Figure. The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) flowchart: selection of relevant studies.
Slieker JC, Daams F, Mulder IM, Jeekel J, Lange JF. Systematic review of the technique of colorectal anastomosis: will a nonstandardized technique stand in evidence-based surgery? JAMA Surg.. 2013. doi:10.1001/2013.jamasurg.33.
eFigure. Key words used in the MEDLINE search.
Slieker JC, Daams F, Mulder IM, Jeekel J, Lange JF. Systematic Review of the Technique of Colorectal Anastomosis. JAMA Surg. 2013;148(2):190-201. doi:10.1001/2013.jamasurg.33
Author Affiliations: Departments of Surgery (Drs Slieker, Daams, Mulder, and Lange) and Neuroscience (Dr Jeekel), Erasmus Medical Center, Rotterdam, the Netherlands.
Many different techniques of colorectal anastomosis have been described in search of the technique with the lowest incidence of anastomotic leak. A systematic review of leak rates of techniques of hand-sewn colorectal anastomosis was conducted to provide a guideline for surgical residents and promote standardization of its technique. Clinical and experimental articles on colorectal anastomotic techniques and anastomotic healing published in the past 4 decades were searched. We included evidence on suture material, suture format, single- vs double-layer sutures, interrupted vs continuous sutures, hand-sewn vs stapled and compression colorectal anastomosis, and anastomotic configuration. In total, 3 meta-analyses, 26 randomized controlled trials, 11 nonrandomized comparative studies, 20 cohort studies, and 57 experimental studies were found. Results show that, for many aspects of the hand-sewn colorectal anastomosis technique, evidence is lacking. A single-layer continuous technique using inverting sutures with slowly absorbable monofilament material seems preferable. However, in contrast to stapled and compression colorectal anastomoses, the technique for hand-sewn colorectal anastomoses is nonstandardized with regard to intersuture distance, suture distance to the anastomotic edge, and tension on the suture. We believe detailed documentation of the anastomotic technique of all colorectal operations is needed to determine the role of the hand-sewn colorectal anastomosis.
Construction of a colorectal anastomosis is a hallmark of surgical training. However, although surgical residents can refer to key publications with evidence-based conclusions for many topics, mere imitation of an experienced surgeon traditionally is considered the basic source for the technique of hand-sewn colorectal anastomosis. The large variety of anastomotic techniques is one of the main difficulties in the interpretation of the literature. Anastomotic leak (AL) following colorectal resection is a major problem of surgical care, with an incidence between 3% and 19%.1- 4 Although accurate prediction of risk is impossible, certain factors are known to contribute to AL, including surgeon-related factors (eg, increased incidence of AL in a colorectal anastomosis constructed after hours5 and the positive role of specialization in reducing the complications of colorectal surgery6) and patient-related risk factors (eg, the inverse relationship between the height of the colorectal anastomosis from the anal verge and the leak rate7- 12). Decades of research have resulted in many studies investigating different techniques for constructing colorectal anastomosis in search of the safest method. Appreciating the conclusions from this extensive research is essential for the quality of colorectal surgery and for the resident being trained in colorectal surgery. Our aims were to perform a systematic review of all aspects of the technique of hand-sewn colorectal anastomosis and compare hand-sewn with mechanical colorectal anastomosis to provide a guideline for residents and promote standardization of the technique.
A literature search was conducted using MEDLINE, EMBASE, and Cochrane databases for studies published between January 1, 1970, and February 1, 2011, using the key words presented in the eFigure. The search was restricted to articles published in English, Dutch, and French. References in the selected publications were searched for additional studies.
Clinical as well as experimental studies were selected to address several aspects of the technique of hand-sewn colorectal anastomosis. These included:
suture format (size of suture bites, in-between distance of bites, suture tension, configuration of the bite, and inverting vs everting sutures),
single- vs double-layer colorectal anastomosis,
interrupted vs continuous sutures,
hand-sewn vs stapled colorectal anastomosis,
hand-sewn vs compression colorectal anastomosis, and
configuration of colorectal anastomosis (end-to-end [ETE], end-to-side, side-to-end, side-to-side, length of the side-limb, and length of the enterotomy).
