|Home | About | Journals | Submit | Contact Us | Français|
Despite evidence that mechanical bowel preparation (MBP) does not reduce the rate of postoperative complications, many surgeons still use MBP before surgery. We sought to appraise and synthesize the available evidence regarding preoperative bowel preparation in patients undergoing elective colorectal surgery.
We searched MEDLINE, EMBASE and Cochrane Databases to identify randomized controlled trials (RCTs) comparing patients who received a bowel preparation with those who did not. Two authors reviewed the abstracts to identify articles for critical appraisal. We used the methods of the United States Preventive Services Task Force to grade study quality and level of evidence, as well as formulate the final recommendations. Outcomes assessed included postoperative infectious complications, such as anastomotic dehiscence and superficial surgical site infections.
Our review identified 14 RCTs and 8 meta-analyses. Based on the quality and content of these original manuscripts, we formulated 6 recommendations for various aspects of bowel preparation in patients undergoing elective colorectal surgery.
Taking into account the lack of difference in postoperative infectious complication rates when MBP is omitted and the adverse effects of MBP, we believe that, based on the literature, MBP before surgery should be omitted.
En dépit de données probantes indiquant que la préparation mécanique de l’intestin (PMI) ne réduit pas le taux de complications postopératoires, beaucoup de chirurgiens utilisent toujours la PMI avant l’intervention. Nous avons cherché à évaluer et résumer les données probantes disponibles sur la préparation préopératoire de l’intestin chez les patients qui subissent une chirurgie colorectale élective.
Nous avons effectué une recherche dans MEDLINE, EMBASE et les bases de données Cochrane pour repérer les essais contrôlés randomisés (ECR) où l’on a comparé les patients qui ont reçu une préparation de l’intestin à ceux qui n’en ont pas reçu. Deux auteurs ont analysé les résumés pour repérer les articles à soumet-tre à une évaluation critique. Nous avons utilisé les méthodes du Groupe de travail sur les services de prévention des États-Unis (United States Preventive Services Task Force pour évaluer la qualité de l’étude et le niveau des éléments probants, et pour formuler des recommandations finales. Les résultats évalués ont inclus les complications infectieuses postopératoires comme la déhiscence de l’anastomose et les infections superficielles du site chirurgical.
Notre étude a permis de repérer 14 ECR et 8 méta-analyses. Compte tenu de la qualité et du contenu de ces manuscrits originaux, nous avons formulé 6 recommandations portant sur divers aspects de la préparation de l’intestin chez les patients qui subissent une chirurgie colorectale élective.
Comme il n’y avait pas de différence au niveau des taux de complications infectieuses postopératoires lorsque la PMI est omise et compte tenu des effets indésirables de la PMI, nous sommes d’avis, en nous basant sur les publications, qu’il faudrait abandonner la PMI avant les interventions chirurgicales.
Mechanical bowel preparation (MBP) before elective colorectal surgery has been the standard in surgical practice for over a century. It is believed that MBP decreases intraluminal fecal mass and presumably decreases bacterial load in the bowel. It has been argued that this decrease in fecal load and bacterial contents reduces the rates of infectious postoperative complications, such as anastomotic dehiscence. These theories, however, have been based largely on clinical experience and expert opinion.1,2 The first study to challenge the need for MBP was published in 1972.3 Since then, there has been mounting level-I evidence indicating that MBP does not reduce the rate of postoperative complications, including anastomotic failure.4–7
Despite this evidence, a survey of colorectal surgeons in the United States published in 2003 revealed that 99% of the surgeons surveyed used MBP before surgery.8 In 2006, a multinational audit of 1082 patients from 295 hospitals in Europe and the United States revealed that 86%–97% (mean 94%) of patients received preoperative MBP.9 These surveys indicate that a large gap exists between the evidence surrounding the use of MBP and surgeon practices. It is unclear why surgeons have not changed practice to parallel the best evidence, since prescribing MBP also results in unnecessary costs (i.e., preadmission of patients, nursing care) as well as increased risks and discomfort for patients. Communication with local experts has indicated that the major hurdles may include lack of awareness of the evidence and, simply, reluctance to change.
