Systematic review Prevention of incisional hernia after midline laparotomy with prophylactic mesh reinforcement: a meta-analysis and trial sequential analysis A. P. Jairam1 , M. López-Cano4 , J. M. Garcia-Alamino6 , J. A. Pereira5 , L. Timmermans2 , J. Jeekel3 , J. Lange3 and F. Muysoms7 1 Department of Surgery, Catharina Hospital Eindhoven, Eindhoven, and 2 Department of Surgery, Maasstad Ziekenhuis Rotterdam and 3 Department of General Surgery, Erasmus University Medical Centre, Rotterdam, the Netherlands, 4 Department of General and Digestive Surgery, Hospital Vall d’Hebrón, Universidad Autónoma de Barcelona, and 5 Department of General and Digestive Surgery, Hospital Del Mar, Barcelona, Spain, 6 Department of Primary Health Care Sciences, University of Oxford, Oxford, UK, and 7 Department of Surgery, Maria Middelares Hospital, Ghent, Belgium Correspondence to: Dr A. P. Jairam, Catharina Hospital Eindhoven, Michelangelolaan 2, 5623 EJ Eindhoven, the Netherlands (e-mail: anjairam@hotmail.com) Background: Incisional hernia is a frequent complication after abdominal surgery. The aim of this study was to assess the efficacy of prophylactic mesh reinforcement (PMR) after midline laparotomy in reducing the incidence of incisional hernia. Methods: A meta-analysis was conducted following PRISMA guidelines. The primary outcome was the incidence of incisional hernia after follow-up of at least 12 months. Secondary outcomes were postoperative complications. Only RCTs were included. A random-effects model was used for the meta-analysis, and trial sequential analysis was conducted. Results: Twelve RCTs were included, comprising 1815 patients. The incidence of incisional hernia was significantly lower after PMR compared with sutured closure (risk ratio (RR) 0⋅35, 95 per cent c.i. 0⋅21 to 0⋅57; P < 0⋅001). Both onlay (RR 0⋅26, 0⋅11 to 0⋅67; P = 0⋅005) and retromuscular (RR 0⋅28, 0⋅10 to 0⋅82; P = 0⋅02) PMR led to a significant reduction in the rate of incisional hernia. The occurrence of seroma was higher in patients who had onlay PMR (RR 2⋅23, 1⋅10 to 4⋅52; P = 0⋅03). PMR did not result in an increased rate of surgical-site infection. Conclusion: PMR of a midline laparotomy using an onlay or retromuscular technique leads to a significant reduction in the rate of incisional hernia in high-risk patients. Individual risk factors should be taken into account to select patients who will benefit most. [Correction added on 19 February 2020, after first online publication: J. García Alamino has been amended to J. M. Garcia-Alamino] Funding information No funding Paper accepted 17 December 2019 Published online 14 February 2020 in Wiley Online Library (www.bjsopen.com). DOI: 10.1002/bjs5.50261 Introduction life and body image13,14 , and treatment of IH represents a financial burden on the healthcare system15 . Incisional hernia (IH) is a frequent complication after Current treatment of IH is mesh repair, which has led abdominal surgery, with an incidence ranging from 11 to to a lower recurrence rate compared with the primary 20 per cent in general surgical populations1 – 4 . The inci- suture technique16 . However, the recurrence rate is still dence of IH can increase up to 40 per cent in high-risk high, even when mesh is used. A Danish nationwide reg- groups, such as patients with an abdominal aortic aneurysm istry study17 reported a cumulative recurrence rate of (AAA) or morbid obesity5 – 12 . IH may be asymptomatic, 37 per cent at 3 years after IH repair. Currently, there but can also lead to serious and potentially fatal compli- is no definitive solution for the high recurrence rates cations, such as incarceration and strangulation of bowel. and complications related to these recurrences. Preven- Furthermore, IH has a high impact on patients’ quality of tion is therefore of paramount importance18 . In the past © 2020 The Authors. BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd BJS Open 2020; 4: 357–368 This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. 358 A. P. Jairam, M. López-Cano, J. M. Garcia-Alamino, J. A. Pereira, L. Timmermans, J. Jeekel et al. Fig. 1 PRISMA diagram for the review Identification Records identified Additional records identified through database searching through other sources n = 2945 n=1 Records after duplicates removed n = 1498 Screening Records screened by title and abstract n = 1498 Records excluded n = 1449 Review articles checked for Eligibility additional references n = 10 Full-text articles assessed for Full-text articles excluded n = 29 eligibility n = 39 Review n = 11 Abstract only n = 8 Follow-up < 12 months n = 2 Duplicate publication n = 3 Retrospective study n = 3 Studies included in qualitative synthesis n = 20 Non-midline incision n = 1 Randomized trial (JADAD score) n = 13 No prophylactic mesh n = 1 Non-randomized trial (MINORS score) n = 7 Included Studies included in quantitative synthesis (meta-analysis) n = 12 Table 1 Risk-of-bias assessment for prevention of incisional hernia with prophylactic mesh reinforcement in midline laparotomy Random Blinding of Blinding of sequence Allocation participants outcome Incomplete Selective generation concealment and personnel assessment outcome data reporting (selection (selection (performance (detection (attrition (reporting Reference bias) bias) bias) bias) bias) bias) Abo-Ryia et al.9 − − − − + Bali et al.10 − − − − + + Bevis et al.8 + + + + + + Caro-Tarrago et al.29 + + + + El-Khadrawy et al.28 − − − − + García-Ureña et al.31 + + + + + Gutiérrez de la Peña et al.6 − − − − + Jairam et al.23 + + + + + Muysoms et al.11 + + + + + Pans et al.5 − − − − + Sarr et al.30 + − − − − + Strzelczyk et al.7 + + + + +, Low risk of bias; −, high risk of bias. © 2020 The Authors. www.bjsopen.com BJS Open 2020; 4: 357–368 BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd Prevention of incisional hernia with mesh 359 Table 2 Summary of findings for included studies on the prevention of incisional hernia with prophylactic mesh reinforcement in midline laparotomies Indication for midline Follow-up Diagnosis of IH Reference n laparotomy Type of mesh Mesh position (months) Outcome measurements (clinical/radiological) Pans et al.5 288 Morbid obesity Polyglactin Intraperitoneal 29⋅8 IH, postoperative Not reported morbidity Gutiérrez de la Peña et al.6 88 Colorectal cancer, gastric Polypropylene Onlay 36 IH, haematoma, seroma, Clinical; if not cancer, cholelithiasis, infection, pain conclusive, CT diverticulosis, Crohn’s disease, pancreatic cystadenoma, gastric ulcer, cancer of small intestine Strzelczyk et al.7 74 Gastric bypass surgery Polypropylene Retromuscular 28 IH, wound leak, bleeding, Clinical plus other surgical ultrasound imaging complication El-Khadrawy et al.28 40 High risk Polypropylene Preperitoneal 36⋅7 IH, seroma Clinical SSI, wound disruption, chronic wound pain, cardiac, pulmonary problems, DVT, ascites Bevis et al.8 80 AAA Polypropylene Retromuscular Mesh 30⋅2 IH, wound infection, hernia Clinical; if doubt, operation ultrasound imaging No mesh 19⋅6 Abo-Ryia et al.9 64 Open bariatric surgery Polypropylene Preperitoneal Mesh 48 Safety and efficacy of Clinical; ultrasound preperitoneal prosthetic imaging in enforcement, seroma, suspected cases infection, partial dehiscence No mesh 49 Caro-Tarrago et al.29 160 Colorectal and general surgery Polypropylene Onlay Mesh 14⋅8 IH, all adverse events, Clinical and CT postoperative complications No mesh 12⋅5 Sarr et al.30 280 Open RYGB Biological Intraperitoneal 24 IH, wound infection, Clinical, phone call, wound dehiscence, primary care wound sinus tract, physician wound erythema, seroma Bali et al.10 40 Open AAA Biological Onlay 36 IH, duration of surgery, Clinical and CT postoperative complications, reoperation rate García-Ureña et al.31 107 Colorectal surgery Polypropylene Onlay 24 IH, incidence of local Clinical and CT complications: SSI, seroma, evisceration, mesh rejection Systemic complications Muysoms et al.11 114 AAA and ASA grade < IV Partially absorbable Retromuscular 24 IH Clinical and, if polypropylene available, ultrasound imaging or CT Postoperative complications SSI Duration of surgery Jairam et al.23 480 Open AAA surgery or midline Polypropylene Onlay (188 patients) 24 IH, postoperative Clinical and laparotomy in patients with complications ultrasound imaging BMI > 27 kg/m2 or CT Quality of life Cost-effectiveness Retromuscular (185 patients) IH, incisional hernia; SSI, surgical-site infection; DVT, deep vein thrombosis; AAA, abdominal aortic aneurysm; RYGB, Roux-en-Y gastric banding. few years, several studies on the prevention of IH with 2015: that PMR to reduce the incidence of IH after elec- prophylactic mesh reinforcement (PMR) have been con- tive midline laparotomy in a high-risk patient be sug- ducted. Most RCTs included a limited number of patients. gested with a weak recommendation. The group also Different surgical techniques of PMR, including mesh stated that larger trials were needed to make a strong placement in onlay, retromuscular or intraperitoneal posi- recommendation. tion, have been studied. The European Hernia Society Since the publication of these EHS guidelines, three (EHS) Guidelines Development Group19 on the closure meta-analyses20 – 22 have been published, together with of abdominal wall incisions made a recommendation in the long-term data from the largest multicentre RCT on © 2020 The Authors. www.bjsopen.com BJS Open 2020; 4: 357–368 BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd 360 A. P. Jairam, M. López-Cano, J. M. Garcia-Alamino, J. A. Pereira, L. Timmermans, J. Jeekel et al. Fig. 2 Forest plots of incisional hernia after prophylactic mesh reinforcement of a midline laparotomy versus primary suture a Overall data IH Reference Mesh Suture Weight (%) Risk ratio Risk ratio Pans et al. 5 33 of 144 41 of 144 16·5 0·80 (0·54, 1·20) Gutiérrez de la Peña et al.6 0 of 44 5 of 44 2·6 0·09 (0·01, 1·60) Strzelczyk et al.7 0 of 36 8 of 38 2·7 0·06 (0·00, 1·04) El-Khadrawy et al.28 1 of 20 3 of 20 4·1 0·33 (0·04, 2·94) Bevis et al.8 5 of 37 16 of 43 11·5 0·36 (0·15, 0·90) Abo-Ryia et al.9 1 of 32 9 of 32 4·6 0·11 (0·01, 0·83) Sarr et al.30 32 of 139 38 of 141 16·4 0·85 (0·57, 1·28) Caro-Tarrago et al.29 2 of 80 30 of 80 7·5 0·07 (0·02, 0·27) Bali et al.10 0 of 20 6 of 20 2·7 0·08 (0·00, 1·28) García-Ureña et al.31 6 of 53 17 of 54 12·0 0·36 (0·15, 0·84) Muysoms et al.11 0 of 56 16 of 58 2·7 0·03 (0·00, 0·51) Jairam et al.23 59 of 373 33 of 107 