Only clinical studies comparing 2 or more colorectal anastomotic techniques with regard to clinical AL were considered relevant. When only 1 comparative study was available on a particular subject, clinical cohort studies were added to the selection. Results were analyzed only if the study groups and results were clearly described with proper statistical analysis.
Experimental studies were selected when comparing 2 or more colorectal anastomotic techniques using objective measurements for anastomotic healing: AL, anastomotic bursting pressure (ABP), anastomotic breaking strength, histologic results, or collagen concentration.
When 2 studies were reported by the same institution, either the better quality study or the most recent publication was included. As with clinical studies, results were analyzed only if the study groups and results were clearly described with proper statistical analysis. However, the lack of statistical analysis of histologic findings in experimental studies was accepted.
Because the healing of small-bowel anastomoses is different and the incidence of AL is lower compared with large-bowel anastomoses, studies including both procedures without differentiating the results and statistical analysis were excluded. Ileocolic anastomoses after right hemicolectomy or ileocecal resection represent healing of the colon and were therefore included. Studies reporting radiologic AL without distinction of clinical AL were excluded, as were studies reporting only on emergency operations, children, and colo-anal anastomosis or pouches. Results of experimental studies measured directly after the construction of colorectal anastomosis were not taken into account because these do not reflect anastomotic healing.
Two physicians (J.C.S. and F.D.) entered data in a database following standard protocols. Seven factors were considered for clinical studies. These included:
first author and year of publication,
level of evidence (following the Centre of Evidence Based Medicine, University of Oxford),
number of patients,
location of anastomosis in the gastrointestinal tract,
definition of outcome by the authors (AL, clinical AL, and radiologic AL), and
results and statistical analysis.
Six factors were considered for experimental studies. These included:
number of animals per group,
outcome factors for anastomotic healing (AL, ABP, breaking strength, histologic results, or collagen concentration), and
The literature search identified 6168 articles; 1443 articles remained after duplicates were removed. The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) flowchart presented in the Figure shows the selection of studies: 117 were included in the systematic review. Included studies and their characteristics are listed in Tables 1, 2, 3, 4, 5, 6, and 7, together with all results of outcome measures. The results per research question are summarized herein.
Decades ago, several materials, such as silk, linen, catgut, polyglactin 910, and nylon, were commonly used for colorectal anastomosis. Today most gastrointestinal anastomoses, including colorectal anastomosis, are constructed with polydioxanone sutures. Ten experimental studies13- 22 were included. Results show that absorbable sutures compared with nonabsorbable or slowly absorbable sutures cause more tissue reaction13- 15; one of these studies13 showed that absorbable sutures dissolve too rapidly, influencing anastomotic strength. Multifilament compared with monofilament sutures cause more tissue damage and easier adherence of material within the interstices of multifilament sutures,16- 19 providing a basis for infection.121 Surprisingly, experimental studies on the healing of colorectal anastomosis constructed with polydioxanone sutures are scarce; only 2 studies17,20 were included, finding equal ABP and histologic characteristics between polydioxanone and polyglycolic acid. Noncomparative experimental studies121- 124 that did not meet the inclusion criteria for the present systematic review have shown that polydioxanone sutures possess all aspects considered important: monofilament, little histologic reaction, slowly absorbable with long preservation of strength, and low adherence of bacteria to the material.
New possibilities for the use of sutures coated with mesenchymal stem cells and doxycycline were explored in 2 experimental studies with promising, but not yet convincing, results.21,22 Two included randomized controlled trials (RCTs)23,24 on suture material failed to achieve a unanimous conclusion because of the small number of patients included and the different suture materials tested that are rarely used today.
In conclusion, on the basis of experimental studies, nonabsorbable or slowly absorbable monofilament sutures seem to be the first choice for colorectal anastomosis. However, there is no level 1 evidence to confirm this hypothesis.