There is some evidence that guidelines can be used as a knowledge translation strategy to target physician awareness.10 This guideline has been prepared for general surgeons and general surgery residents who are involved in the preoperative management of patients undergoing elective colorectal surgery. The question addressed by this guideline is this: In patients undergoing elective colorectal surgery, do MBP, dietary modifications and enemas reduce the risk of infectious complications, such as superficial surgical site infections (SSIs) and anastomotic leaks?
Bowel preparation before elective colorectal surgery can include a variety or combination of interventions. For the purposes of this guideline, MBP refers to the use of an oral laxative solution used to cleanse the colon of fecal contents (e.g., polyethylene glycol, sodium phosphate, sodium picosulphate, magnesium citrate). Preoperative dietary modifications and the use of enemas are also addressed as separate components of bowel preparation. The use of a normal diet refers to allowing patients a regular, unrestricted diet on the day before surgery. This can be replaced with a clear-fluid diet, which restricts patients from eating solid food. An enema is the administration of liquid in the rectum to evacuate stool.
We performed 2 searches with the assistance of a medical librarian. The first search identified articles evaluating postoperative complications in patients who did and did not receive bowel preparation (including MBP, dietary restrictions and enemas). The second search identified articles describing adverse effects related to the use of MBP. These search strategies complete with medical subject headings are outlined in Table 1.
We searched MEDLINE, EMBASE and Cochrane databases to identify relevant articles published between January 1950 and February 2009 that compared adult patients who received bowel preparation or no bowel preparation and reported postoperative infectious complications as an outcome (search 1). The search was limited to randomized controlled trials (RCTs) involving adult human participants using the sensitivity strategy of Robinson and Dickersin.11 We excluded nonrandomized controlled trials and studies including patients undergoing emergency colorectal surgery.
We also searched MEDLINE, EMBASE and Cochrane databases to identify relevant articles pertaining to adverse effects (search 2). The search strategy was not limited to publication type. We manually searched the reference lists of selected manuscripts from each literature search to further identify relevant research studies. Two reviewers (C.E., S.S.F.) independently assessed all titles and abstracts to select the studies to be included in this guideline. Dis-agreement on selection was resolved by consensus.
Quality appraisal was performed independently by 2 authors (C.E., S.S.F.). The selected manuscripts were reviewed and a quality assessment was performed using the criteria of the U.S. Preventive Services Task Force (USPSTF).12 The RCTs received a good rating provided patients in the intervention and control groups were comparable at the start of the trial, there was no crossover between the 2 groups, minimum follow-up of 80% was reported, interventions were clearly defined, well-defined and reproducible outcome assessments were used equally in both groups, outcome assessors were blinded, intention-to-treat analysis was employed and appropriate attention was given to confounders in the analysis. Studies were deemed to be of poor quality if they had any one of the following: gross differences between the intervention and control groups at the start of the study, greater than 10% crossover between the 2 groups, substantial (> 20%) loss to follow-up, lack of a power calculation, or interventions that were not clearly defined. Studies with minor methodological flaws received a fair rating. Meta-analyses received a good rating if they were published within the last 3 years, included a comprehensive literature search, duplicated study selection and/or data extraction, used relevant selection criteria, provided characteristics of the included studies, documented and used a quality assessment to formulate conclusions, used statistical methods to combine study findings described (i.e., pooled analysis, tests for heterogeneity), assessed the likelihood of publication bias and stated conflicts of interest.
After critical appraisal of the methodology and evidence of the included studies, we made recommendations using the criteria established by the USPSTF.12 Outcomes assessed included anastomotic dehiscence and superficial SSIs. These outcomes were reviewed for all patients undergoing elective colorectal surgery as well as for the following subgroups: patients undergoing low anterior resections with or without diverting ileostomies and patients undergoing laparoscopic colorectal surgery. Recommendations are also made regarding preoperative dietary modifications and the use of preoperative enemas. Finally, the Canadian Society of Colon and Rectal Surgeons endorsed this guideline.
Our search identified 14 unique RCTs.4,5,13–24 One trial15 was published twice and included only once. Another trial24 was published as both an interim and final analysis; we included only the final analysis. Two trials published subgroup analyses as separate manuscripts and were excluded from further review to eliminate duplicate results.4,19 A summary of our quality assessment of the RCTs is shown in Table 2. We did not assess the quality of 2 trials15,17 because they were not published in English.
