Retromuscular (sublay) synthetic mesh reinforcement versus no mesh at end-colostomy creation to prevent parastomal hernia: a GRADE-assessed systematic review and meta-analysis of randomized controlled trials featuring subgroup analysis by CT-mandated versus clinical detection

Abstract

Background

Parastomal hernia is a major complication following end-colostomy, affecting 30–70% of patients and imposing significant healthcare burdens. Prophylactic mesh reinforcement has emerged as a preventive strategy; however, evidence synthesis has been limited by methodological heterogeneity across diverse anatomical approaches to hernia repair.

Methods

This GRADE-assessed systematic review and meta-analysis was prospectively registered in the PROSPERO database (ID: CRD420251143346). We searched MEDLINE, Embase, Cochrane CENTRAL, Scopus, Web of Science, ClinicalTrials.gov, and WHO ICTRP until August 2025 for randomized controlled trials comparing retromuscular synthetic mesh reinforcement versus no mesh during end colostomy creation. The primary outcomes were clinical parastomal hernias and safety parameters. Random-effects meta-analyses were performed, and comprehensive sensitivity and subgroup analyses were included.

Results

Seven randomized trials (956 patients) were included in this meta-analysis. Prophylactic retromuscular mesh reduced clinical parastomal hernia rates (OR 0.44, 95% CI 0.22–0.89; P = 0.02), although heterogeneity was substantial (I²=73%). The control group parastomal hernia rate was 312 per 1,000 patients; prophylactic mesh reduced this to 213 per 1,000 (99 fewer hernias per 1,000 patients; 95% CI 28–213 fewer). However, in the leave-one-out analysis, the reduction in clinical parastomal hernia was no longer statistically significant after exclusion of one influential trial (OR 0.58, 95% CI 0.31–1.10), indicating that the pooled estimate was sensitive to study selection. Safety outcomes showed no significant differences in the incidence of composite complications (OR 1.15, 95% CI 0.84–1.59), stoma necrosis (OR 0.79, 95% CI 0.31–2.07), reoperation (OR 0.88, 95% CI 0.52–1.48), or surgical site infection (OR 0.94, 95% CI 0.57–1.55). Operative time (MD + 8.52 min, 95% CI − 8.38 to + 25.42) and hospital stay (MD + 3.69 days, 95% CI − 1.52 to + 8.90) also showed no significant differences. The certainty of evidence for the primary outcome was low because of substantial inconsistency, indicating limited confidence in the precision and robustness of the effect estimates.

Conclusion

This GRADE-assessed systematic review and meta-analysis suggests that prophylactic retromuscular synthetic mesh may reduce clinically detected parastomal hernia after end-colostomy creation; however, the certainty of this benefit remains low because the pooled signal was heterogeneous, attenuated in trials using standardized CT-based assessment, and sensitive to individual-study removal. These findings support cautious and selective use rather than routine universal implementation.

Introuction

Parastomal hernia (PSH) is one of the most frequent long-term complications after permanent end-colostomy and remains a major source of morbidity, impaired stoma function, and reduced quality of life. Its reported incidence varies substantially according to follow-up duration, diagnostic method, and patient selection; however, the burden is consistently high enough to make prevention a major surgical priority [1]. Updated randomized evidence suggests that prophylactic mesh may reduce PSH occurrence, although the magnitude and durability of the benefit remain uncertain because more recent trials have yielded less consistent results [2]. Mechanistically, PSH reflects the interaction between a surgically created abdominal wall defect, chronic tangential forces at the stoma aperture, progressive tissue remodeling, and patient-related susceptibility factors, all of which complicate prevention and repair [3].

This pathophysiological rationale has driven interest in prophylactic reinforcement during stoma formation. Broader experience in abdominal wall surgery shows that permanent synthetic mesh can reduce subsequent hernia formation in high-risk fascial defects, including stoma-related settings, without necessarily increasing early wound morbidity [4]. In permanent end-colostomy populations, updated meta-analyses have suggested a protective effect of prophylactic meshes; however, pooled estimates remain limited by heterogeneity in the mesh plane, operative technique, and outcome definition [5]. Economic evaluations further suggest that prophylactic reinforcement may be cost-effective under selected assumptions, although such models depend heavily on the baseline hernia risk and long-term durability of the benefit [6].

Notwithstanding these promising results, the clinical problem remains unresolved. Established PSH is difficult to manage, and even retromuscular synthetic mesh repair is challenged by wound morbidity, recurrence risk, and technical complexity of reconstructing the abdominal wall around a functioning stoma [7]. Preventive strategies have also evolved with newer constructs, such as three-dimensional meshes; however, these innovations add further heterogeneity to an already fragmented evidence base [8]. More recent systematic reviews, meta-analyses, and network meta-analyses suggest that mesh-based prevention may be beneficial overall; however, they combine different mesh materials, placement planes, stoma types, and study designs, thereby limiting direct inference for retromuscular synthetic meshes specifically [9].

Accordingly, an important evidence gap remains regarding whether retromuscular synthetic mesh reinforcement specifically provides a reproducible advantage over no mesh during permanent end-colostomy creation. This distinction is clinically important because evidence from stoma-site closure and related hernia prevention settings cannot be directly extrapolated to index colostomy PSH prevention [10]. To address this gap, the present study was designed as the first systematic review and meta-analysis focused specifically on retromuscular synthetic mesh versus no mesh, incorporating sensitivity analyses, subgroup analyses, and GRADE-based certainty assessments to provide a robust and clinically relevant synthesis of the available evidence.

Methods

Study design

We conducted a systematic review and meta-analysis of randomized controlled trials (RCTs) comparing retromuscular (sublay/retro-rectus) synthetic mesh reinforcement versus no mesh at the time of end-colostomy creation for the prevention of parastomal hernia (PSH). The review followed the PRISMA 2020 guidelines and adhered to the Cochrane methodological standards for study identification, screening, data collection and synthesis. Only RCTs were eligible; quasi-randomized and observational studies were excluded from the review. Our analytic plan, including outcomes, effect measures, and sensitivity/subgroup analyses, was specified in advance and prospectively registered in the PROSPERO database (ID: CRD420251143346).

