Incidence, risk factors and prevention of stoma site incisional hernias: a systematic review and meta‐analysis

Stoma reversal might lead to a stoma site incisional hernia. Recently, prophylactic mesh reinforcement of the stoma site has gained increased attention, supporting the need for accurate data on the incidence of and risk factors for stoma site incisional hernia and to identify high‐risk patients. The aim of this study was to assess incidence, risk factors and prevention of stoma site incisional hernias.

Recent research has shown that prophylactic mesh reinforcement (PMR) in midline laparotomies in highrisk patients significantly decreases the incidence of MIH [16,17], and PMR at the stoma site during permanent stoma construction has been considered to reduce rates of parastomal hernia [18][19][20][21]. Considering the largely comparable pathophysiology, PMR during temporary ostomy takedown to prevent SSIH could also be beneficial by potentially obviating complications and re-operations, and has gained increased attention amongst surgeons [17]. Accurate data on incidence and risk factors for the development of SSIH are of importance to correctly assess the clinical value of PMR to prevent SSIH, to facilitate selection of high-risk patients and to aid clinical and shared decision-making [22]. Therefore, the aims of this study were to systematically investigate the literature regarding the incidence of SSIH after stoma reversal, to evaluate potential risk factors for SSIH and to assess the effectiveness of PMR in preventing SSIH.

Method
The protocol of this study was registered in PROS-PERO (CRD42016053347). This study was conducted following the MOOSE guidelines and PRISMA statement [23,24]. Furthermore, decisions on the content were based on items proposed by Wille-Jørgensen et al. [25].

Study design and participants
Randomized controlled trials (RCTs) and prospective or retrospective cohort or case-control studies providing data on the incidence of SSIH were included. Case reports, reviews, letters, abstracts or comments were excluded. Studies were included if they met the Full-text articles excluded (n = 141) Non-English (2) No data on incidence (38) <16 years of age (18) No mention of follow-up (27) Abstract, letter, case series or comment (41) Full text not available (3) Abdominal wall trauma (9) Only data of midline hernia (3) Records excluded (n = 1266) Additional records identified through other sources (n = 0)    per database are shown in Appendix S1 in the online Supporting Information. There was no limit on publication date. Identified articles were reviewed independently by two reviewers (GS and DL) after duplicates were removed on title and abstract, followed by full-text review using EndNote X7â. Differences in article selection were discussed and inclusion or exclusion was performed after consensus was reached between reviewers.

Data extraction
Data extraction was performed by two researchers (GS and DL) and checked by a third independent researcher

Data synthesis
A Mantel-Haenszel random-effects model was used to calculate pooled odds ratios (ORs), while taking between-study variance and within-study variance into account. ORs with 95% CIs were calculated to assess outcome differences after ileostomy or colostomy reversal. Q statistics and I 2 were calculated to evaluate heterogeneity. All analyses were performed with Rev-Man 5.3 (Cochrane Centre, Denmark), except for the cumulative meta-analysis, which was performed using R (version 3.4.1.).

Search
A PRISMA flow diagram of the full search results is shown in Fig. 1. After fulfilling the search, a total of 2458 articles were identified, of which 1440 remained after removal of duplicates. After screening on title and abstract and full-text reading, 33 articles were included for qualitative and quantitative analyses [3][4][5][6][7][8]. Four articles provided data on outcomes after PMR for prevention of SSIH [45,46,56,57], of which two had a nonmesh control group and could therefore be included in quantitative synthesis [45,46].

Study characteristics
Study characteristics are shown in Table 1. Two articles were RCTs, 7 were prospective, 23 were retrospective cohort studies and 1 study was case-matched. A total of 6594 nonmesh and 77 mesh patients were available.
The majority of studies (20/33) did not specify the SSIH detection method. Two studies specifically mentioned the use of clinical examination and 11 reported on imaging [ultrasound (US), CT or MRI].

Stoma characteristics
An overview of stoma characteristics is shown in Table 2. Overall, 5289 stomas were constructed, of which 4679 (88.5%) were closed. In three studies, the type of colostomy or ileostomy was not clearly described and was therefore reported as total number of colostomies or ileostomies. In all other studies, LI was the most investigated stoma type (28/30), followed by LC (8/30), EC (6/30) and EI (5/30).  Figure 2 shows a forest plot of seven studies from which data could be used to compare SSIH rates after ileostomy and colostomy reversal. No difference in SSIH risk was found (OR 0.82, 95%CI 0.40-1.69, I 2 0%). Publication bias seemed unlikely, because the study distribution was justifiably symmetrical in an additional funnel plot (Fig. 3). In addition, no differences were found in cumulative meta-analysis (cumulative OR 0.87, 95%CI 0.44-1.75), as shown in Figure S2.

