The impact of stapled compared to handsewn repair on anastomotic outcomes in trauma patients: a systematic review and meta‐analysis

Approach to enteric anastomotic technique has been a subject of debate, with no clear consensus as to whether handsewn or stapled techniques are superior in trauma settings, which are influenced by unique perturbances to important processes such as immune function, coagulation, wound healing and response to infection. This systematic review and meta‐analysis compares the risk of anastomotic complications in trauma patients with gastrointestinal injury requiring restoration of continuity with handsewn versus staples approaches.


Background
Anastomotic failure is a major cause of morbid and mortality for the patient requiring enteric anastomosis. 1 Anastomotic failure exists on a spectrum, notably the brakdown of enteric anastmoses leading to leakage of contents which is commonly referred to as anastomotic leak (AL), as well as the formation of peri-anastomotic fistulae and collections.For many years the influence of anastomotic technique, either handsewn or stapled, on anastomotic complications has been debated, with two Cochrane reviews having explored these approaches in elective colorectal and ileocolic anastomoses. 2,3The findings of these reviews were variable.In particular Choy et al. demonstrated that stapled anastomosis may be associated with fewer anastomotic leaks compared to handsewn anastomoses in patients requiring ileocolic anastomosis, however another Cochrane review by Neutzling et al. found no difference between these two approaches with respect to post-operative outcomes. 2,3owever, these reviews of elective surgical patients may not be directly applicable to the trauma patients requiring enteric resection and anastomosis.5][6] The response to this is often a staged approach to resection and restoration, which often involves anastomosis using oedematous or thickened bowel, in a milieu of potentially traumatized adjacent organs.Furthermore, contamination is common, as are other previously described risks for anastomotic failure, such as shock and transfusion, and mechanical and antibiotic bowel preparation, which may have an influence on anastomotic complications, are not possible.Finally, anastomotic staplers have been designed to oppose normal bowel, whereas this is not guaranteed in traumatic resections, particularly in damage control situations.
Currently there is limited evidence exploring anastomotic outcomes in the trauma patient.Given the complexities of altered physiology, emergent timing of operative intervention and often variable stability of trauma patients, we conducted systematic review and meta-analysis to synthesize available evidence upon this topic.

Search strategy
This review was performed in adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.The review protocol was prospectively registered in the PROSPERO database (PROSPERO ID: CRD42023458237).
A computer-assisted literature search of databases Medline, Embase and Cochrane Central was conducted on 27 August 2023.Reference lists of articles identified during this process were handsearched to capture additional relevant studies.The search strategy combined medical subject headings (MeSH) and keywords related to trauma laparotomy, damage control laparotomy and anastomosis.Subject terms, truncations and Boolean operators were used during this process to find all relevant articles.

Inclusion and exclusion criteria
Full-text peer-reviewed publications available in English language which evaluated the anastomotic outcomes following handsewn (sutured) versus stapled gastrointestinal anastomoses in the emergent trauma setting were considered for this review.
Articles were included if they were 1. Original and peerreviewed randomized-controlled trials, prospective non-randomized trials and retrospective non-randomized trials 2. Included adults (≥18 years) who sustained abdominal injury requiring gastrointestinal anastomosis following blunt or penetrating trauma 3. Evaluated the role of handsewn compared to stapled anastomosis with respect to anastomotic outcomes.
Articles were excluded from this review if they were 1.Of the following study types: reviews, meta-analyses, conference papers, letters and editorials, abstract only, commentaries, case reports and case series, opinion pieces and were non-human trials 2. Had incomplete data (which did not allow appropriate pooling for analysis to take place) 3. Evaluated other cohorts that would require emergency gastrointestinal anastomoses such as emergency general surgery populations 4. Underwent rectal anastomosis 5. Underwent primary repair without anastomosis and 6.Did not explore the primary outcomes of interest for this review.Notably, articles that reported on patients undergoing anastomosis and defunctioning were not excluded from analysis.

