( 16 of 278 patients in Group A) compared to 2.5% (7 of 275 patients in Group B); Peto OR 2.28, 95% CI: 0.99 to 5.25) (p=0.05) - no statistical heterogeneity (Brownson 1992, Fillmann 1995, Miettinen 2000, Tabusso 2002); 5) Reoperation: 4.0% ( 16 of 393 patients in Group A) compared to 2.2% (9 of 392 patients in Gr oup B); Peto OR 1.80, 95% CI: 0.81 to 3.98) (non-significant) - no statistical heterogeneity (Bucher 2003 , Burke 1994 , Fillmann 1995 , Miettinen 2000 , Santos 1994,Tabusso 2002); 6) Wound infection: 7.4% (59 of 789 patients in Group A) com- paredto 5.4% (43of 803 patientsin GroupB); PetoOR 1.46, 95% CI: 0.97 to 2.18 (p=0.07) - no statistical heterogeneity (Brownson 1992, Bucher 2003, Burke 1994, Fillmann 1995, Miettinen 2000, Santos 1994, Tabusso 2002, Zmora 2003); 7) Infectious extra-abdominal complication: 8.3% ( 14 of 168 patients in Group A) compared to 9.4% (15 of 159 patients in Group B); Peto OR, 95%: 0.87 (0.41 to 1.87) (non-significant) - no statistical heterogeneity ( Fillmann 1995, Miettinen 2000); 7Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaborat ion. Published by John Wi ley & Sons, Ltd. 8) Non-infectious extra-abdominal complication: 16.8% ( 73 of 433 patients in Group A) compared to 16.1% (71 of 439 patients in Group B); Peto OR 1.19, 95% CI: 0.61 to 2.32 (non-signifi- cant) - no statistical heterogeneity ( Burke 1994, Fillmann 1995, Miettinen 2000); 9) Surgical site infection: 9.8% (31 of 325 patients in Group A ) compared to 8.3% (27 of 322 patients in Group B); Peto OR 1.20, 95% CI: 0.70 to 2.05 (non-significant) - no statistical het- erogeneity ( Miettinen 2000, Zmora 2003); SENSITIVITY ANALYSES: Applying the random effects model in the only statistical signifi- cant outcome still shows significant difference in f avour of avoid- ing cleansing. OR 2.09, CI: 1.16 to3.78, p =0.01. 10) Excluding the four studies where the allocation procedure was considered unclear did not change the Peto OR substantially for the two clinical most important outcomes (anastomosis leakage and wound infection), although the significance disappeared due to the smaller total the sample size (p = 0.1 and 0.14 respectively). 11) Excluding the two studies only presented as abstracts substan- tially changed neither the Peto OR nor the level of significance for the two analysed outcomes. 12) Excluding the study which included children did not change the significant higher incidence of anastomotic leakage in the me- chanical bowel preparation group, but the potential negative effect of cleansning on wound infection became smaller. 13) Excluding the studies that included patients without anasto- mosis for the outcome anastomosis leakage, the Peto OR was 2.14 (p = 0.03) compared with the Peto OR of 2.29 (p = 0.002) before these studies were excluded. There was no substantially difference for the wound infection outcome. D I S C U S S I O N In 1987 Irving ( Irving 1987) questioned the efficacy of mechanical bowel cleansing. The study was criticised by the editor (Johnston 1987 ): “the paper which challenges accepted surgical practice, is a veritable little bomb of a paper, brief, iconoclastic, and disrespect- ful of hallowed tradition in colorectal surgery”. At th at time, the mechanical bowel preparation was an incontestable routine - and still is according to guidelines from some surgical associations and scientific societies (ASCGBI 2001; Moore 1999; SIGN 1997), while other guidelines are now more up to date ( Kronborg 2002). The nine included trials were all prospective andrandomised. Typ- ically for studies of surgical practice, the allocation procedure was not very well described, but was considered adequate in half of the studies. Most of the studies were performed before the im- portance of allocation concealment ( Schulz 1996) became general knowledge. Only one of the studies trie d to include some kind of blinding ( Fillmann 1995) - an almost impossible task in trials of this kind. Despite these methodological flaws, the included stud- ies must be considered of such a scientific value that their conclu- sions should be taken into consideration, when trying to answer the question stated under ’objectives’. We found no convincing evidence that mechanical bowel prepa- ration before elective colorectal surgery reduces the incidence of postoperative complications. When looking at the primary out- come - anastomosis leakage - mechanical bowel preparation was dangerous when looking at colorectal surgery as a whole (statis- tically significant result). The subgroup analyses did not alter the direction of association, although the statistical significance disap- peared. The outcome anastomosis leakage was split into leakage after low anterior resection and leakage after colonic surgery. It was only possible to obtain results from four authors ( Burke 1994, Santos 1994 , Miettinen 2000 , Zmora 2003 ). After this stratification, the results tended to favour the group without mechanical bowel preparation. Some of the studies included patients in whom bowel continuity was not restored when analysing the outcome anastomosis leakage ( Tabusso 2002; Fillmann 1995; Santos 1994). Because the number of non-anastomotic patients were equally distributed between the groups so we do not feel this potential bias to be of significance. None of the studies included an intention to treat-analysis nor had any of theauthors calculated the sample size before the study. Seven of the studies must be considered underpowered from the begin- ning - only the Peruvian study ( Tabusso 2002) showed its own significance in favour of no cleansing. In this respect, the meta- analysis is a good tool, and when there is no he terogeneity among the studies the