Literature search strategy

We systematically searched major bibliographic databases from inception to August 7, 2025, without any restrictions on language or publication status. Controlled vocabulary (e.g., MeSH/Emtree) and keywords were combined with Boolean operators and an RCT filter to identify the relevant studies. Reference lists of relevant trials and reviews were hand-searched, and trial registries were queried to identify ongoing or unpublished studies. Duplicates were removed prior to screening. The search details and record yields are presented in Table 1.

Table 1.

Electronic search strategy and record yields

Database / source	Key search terms used (exact syntax shortened for display)	Results returned*
MEDLINE (via PubMed)	(colostomy[MeSH] OR colostom*[tiab] OR “end colostomy“[tiab]) AND (“parastomal hernia“[tiab] OR “stoma hernia“[tiab] OR “paracolostomy hernia“[tiab]) AND (mesh*[tiab] OR prosthes*[tiab]) AND (prophylax*[tiab] OR prevent*[tiab]) AND (random*[tiab] OR randomized controlled trial[pt])	≈ 400
Embase (Emtree + keywords)	(‘colostomy’/exp OR colostom*:ti, ab) AND (‘parastomal hernia’/exp OR ‘stoma hernia’:ti, ab) AND (mesh*:ti, ab OR prosthesis/exp) AND (prophylaxis/exp OR prevent*:ti, ab) AND (‘randomized controlled trial’/exp)	≈ 600
Cochrane CENTRAL	Colostom* AND (parastomal OR stoma) AND (mesh OR prosthesis)	≈ 120
Scopus	TITLE-ABS-KEY(colostom* AND (parastomal OR stoma) AND (mesh OR prosthesis) AND (prophylax* OR prevent*) AND random*)	≈ 650
Web of Science Core Collection	TS=(colostom* AND (parastomal OR stoma) AND (mesh OR prosthesis) AND (prophylax* OR prevent*) AND random*)	≈ 500
ClinicalTrials.gov	Condition/terms: parastomal hernia OR stoma hernia; Other terms: colostomy AND mesh; Study type: Interventional; Phase: All	≈ 20
WHO ICTRP	parastomal hernia OR stoma hernia AND colostomy AND mesh	≈ 20
Google Scholar	“parastomal hernia” prophylactic mesh randomized colostomy	≈ 3,000 (first 300 screened)

Inclusion and exclusion criteria

We included parallel-group randomized controlled trials enrolling adult patients (aged ≥ 18 years) who underwent permanent end colostomy formation during index colorectal surgery. Eligible trials compared prophylactic retromuscular (sublay/retro-rectus) synthetic mesh reinforcement with no mesh (standard stoma construction). We excluded non-randomized, quasi-randomized, crossover, cluster-randomized (without appropriate analysis), and observational designs; trials using non-retromuscular planes (e.g., intraperitoneal onlay/keyhole-IPOM or Sugarbaker-IPOM), biologic mesh only, loop stomas/ileostomies/urostomies/temporary stomas, pediatric populations, animal studies, modeling studies, narrative reviews, case reports/series, and conference abstracts without extractable data. Trials mixing end colostomy with other stoma types were excluded unless end colostomy-specific results were separable. For the three-arm Stoma-Const trial comparing prophylactic mesh versus cruciate fascial incision versus circular fascial incision, we retained the mesh arm and combined the two non-mesh arms into a single pooled comparator by summing events and sample sizes ((n) pooled = (n) cruciate + (n) circular; (events) pooled = (events) cruciate + (events) circular), following the Cochrane Handbook Sect.  23.3.4 to avoid unit-of-analysis error from double counting. When multiple reports described the same cohort, we used the most complete or longest follow-up period for each endpoint.

Data extraction and outcome measures

Two reviewers (W.M. and A.K.) independently and in duplicate screened, selected, and extracted study- and arm-level data using a pilot template. We captured trial identifiers, setting (CRC/IBD/diverticular), surgical approach (laparoscopic/open), mesh specification (type/brand/weight), mesh plane and fixation method, trephine technique, PSH ascertainment method, follow-up schedule, and all prespecified outcomes with their denominators and time windows. For dichotomous outcomes, we extracted events and totals (n/N) per arm; for continuous outcomes, we extracted the mean ± standard deviation (SD) with the corresponding units. When only medians and dispersion (IQR/range) were available, we converted to mean/SD using Wan et al. and Luo et al. formulae, noting transformations in the dataset, and conducted sensitivity analyses by excluding transformed studies. We preferentially used intention-to-treat counts; when trials reported alternative analysis sets, we recorded both and used the set closest to the ITT for primary analyses. For outcomes reported at multiple time points, we extracted the measurement closest to the prespecified window (12 ± 3 months [i.e., 9–15 months] for late outcomes; 30–90 days for early outcomes) and recorded the exact timing.

Our primary outcomes were clinical parastomal hernia at 12 months (± 3 months), ascertained by the trial’s primary detection method categorized as either mandated CT imaging (performed in all patients regardless of symptoms) or clinical assessment ± selective CT (physical examination with imaging only when clinically indicated), accepting each trial’s operative hernia definition; any early postoperative complication (composite) within 30–90 days; stoma necrosis within 30–90 days; early reoperation within 30–90 days; and surgical site infection within 30–90 days. Secondary outcomes included stoma stenosis/stricture, stoma prolapse, peristomal infection (all ≤ 30–90 days), midline incisional hernia at ≥ 12 months, length of hospital stay (days), operative time (minutes), and quality of life scores at ≥ 12 months (pooled with SMD if the instruments differed).