The literature review identified 8 meta-analyses.6,25–31 These meta-analyses reported different combinations of the 14 published RCTs. The largest meta-analysis published in 2009 combined the results of the 14 RCTs.31 The Cochrane review was published in 2003 and was updated in 2005.26 We included the most current version. The Cochrane review was also published in another source separately by the same authors, and we excluded this duplicate publication. A summary of the quality assessment of these 8 meta-analyses is shown in Table 3.
All 14 trials compared anastomotic leak rates in patients receiving MBP and those not receiving MBP. The results for anastomotic leak rates in these trials are summarized in Table 4. Two of the 14 trials found significant differences in anastomotic leak rates in favour of the omission of MBP.13,17 The other 12 trials found no significant differences in the anastomotic leak rates. Two of these trials were large and are described in further detail below.22,23 The main flaw in the other trials was that they were underpowered.
An RCT by Contant and colleagues22 published in 2007 was a multicentre trial where investigators from 13 hospitals in the Netherlands randomly assigned 670 patients to receive MBP and 684 patients to no MBP. Those patients receiving MBP were prescribed either polyethylene glycol with bisacodyl or a sodium phosphate solution. There was no significant difference in anastomotic leaks (difference 0.6%, 95% confidence interval [CI] −1.7% to 2.9%, p = 0.69).22 This was a fair-quality RCT with one of its strengths being its large sample size. However, like many of the RCTs performed on this topic, outcome assessment was not blinded. Furthermore, the 2 groups were not comparable at the beginning of the trial; there was a larger proportion of smokers and patients with inflammatory bowel disease in the MBP group.
In the next RCT by Jung and colleagues,23 all Swedish centres and 1 German colorectal unit participated. In all, 686 patients were randomly assigned to receive MBP and 657 patients to no MBP;23 47% of patients in the MBP group were prescribed a polyethylene glycol preparation and 48.5% received a sodium phosphate preparation. There were no significant differences between the 2 groups for the primary outcomes of cardiovascular, general infectious and surgical-site complications. Specifically, anastomotic dehiscence was seen in 2.3% of patients in the MBP group and 2.6% of patients in the no MBP group. Six patients in each group died (p = 0.94).23
The authors examined the generalizability of the results and potential selection bias by comparing study participants to those patients who were not enrolled in the study at 3 participating centres. They found no statistically significant differences in the demographics or the outcomes between these 2 groups of patients. This study did not show a significant difference but was also underpowered in that it was powered to detect a 50% difference in complication rates. However, it is unlikely that the addition of 57 patients (for a total of 1400 patients as required by the reported sample size calculation) would change the conclusion.23 For these reasons, this was not deemed a fatal flaw and we gave the trial a fair rating.
Our review of the included meta-analyses revealed that 1 meta-analysis provided no pooled data and reported only a descriptive analysis of the included studies.29 Of the remaining 7 meta-analyses, 4 reported statistically significant differences in the pooled results for anastomotic leakage.26–29 Of these 4 meta-analyses showing a difference, the largest and most recent was the Cochrane review published in 2005.26 Three of the 7 meta-analyses found no significant difference between the MBP and the no MBP groups.25,30,31 Of the 3 meta-analyses that reported no difference in anastomotic leak rates, 1 was the oldest review,30 including only 3 trials, and the other 225,31 were the most recent reviews.
The 2 most recent fair-quality meta-analyses were published by Guenaga and colleagues in 200526 as a Cochrane systematic review and by Slim and colleagues31 in 2009. The fair-quality review by the former group was an update of the first Cochrane review published in 2003 and included 9 trials with a total of 1592 patients.26 Of these patients, 789 were allocated to the MBP group and 803 to the no MBP group. The main outcome was anastomotic leakage; other outcomes evaluated included mortality, superficial SSIs, peritonitis and reoperation. The overall anastomotic leakage in both groups indicated that MBP was associated with a higher rate of anastomotic leakage (odds ratio [OR] 2.03, 95% CI 1.276–3.26, p = 0.003).26 The authors of this review concluded that MBP for patients undergoing elective colorectal surgery has not proven valuable and the procedure should be omitted as it may increase the risk of anastomotic dehiscence. The primary strength of this meta-analysis was the thorough discussion of the quality and methodology of the included articles.