Risk factors
Eight studies reported on risk factors for development of SSIH (Table S2) [3][4][5][30][31][32]45,52]. In univariate Bhangu et al. [30] found no significant differences in age or gender for patients with SSIH versus no SSIH. Moreover, no difference in MIH between patients with and without SSIH was found (50% vs 41%, P = 0.51). Age, SSI, stoma type, gender, BMI and time to closure did not significantly increase the risk of SSIH in the study by Cingi et al. [31]. However, patients with a MIH had an increased risk (OR 4.4) of SSIH.
De Keersmaecker et al. [32] assessed a number of potential patient-and surgery-related risk factors but did not find any significant differences in univariate analysis.
Two other studies, by Liu et al. and Maggiori et al., were comparative cohort studies, including 83 and 94 patients, respectively [45,46]. In the retrospective study by Liu et al. [45], consecutive patients undergoing ileostomy closure were included, of whom 47 (56.6%) had PMR with polypropylene mesh (Ultrapro, Ethicon Inc.) placed in an onlay position by the same surgeon in all patients. During median follow-up of 18.2 months (IQR 11.7-30.8), three SSIHs (6.4%) were detected in mesh patients, whereas 13 SSIHs (36.1%) were found in control patients (OR 8.29, 95% CI 2.14-32.08, P = 0.001). SSIH in the mesh group was small and asymptomatic, and did not require repair, compared with 23% SSIH repairs in control patients. In the matched case-control study by Maggiori et al. [46], 30 consecutive patients were individually matched to patients from a prospective database. In these patients, a biological mesh (noncross-linked collagen, porcine dermal matrix; Meccellis BioTech, France) was placed in a retromuscular position. At 1-year CT follow-up, SSIH incidence was lower in mesh patients than the control group (3% vs 19%, P = 0.04), while postoperative morbidity was similar in both groups (17% vs 11%, P = 0.51). SSIH repair was needed in eight control patients (13% vs 0%, P = 0.05).