The primary efficacy endpoint was a clinically detected parastomal hernia at a prespecified follow-up, defined according to each trial’s clinical assessment framework. Imaging-detected parastomal hernias identified on protocol-mandated CT but not meeting the clinical endpoint definition were recorded descriptively. Because the ascertainment methods differed across trials, we prespecified an exploratory subgroup analysis comparing studies using mandated CT surveillance in all participants with those relying on clinical assessment with or without selective CT.

Quality (risk-of-bias) assessment

Two reviewers (W.M. and M.E.K.) independently assessed the risk of bias at the outcome level using the Cochrane RoB 2 tool, evaluating the randomization process, deviations from the intended interventions, missing outcome data, measurement of the outcome, and selection of reported results. Disagreements were resolved by consensus agreement. We summarized the judgments as low risk, some concerns, or high risk, and prepared domain-level and overall RoB figures. Risk-of-bias judgments informed sensitivity analyses and GRADE certainty assessment, with Domain 4 (measurement of outcome, specifically unblinded clinical PSH assessment) weighted most heavily in determining the overall bias impact.

Statistical analysis

All meta-analyses were conducted using random-effects models. Between-study variance (τ²) was estimated using the Paule–Mandel method as the primary estimator, given its favorable performance in meta-analyses with relatively few studies. For dichotomous outcomes, pooled odds ratios (ORs) with 95% confidence intervals were calculated using the Mantel–Haenszel random effects method. ORs were selected as the primary effect measure because baseline event rates varied substantially across studies (control group PSH risk approximately 15%-50%), making ORs more stable for pooled estimation. To improve clinical interpretability, the pooled OR for the primary outcome was translated into an absolute effect per 1,000 patients using a representative baseline control risk derived from the included trials, consistent with the GRADE framework. For continuous outcomes, mean differences (MDs) were used when outcomes were measured on identical scales, and standardized mean differences (SMDs) were used when the scales differed.

Statistical heterogeneity was quantified using I² and τ. For outcomes with at least five studies, the Hartung–Knapp–Sidik–Jonkman (HKSJ) adjustment was applied to provide more conservative confidence intervals. For dichotomous outcomes with zero events in one arm, a continuity correction of 0.5 was applied in accordance with Cochrane guidance; double-zero studies were retained in the primary Mantel–Haenszel analysis and were additionally explored in sensitivity analyses using treatment-arm continuity correction or study exclusion. The preplanned analytic procedures also included combining multiple control arms where applicable, using the longest follow-up nearest to the prespecified outcome window, and conducting leave-one-out analyses to assess the influence of the study.

Sensitivity analysis, subgroup analysis, and GRADE-certainty of evidence

Sensitivity analysis used the leave-one-out approach, in which each study was sequentially removed, and the meta-analysis was rerun to identify influential studies and assess the robustness of the pooled estimates and heterogeneity. This approach informed our interpretation of the results by revealing whether the overall conclusions depended on a single trial. In the three-arm Stoma-Const trial, we consistently used the combined non-mesh arms to avoid double counting in all analyses.

Subgroup analysis was prespecified only for the primary endpoint (clinical PSH), stratified by detection method (mandated CT [standardized imaging in all patients] vs. clinical assessment ± selective CT [imaging only when indicated]), with random-effects pooling within subgroups, χ² test for subgroup differences (p < 0.10 threshold), and the requirement of ≥ 2 studies per stratum. This subgroup analysis directly informed our GRADE assessment of risk of bias (Domain 4) and inconsistency: trials using mandated CT were judged to have a lower risk of detection bias, and we hypothesized that CT-mandated trials would show smaller, more conservative effects than clinical-only trials because the superior sensitivity of standardized imaging detects more asymptomatic hernias in both groups, attenuating the apparent mesh benefit, whereas clinical assessment may selectively identify symptomatic hernias (more likely in controls), amplifying the observed treatment effect. All subgroup findings were deemed exploratory and interpreted cautiously without multiplicity adjustments. Residual unexplained heterogeneity after subgroup stratification contributed to GRADE downgrading for inconsistency.

Certainty was rated with GRADE at the outcome level (RCTs start high), downgrading for risk of bias (one level if ≥ 50% weight from “some concerns” studies or any “high risk” study contributing ≥ 20% weight, with Domain 4 [measurement] weighted most heavily; two levels if ≥ 50% weight from “high risk” studies), inconsistency (one level if I²≥50% with poor CI overlap but consistent direction; two levels if I²≥75% with contradictory directions), indirectness (one level for deviation from PICO: adult permanent end-colostomy, retromuscular synthetic mesh vs. no mesh, clinical PSH at ~ 12 months), imprecision (one level if 95% CI crossed clinically important thresholds [OR 0.75 or 1.25] and optimal information size not met [~ 750 patients for 80% power to detect OR 0.50 assuming 30% control rate]; two levels if CI crossed both benefit and harm or sample < 50% OIS), and publication bias (assessed only if ≥ 10 studies; not met). For dichotomous outcomes, we reported pooled ORs and calculated absolute effects per 1,000 participants using the pooled control risk (total events/total patients in the control arms). For continuous outcomes, we used MD or SMD.

Results

Of the 2,310 records identified across databases and trial registers, 1,170 were removed before screening (duplicates/automation/other), leaving 1,140 unique records. After excluding 1,010 studies based on title/abstract, we sought 130 full texts; 126 were retrieved and assessed, of which 116 were excluded primarily for non-randomized design or non-retromuscular/biologic mesh interventions. Ultimately, seven RCTs met the inclusion criteria and were included in the analysis (Fig. 1).

Fig. 1.

PRISMA 2020 flow diagram of study selection for the RCT-only review of retromuscular (sublay) mesh vs. no mesh at end-colostomy creation

Methodological quality and study integrity

The traffic-light plot shows one study at high overall risk, Jänes [11], driven mainly by missing outcome data (D3) and non-blinded outcome measurement (D4); all other six RCTs are “some concerns” overall, with generally low risk in randomization (D1) and deviations from intended interventions (D2), particularly in the multicenter trials (PREVENT, STOMAMESH, GRECCAR-7, Stoma-Const) (Fig. 2). The summary bars confirm this pattern: D1 and D2 are predominantly low risk, D3 shows a mix of some concerns with a small high-risk fraction, and D4 and D5 (measurement and selective reporting) are mostly some concerns reflecting clinical (not always blinded) outcome assessments and variable reporting practices (Fig. 3).