16·7 0·51 (0·36, 0·74) Total 139 of 1034 222 of 781 100·0 0·35 (0·21, 0·57) Heterogeneity: τ2 = 0·35; χ2 = 35·62, 11 d.f., P < 0·001; I2 = 69% Test for overall effect: Z = 4·18, P < 0·001 0·01 0·1 1 10 100 Favours mesh Favours suture b According to risk of bias IH Reference Mesh Suture Weight (%) Risk ratio Risk ratio Low risk of bias Strzelczyk et al.7 0 of 36 8 of 38 2·7 0·06 (0·00, 1·04) Bevis et al.8 5 of 37 16 of 43 11·5 0·36 (0·15, 0·90) Caro-Tarrago et al.29 2 of 80 30 of 80 7·5 0·07 (0·02, 0·27) García-Ureña et al.31 6 of 53 17 of 54 12·0 0·36 (0·15, 0·84) Muysoms et al.11 0 of 56 16 of 58 2·7 0·03 (0·00, 0·51) Jairam et al.23 59 of 373 33 of 107 16·7 0·51 (0·36, 0·74) Subtotal 72 of 635 120 of 380 53·1 0·23 (0·10, 0·52) Heterogeneity: τ2 = 0·58; χ2 = 16·97, 5 d.f., P = 0·005; I2 = 71% Test for overall effect: Z = 3·53, P < 0·001 High risk of bias Pans et al.5 33 of 144 41 of 144 16·5 0·80 (0·54, 1·20) Gutiérrez de la Peña et al.6 0 of 44 5 of 44 2·6 0·09 (0·01, 1·60) El-Khadrawy et al.28 1 of 20 3 of 20 4·1 0·33 (0·04, 2·94) Abo-Ryia et al.9 1 of 32 9 of 32 4·6 0·11 (0·01, 0·83) Sarr et al.30 32 of 139 38 of 141 16·4 0·85 (0·57, 1·28) Bali et al.10 0 of 20 6 of 20 2·7 0·08 (0·00, 1·28) Subtotal 67 of 399 102 of 401 46·9 0·57 (0·32, 1·02) Heterogeneity: τ2 = 0·18; χ2 = 9·91, 5 d.f., P = 0·08; I2 = 50% Test for overall effect: Z = 1·90, P = 0·06 Total 139 of 1034 222 of 781 100·0 0·35 (0·21, 0·57) Heterogeneity: τ2 = 0·35; χ2 = 35·62, 11 d.f., P = 0·001; I2 = 69% Test for overall effect: Z = 4·18, P < 0·001 0·01 0·1 1 10 100 Test for subgroup difference: χ2 = 3·14, 1 d.f., P = 0·08; I2 = 68·1% Favours mesh Favours suture a Overall data from all 12 included studies; b data for studies with a low or high risk of bias. A Mantel–Haenszel random-effects model was used for meta-analysis. Risk ratios are shown with 95 per cent confidence intervals. IH, incisional hernia. © 2020 The Authors. www.bjsopen.com BJS Open 2020; 4: 357–368 BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd Prevention of incisional hernia with mesh 361 Fig. 3 Funnel plot for low and high risk of bias studies Study selection, data extraction and quality assessment 0 Three reviewers independently screened all records by title and abstract for eligibility. After this first screening, 0·5 the full text of records was assessed. Only eligible RCTs s.e.(log[RR]) were included. The methodological quality of RCTs was 1·0 assessed using SIGN (Scottish Intercollegiate Guidelines Network) checklists. Risk-of-bias assessment was done 1·5 with the Cochrane Collaboration tool25 , in which the fol- Low risk of bias High risk of bias lowing aspects are assessed: random sequence generation, allocation concealment, blinding of patients, personnel or 2·0 0·01 0·1 1 10 100 outcome assessors, incomplete outcome data and selective Risk ratio reporting. Assessment of both methodological quality and risk of bias was performed by three independent reviewers. RR, risk ratio. Studies were assessed as having either a low or high risk of bias. RCTs were included if they met the following inclusion IH prevention after midline laparotomy, comparing PMR criteria: patients aged at least 18 years, undergoing mid- with primary suture (the PRIMA trial)23 . However, in one line laparotomy, for all types of indication, with all types of these meta-analyses20 , RCTs and observational studies of mesh and all types of mesh position, and follow-up of at were mixed. In the meta-analysis of Wang and colleagues21 , least 12 months. The primary outcome was the incidence some studies on non-midline incisions were included and of IH. Secondary outcomes were postoperative complica- none of the meta-analyses included long-term data from tions: seroma, surgical-site infection (SSI), haematoma and the PRIMA trial. burst abdomen. No language restrictions were used. The aim of this meta-analysis was to assess the safety and All required data were extracted and collected in a efficacy of PMR, both onlay and retromuscular, in reducing standard manner by at least two authors independently. the IH incidence after elective midline laparotomy. Any disagreements during the data extraction phase were resolved through discussion and by consulting a third Methods investigator. A summary-of-findings table was created, in A meta-analysis was conducted and reported follow- which the following information was collected: study char- ing the PRISMA guidelines24 . This meta-analysis was acteristics (title, year of publication, study design, number registered prospectively at the Prospero database on 5 of included patients), indication for midline laparotomy, November 2015 (CRD42015027079) with the acronym description of intervention and description of the com- MARIA review. The meta-analysis was finalized after pared intervention (‘control group’), type of mesh used, publication of the final results of the PRIMA trial on mesh placement, length of follow-up and outcome mea- 19 June 2017. surements. When a paper included data for different mesh positions, the data for these were described separately per group in the summary-of-findings table. For duplicate data Information sources and search terms reported by the same author(s), the article with the