Literature screening
Initial screening by title and abstract of articles captured by the search strategy was performed by two independent investigators (KL and DR).Titles and abstracts that did not provide sufficient information with respect to aforementioned eligibility criteria proceeded to full-text analysis.Following title and abstract screening, the same two investigators (KL, DR) independently performed a full-text analysis of studies for eligibility.Disagreement during the process was resolved by discussion and consensus.

Outcomes
Primary outcome of interest was the presence of anastomotic leak.Secondary outcome of interest was a composite anastomotic complication (CAC) consisting of anastomotic leak, enterocutaneous fistula (ECF) and deep abscess.Combining these three outcomes, which exist on a similar spectrum of intra-abdominal anastomosisrelated complications, allowed more robust power, statistical efficiency and stronger evidence in our meta-analysis than when these outcomes were considered as individual entities and additionally, some mitigation against the variability in definition of anastomotic leak that exist in the literature.

Data extraction
Eligible studies were extracted for data including study design, publication year, country of publication, demographic data of respective cohorts (age, sex, and number of patients) and data relevant to our outcomes of interest including number of patients with handsewn anastomoses, number of patients with stapled anastomoses and the aforementioned postoperative outcomes.

Quality assessment
Quality assessment of methodological rigour was performed by two independent investigators (KL, DR) using the Risk Of Bias In Nonrandomized Studies of Interventions (ROBINS-I) tool. 7Quality was considered by the following schema: low risk, moderate risk, serious risk and critical risk of bias.Disagreement during this process was resolved by discussion and consensus.Publication bias was assessed with funnel plot as part of statistical analysis of included articles provided sufficient number of articles were captured (≥25).

Statistical analysis
Statistical analysis was performed with the assistance of Review Manager 5.4 (RevMan 5.4) (Cochrane, London, United Kingdom).Odds ratios (OR) and their 95% confidence intervals (95% CI) of postoperative outcomes for handsewn compared to stapled anastomoses were extracted from the included studies and meta-analysed.
Heterogeneity between studies was evaluated with the Higgins I 2 test. 8,9Values of I 2 at 25%, 50%, and 75% were graded as low, moderate and high heterogeneity respectively.Fixed-effects model was used if substantial heterogeneity was absent and randomeffects model was used if substantial heterogeneity was found.A P-value of <0.05 was considered to be statistically significant.To determine sources of heterogeneity, subgroup analyses and metaregression analyses where relevant were performed.

Subgroup analyses
Subgroup analyses was planned to assess for differences in the effect of anastomotic method for the following subgroups 1 analysis of anatomical location of anastomosis (small bowel to small bowel compared to small bowel to colon anastomosis) 2 analysis of anastomoses performed on initial laparotomy or delayed (including damage control laparotomy) 3 analysis of anastomotic outcomes in patients undergoing damage control procedures 4 analysis of subgroups following removal of studies at high risk of bias, namely those at serious or critical risk as determined by ROBINS-I assessment.

Overview of literature search
Two hundred and seventy-one articles were identified from a computer-assisted search of Medline, Embase and Cochrane Central databases.After accounting for duplicates, a total of 229 articles progress to title and abstract screening by two independent investigators (KL, KT).One hundred and eighty-two articles were excluded on the basis of eligibility and 47 articles progressed to full-text analysis by the same two independent investigators (KL, KT).Thirty nine studies were subsequently excluded with 8 studies being eligible to being included in this review.The PRISMA flowchart is shown in Figure 1.

Description of included studies and patient characteristics
Study characteristics are presented in Table 1.A total of 8 studies, concerning 931 patients were included for quantitative analysis [10][11][12][13][14][15][16][17] (Table 1).Of the total number of patients, data from 790 patients were available for analysis.All included studies were from the year 2000 onwards and the most frequent study type were retrospective cohort studies (7 studies, 87.5%).All of the included studies were from North America, specifically the United States of America (USA) (n = 8) and two of these studies were multicentre.Sample sizes of studies were highly variable, ranging between 29 and 257 patients, the latter of which only 120 patients were analysed.Demographic factors including age, and sex were unable to be pooled and quantitatively analysed due to lack of reported information in three studies.Of the included patients, 310 underwent handsewn anastomosis, 431 underwent stapled anastomosis, 33 underwent both handsewn and stapled anastomosis and 16 had unknown anastomosis of their gastrointestinal injuries.