Fig. 2.

Cochrane RoB 2 “traffic-light” assessment by study and domain (D1–D5) with an overall judgment for each included RCT

Fig. 3.

Cochrane RoB 2 domain summary across studies—proportion judged low risk, some concerns, or high risk, plus overall RoB

Characteristics of included studies

Across seven European RCTs (433 mesh vs. 523 controls), the included studies were largely multicenter and contemporary: PREVENT (Netherlands), STOMAMESH (Sweden), GRECCAR-7 (France), and Stoma-Const (Sweden/Denmark), with two earlier single-center trials from Sweden and Spain (Table 2). Allocation concealment was adequate in the later trials, and true blinding was achieved in the STOMAMESH and GRECCAR-7 trials. PSH assessment was standardized radiologically in PREVENT (mandated CT) and STOMAMESH (prone CT), while others relied on clinical ± CT confirmation, with primary follow-up most commonly at 12 months (GRECCAR-7, 24 months; Lambrecht, median 36 months) (Table 2). Baseline demographics were typical for permanent end-colostomy: median age ~ 64–70 years, BMI ~ 25–27 kg/m², female proportion ~ 20–45%, and predominant oncologic indication (≥ 80–90% where reported); urgent/emergency cases were rare, limiting generalizability to emergent stomas (Table 3). Operative details were strikingly consistent: all trials used synthetic polypropylene mesh placed retromuscular (sublay) in a keyhole configuration at index colostomy, with fixation most commonly retrorectus lateral-corner sutures to the rectus sheath (posterior-sheath suturing in GRECCAR-7); approaches were mainly open, with limited laparoscopy in GRECCAR-7 and Stoma-Const, and the three-arm Stoma-Const compared mesh against two no-mesh fascial-incision techniques (cruciate vs. circular) (Table 4). Collectively, these features indicate a broadly comparable population and standardized prophylactic technique, supporting pooling while acknowledging heterogeneity in detection methods and follow-up periods (Tables 2, 3 and 4).

Table 2.

Characteristics of randomized controlled trials comparing retromuscular/sublay prophylactic synthetic mesh versus no mesh (or non-mesh techniques) at end-colostomy creation for prevention of parastomal hernia (PSH)

Study (first author, year)	Country / centers	Study design	Randomization & allocation concealment	Blinding / outcome assessment	Experimental arm (mesh details)	n (exp)	Control arm	n (ctrl)	Key inclusion criteria	Key exclusion criteria (as reported)	Procedure details (both arms)	Primary outcomes reported	Secondary outcomes reported	Analysis set / notes
Jänes 2004 [11]	Sweden; single center	Parallel-group RCT	Computer randomization; sealed opaque envelopes used for allocation; 1:1	Not blinded; outcomes by clinical exam	Retromuscular (sublay) key-hole synthetic mesh (Vypro; ~10 × 10 cm) placed around colostomy at index surgery	27	Standard end colostomy without mesh	27	Adults undergoing permanent end colostomy	Not fully detailed in text (emergency cases rarely included)	Open colostomy creation in both arms; mesh key-hole around bowel in retrorectus plane	Parastomal hernia (PSH) / stoma complications at ~ 12 mo	Wound complications, infections, reoperation	ITT not explicitly stated; small single-center trial
Serra-Aracil 2009 [27]	Spain; single center	Parallel-group RCT	Randomization method not fully detailed in abstract; 1:1	Not blinded; outcome assessed clinically and by CT classification	Retromuscular (sublay) polypropylene (light-weight) key-hole mesh (“Ultrapro” reported in trial series)	27	Standard end colostomy without mesh	27	Permanent end colostomy (mostly APR)	Not clearly specified	Mesh key-hole in retrorectus space at index APR/colostomy vs. identical surgery with no mesh	PSH rate (clinical & CT) at ≈ 12 mo	Morbidity/mortality, stoma complications	Per-protocol and stated totals as reported
Lambrecht 2015 (Colorectal Dis.) [12]	Norway; multicenter (≥ 2 hospitals)	Parallel-group RCT	Block randomization; allocation via sealed envelopes	Not blinded; outcomes assessed clinically; CT used for orifice measures	Retromuscular (sublay) lightweight polypropylene key-hole mesh at index surgery	32	Standard end colostomy without mesh	26	Adults planned for permanent end colostomy	Not fully detailed	Mesh key-hole in retrorectus vs. identical technique without mesh	PSH occurrence	Incisional hernia, complications, reoperations; stoma orifice area change	Small multicenter; reported ORs and long follow-up
Brandsma 2017 (Dutch PREVENT RCT) [24]	Netherlands; multicenter	Parallel-group RCT	Central randomization; 1:1	Outcome assessors (radiology) blinded to group per protocol	Retromuscular (sublay) polypropylene key-hole mesh during colostomy formation	72	Standard end colostomy without mesh	78	Adults scheduled for permanent end colostomy	Standard contraindications to mesh	Uniform technique across centers; mesh retrorectus key-hole vs. no mesh	PSH at 12 mo	Surgical site infection, reoperation, QoL, costs	ITT; predefined radiologic criteria
Odensten 2019 (STOMAMESH) [26]	Sweden; multicenter	Double-blind parallel-group RCT	Central computer randomization; block stratified by center	Patients and radiologic assessors blinded	Retromuscular (sublay) lightweight polypropylene key-hole mesh	114	Standard end colostomy without mesh	118	Adults undergoing permanent end colostomy	None specific beyond safety; emergency cases excluded	Same open technique; mesh retrorectus around bowel vs. no mesh	PSH at 12 mo	30-day & 12-mo complications; reoperations; QoL	ITT
Prudhomme 2021 (GRECCAR-7) [18]	France; 18 hospitals	Prospective, double-blinded, multicenter RCT	Block randomization, stratified by center	Patients and assessors blinded	Preperitoneal/retromuscular (sublay) synthetic mesh placed at index colostomy	98	Standard end colostomy without mesh	101	Adults with end colostomy formation	Not detailed beyond safety criteria	Standardized formation; mesh retrorectus around stoma vs. no mesh	PSH (time-to-event analysis)	Post-op complications, readmissions, QoL	ITT; multicenter
Correa-Marinez 2021 (Stoma-Const) [25]	Sweden & Denmark; 3 hospitals	Three-arm randomized multicenter trial	Central randomization 1:1:1	Not blinded; radiology standardized	Arm C: retromuscular (sublay) prophylactic mesh. Arm A: Cruciate fascial incision (no mesh). Arm B: circular fascial incision (no mesh)	63	Cruciate or circular fascial incision without mesh	146 (74 + 72)	Adults scheduled for end colostomy	Not specifically detailed in abstract	Open colostomy; mesh key-hole retrorectus in Arm C; same stoma construction otherwise	PSH within 12 mo	Post-op complications, readmissions, reoperations; risk factors	ITT; three-arm design (mesh vs. two non-mesh techniques)