The meta-analysis published by Slim and colleagues31 in 2009 included 14 trials with a total of 4859 patients and provided different results. This meta-analysis included substantially more patients because of the inclusion of the trials by Jung and colleagues23 and Contant and colleagues,22 which were published after the meta-analysis by Guenaga and colleagues.26 This meta-analysis was given a fair quality rating because conflicts of interest were not reported. In all, 2452 patients were randomly assigned to the MBP group and 2407 to the no MBP group. The outcomes reported were rates of anastomotic leakage and superficial SSIs. The pooled results revealed no significant difference in anastomotic leakage rates between the 2 groups with a fairly narrow 95% CI (OR 1.12, 95% CI 0.824–1.532, p = 0.46).31 Although these results differed from the results of the Cochrane review, these authors again concluded that there is no benefit to using MBP in patients undergoing elective colorectal surgery.31
All 14 RCTs included superficial SSIs as another end point, and these results are summarized in Table 5. In all 14 trials, there were no significant differences in the rates of superficial SSIs in the MBP and no MBP groups.4,5,13–24 One of the 7 meta-analyses reported a significant difference in superficial SSIs between the 2 groups, with an increased rate of superficial SSIs in patients who received MBP (difference 3.4%, 95% CI −1.6% to 8.4%, p = 0.002).30 The other 6 meta-analyses found no difference in the rates of superficial SSIs when comparing patients who did and did not receive MBP.6, 25–8 Guenaga and colleagues26 reported rates of superficial SSIs as 7.4% (59/789) in the MBP group and 5.4% (43/803) in the no MBP group (OR 1.46, 95% CI 0.97–2.18, p = 0.07). In the meta-analysis by Slim and colleagues,31 the rate of superficial SSIs in the MBP group was 9.5% compared with 8.3% in the no MBP group (OR 1.17, 95% CI 0.96–1.44, p = 0.11).
It has been well documented that the risk of anastomotic dehiscence is greater following low colorectal or coloanal anastomoses, and these low anastomoses have been associated with high rates of morbidity and mortality.32 For this reason, many surgeons performing these operations opt to protect the anastomosis with a diverting stoma. The use or omission of MBP in patients undergoing low anterior resection (LAR) with or without diverting stoma in particular poses a difficult dilemma and raises important concerns. Surgeons may hesitate to omit MBP in these patients because it would leave a column of stool between the stoma and the anastomosis. In the event that such patients experience an anastomotic leak, there would still be a risk of fecal contamination despite the anastomosis having been protected. In patients who do not receive a diverting stoma, surgeons may also be concerned with the potentially increased morbidity associated with an anastomotic leak.
Patients undergoing an LAR with a diverting ileostomy were poorly represented in the 14 RCTs included in our review for 2 main reasons. Some RCTs (2 of 14) excluded patients who underwent LAR or LAR with anastomoses below the peritoneal reflection.19,20 Others (5 of 14) excluded patients who had planned diverting stomas.4,18,22–24 Finally, in some RCTs (3 of 14) the level of the anastomosis and whether the patients had diverting stomas was unclear.13,15,17
Five RCTs included patients undergoing LAR, and the results of 4 of them4,5,14,16 were included in a subgroup analysis reported in the Cochrane review.26 In one of these RCTs,14 whether patients received diverting stomas was not mentioned. In another,4 patients with diverting stomas were excluded, and the other 2 studies5,16 clearly state that patients did not receive diverting stomas. When the results of this subgroup of LAR patients from these 4 RCTs were pooled in the Cochrane review, the rate of anastomotic leakage for LAR was 9.8% (11 of 112) in patients in the MBP group compared with 7.5% (9 of 119) in patients in the no MBP group.26 The OR was 1.45 (95%CI 0.57–3.67, p = 0.40) and was not statistically significant, with wide 95% CIs, likely because of the small sample size.26