Anastomotic outcomes
All studies compared handsewn to stapled anastomosis of gastrointestinal injury in the trauma setting.The summary of findings is available in Table 2.

Anastomotic leak
The analysis of anastomotic leak (AL) involved 5 studies in which a random model was employed to analyse stapled compare to handsewn anastomosis (P = 0.66, I 2 = 0.40) given low statistical heterogeneity but high level of clinical heterogeneity (Fig. 2).The pooled analysis of 567 patients (311 stapled, 256 handsewn) revealed a non-significant association between stapled anastomosis and reduced anastomotic leak when compared to handsewn anastomosis (OR = 0.77; 95% CI 0.24-2.45;P = 0.66).

Composite anastomotic complications
Eight studies reported on composite anastomotic complications (CAC) including anastomotic leak, ECF and deep abscess.This included analysis of a pooled 741 patients (431 stapled, 310 handsewn).A random-effects model was employed to analyse the impact of stapled compared to handsewn anastomosis on deep space infection (P = 0.90, I 2 = 0.45) given low statistical heterogeneity but high level of clinical heterogeneity (Fig. 3).The pooled analysis demonstrated a non-significant increase in the odds of CAS in stapled anastomosis when compared to handsewn anastomosis (OR = 1.05; 95% CI 0.53-2.09;P = 0.90).

Quality and risk of bias assessment
Methodological quality and risk of bias assessment was performed on the 8 included studies utilizing the ROBINS-I tool (Fig. 4).Overall, 2 articles (25%) were considered to have low risk of bias, 4 articles (50%) were considered to have moderate risk of bias and 2 articles (25%) were considered to have serious risk of bias.Key areas that were considered to introduce bias including domains of confounding factors, selection bias and bias secondary to missing data.A funnel plot analysis of anastomotic leak and CAS was not performed given the low number of articles.

Subgroup analysis
Two studies had sufficient data for subgroup analysis on CAS in the context of damage control laparotomy.This subgroup analysis included a total of 66 patients (28 stapled and 38 handsewn).A random-effects model was employed to analyse the impact of stapled compared to handsewn anastomosis on deep space infection (P = 0.92, I 2 = 0.00) given low statistical heterogeneity but high level of clinical heterogeneity (Fig. 5).The pooled analysis demonstrated a non-significant reduction in the odds of CAS in stapled anastomosis when compared to handsewn anastomosis (OR = 0.93; 95% CI 0.24-3.59;P = 0.92).
Further, subgroup analyses were performed following removal of articles that were deemed at high risk of bias (serious or critical risk as per ROBINS-I assessment).Subgroup of anastomotic leak utilizing a random effects model in this way included 4 studies accounting for 447 patients (218 staples, 229 handsewn) (P = 0.97, I 2 = 0.00).A non-significant increase in the odds of anastomotic leak in stapled anastomosis was demonstrated (OR = 1.02; 95% CI 0.45-2.31;P = 0.97) (Fig. 6a).Similarly, a random effects model assessed CAS utilizing a random effects model which included 6 studies accounting for 558 patients (288 stapled, 270 handsewn) (P = 0.22, I 2 = 0.28).A non-significant increase in the odds of CAS in stapled anastomosis was demonstrated (OR = 1.21; 95% CI 0.68-2.15;P = 0.22) (Fig. 6b).