APR abdominoperineal resection, CT computed tomography, exp experimental, ITT intention-to-treat, n number of patients, PSH parastomal hernia, QoL quality of life, RCT randomized controlled trial, sublay/retromuscular mesh placed posterior to rectus muscle and anterior to posterior rectus sheath, Ultrapro/Vypro are polypropylene-based meshes, PREVENT Dutch multicenter RCT, STOMAMESH Swedish multicenter RCT, GRECCAR-7 French multicenter RCT, Stoma-Const Swedish/Danish three-arm RCT of colostomy construction methods, including a prophylactic mesh arm

Table 3.

Patient demographics and clinical characteristics of participants in the seven included randomized trials comparing retromuscular sublay synthetic mesh reinforcement versus no mesh during end colostomy creation

Study (year)	N (Mesh / Control)	Age (yrs)*	Female (%)	BMI (kg/m²) *	ASA ≥ III (%)	Smokers (%)	Diabetes (%)	COPD (%)	Cancer indication (%)	Emergency cases (%)
Janes 2004 [11]	27 / 27	70 (95% CI 64–75) / 71 (67–76)	44.4 / 40.7	26 (95% CI 24–28) / 27 (25–29)	NR	NR	22.2 / 25.9	NR	92.6 / 81.5	3.7 / 14.8
Serra-Aracil 2009 [27]	27 / 27	67.5 ± 8.8 / 67.2 ± 9.7	18.5 / 29.6	25.6 ± 2.9 / 27.3 ± 3.5	NR	NR	NR	NR	NR	NR
Lambrecht 2015 [12]	32 / 26	64 ± 4.0 / 63 ± 4.1	31 / 19	24.6 ± 0.6 / 25.5 ± 0.8	NR	NR	NR	NR	NR	NR
Brandsma (PREVENT) 2015 [24]	72 / 78	63.6 / 63.1	40.3 / 37.2	26.7 / 26.5	NR	NR	8.3 / 9.0	NR	87.5 / 88.5	NR
Odensten (STOMAMESH) 2019 [26]	114 / 118	69.7 [41–86] / 69.9 [35–89]	35 / 48	26.1 [16.7–37.8] / 26.3 [18.5–43.7]	26 / 28	11 / 7	NR	NR	93 / 90	1 / 0 (planned 99 / 100)
Prudhomme (GRECCAR-7) 2021 [18]	98 / 101	67.2 ± 12.4 / 70.5 ± 11.1	41.8 / 43.6	25.6 ± 4.6 / 24.8 ± 4.7	NR	30.6 / 13.9	NR	7.1 / 5.0	83.7 / 89.1	NR
Correa-Marínez (Stoma-Const) 2021 [25]	63 / 146	68 [33–91] / Controls: Cruciate 72 [34–94]; Circular 68 [34–87]	33 / Controls: 42; 49	26.3 [17.3–40.4] / Controls: 24.7 [18.7–49.4]; 25.6 [16.7–39.6]	NR	NR	NR	NR	NR	NR

APR abdominoperineal resection, AR-ACP abdominal resection with anal canal preservation, ASA American Society of Anesthesiologists physical status, BMI body mass index, CI confidence interval, COPD chronic obstructive pulmonary disease, Mesh prophylactic (retromuscular sublay) synthetic mesh reinforcement, No-mesh standard colostomy creation without mesh, APE abdominoperineal excision (APR)

Table 4.

Procedural and trial-level characteristics of the included randomized controlled trials comparing retromuscular (sublay/retro-rectus) synthetic mesh versus no mesh at end-colostomy creation