longest A systematic computerized literature search was per- follow-up period was selected. formed until 1 January 2017, using 12 databases: Embase, MEDLINE, Web of Science, SCOPUS, Cochrane Statistical analysis Library, CINAHL, PubMed Publisher, LILACS (Latin American and Caribbean Literature on Health Sci- A meta-analysis, pooling the results of the retrieved stud- ences), SciELO (Scientific Electronic Library Online), ies, was performed. A sensitivity analysis was conducted to ScienceDirect, ProQuest and Google Scholar. The reduce the risk of possible bias for primary and secondary Biomedical Information Specialist of the Medical Library outcomes. Meta-analyses that combined other subgroups (Erasmus University Medical Centre, Rotterdam, the (mesh position) were also performed. A random-effects Netherlands) prepared the search strategy. The syntax model was used and presented as risk ratios (RRs) with with search terms is shown in Appendix S1 (supporting 95 per cent confidence intervals. Effects were consid- information). ered statistically significant if the 95 per cent c.i. of the © 2020 The Authors. www.bjsopen.com BJS Open 2020; 4: 357–368 BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd 362 A. P. Jairam, M. López-Cano, J. M. Garcia-Alamino, J. A. Pereira, L. Timmermans, J. Jeekel et al. Fig. 4 Forest plots of incisional hernia after prophylactic mesh reinforcement of a midline laparotomy versus primary suture a Onlay mesh IH Reference Mesh Suture Weight (%) Risk ratio Risk ratio Caro-Tarrago et al.29 2 of 80 30 of 80 22·9 0·07 (0·02, 0·27) García-Ureña et al.31 6 of 53 17 of 54 34·2 0·36 (0·15, 0·84) Jairam et al.23 25 of 188 33 of 107 42·9 0·43 (0·27, 0·68) Total 33 of 321 80 of 241 100·0 0·26 (0·11, 0·67) Heterogeneity: τ2 = 0·46; χ2 = 7·16, 2 d.f., P = 0·03; I2 = 72% Test for overall effect: Z = 2·82, P = 0·005 0·01 0·1 1 10 100 Favours mesh Favours suture b Retromuscular mesh IH Reference Mesh Suture Weight (%) Risk ratio Risk ratio Strzelczyk et al. 7 0 of 36 8 of 38 10·9 0·06 (0·00, 1·04) Bevis et al.8 5 of 37 16 of 43 34·8 0·36 (0·15, 0·90) Muysoms et al.11 0 of 56 16 of 58 11·0 0·03 (0·00, 0·51) Jairam et al.23 34 of 185 33 of 107 43·4 0·60 (0·39, 0·90) Total 39 of 314 73 of 246 100·0 0·28 (0·10, 0·82) Heterogeneity: τ2 = 0·63; χ2 = 8·94, 3 d.f., P = 0·03; I2 = 66% Test for overall effect: Z = 2·33, P = 0·02 0·01 0·1 1 10 100 Favours mesh Favours suture a Data from studies using onlay mesh reinforcement; b data from studies using retromuscular mesh reinforcement. A Mantel–Haenszel random-effects model was used for meta-analysis. Risk ratios are shown with 95 per cent confidence intervals. IH, incisional hernia. overall effect estimate did not overlap. The I 2 statistic relative risk of the specified outcome in the treatment was used to assess heterogeneity. Groups with zero events group, compared with the control group. TSA was planned were adjusted with a constant continuity adjustment of for all retrieved studies, and for the group of studies with 0⋅5 in each arm (as per the default adjustment in the low risk of bias. TSA was performed using the TSA soft- software used). Publication bias was assessed by a funnel ware v0.9 (http://www.ctu.dk/tsa/downloads.aspx). plot. Analyses were performed using Review Manager soft- ware (RevMan version 5.3; The Nordic Cochrane Centre, Results Copenhagen, Denmark). Two-sided P < 0⋅050 was consid- ered statistically significant. A total of 1498 records were identified after removal of Conducting a meta-analysis can lead to type I errors duplicates. After screening of title and abstract, 49 articles (false-positives) or overestimation of treatment effects due were found relevant for full-text assessment. After full-text to systematic errors (bias) and random errors (play of assessment, 29 articles were excluded, leaving 13 RCTs and change). To avoid this, trial sequential analysis (TSA) was seven non-randomized trials for the qualitative synthesis conducted. TSA can provide a required information size (Fig. 1). In total, 13 RCTs5 – 11,23,28 – 32 fulfilled the inclusion (RIS). The RIS is the required number of patients that criteria, but the study of Timmermans and colleagues32 needs to be included in the meta-analysis to provide firm was excluded because only the short-term results (postop- evidence26,27 . Control event rate (CER) and relative risk erative complications in the first month) were discussed; reduction (RRR) values were calculated. CER is the pro- thus 12 RCTs were analysed. Six studies7,8,11,23,29,31 were portion of participants in the control group who have the considered to have a low risk of bias, and six5,6,9,10,28,30 to outcome. RRR can be interpreted as the reduction in the have a high risk of bias (Table 1). © 2020 The Authors. www.bjsopen.com BJS Open 2020; 4: 357–368 BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd Prevention of incisional hernia with mesh 363 Fig. 5 Trial sequential analysis curves of the incidence of incisional hernia after prophylactic mesh reinforcement of a midline laparotomy versus primary suture a Overall data RIS = 664 8 7 Trial sequential monitoring boundary Z-score Favours mesh 6 5 4 3 Cumulative Z-score 2 1 0 634 No. of patients (linear scale) –1 –2 Favours suture –3 –4 –5 –6 –7 –8 b Low-risk-of-bias data RIS = 358 8 7 6 Favours mesh 5 4 3 Cumulative Z-score 2 1 0 314 No. of patients (linear scale) –1 –2 Favours suture –3 –4 –5 –6 –7 –8 a Overall data from all 12 included studies; b data from studies with a low risk of bias. a,b Both graphs are two-sided. RIS, required information size. The 12 RCTs comprised a total of 1815 patients. AAA8,10,11,23 , morbid obesity5,7,9,30 , colorectal cancer Study and patient characteristics are shown in Table 2. surgery31 or a mixture of operative indications6,28,29 . Most Inclusion criteria for PMR of the midline laparotomy studies placed a polypropylene mesh in an onlay6,10,23,29,31 in the individual RCTs were either the presence of an or retromuscular7,8,11,23 position. Two studies10,30 used © 2020 The Authors. www.bjsopen.com BJS Open 2020; 4: 357–368 BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd 364 A. P. Jairam, M. López-Cano, J. M. Garcia-Alamino, J. A. Pereira, L. Timmermans, J. Jeekel et al. Fig. 6 Forest plots of postoperative seroma after prophylactic mesh reinforcement of a midline laparotomy versus primary suture a Onlay mesh Seroma Reference Mesh Suture Weight (%) Risk ratio Risk ratio Caro-Tarrago et al.29 23 of 80 9 of 80 39·5 2·56 (1·26, 5·17) García-Ureña et al.31 7 of 53 7 of 54 29·1 1·02 (0·38, 2·71) Jairam et al.23 34 of 188 5 of 107 31·4 3·87 (1·56, 9·60) Total 64 of 321 21 of 241 100·0 2·23 (1·10, 4·52) Heterogeneity: τ2 = 0·20; χ2 = 4·09, 2 d.f., P = 0·13; I2 = 51% Test for overall effect: Z = 2·22, P = 0·03 0·01 0·1 1 10 100 Favours mesh Favours suture b Retromuscular mesh Seroma Reference Mesh Suture Weight (%) Risk ratio Risk ratio Strzelczyk et al. 7 5 of 36 4 of 38 35·1 1·32 (0·38, 4·53) Bevis et al.8 2 of 37 0 of 43 5·9 5·79 (0·29, 116·89) Muysoms et al.11 2 of 56 0 of 58 5·9 5·18 (0·25, 105·47) Jairam et al.23 13 of 185 5 of 107 53·1 1·50 (0·55, 4·10) Total 22 of 314 9 of 246 100·0 1·67 (0·81, 3·47) Heterogeneity: τ2 = 0·00; χ2 = 1·41, 3 d.f., P = 0·70; I2 = 0% Test for overall effect: Z = 1·38, P = 0·17 0·01 0·1 1 10 100 Favours mesh Favours suture a Data from studies using onlay mesh reinforcement; b data from studies using retromuscular mesh reinforcement. A Mantel–Haenszel random-effects model was used for meta-analysis. Risk ratios are shown with 95 per cent confidence intervals. biological meshes, and one5 used a rapidly absorbable retromuscular PMR led to a significant reduction of intraperitoneal mesh. the IH incidence compared with primary suture, with RRs of 0⋅26 (P = 0⋅005) and 0⋅28 (P = 0⋅02) respectively (Fig. 4). Outcome measurements TSA (for the primary outcome) was done for all 12 The primary outcome was the incidence of IH after included studies. The CER proportion was 28 per cent, the follow-up of at least 12 months. Twelve RCTs were RRR was 65 per cent, and a constant continuity adjustment included in the overall quantitative analysis for the was set at 0⋅5 events per group. The accrued information primary outcome and publication bias33 was evalu- size (n = 1815) was 273⋅3 per cent of the estimated RIS ated. The meta-analysis showed a significant reduction (n = 664). This means that firm evidence was available. For of IH in patients with PMR compared with that in low-risk-of-bias studies (6 RCTs), the CER proportion was patients who had a primary suture (RR 0⋅35; P < 0⋅001) 31 per cent and RRR 77 per cent. The accrued information (Fig. 2a). The funnel plot was slightly asymmetrical, size (n = 1015) was 283⋅5 per cent of the estimated RIS indicating a possible publication bias for studies favour- (n = 358) (Fig. 5). ing mesh prophylaxis (Fig. 3). Analysis of the primary Secondary outcomes were analysed for the outcome for studies with a low risk of bias (6 RCTs) low-risk-of-bias studies (6 RCTs). Patients with an onlay showed that the occurrence of IH was significantly PMR had a higher risk of developing seroma (RR 2⋅23; lower (RR 0⋅23; P < 0⋅001) in the PMR group than P = 0⋅030) compared with patients who had a primary in the primary suture group (Fig. 2b). Both onlay and suture. Patients who had a retromuscular PMR had no © 2020 The Authors. www.bjsopen.com BJS Open 2020; 4: 357–368 BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd Prevention of incisional hernia with mesh 365 Fig. 7 Forest plots of surgical-site infection after prophylactic mesh reinforcement of a midline laparotomy versus primary suture a Onlay mesh SSI Reference Mesh Suture Weight (%) Risk ratio Risk ratio Caro-Tarrago et al. 29 8 of 80 8 of 80 19·1 1·00 (0·39, 2·53) García-Ureña et al.31 10 of 53 18 of 54 36·4 0·57 (0·29, 1·11) Jairam et al.23 25 of 188 14 of 107 44·5 1·02 (0·55, 1·87) Total 43 of 321 40 of 241 100·0 0·82 (0·55, 1·23) Heterogeneity: τ2 = 0·00; χ2 = 1·82, 2 d.f., P = 0·40; I2 = 0% 0·01 0·1 1 10 100 Test for overall effect: Z = 0·96, P = 0·33 Favours mesh Favours suture b Retromuscular mesh SSI Reference Mesh Suture Weight (%) Risk ratio Risk ratio Strzelczyk et al. 7 5 of 36 4 of 38 18·7 1·32 (0·38, 4·53) Bevis et al.8 2 of 37 2 of 43 7·8 1·16 (0·17, 7·85) Muysoms et al.11 1 of 56 3 of 58 5·7 0·35 (0·04, 3·22) Jairam et al.23 19 of 185 14 of 107 67·8 0·78 (0·41, 1·50) Total 27 of 314 23 of 246 100·0 0·85 (0·50, 