Discussion
This meta-analysis has shown insufficient evidence to reject the null hypothesis of a difference in the odds of anastomotic leak (OR = 0.77; 95% CI 0.24-2.45;P = 0.66) or CAS (OR = 1.05; 95% CI 0.53-2.09;P = 0.90) between handsewn and stapled anastomoses in patients suffering trauma resulting in enteric injury requiring resection and anastomosis.However, this recommendation is made with low certainty as per our GRADE assessment, primarily attributed to the clinical heterogeneity of studies leading to inconsistent results and the risk of bias of included studies (Table 2).
As there was substantial clinical heterogeneity between studies, we attempted to perform two clinically relevant subgroup analyses.The first was by anatomical site of anastomosis, namely enteroenteric, ileo-colic and colo-colic.We considered anastomoses involving rectum clinically different due to their higher anastomotic leak rates, and hence an a priori exclusion for this study.The second attempted subgroup analysis was those who had a delayed anastomosis versus immediate on first laparotomy.This distinction was considered particularly clinically relevant as those with delayed anastomoses could represent a differing clinical picture, being more likely to have passed through a damage control phase, with substantial metabolic derangement.Furthermore, enteric stapling devices were designed primarily for the elective setting, and concerns have been raised that the oedema resulting from splanchnic hypoperfusion followed by subsequent reperfusion makes staples theoretically less appropriate. 18In both cases, there was insufficient data from our included studies to adequately perform these subgroup analyses for anastomotic leak.However, a subgroup analysis was able to be performed on patients undergoing damage control laparotomy with respect to CAS when comparing stapled to handsewn techniques, but once again no superiority for either technique established (OR = 0.93; 95% CI 0.24-3.59;P = 0.92).This is consistent with the finding of the majority of the individual studies' findings.
The influence of technique upon anastomotic outcomes has long been a topic for debate in the surgical literature.In the elective setting the literature is disparate: studies have either demonstrated that stapled anastomosis may be associated with a two-fold increase in the rate of anastomotic failure, or conversely, have no impact on outcomes at all. 2,18,19More robust analysis, coming from two Cochrane review within the past 2 decades comparing stapled and handsewn anastomosis in general surgery and colorectal cancer surgery similarly showed mixed results. 2,3Overall, these studies are confounded by the differing spectrum of management considerations when compared with the acute trauma setting.Furthermore, these less acute study populations poorly reflect the complexities of trauma physiology.It is arguable that patients undergoing resection and anastomosis for emergency conditions could be a more comparable population to the trauma patient than the elective patient.The single prospective randomized trial showed no difference in leak rate for handsewn (5.2%) compared to stapled (6.6%) anastomoses. 20A similar equivalence was described from a multicentre US observational study, which showed a potential difference with AL rates for handsewn (15.4%) and stapled (10.6%), but odds ratio of 0.92 (95% CI 0.51-1.53)after adjustment for confounders. 21This differs from another observational study which has shown a 2.6 fold increase (95% CI 1.1-6.5;P = 0.03) in the adjusted odds of anastomotic failure. 19A more recent effort aimed at characterizing approaches to anastomosis in emergency settings have included analysis of emergency general  surgery cohorts and trauma cohorts together. 18From these analyses, there is no justifiable superior method to anastomosis that could be concluded, noting the paucity of data and lack of rigorous methodological quality.
Further, we had intended to explore the influence of anastomotic technique in trauma patients requiring a delayed (i.e., not at initial laparotomy) repair, however the data was insufficient to draw upon a robust conclusion.3][24] It is also feasible that such patients are more likely to have other described risks for anastomotic or enteric repair leak in trauma, namely blood transfusion, shock, inotrope usage and contamination.Intestinal oedema encountered in relook laparotomy, as a consequence of the systemic inflammatory response syndrome, and potentially worsened by overzealous transfusion with crystalloid solutions common in older resuscitation strategies is another key difference in the delayed anastomotic trauma group.A study from 2011 showed a five-fold increase in AL rate when resuscitation with crystalloids reach over 10 L. 25 Although such volumes are less commonly utilized in modern resuscitation, there remains concern about the suitability of gastrointestinal staplers on oedematous bowel, noting that they were designed for normal bowel in elective surgery.Whilst it is likely that bowel oedema may increase anastomotic leak rate, it is less clear if this is due to a direct effect of bowel wall oedema on anastomotic technique, or if the oedema is a surrogate for another factor. 14inally, there are pragmatic surgeon specific considerations that should be mentioned.Primarily, differing surgeons will have  differing levels of comfort with either technique.An unmeasured bias in the observational studies included is that it is very likely the technique was chosen by the surgeon as it is their preferred technique.Differences in time taken to perform hand seen versus stapled anastomoses is described but once again in an observational dataset the same bias persists. 26Although operative time is a critical consideration in trauma, it could be argued that an individual surgeon will be quicker and better with a technique they are most familiar with.Additionally, an important point to note is that aside from two studies which were performed in the last decade, the majority of studies, all of whcih were observational, occurred in the first decade of the 2000s and therefore representing a different era in terms of trauma surgery and resuscitation.Although not immediately apparent in terms of the role of the historical context on anastomotic outcomes in the trauma setting, it is likely these contexts would have a confounding role in surgical preference with respect to anastomotic technique, as well as indirectly through the means by which patients are resuscitated as this may influence outcomes through mechanisms such as bowel wall oedema.
To the authors' knowledge, this is the first systematic review and meta-analysis on this topic.The study protocol was robust and in accordance with the PRISMA guidelines, with the prevalence of homogeneous data within included studies allowing for appropriate meta-analysis.Given the heterogeneity of the definition of anastomotic leak, we also analysed a composite of anastomotic complications rather than individually analysing anastomotic leak, deep intra-abdominal abscess and intra-abdominal fistula. 27owever, there are some limitations to note within this meta-analysis.First, the majority of studies (n = 7; 87.5%) were non-randomized retrospective studies with small sample sizes.Additionally, 75% of the articles (n = 6) were consider of moderate to serious risk of bias.Overall, the foundation of studies included were of lower quality and did not include the key components of rigorous study design including randomisation, allocation concealment, cohort matching and blinding.Therefore, the results of this meta-analysis should be considered with caution.Second, there was paucity of data with respect to the number of individual anastomoses per patient, consequently, analysis of outcomes by individual anastomosis was not possible.This is important to recognize as the elective patient will usually have a single gastrointestinal anastomosis whereas in the trauma patient, multiple resections and anastomoses are common.Third, given the eligibility criteria of this review only including articles in English language, there is the potential for missing data potentially from published literature from culturally and linguistically diverse regions.Fourthly, outcome ascertainment, specifically the definition of 'anastomotic leak', varied in both definition and terminology, with some studies using the term 'anastomotic failure'.Disappointingly, there was limited data for which to draw conclusions upon anastomotic technique in the delayed anastomotic or damage control group, a highly pertinent clinically question.Lastly, given there was often inconsistent reporting of demographic data within the included articles, it was not possible to be able to control for confounding demographic factors in the analysis of outcomes within this meta-analysis.