Since Lembert125 described the construction of intestinal anastomoses in dogs using suture bites with 5-mm distance to the cut edge nearly 2 centuries ago, this aspect seems to have become less clear in surgical literature. One experimental study25 was found for this systematic review that investigated the difference in anastomotic strength in rats with sutures placed between 3 mm and 1.5 mm from the cut edges. Results showed lower breaking strength for small bites, measured at day 2. One RCT by Greenall et al26 reporting on the distance of the suture to the wound edge matched the inclusion criteria. They randomly allocated patients to have bowel sutures placed either 5 or 10 mm from the cut edges, with no significant differences in AL. Because it is not possible to extrapolate the distances used in a rat model to the clinical situation, we can only conclude from one level 1b RCT that distances of 5 and 10 mm from the cut edge will probably give adequate results.
Lembert described in 1826125 an in-between distance of approximately 1 cm between sutures. One experimental study conducted by Waninger et al27 investigating the distance between sutures in rats was included in our review. It concluded that a small distance between sutures (1.5 mm) improves apposition compared with a larger distance (2.5 mm). Neither clinical comparative studies nor cohort studies were found. Again, distances in a rat model are difficult to extrapolate to the patient. Because clinical studies on this topic are lacking, no precise maxim can be distilled from the literature.
In routine clinical practice, 2 undefined schools of thought seem to exist: the first believes that sutures should be tightened to prevent dehiscence of the anastomosis, and the second considers that sutures should be applied more loosely, allowing maximal perfusion of the cut edges. Only one rat study27 investigated this, with moderate tension giving the best histologic and microangiographic results. Whether tension on knots could influence the incidence of AL in a clinical setting has not been investigated for interrupted or continuous suturing. On the basis of the literature evaluated in the present review, nothing can be concluded on the proper tension on the thread or the knot.
Historically, all opinion leaders proposed their own configuration of gastrointestinal sutures. Anatomic apposition of all layers promoting primary healing was thought to be important. These days, most surgeons use a simple through-all-layers technique. From ex vivo studies,126,127 it is known that sutures through the mucosal layer do not contribute to anastomotic strength.
The present review included 2 experimental studies on rat colon, comparing histologic results of full-thickness sutures with those of serosubmucosal sutures. Houdart et al28 found no significant histologic differences, but Krasniqi et al29 found better histologic results for full-thickness sutures with equal anastomotic strength. No comparative clinical studies were found on the configuration of the bite. Because of this lack of evidence, we have included cohort studies,30- 34 reporting low rates of AL for both serosubmucosal and full-thickness suture formats (AL, 0%-4.4%).
We can only conclude, using scarce level 2b evidence from the cohort studies evaluated, that both serosubmucosal and full-thickness sutures seem to provide low rates of AL. It is clear that the configuration of the suture bite is considered of little interest in studies regarding AL.
Since the publication of Lembert,125 surgeons generally have advocated an inverting technique of gastrointestinal anastomosis because it is believed that protruding mucosa will lead to AL. However, in the 1960s, 2 clinical studies128,129 showed good healing of everting anastomoses with a low incidence of AL. Between 1960 and 1970, these 2 noncomparative studies were followed by many experimental publications comparing everting with inverting techniques. They failed to achieve a unanimous conclusion on anastomotic healing; however, they were consistent in showing that everting anastomoses cause more adhesions but less stenosis.35- 37,130- 134 All 3 experimental studies35- 37 published after 1970 included in the present review seem to show improved anastomotic healing for inverted anastomoses. The only clinical study matching the inclusion criteria was an RCT38 showing a 5-fold increased incidence of AL in patients receiving an everting colorectal anastomosis compared with those receiving an inverting colorectal anastomosis. No cohort studies matching our inclusion criteria were found. Therefore, on the basis of available experimental studies and a level 1b clinical study, there seems to be an advantage of inverting over everting colorectal anastomosis; nonetheless, level 1a evidence is lacking.