1·45) Heterogeneity: τ2 = 0·00; χ2 = 1·28, 3 d.f., P = 0·73; I2 = 0% Test for overall effect: Z = 0·59, P = 0·55 0·01 0·1 1 10 100 Favours mesh Favours suture a Data from studies using onlay mesh reinforcement; b data from studies using retromuscular mesh reinforcement. A Mantel–Haenszel random-effects model was used for meta-analysis. Risk ratios are shown with 95 per cent confidence intervals. SSI, surgical-site infection. greater chance of developing seroma than those with a pri- the variability of surgical technique and methodological mary suture (RR 1⋅67; P = 0⋅17) (Fig. 6). The occurrence approach across studies. of SSI was not significantly higher for onlay PMR (RR Although the technique of PMR in the treatment arm of 0⋅82; P = 0⋅33) or retromuscular PMR (RR 0⋅85; P = 0⋅55) the RCTs is often well described, there is less information than for primary suture (Fig. 7). There were insufficient on the control group undergoing primary suture. Protocols data to analyse the incidence of haematoma and burst often describe the use of a suture : wound length ratio abdomen. (SL : WL) greater than 4 : 1, but data for the SL : WL ratio were recorded and reported in only a few studies11 , so Discussion that adherence to the optimal primary suture technique was unclear in most studies. Moreover, the short-stitch This meta-analysis has shown that the use of PMR in technique, which is currently considered the best evidenced patients undergoing midline laparotomy leads to a signif- technique with the lowest incidence of IH3 , was not used in icantly lower occurrence of IH compared with a primary any of the RCTs in this meta-analysis. Therefore, it could suture closure, with TSA indicating that the evidence was be argued that the treatment effect of PMR is increased firm. This significant effect was shown for both onlay and because of a suboptimal suturing technique in the control retromuscular PMR. PMR was found to be safe, with no groups. increase in SSI, but an increased risk of seroma formation Of the 12 RCTs included in this meta-analysis, half were for onlay PMR only. considered to have a high risk of bias. Owing to the use This meta-analysis has a few limitations. Substantial sta- of sensitivity analysis, however, the treatment effect was tistical heterogeneity was seen in the studies regarding the maintained, and was even greater when only studies with primary outcome (I 2 = 71 per cent). This probably reflects a low risk of bias were analysed. © 2020 The Authors. www.bjsopen.com BJS Open 2020; 4: 357–368 BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd 366 A. P. Jairam, M. López-Cano, J. M. Garcia-Alamino, J. A. Pereira, L. Timmermans, J. Jeekel et al. Only RCTs with a minimum follow-up of 12 months an inclusion criterion for PMR, but further research is were included in the meta-analysis, with all but one29 hav- needed to identify an appropriate BMI cut-off point at ing follow-up of at least 24 months. This is still too short, which the risk of developing IH increases. Data from large however, to evaluate the long-term efficacy or potential late clinically oriented prospective registries34 might be helpful adverse effects of PMR. to explore which factors might lead to an increased risk In most studies, physical examination was used to detect of IH. IH during follow-up. Some added selective or systematic PMR can be considered safe for elective laparotomy. medical imaging for the evaluation of IH, using either The only adverse event detected in this meta-analysis was ultrasound imaging or CT. Imaging will increase the num- an increased rate of seroma formation after onlay PMR, ber of patients diagnosed with IH by detection of subclin- related to the subcutaneous dissection required. How- ical IHs, and this might overestimate the importance of ever, even though the occurrence of seroma was higher PMR in providing clinical benefit. in the onlay PMR group, there was no increase in SSI. Most of the 12 included studies used a polypropylene There were insufficient data to analyse the number of other mesh in either an onlay or a retromuscular position; adverse events, such as haematoma or burst abdomen. two10,30 used a biological mesh and one5 used an absorbable One of the main strengths of this meta-analysis is the fact synthetic mesh. Two of these studies did not show PMR to that RCTs with a low risk of bias were analysed separately35 . be effective, and all three were considered to have a high Further, the most up-to-date RCTs were included23 . TSA risk of bias. All studies with a low risk of bias used either an showed firm evidence in favour of PMR for midline laparo- onlay or a retromuscular mesh position, and all included tomy in high-risk patients, suggesting that no further trials trials were performed in an elective surgery setting. The are required to address the effects of PMR in this popu- results of this meta-analysis on PMR can therefore be con- lation. RCTs in other patient populations with differing sidered valid only for synthetic, non-absorbable mesh in levels of risk would, however, be helpful to evaluate the either an onlay or a retromuscular position in an elective effectiveness of PMR. setting. Laparotomies are performed by a variety of