Conclusion
This meta-analysis suggests that there is no evidence to propose superiority of either staples or handsewn technique in trauma for anastomotic outcomes.However, the evidence for this still remains poorly characterized and must be interpreted with caution, as the majority of evidence supporting this finding were from retrospective studies of small sample.More robust and randomized prospective clinical trials are needed to better inform our understanding of how anastomosis techniques influence outcomes in patients, to allow more effective, safe and standardized approaches to the complex environment of trauma.Further studies, either prospective, or at least involving robust statistical adjustment for relevant confounders, would assist in this space.No conclusion can be drawn about the clinically relevant question of the usage of gastrointestinal staplers in the patients requiring restoration of continuity at a later rather than initial laparotomy.

Fig. 1 .
Fig. 1.PRISMA flowchart for search and selection of eligible articles in this review.

Fig. 4 .
Fig. 4. Quality assessment of included articles utilizing the ROBINS-I assessment tool.

Fig. 5 .
Fig. 5. Forrest plot for subgroup analysis of composite anastomotic complications (CAS) in patients undergoing damage control laparotomy.

Fig. 6 .
Fig. 6.(a) Forrest plot for subgroup analysis of anastomotic leak after removal of studies at high risk of bias.(b) Forrest plot for subgroup analysis of composite anastomotic outcomes after removal of studies at high risk of bias.

Table 1
Study demographics and frequency of handsewn and staples anastomoses

Table 2
Summary of findingsLow GRADE assessment: downgrade 1 point for risk of bias and downgrade 1 point for inconsistency secondary to clinical heterogeneity.There were too few studies to assess publication bias.