Study (year)	N (Mesh / Ctrl)	Setting (Cancer / IBD / Diverticular) *	Approach (Lap/Open)	Mesh type (brand if stated)	Mesh plane	Fixation	Trephine technique	PSH detection method	Primary follow-up (mo)	Pre-op stoma marking	Index operation mix (APR / Hartmann / Colostomy-only)
Janes 2004 [11]	27 / 27	Malignancy (47)	Open	Lightweight polypropylene	Retromuscular (sublay), key-hole	Retrorectus; lateral corner sutured to rectus sheath	NR	Clinical (CT if indicated)	12	NR	NR
Serra-Aracil 2009 [27]	27 / 27	Benign (7) Malignancy	Open	Lightweight polypropylene	Retromuscular (sublay), key-hole	Retrorectus; not described	NR	Clinical + CT (Moreno-Matias’s grading)	≈ 12	NR	APR 85.2% / 81.5%; AR-ACP 14.8% / 18.5%
Lambrecht 2015 [12]	32 / 26	Malignancy	Open	Lightweight polypropylene	Retromuscular (sublay), key-hole	Retrorectus; lateral corner sutured to rectus sheath	NR	Clinical (CT used for orifice area)	Median 36 (6–87)	NR	NR
Brandsma (PREVENT) 2017 [24]	72 / 78	Malignancy (132)	Open	Polypropylene	Retromuscular / pre-peritoneal, key-hole	Retrorectus; lateral corners sutured	NR	CT (mandated at 12 mo) + clinical, blinded reading	12	NR	NR
Odensten (STOMAMESH) 2019 [26]	114 / 118	Malignancy (212) Benign (20)	Open	Lightweight polypropylene	Retromuscular (sublay), key-hole	Retrorectus; lateral corners sutured	NR	CT (prone) + clinical (blinded)	12	NR	NR
Prudhomme (GRECCAR-7) 2021 [18]	98 / 101	Malignancy (172) Benign (47)	Both	Synthetic polypropylene	Retromuscular (sublay), key-hole	Retrorectus; sutured to posterior rectus sheath	NR	Clinical (scheduled); imaging if indicated (blinded)	24	≈ 96% both arms	NR
Correa-Marínez (Stoma-Const) 2021 [25]	63 / 146 (cruciate 74 + circular 72)	Malignancy (170) Other (42)	Both	Polypropylene	Retromuscular (sublay), key-hole	Retrorectus; not described	Controls define fascial incision: cruciate vs. circular	CT (prone) + clinical (centralized)	12	≈ 100% all arms	Mesh: APE 71% / Hartmann 19% / Colostomy-only 10% (controls reported separately)

APR abdominoperineal resection, APE abdominoperineal excision (synonymous with APR in these trials), AR-ACP abdominal resection with anal canal preservation, CRC colorectal cancer, IBD inflammatory bowel disease, Lap laparoscopic, NR not reported, PSH parastomal hernia, CT computed tomography

Primary outcomes

Clinical parastomal hernia

Prophylactic retromuscular mesh was associated with a lower pooled odds of clinical PSH than no mesh (random-effects OR 0.44, 95% CI 0.22–0.89, P = 0.02), although heterogeneity was substantial (I² = 73%). Study-level effects were mixed: four trials had point estimates favoring the mesh, whereas three did not, and several confidence intervals were wide and crossed unity (Fig. 4).

Fig. 4.

Forest plot of clinical parastomal hernia (PSH), mesh versus no mesh; random-effects OR (95% CI). OR < 1 favors mesh

Any postoperative complication (less than or equal to 30 to 90 days, composite)

There was no difference in composite early complications (OR 1.15, 95% CI 0.84–1.59, P = 0.39; I² = 0%). The point estimates varied around the null hypothesis without evidence of between-study heterogeneity (Fig. 5).

Fig. 5.

Forest plot of any postoperative complication (early ≤ 30–90 d), mesh vs. no mesh; random-effects OR (95% CI). OR < 1 favors mesh

Stoma necrosis

The use of mesh did not change the risk of early stoma necrosis (OR 0.79, 95% CI 0.31–2.07, P = 0.64; I² = 0%). The estimates were imprecise owing to the low event counts (Fig. 6).

Fig. 6.

Forest plot of stoma necrosis (early), mesh vs. no-mesh; random-effects OR (95% CI). OR < 1 favors mesh

Reoperation (early, ≤ 30–90 d)

No significant difference was observed in the early reoperation rate (OR 0.88, 95% CI 0.52–1.48, P = 0.63; I² = 0%). The directions of the effect were mixed but consistently crossed the unity (Fig. 7).

Fig. 7.

Forest plot of reoperation (early), mesh vs. no mesh; random-effects OR (95% CI). OR < 1 favors mesh

Surgical site infection

Early SSI rates were similar between the groups (OR 0.94, 95% CI 0.57–1.55, P = 0.81; I² = 0%), arguing against a short-term infectious penalty with the mesh (Fig. 8).

Fig. 8.

Forest plot of surgical site infection (incisional; early), mesh vs. no mesh; random-effects OR (95% CI). OR < 1 favors mesh

Secondary outcomes

Stoma stenosis/stricture

No significant effect was observed (OR 1.49, 95% CI 0.48–4.59, P = 0.49; I² = 0%). The precision was limited by the number of events and contributing studies (Fig. 9).

Fig. 9.

Forest plot of stoma stenosis/stricture (early), mesh vs. no mesh; random-effects OR (95% CI). OR < 1 favors mesh

Stoma prolapses

The rates of early prolapse did not differ (OR 0.76, 95% CI 0.31–1.85, P = 0.55; I² = 0%). All study-level CIs crossed one (Fig. 10).

Fig. 10.

Forest plot of stoma prolapse (early), mesh vs. no mesh; random-effects OR (95% CI). OR < 1 favors mesh

Peristomal infection

No signal for increased peristomal infection with mesh (OR 0.82, 95% CI 0.29–2.32, P = 0.72; I² = 0%). Events were uncommon, yielding wide confidence intervals (CIs) (Fig. 11).

Fig. 11.

Forest plot of peristomal infection (early), mesh vs. no mesh; random-effects OR (95% CI). OR < 1 favors mesh

Incisional hernia—midline/main laparotomy (≥ 12 mo)

A non-significant trend favored the use of mesh for midline incisional hernia prevention (OR 0.45, 95% CI 0.16–1.26, P = 0.13; I² = 0%). The estimates were driven by two trials with longer follow-up periods (Fig. 12).

Fig. 12.

Forest plot of incisional hernia (midline, ≥ 12 months), mesh vs. no mesh; random-effects OR (95% CI). OR < 1 favors mesh

Length of hospital stay (days)

The mean stay was not different overall (MD + 3.69 days, 95% CI − 1.52 to 8.90, P = 0.16), with considerable heterogeneity (I² = 92%) reflecting varied practice patterns and imputation from medians (Fig. 13).

Fig. 13.