surgical The slightly asymmetrical funnel plot for IH indicates specialties, such as vascular surgeons, colorectal surgeons, possible publication bias towards studies that favour PMR, gynaecologists and urologists. Many of those surgeons have resulting in an overestimation of the underlying beneficial little experience in treating abdominal wall hernias with effect of the intervention. mesh, particularly for retromuscular mesh placement. This Most studies included only patients considered at high meta-analysis has shown that onlay PMR, which is easier to risk for developing an IH. It is difficult to identify the indi- perform, is also effective, and likely to be more acceptable vidual risk factors whereby patients would benefit from to these surgeons. PMR. The guidelines on the closure of abdominal wall This meta-analysis has provided evidence in favour of incisions from the EHS19 state that the evidence is weak closure of midline laparotomies with PMR in high-risk for the use of PMR in patients at high risk of IH devel- patients. Individual risk factors should be taken into opment. Such a guideline could be implemented only if account to select patients who will benefit (most) from the EHS Guidelines Development Group also described PMR, which should become standard treatment for the exact criteria for considering patients to be at high high-risk groups. risk. From this perspective, the study performed by Fischer et al.18 is interesting, as these authors stratified patients Acknowledgements into four IH risk groups (low, moderate, high, extreme), based on characteristics of the patient and the surgical A.P.J. and M.L.-C. contributed equally to this study. procedure. From this meta-analysis, it seems clear that Disclosure: The authors declare no conflict of interest. patients undergoing AAA repair and those having bariatric surgery through a midline incision will benefit from PMR, References although these are now less common as many patients 1 Höer J, Lawong G, Klinge U, Schumpelick V. Factors with an AAA are treated with endovascular procedures, influencing the development of incisional hernia. A and bariatric surgery is often performed by a laparoscopic retrospective study of 2983 laparotomy patients over a period approach. of 10 years. Chirurg 2002; 73: 474–480. Selection based on patient characteristics needs cut-off 2 Mudge M, Hughes LE. Incisional hernia: a 10 year values. The appropriate cut-off value for BMI is unclear. prospective study of incidence and attitudes. Br J Surg 1985; The PRIMA trial23 used a BMI of at least 27 kg/m2 as 72: 70–71. © 2020 The Authors. www.bjsopen.com BJS Open 2020; 4: 357–368 BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd Prevention of incisional hernia with mesh 367 3 Millbourn D, Cengiz Y, Israelsson LA. Effect of stitch length 16 Burger JW, Luijendijk RW, Hop WC, Halm JA, on wound complications after closure of midline incisions: a Verdaasdonk EG, Jeekel J. Long-term follow-up of a randomized controlled trial. Arch Surg 2009; 144: randomized controlled trial of suture versus mesh repair of 1056–1059. incisional hernia. Ann Surg 2004; 240: 578–583. 4 Bosanquet DC, Ansell J, Abdelrahman T, Cornish J, 17 Helgstrand F, Rosenberg J, Kehlet H, Strandfelt P, Harries R, Stimpson A et al. Systematic review and Bisgaard T. Reoperation versus clinical recurrence rate after meta-regression of factors affecting midline incisional hernia ventral hernia repair. Ann Surg 2012; 256: 955–958. rates: analysis of 14 618 patients. PLoS One 2015; 10: 18 Fischer JP, Basta MN, Mirzabeigi MN, Bauder AR, Fox JP, e0138745. Drebin JA et al. A risk model and cost analysis of incisional 5 Pans A, Elen P, Dewé W, Desaive C. Long-term results of hernia after elective, abdominal surgery based upon 12 373 polyglactin mesh for the prevention of incisional hernias in cases: the case for targeted prophylactic intervention. Ann obese patients. World J Surg 1998; 22: 479–482. Surg 2016; 263: 1010–1017. 6 Gutiérrez de la Peña C, Medina Achirica C, 19 Muysoms FE, Antoniou SA, Bury K, Campanelli G, Conze J, Domínguez-Adame E, Medina Díez J. Primary closure of Cuccurullo D et al.; European Hernia Society. European laparotomies with high risk of incisional hernia using Hernia Society guidelines on the closure of abdominal wall prosthetic material: analysis of usefulness. Hernia 2003; 7: incisions. Hernia 2015; 19: 1–24. 134–136. 20 Borab ZM, Shakir S, Lanni MA, Tecce MG, MacDonald J, 7 Strzelczyk JM, Szymański D, Nowicki ME, Wilczyński W, Hope WW et al. Does prophylactic mesh placement in Gaszynski T, Czupryniak L. Randomized clinical trial of elective, midline laparotomy reduce the incidence of postoperative hernia prophylaxis in open bariatric surgery. Br incisional hernia? A systematic review and meta-analysis. J Surg 2006; 93: 1347–1350. Surgery 2017; 161: 1149–1163. 8 Bevis PM, Windhaber RA, Lear PA, Poskitt KR, Earnshaw 21 Wang XC, Zhang D, Yang ZX, Gan JX, Yin LN. Mesh JJ, Mitchell DC. Randomized clinical trial of mesh versus reinforcement for the prevention of incisional hernia sutured wound closure after open abdominal aortic aneurysm formation: a systematic review and meta-analysis of surgery. Br J Surg 2010; 97: 1497–1502. randomized controlled trials. J Surg Res 2017; 209: 17–29. 