The technique developed by Lembert125 and later modified by Czerny135 is based on a double-layer inverting anastomotic technique. In the 19th and the greater part of the 20th centuries, this was the criterion standard for gastrointestinal anastomosis; in the second half of the 20th century, however, the single-layer anastomosis regained attention through the favorable results obtained by Halstead,136 Gambee,137 and Gambee et al.138 The 13 included experimental studies36,39- 50 came to the same conclusion: double-layer anastomoses are inferior to single-layer anastomoses because of increased inflammation and diminished circulation. One RCT51 matched the inclusion criteria, showing no significant differences in AL between single- and double-layer colorectal anastomosis in 92 patients. This RCT conducted a subgroup analysis of 25 low colorectal anastomoses, finding a significantly higher incidence of AL in colorectal anastomosis created with the double-layer technique. None of the 3 nonrandomized comparative studies47,52,53 included in this review found a significant difference in AL between the 2 techniques. In conclusion, these results add to the knowledge that single-layer anastomoses take significantly less time to construct and are less costly139 and are in favor of single anastomoses on the basis of level 1b evidence.
The question on whether to use interrupted or continuous sutures arose when single-layer anastomoses became common practice. Six experimental studies were included, showing equivocal results: better serosal apposition54 and blood flow in continuous sutures,55 with equal results on ABP and histologic examination.28,56,57,73 Randomized controlled trials investigating interrupted and continuous sutures for colorectal anastomosis are lacking; therefore, only 1 small, nonrandomized, comparative clinical study finding no significant differences was included,58 and noncomparative cohort studies were selected on continuous and interrupted suturing, finding equally low leak rates.31,59- 72 Clinical and experimental studies have not concluded that one technique is superior to the other, and a high level of evidence is lacking (limited here to level 2b); however, from a technical and time-consuming point of view, a continuous suture is preferable over interrupted sutures for creating colorectal anastomosis.
After the introduction of stapled colorectal anastomosis in the 1980s, both techniques have become prevalent, without defined indications but for the lower rectal anastomoses. Most surgeons apply both techniques, although often with a personal preference.
Thirteen RCTs74- 86 and 3 meta-analyses87- 89 were included in the present review. Lustosa et al87 published a Cochrane meta-analysis of 9 RCTs conducted between 1981 and 1991. In this group of 1233 patients, there was no significant difference in mortality, AL, strictures, or reoperation between stapled and hand-sewn colorectal anastomosis. An earlier meta-analysis,88 conducted in 1998, combined 13 RCTs concerning patients with colorectal anastomosis and found similar results: no significant differences in AL or mortality. The Cochrane review conducted by Choy et al89 included studies on colorectal anastomosis after right hemicolectomy. This review showed significantly less overall AL in the stapled group; however, when clinical AL was used as the only outcome measure, this difference did not reach statistical significance. An interesting subgroup analysis made by Friend et al82 found more AL in hand-sewn colorectal anastomosis when the anastomoses performed by residents were analyzed separately. Their conclusion was that stapling seems to have an advantage in less-experienced hands. Of 7 included nonrandomized cohort studies included in this review, 5 found no superiority of one technique.100- 104 Two studies found significantly more AL in stapled compared with hand-sewn anastomoses.98,99 However, one of these had significantly more patients with corticosteroids in the stapled group,99 while the other included 505 hand-sewn compared with 28 stapled colorectal anastomoses in 20 years.98 Thirteen experimental studies41,45,50,90- 97,105,106 included herein found results approximately similar to those in the clinical setting: no significant differences in AL, with equal or higher ABP in stapled colorectal anastomosis.
In conclusion, the field of hand-sewn vs stapled colorectal anastomosis has been well studied. On the basis of level 1a evidence, no superiority of stapled over hand-sewn colorectal anastomosis exists.
Denans described the first technique to create intestinal anastomoses by compression in 1827,140 followed by other devices, such as the Murphy button, in 1892.141 Today the biofragmentable anastomotic ring, made of absorbable polyglycolic acid, is used most often. Four included experimental studies107- 110 showed that compression colorectal anastomosis leads to acceptable healing and strength; 6 included RCTs111- 116 provide equivalent conclusions, finding no significant differences between hand-sewn and compression colorectal anastomosis. Also, noncomparative clinical cohort studies142- 145 including up to 1360 patients have reported incidences of AL between 0.7% and 5%. Although few gastrointestinal surgeons routinely use compression colorectal anastomosis, it seems a safe method. On the basis of 6 level 1b studies, no superiority of compression over hand-sewn colorectal anastomosis exists when comparing leak rates.