Forest plot of length of hospital stay (days), mesh vs. no-mesh; random-effects MD (95% CI). MD < 0 favors mesh

Operative time (minutes)

Mesh use did not significantly prolong operative time (MD + 8.52 min, 95% CI − 8.38 to 25.42, P = 0.32; I² = 62%). The differences were small and imprecise across the trials (Fig. 14).

Fig. 14.

Forest plot of operative time (minutes), mesh vs. non-mesh; random-effects MD (95% CI). MD < 0 favors mesh

Only clinical parastomal hernias showed a statistically significant benefit from mesh use (random-effects OR 0.44, 95% CI 0.22–0.89; Fig. 4). All other safety outcomes were not significant: any postoperative complication (Fig. 5), stoma necrosis (Fig. 6), reoperation (Fig. 7), surgical site infection (Fig. 8), stoma stenosis/stricture (Fig. 9), stoma prolapse (Fig. 10), and peristomal infection (Fig. 11). The estimate for midline incisional hernia suggested a reduction, but did not reach significance (OR 0.45, 95% CI 0.16–1.26; Fig. 12). Perioperative metrics were also not different: length of stay (MD + 3.69 days, Fig. 13) and operative time (MD + 8.52 min, Fig. 14). Heterogeneity was substantial for PSH and length of stay but low or absent for most early complications.

Sensitivity analysis

The leave-one-out analysis showed that the primary clinical PSH result was sensitive to the exclusion of Lambrecht [12]. After removal of this trial, the pooled effect was attenuated and no longer statistically significant (OR 0.58, 95% CI 0.31–1.10; I²=66% vs. 73% in the primary analysis), indicating fragility of the main efficacy signal (Table 5). Lambrecht [12] appeared influential because it contributed a moderate sample size, a relatively favorable effect estimate, and clinically assessed outcome ascertainment within a dataset otherwise characterized by heterogeneous follow-up durations and detection methods. Regarding operative time, excluding the small early trial, Serra-Aracil et al. [27] eliminated heterogeneity (I²=0%) and revealed a significant increase in operative time with mesh use (MD + 15.31 min, 95% CI 4.32–26.30 min). For length of stay, omitting Correa-Marínez [25] markedly reduced heterogeneity (I²=54% vs. 92%) and shifted the pooled effect toward the null without statistical significance (MD + 0.38 days, 95% CI − 1.85 to 2.61). Overall, the safety findings remained stable, but the apparent benefit of clinical PSH and perioperative time outcomes showed sensitivity to study selection.

Table 5.

Leave-one-out sensitivity analyses for outcomes with higher heterogeneity (random-effects models); “Before” = full set; “After” = pooled result after removing the specified studies. An OR < 1 and MD < 0 favored the mesh group

Outcome	Effect Estimate (Before)	P-value (Before)	I² % (Before)	Study Removed	Effect Estimate (After)	P-value (After)	I² % (After)
Clinical parastomal hernia (clinical)	OR 0.44 [0.22, 0.89]	0.02	73	Lambrecht et al., 2015 [12]	OR 0.58 [0.31, 1.10]	0.09	66
Operative time (min)	MD + 8.52 [− 8.38, 25.42]	0.32	62	Serra-Aracil et al., 2009 [27]	MD + 15.31 [4.32, 26.30]	0.006	0
Length of hospital stay (days)	MD + 3.69 [− 1.52, 8.90]	0.16	92	Correa-Marínez et al., 2021 [25]	MD + 0.38 [− 1.85, 2.61]	0.74	54

Subgroup analysis

The mesh reduced the overall incidence of clinically detected parastomal hernia (random-effects OR 0.44, 95% CI 0.22–0.89), but with substantial heterogeneity (I² = 73%; τ² = 0.57). In the exploratory subgroup analysis, the apparent effect was larger in studies using predominantly clinical assessment (OR 0.23, 95% CI 0.05–1.12; I² = 81%) and attenuated toward the null in studies with mandated CT surveillance (OR 0.65, 95% CI 0.30–1.42; I² = 69%). The subgroup interaction was not statistically significant (p = 0.25); therefore, these findings should be interpreted as hypothesis-generating rather than confirmatory (Fig. 15).

Fig. 15.

Subgroup analysis for clinical parastomal hernia by detection method— clinical assessment (± selective CT) vs. Mandated CT (standardized imaging)—random-effects ORs (95% CI); OR < 1 favors mesh

GRADE assessed certainty of evidence

Evidence certainty for the primary endpoint indicated a low-certainty reduction in clinical parastomal hernia with mesh (OR 0.44), translating to approximately 99 fewer PSH per 1,000 patients versus mesh nonuse. Downgrades were mainly due to the risk of bias (several unblinded trials) and inconsistency (I²≈73%) (Table 6). For early safety outcomes (any complication, reoperation, SSI, necrosis, stenosis/stricture, prolapse, and peristomal infection), the certainty ranged from low to very low, chiefly due to the risk of bias and imprecision with few events and CIs spanning benefits and harms, yielding small absolute differences per 1,000 patients. Incisional hernia (midline) suggested a possible reduction, but the certainty was low given only two trials and the imprecision of the results. Perioperative metrics (length of stay and operative time) were graded as very low because of serious/very serious heterogeneity and imprecision; therefore, any effect was uncertain. Overall, the GRADE profile supports the use of mesh as probably beneficial for PSH prevention without clear evidence of increased complications but emphasizes the need for further well-blinded, standardized, CT-mandated RCTs to solidify these estimates (Table 6).

Table 6.