There is 1 RCT published by Platell and colleagues21 that included a substantial proportion of patients having LAR with diverting stomas. This study was underpowered to show equivalence, although it did reveal statistically significant differences in some secondary outcomes. Therefore, we gave this study a fair rating. Patients were randomly assigned to receive oral MBP (polyethylene glycol) or a single phosphate enema only. For the purpose of this guideline, we considered the enema group to be the no MBP group because none of these patients received an oral MBP. In all, 147 patients were randomly assigned to MBP and 147 patients to no MBP.21 Sixty-four percent (94 of 147) of patients in the MBP group and 55% (81 of 147) of patients in the no MBP group underwent an anterior resection.21 Furthermore, 39% (57 of 147) of patients in the MBP group and 32% (47 of 147) of patients in the no MBP group had a diverting stoma. The authors stated that patients undergoing a low or ultra-low anterior resection were “routinely covered with a defunctioning loop ileostomy.”21 There were 3 anastomotic leaks in the MBP group and 7 in the no MBP group (2% and 4.8%, respectively, p = 0.20).21 However, none of the patients in the MBP group compared with the 6 patients in the no MBP group required reoperation (0% and 4.1%, respectively, p = 0.013).21 These results led to the trial being closed prematurely. The mortality rate in the MBP group was 2.7% compared with 0.7% in the no MBP group (OR 1.62, 95%CI 0.45–36.98, p = 0.18). There was no significant difference in the rate of superficial SSIs between the MBP and no MBP groups.21
These results are in contrast to those of all the other RCTs and meta-analyses. However, this trial differs in that patients in the no MBP group received an enema. To make further conclusions about the use of enemas in the pre-operative preparation of patients undergoing elective colorectal surgery, an RCT examining only the enema intervention would be required. We included this trial in this guideline because many surgeons who disagree with the omission of MBP cite this article as an example of increased complications when no MBP is prescribed. However, as demonstrated above, it is important to distinguish this study from the others as it compares a different intervention in addition to comparing MBP versus no MBP.
Although there are no studies examining the effect of MBP in patients undergoing elective laparoscopic surgery, the evidence presented in this guideline likely can be extrapolated to this population. There is no clinical reason why patients having laparoscopic colorectal surgery would be more likely to develop postoperative infectious complications. Some argue that MBP may be required in patients with small tumours that may not be appreciated laparoscopically, thus requiring intraoperative colonoscopy, but preoperative tattooing of the lesion would obviate such a need. Some surgeons have also indicated that the unprepared colon may be slightly heavier and thus difficult to manipulate laparoscopically.
Our search strategy identified 1 RCT of fair quality that examined the adverse histological effects of MBP. There were many other citations in the form of letters to the editor and case reports describing the adverse effects related to MBP. The RCT published by Bucher and colleagues33 reported the histological changes in intestinal mucosa in 25 patients who had MBP with polyethylene glycol compared with 25 patients who did not receive MBP. There was a significant difference in the loss of superficial mucous (p < 0.001), loss of epithelial cells (p < 0.01), edema of the lamina propria (p < 0.01), lymphocyte infiltration (p < 0.02) and polymorphonuclear cell infiltration (p < 0.02) when the 2 groups were compared. These changes were all more frequent in those patients who had received MBP. Although it is unclear if these morphological changes are clinically relevant, they could potentially result in bacterial translocation and anastomotic disruption.34,35
We reviewed 13 other selected articles describing the adverse effects of MBP.36–48 The details of these manuscripts can be seen in Table 6. In brief, these case reports revealed that many of the different types of MBP, such as sodium picosulfate, polyethylene glycol, sodium phosphate and magnesium citrate, were associated with adverse effects.37–43,45,47,48 The primary adverse effects were related to electrolyte and volume disturbances in both healthy patients and patients with underlying cardiac or renal disease. Furthermore, these electrolyte disturbances led to seizures, syncope, coma and even death in some patients. Finally, there have also been reports of MBP-associated ischemic colitis, pancreatitis and esophageal perforation.36,44,46
None of the 14 RCTs included in this review performed a direct comparison of different dietary modifications before surgery. Table 7 describes the specific MBP, dietary modifications and enemas that were used in each group in each RCT. Nine of the 14 RCTs stipulated no dietary restrictions before surgery, and patients in the no MBP arm received a normal or low-residue diet on the day before surgery. Since most of these trials allowed patients in the no MBP arm to have a normal diet before surgery and these patients did not have increased postoperative infectious complications, it is likely safe to omit dietary modifications in the preoperative management of patients undergoing elective colorectal surgery.