overall result can be accepted as valid. Allthough no statistical heterogeneity was found between the outcomes of the individual studies, some methodological and clinical hetero- geneity exists. Wheth er or not this should modify the conclusions is debatable. We have tried with sensitivity-analysis to elucidate the consequences of th e heterogeneity, and none of the analysis led to the conclusion that preparation would be of benefit for the patient. The significance for the primary outcome although dis- appers in some of the analyses, but the te ndency is still strong, and always in the same direction - preparation might lead to more anastomotic leakage. When performing the sensitivity-and sub- group-analyses the reduced volume of material makes the analyses statistically underpowered. This increases the risk of a type II error when evaluating the primary outcome. The Peto Odds Ratio re- mains almost unchanged during the sensitivity-analyses although the significance disappears. This strongly supports the conclusion: Mechanical bowel cleansing leads to more anastomotic dehiscence in colorectal surgery. A stratified analysis between colonic and rectal surgery was only feasible for four studies, andthe results were inconclusive, although the tendency goes in the same direction as the overall results - bowel preparation cannot be recommendedin patientsundergoing elective colorectal surger y. 8Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaborat ion. Published by John Wi ley & Sons, Ltd. [...]... 2.15] Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaboration Published by John Wiley & Sons, Ltd 19 Analysis 1.1 Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 1 Anastomosis leakage stratified for colonic or rectal surgery Review: Mechanical bowel preparation for elective colorectal surgery Comparison: 1 Mechanical bowel. .. preparation Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaboration Published by John Wiley & Sons, Ltd 1 10 100 Favors control 21 Analysis 1.2 Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 2 Overall anastomotic leakage for colorectal surgery Review: Mechanical bowel preparation for elective colorectal surgery Comparison: 1 Mechanical. .. Favors preparation 1 10 100 Favors control Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaboration Published by John Wiley & Sons, Ltd 20 Review: Mechanical bowel preparation for elective colorectal surgery Comparison: 1 Mechanical bowel preparation versus no preparation Outcome: 1 Anastomosis leakage stratified for colonic or rectal surgery. .. 0.74); I2 =0.0% Test for overall effect: Z = 0.79 (P = 0.43) 0.2 0.5 Favours treatment 1 2 5 Favours control Mechanical bowel preparation for elective colorectal surgery (Review) Copyright © 2008 The Cochrane Collaboration Published by John Wiley & Sons, Ltd 33 Review: Mechanical bowel preparation for elective colorectal surgery Comparison: 1 Mechanical bowel preparation versus no preparation Outcome:... 16 (Preparation) , 9 (No preparation) Heterogeneity: Chi2 = 3.84, df = 4 (P = 0.43); I2 =0.0% Test for overall effect: Z = 1.44 (P = 0.15) 0.01 0.1 1 Favours preparation 10 100 Favours control Analysis 1.6 Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 6 Wound infection Review: Mechanical bowel preparation for elective colorectal surgery Comparison: 1 Mechanical bowel preparation. .. events: 5 (Preparation) , 3 (No preparation) Heterogeneity: Chi2 = 1.55, df = 1 (P = 0.21); I2 =35% Test for overall effect: Z = 0.77 (P = 0.44) 0.01 0.1 1 Favours preparation 10 100 Favours control Analysis 1.4 Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 4 Peritonitis Review: Mechanical bowel preparation for elective colorectal surgery Comparison: 1 Mechanical bowel preparation. .. 73 (Preparation) , 71 (No preparation) Heterogeneity: Chi2 = 1.43, df = 3 (P = 0.70); I2 =0.0% Test for overall effect: Z = 0.45 (P = 0.65) 0.01 0.1 1 Favours preparation 10 100 Favours control Analysis 1.9 Comparison 1 Mechanical bowel preparation versus no preparation, Outcome 9 Surgical site infections Review: Mechanical bowel preparation for elective colorectal surgery Comparison: 1 Mechanical bowel. .. df = 3 (P = 0.62); I2 =0.0% Test for overall effect: Z = 0.78 (P = 0.43) 0.01 0.1 1 Favors preparation Review: 10 100 Favors control Mechanical bowel preparation for elective colorectal surgery Comparison: 1 Mechanical bowel preparation versus no preparation Outcome: 1 Anastomosis leakage stratified for colonic or rectal surgery Study or subgroup Preparation n/N No preparation Peto Odds Ratio n/N Peto,Fixed,95%... Review: Mechanical bowel preparation for elective colorectal surgery Comparison: 1 Mechanical bowel preparation versus no preparation Outcome: 10 Sensitivity analysis 1 - Studies with dubious randomisation procedure excluded Study or subgroup Preparation n/N No Preparation Peto Odds Ratio n/N Weight Peto,Fixed,95% CI Peto Odds Ratio Peto,Fixed,95% CI 1 Overall anastomotic leakage for colorectal surgery. .. 27 Review: Mechanical bowel preparation for elective colorectal surgery Comparison: 1 Mechanical bowel preparation versus no preparation Outcome: 10 Sensitivity analysis 1 - Studies with dubious randomisation procedure excluded Study or subgroup Preparation No Preparation n/N Peto Odds Ratio n/N Peto,Fixed,95% CI Peto Odds Ratio Peto,Fixed,95% CI 1 Overall anastomotic leakage for colorectal surgery Fillmann . is necessary before elective colorectal surgery should be reconsidered. 2Mechanical bowel preparation for elective colorectal surgery (Review) Copyright. Preoperative mechanical bowel preparation before colorectal sur g ery does not reduce anastomotic leakage. Preoperative mechanical bowel preparation before colorectal