Discussion
This study shows an overall incidence of SSIH of 6.5% [range 0%-38.5%, median follow-up 27.5 (17.54-36) months], which is in accordance with the review by Bhangu et al. [58], who reported an overall hernia rate of 7% (range 0%-48%, median follow-up 36 months). However, this study was based on a smaller number of patients (n = 2698) than the present study (n = 4602). Both previous studies, by Bhangu [58,59]. To reduce this heterogeneity, several inclusion and exclusion criteria were used during our systematic literature search (Fig. 1). Most importantly, to be included, studies had to mention follow-up duration, since hernia rates increase over time and might vary between different durations. Furthermore, studies with > 10% of patients with abdominal trauma were excluded, as earlier reports showed these patients to be more prone to hernia development [60,61].
To compare the SSIH rate between ileostomy and colostomy reversal, seven studies were eligible for analysis. Whereas the previous review of Bhangu et al. [58] showed a significantly different lower SSIH rate after ileostomy (OR 0.28, 95% CI 0.12-0.65), this review found no significant difference in the risk of SSIH between ileostomy and colostomy (OR 0.82, 95% CI 0.40-1.69), which was also not found in an additional cumulative meta-analysis (cumulative OR 0.87, 95% CI 0. 44-1.75).
In this study, only one-third (11/33) of included studies assessed SSIH incidence as the primary outcome. Twenty studies did not mention detection methods and, therefore, it seems likely to assume that imaging was not used in these studies. To investigate potential underestimation, the overall incidence of SSIH from the 11 studies with SSIH as the primary outcome was calculated (17.7%, range 1.7%-36.1%) [4,5,[30][31][32]37,45,46,48,51,52]. These rates indeed support the hypothesis that the overall incidence of hernia from all included studies (6.5%), as from those only reporting on SSIH as a secondary outcome (3.6%), is an underestimation. The potential risk of underestimation by not using imaging for detection of SSIH is further supported by the higher incidence in studies that used imaging, compared with studies that did not use, or did not mention the use of imaging as a detection method (15.3% vs 3.7%, respectively). Indeed, from the literature on incisional hernias it is known that prevalence rates vary substantially, through differences in diagnostic modalities, observer, definition and diagnostic protocol [62]. The use of imaging, which led to higher SSIH rates, might have identified asymptomatic or occult hernias. Therefore, the overall SSIH rate of 6.5% seems to be lower but more clinically relevant, and thus it remains debatable if PMR might even be necessary at all. Hence, it is important to identify high-risk patients, in whom PMR might still be of added value and if in these patients its risks outweigh its benefits. Eight studies were identified that reported on potential risk factors for development of SSIH. Three studies [5,45,52] performed a multivariate analysis, from which BMI, primary surgery for malignant disease and diabetes mellitus were identified as potential risk factors. BMI is known to affect midline incisional and parastomal hernia rates [16,[63][64][65][66], which might be explained by higher intra-abdominal pressure and consequent higher abdominal wall tension [16,67]. Additionally, obesity and diabetes are associated with wound healing complications due to local hypoxia, caused by a decreased vascularization of adipose tissue and other microvascular changes, impairing collagen synthesis and having a negative effect on the overall healing process [16,68].
Smoking has a comparable negative effect on wound healing and is considered a risk factor for incisional hernia [69]. However, none of the included studies has shown a significant effect on occurrence of SSIH. Moreover, with regard to primary surgery for malignant disease, factors as malnutrition, poor general health and immunosuppressive effects of chemotherapy are thought to negatively affect the normal healing process [45,68,70]. Wound infections are known to increase the risk of hernia formation [63,71]; however, in the present literature review SSIs were not found to be independently associated with an increased risk of SSIH. Overall, the study by Oriel et al. [4] was the only study to identify superficial SSI as a factor contributing to future SSIH formation. The data on risk factors in this review might help with the selection of high-risk patients and therefore help guide clinical decision-making, potentially involving PMR. Moreover, since factors such as obesity and smoking can potentially be minimized, it might be of interest to focus not only on PMR but also on lifestyle interventions such as preoperative weight loss, smoking cessation and nutritional optimization for the prevention of SSIHs. However, to date no evidence is available on the efficacy or effectiveness of these lifestyle interventions with regard to incidence of SSIH.
Four studies reported on PMR for SSIH prevention [45,46,56,57]. These studies had several methodological limitations that made it difficult to draw conclusions about the potential added value of PMR. Two of the studies reported on a very limited number of patients (n < 10), decreasing their generalizability [56,57]. Furthermore, these studies had no control (nonmesh) group. Two other studies on PMR were of better quality because they included larger numbers of patients and as well as control patients [45,46]. Liu et al. [45] stated that mesh placement significantly reduced the incidence of SSIH following ileostomy closure, without an increase in complications. Maggiori et al. [46] reported a significant difference in SSIH on 1-year follow-up CT in favour of PMR. Nevertheless, all four studies recognized the need for RCTs to further evaluate the beneficial effects, safety and (cost-) effectiveness of PMR. Efforts have already been made by several research groups, and further trial results are awaited. A feasibility study by the Reinforcement of Closure of Stoma Site (ROCSS) Collaborative has recently been published and reported their study protocol to be feasible, without early safety concerns [72]. Based on their data, progression towards their ROCCS trial (ClinicalTrials.gov identifier NCT02238964) has continued [72,73]. Several other trials have been initiated, such as the MEMBO trial (NCT02576184), the ILEOCLOSE trial (NCT02226887) and the LISTO-trial (NCT02669992). Next to ileostomy closure, only the ROCCS trial also includes patients undergoing colostomy closure, and none of these trials focuses on a specific risk group, such as obese patients. However, since obesity seems to increase the risk of SSIH after stoma closure, this group of patients might potentially benefit more from PMR, although, paradoxically, these patients, especially in case of diabetes, might at the same time also be at higher risk of developing meshrelated complications [74,75]. Therefore, it would be interesting to see the results of PMR in these patients specifically. With regard to the efficacy and (cost-)effectiveness of PMR, it is still debatable as to what would be a clinically significant reduction in SSIH rates. In the case of the ROCCS trial, sample size calculation of the full Phase III study was based on a 40% reduction (25% to 15%) in the 2-year clinical hernia rate [72]. In the study by Maggiori et al. [46], a 16% difference was found (19% vs 3%, P = 0.043), which might have been used for the sample size calculation of the MEMBO trial. However, further data on sample size calculations and risk reduction were not available. Unfortunately, robust conclusions cannot yet be drawn on its risks and benefits from the available literature on PMR. If PMR is proven to be beneficial in these studies, further implications for practice should be made sufficiently clear (e.g. patient selection) in order to overcome the barriers of implementing these findings [76].
The low level of evidence and the vast heterogeneity of the included studies are two important limitations of this study. Nevertheless, inclusion of these studies was still deemed necessary as they allowed a more comprehensive overview of potential risk factors, as well as more detailed analyses of SSIH and repair rates. The lack of a predefined time period from which studies could be included might also have been a limitation of this review, because important changes in operative and perioperative care of patients have been introduced in recent decades (e.g. laparoscopy). However, this effect is presumably largely negligible since the majority of included studies were published in the previous decade (Table 1).
In conclusion, this review shows an overall incidence of SSIH of 6.5% (range 0%-38.5%), as well as an incidence of 17.7% (range 1.7%-36.1%) from 11 studies assessing SSIH as the primary outcome. Furthermore, potential risk factors have been identified, of which BMI, malignant disease and diabetes were considered to be the most important. Lastly, early results from four studies on PMR were identified, but no robust conclusions could be drawn. Results of ongoing trials are awaited.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Figure S1. Risk of bias assessment. Figure S2. Cumulative meta-analysis of studies reporting on SSIH rates of ileostomies and colostomies. Table S1. SSIH repair rates (subdivided by stoma type). Table S2. Risk factors for SSIH. Appendix S1. Details of the search strategy.