9 Abo-Ryia MH, El-Khadrawy OH, Abd-Allah HS. 22 Payne R, Aldwinckle J, Ward S. Meta-analysis of randomised Prophylactic preperitoneal mesh placement in open bariatric trials comparing the use of prophylactic mesh to standard surgery: a guard against incisional hernia development. Obes midline closure in the reduction of incisional herniae. Hernia Surg 2013; 23: 1571–1574. 2017; 21: 843–853. 10 Bali C, Papakostas J, Georgiou G, Kouvelos G, Avgos S, 23 Jairam AP, Timmermans L, Eker HH, Pierik REGJM, van Arnaoutoglou E et al. A comparative study of sutured versus Klaveren D, Steyerberg EW et al.; PRIMA Trialist Group. bovine pericardium mesh abdominal closure after open Prevention of incisional hernia with prophylactic onlay and abdominal aortic aneurysm repair. Hernia 2015; 19: sublay mesh reinforcement versus primary suture only in 267–271. midline laparotomies (PRIMA): 2-year follow-up of a 11 Muysoms FE, Detry O, Vierendeels T, Huyghe M, multicentre, double-blind, randomised controlled trial. Miserez M, Ruppert M et al. Prevention of incisional hernias Lancet 2017; 390: 567–576. by prophylactic mesh-augmented reinforcement of midline 24 Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA laparotomies for abdominal aortic aneurysm treatment: a Group. Preferred reporting items for systematic reviews and randomized controlled trial. Ann Surg 2016; 263: 638–645. meta-analyses: the PRISMA statement. PLoS Med 2009; 6: 12 Deerenberg EB, Harlaar JJ, Steyerberg EW, Lont HE, van e1000097. Doorn HC, Heisterkamp J et al. Small bites versus large bites 25 Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, for closure of abdominal midline incisions (STITCH): a Oxman AD et al.; Cochrane Bias Methods Group; Cochrane double-blind, multicentre, randomised controlled trial. Statistical Methods Group. The Cochrane Collaboration’s Lancet 2015; 386: 1254–1260. tool for assessing risk of bias in randomised trials. BMJ 2011; 13 van Ramshorst GH, Eker HH, Hop WC, Jeekel J, Lange JF. 343: d5928. Impact of incisional hernia on health-related quality of life 26 Thorlund K, Imberger G, Walsh M, Chu R, Gluud C, and body image: a prospective cohort study. Am J Surg 2012; Wetterslev J et al. The number of patients and events 204: 144–150. required to limit the risk of overestimation of intervention 14 van Ramshorst GH, Eker HH, van der Voet JA, Jeekel J, effects in meta-analysis – a simulation study. PLoS One 2011; Lange JF. Long-term outcome study in patients with 6: e2549. abdominal wound dehiscence: a comparative study on quality 27 Thorlund K, Engstrøm J, Wetterslev J, Brok J, Imberger G, of life, body image, and incisional hernia. J Gastrointest Surg Gluud C. User Manual for Trial Sequential Analysis (TSA). 2013; 17: 1477–1484. Copenhagen Trial Unit, Centre for Clinical Intervention 15 Gillion JF, Sanders D, Miserez M, Muysoms F. The Research: Copenhagen, 2011. economic burden of incisional ventral hernia repair: a 28 El-Khadrawy OH, Moussa G, Mansour O, Hashish MS. multicentric cost analysis. Hernia 2016; 20: 819–830. Prophylactic prosthetic reinforcement of midline © 2020 The Authors. www.bjsopen.com BJS Open 2020; 4: 357–368 BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd 368 A. P. Jairam, M. López-Cano, J. M. Garcia-Alamino, J. A. Pereira, L. Timmermans, J. Jeekel et al. abdominal incisions in high-risk patients. Hernia 2009; 13: 32 Timmermans L, Eker HH, Steyerberg EW, Jairam A, de 267–274. Jong D, Pierik EG et al. Short-term results of a randomized 29 Caro-Tarrago A, Olona Casas C, Jimenez Salido A, Duque controlled trial comparing primary suture with primary Guilera E, Moreno Fernandez F, Vicente Guillen V. glued mesh augmentation to prevent incisional hernia. Ann Prevention of incisional hernia in midline laparotomy with Surg 2015; 261: 276–281. an onlay mesh: a randomized clinical trial. World J Surg 33 Schünemann H, Brozek J, Guyatt G, Oxman A (eds). 2014; 38: 2223–2230. GRADE Handbook for Grading Quality of Evidence and 30 Sarr MG, Hutcher NE, Snyder S, Hodde J, Carmody B. A Strength of Recommendations; 2013. http://gdt prospective, randomized, multicenter trial of Surgisis Gold, a .guidelinedevelopment.org/app/handbook/handbook.html biologic prosthetic, as a sublay reinforcement of the fascial [accessed 1 July 2017]. closure after open bariatric surgery. Surgery 2014; 156: 34 Sedrakyan A, Campbell B, Graves S, Cronenwett JL. 902–908. Surgical registries for advancing quality and device 31 García-Ureña MÁ, López-Monclús J, Hernando LA, Montes surveillance. Lancet 2016; 388: 1358–1360. DM, Valle de Lersundi AR, Pavón CC et al. Randomized 35 Roberts I, Ker K, Edwards P, Beecher D, Manno D, controlled trial of the use of a large-pore polypropylene Sydenham E. The knowledge system underpinning mesh to prevent incisional hernia in colorectal surgery. Ann healthcare is not fit for purpose and must change. BMJ 2015; Surg 2015; 261: 876–881. 350: h2463. Supporting information Additional supporting information can be found online in the Supporting Information section at the end of the article. © 2020 The Authors. www.bjsopen.com BJS Open 2020; 4: 357–368 BJS Open published by John Wiley & Sons Ltd on behalf of BJS Society Ltd