Again, none of the 14 RCTs included in this review performed a direct comparison of enema versus no enema before surgery. Three of the 14 RCTs prescribed enemas for left-sided or rectal resections in patients in the no MBP group.4,18,21 Also, in 5 of the 14 RCTs, patients in the MBP group also had an enema.4,5,18,23,24 Applying this evidence, it is difficult to draw conclusions and make recommendations regarding the use or omission of enemas in patients undergoing elective colorectal surgery.
Most of the evidence supports the omission of MBP and reveals that MBP is not associated with an increased risk of anastomotic dehiscence. Furthermore, there appears to be no difference in other postoperative complications, such as superficial SSIs. Based on the population of patients in these trials, these results can be applied to patients undergoing elective, open right-sided and left-sided colorectal resections. Mechanical bowel preparation is generally safe, but it has been associated with serious complications in patients with existing cardiac and renal disease as well as previously healthy patients. Furthermore, most patients find MBP to be unpleasant. Thus, the use of MBP has not been shown to be beneficial, but rather has been shown to be associated with rare but serious adverse effects.
There is less evidence regarding patients undergoing LAR with or without a diverting ileostomy. After thorough assessment of the included RCTs, only 1 provided a comparison of MBP and no MBP in this specific population, and all others excluded this group of patients. This fair-quality RCT revealed that patients receiving MBP had lower rates of anastomotic dehiscence, but this was not statistically significant.21 This study was designed to be an equivalence study but was ended early owing to the need for reoperations in patients who experienced a leak. However, all patients in the no MBP group received a phosphate enema, which might account for the differences seen between the 2 groups. Furthermore, the Cochrane review included a subgroup analysis of patients undergoing LAR and showed no statistically significant difference in anastomotic leak rates between the MBP and no MBP groups.26
Patients undergoing laparoscopic colorectal resections are not included in any of the RCTs discussed in this guideline. The results from the included RCTs where patients underwent open procedures, however, likely can be generalized to this patient population.
A synthesis of the level-I evidence reveals that there is good evidence supporting the omission of MBP in the preoperative management of patients undergoing elective right-sided and left-sided colorectal surgical resections (grade A recommendation). Examining the data specifically for patients undergoing LAR with or without diverting stomas has revealed that there is insufficient evidence to support or refute the omission of MBP in the preoperative management of these patients (grade I recommendation). There is no specific evidence regarding patients undergoing laparoscopic colorectal surgery. Therefore, there is insufficient evidence to support or refute the omission of MBP in the preoperative management of patients undergoing elective laparoscopic colorectal surgery (grade I recommendation).
Although there is some heterogeneity when evaluating dietary modifications before elective colorectal surgery, most RCTs allowed patients in the no MBP group to consume a regular diet until midnight on the day before surgery. These interventions have revealed that there is fair evidence to recommend normal diet until midnight the day before surgery in the preoperative management of patients undergoing elective colorectal surgery (grade B recommendation). Finally, there is insufficient evidence to support or refute the use of enemas in the preoperative management of patients undergoing elective colorectal surgery (grade I recommendation).
These recommendations are driven mostly by the 2 large RCTs22,23 and the 3 recent meta-analyses.25,26,31 Although the primary RCTs have not shown a statistically significant difference in postoperative complications when comparing the MBP and no MBP groups, the common flaw in these studies is inadequate sample size and power. The utility of the meta-analyses is directed at this particular problem. Furthermore, the reports surrounding adverse effects of MBP reveal that although complications are rare and more common in individuals with underlying cardiac and renal disease, these complications are extremely serious. Taking into account the lack of difference in postoperative infectious complication rates when MBP is omitted and the adverse effects of MBP, we believe that we are justified in making a strong recommendation based on the literature.
Competing interests: None declared.
Contributors: All authors helped design the study, review and article and approved its publication. Drs. Eskicioglu and Forbes acquired and analyzed the data. Dr. Eskicioglu wrote the article.