Studies regarding the configuration of the afferent and efferent ileal, colonic, or rectal loops are heterogeneous in patient selection and configuration and often concentrate on stapled pouches for very low anastomoses with outcome variables other than AL. Only 2 experimental studies matched the inclusion criteria; one study117 found no significant difference in blood flow between ETE or side-to-side anastomosis after rectal resection in pigs, and the other118 found better blood flow in ETE compared with side-to-end anastomosis after rectal resection in dogs. The included RCTs are also scarce: one119 on ETE vs end-to-side found more AL in ETE, and the other120 on the optimum side limb for side-to-end colorectal anastomosis found no significant difference between 3- and 6-cm sized limbs. No studies investigating the ideal length of the enterotomy were identified.
It is difficult to draw a conclusion from this small number of studies; there is one level 1b study showing a lower incidence of AL with end-to-side colorectal anastomosis and one level 1b study indicating that a 3-cm or a 6-cm side limb does not affect the incidence of AL.
In the clinical setting, healing of colorectal anastomosis is obscured from direct postoperative inspection. When AL occurs, diagnosis can be made only after the patient has become ill, making it a feared complication with high morbidity and mortality.1,4,146- 148 This systematic review of all aspects of hand-sewn colorectal anastomosis and the comparison of hand-sewn with mechanical anastomosis provides an overview of the existing colorectal anastomotic techniques combined with the available scientific evidence on anastomotic healing. Evaluation of colorectal anastomosis with clinical AL as the outcome measure and proper statistics produced very little level 1 evidence for all aspects of hand-sewn colorectal anastomosis. Nevertheless, we can formulate a conclusion using experimental results combined with clinical results for many aspects: the single-layer continuous suture technique by an inverting technique with slowly absorbable monofilament material seems preferable on the basis of level 1b evidence. However, for the other aspects of the technique, such as how far to place the suture from the anastomotic edge, the intersuture distance in relationship to the distance to the edge, which layers to include in the bite, how high the tension on the suture should be, and through what configuration the anastomosis should be made, surgeons probably rely on their teachers and instinct rather than on scientific evidence.
Large cohort studies that are available, describing low rates of AL for the used anastomotic technique, might indicate that dedicated, high-volume colorectal surgery has a role in lowering the incidence of AL because of a surgeon's familiarity with a certain technique.
Considering mechanical colorectal anastomosis, level 1a evidence indicates that stapling and hand-sewn anastomoses give equal results with regard to clinical AL, and level 1b evidence determines that compression and hand-sewn colorectal anastomosis have similar AL rates. In contrast to all possible variations that exist when sewing an anastomosis by hand, the technique of a stapled technique is much more uniform in the hands of surgeons. This could lead to standardizing colorectal anastomosis and prevent the nonscientific practice of the preferences of individual surgeons from being handed down from teacher to student without documentation of their exact properties and incidence of AL.
We can conclude from this review that, as of now, hand-sewn colorectal anastomosis is constructed following a largely undefined technique. The circumstances of RCTs do not reflect daily practice; therefore, routine, detailed documentation of anastomotic technique of all colorectal operations will be instrumental in formulating a definitive conclusion on the role of the unstandardized hand-sewn colorectal anastomosis.
Correspondence: Juliette C. Slieker, MD, Department of Surgery, Erasmus Medical Center, Laboratory of Experimental Surgery, Room Ee-173, Postbus 2040, 3000 CA Rotterdam, the Netherlands (email@example.com).
Accepted for Publication: June 25, 2012.
Author Contributions:Study concept and design: Slieker, Daams, Jeekel, and Lange. Acquisition of data: Slieker and Daams. Analysis and interpretation of data: Slieker, Daams, Mulder, Jeekel, and Lange. Drafting of the manuscript: Slieker, Daams, and Jeekel. Critical revision of the manuscript for important intellectual content: Slieker, Daams, Mulder, Jeekel, and Lange. Statistical analysis: Slieker. Obtained funding: Jeekel. Administrative, technical, and material support: Mulder and Jeekel. Study supervision: Jeekel and Lange.
Conflict of Interest Disclosures: None reported.
This article was corrected for missing online-only text on February 20th, 2013.