GRADE assessment of the certainty of evidence for retromuscular (sublay) synthetic mesh versus no mesh at end-colostomy creation

Outcome	Risk of Bias	Inconsistency	Indirectness	Imprecision	Publication Bias	Effect estimate (95% CI)	Absolute effect (per 1,000)	GRADE certainty
Clinical parastomal hernia (12 mo)	Serious	Serious (I²≈73%)	Not serious	Not serious	Not serious	OR 0.44 (0.22–0.89)	Control: 312 → Mesh: 212 (Δ − 99/1,000)	⨁⨁◯◯ Low
Any postoperative complication (≤ 90 d)	Serious	Not serious (I²≈0%)	Not serious	Serious	Not serious	OR 1.15 (0.84–1.59)	435 → 423 (Δ − 12/1,000)	⨁⨁◯◯ Low
Stoma necrosis	Serious	Not serious (I²≈0%)	Not serious	Very serious	Not serious	OR 0.79 (0.31–2.07)	37 → 33 (Δ − 4/1,000)	⨁◯◯◯ Very low
Reoperation (≤ 90 d)	Serious	Not serious (I²≈0%)	Not serious	Serious	Not serious	OR 0.88 (0.52–1.48)	123 → 88 (Δ − 35/1,000)	⨁⨁◯◯ Low
Surgical-site infection	Serious	Not serious (I²≈0%)	Not serious	Serious	Not serious	OR 0.94 (0.57–1.55)	156 → 146 (Δ − 10/1,000)	⨁⨁◯◯ Low
Stoma stenosis/stricture	Serious	Not serious (I²≈0%)	Not serious	Very serious	Not serious	OR 1.49 (0.48–4.59)	24 → 40 (Δ + 15/1,000)	⨁◯◯◯ Very low
Stoma prolapse	Serious	Not serious (I²≈0%)	Not serious	Serious	Not serious	OR 0.76 (0.31–1.85)	84 → 98 (Δ + 14/1,000)	⨁◯◯◯ Very low
Peristomal infection	Serious	Not serious (I²≈0%)	Not serious	Very serious	Not serious	OR 0.82 (0.29–2.32)	49 → 36 (Δ − 13/1,000)	⨁◯◯◯ Very low
Incisional hernia—midline (> 12 mo)	Serious	Not serious (I²≈0%)	Not serious	Serious	Not serious	OR 0.45 (0.16–1.26)	115 → 67 (Δ − 48/1,000)	⨁⨁◯◯ Low
Length of hospital stay (days)	Serious	Very serious (I²≈92%)	Not serious	Serious	Not serious	MD + 3.69 days (− 1.52 to + 8.90)	—	⨁◯◯◯ Very low
Operative time (min)	Serious	Serious (I²≈62%)	Not serious	Serious	Not serious	MD + 8.52 min (− 8.38 to + 25.42)	—	⨁◯◯◯ Very low

Discussion

This systematic review and meta-analysis provides a comprehensive GRADE-assessed evaluation of retromuscular synthetic mesh reinforcement for parastomal hernia prevention during end-colostomy creation. Our findings suggest that prophylactic retromuscular mesh placement may reduce clinical parastomal hernia rates (OR 0.44, 95% CI 0.22–0.89), translating to an estimated 99 fewer hernias per 1,000 treated patients, although the certainty of this evidence is low due to substantial heterogeneity (I²=73%, τ²=0.57) and risk of bias from predominantly unblinded outcome assessment. The wide 95% prediction interval (OR 0.08–2.43) indicates that the true effect in a new clinical setting could range from substantial benefit to potential harm, reflecting considerable variability among studies. Safety outcomes showed no statistically significant differences, although the confidence intervals were wide and could not exclude clinically meaningful increases in complications. These findings add to the growing body of evidence regarding preventive mesh strategies while highlighting the substantial morbidity and healthcare burden associated with parastomal hernia development [1–3]. The observed protective effect requires cautious interpretation, given the methodological heterogeneity across the included trials, attenuation of effects in trials using mandated CT imaging, and sensitivity of the findings to individual study removal.

The safety profile observed in our analysis addresses critical concerns regarding mesh-related complications, which have historically limited the adoption of prophylactic surgical techniques in the past. Evaluation of early postoperative outcomes, including composite complications, stoma necrosis, reoperation rates, and surgical site infections, revealed no statistically significant increase in morbidity with mesh use, although the confidence intervals were wide. For rare events, such as stoma necrosis (OR 0.79, 95% CI 0.31–2.07), the evidence was of very low certainty due to serious imprecision, and clinically important increases in risk could not be excluded from the analysis. These findings are generally consistent with those of recent large-scale studies on synthetic meshes in abdominal wall reconstruction [4–6]. All included trials used macroporous monofilament polypropylene meshes placed retromuscularly; external evidence suggests that microporous materials may be more infection-prone, whereas large-pore polypropylene demonstrates a lower infection risk and favorable salvage characteristics [7–9]. The non-significant trend toward reduced midline incisional hernia rates (OR 0.45, 95% CI 0.16–1.26) suggests potential additional benefits of mesh reinforcement beyond parastomal hernia prevention, although the evidence remains imprecise and is derived from only two trials [10, 13, 14].

The anatomical rationale for retromuscular mesh placement represents an evolution from earlier intraperitoneal approaches, offering theoretical advantages, including superior biomechanical properties, improved mesh–tissue integration, reduced bowel contact, and optimized tensile strength distribution [15–17]. This focused evaluation of retromuscular placement distinguishes our analysis from previous reviews that pooled heterogeneous mesh positions. The landmark GRECCAR 7 trial demonstrated significant hernia reduction with acceptable morbidity in a large multicenter population, establishing retromuscular mesh placement as a widely studied approach [18, 19]. Contemporary innovations in three-dimensional mesh configurations and stapled reinforcement techniques are based on these principles, although long-term comparative data remain limited [20–22].

The substantial heterogeneity observed in the primary efficacy outcomes likely reflects both methodological and trial size differences. Variations in parastomal hernia detection methods appear to be a major driver, with trials using mandated CT-based assessment generally showing smaller and non-significant effects, whereas studies relying predominantly on clinical assessment reported larger apparent benefits, although subgroup differences were not statistically significant [11, 23, 24]. Notably, the apparent benefit was driven more by smaller and earlier studies, whereas larger contemporary trials with more standardized assessments generally reported attenuated or non-significant differences [12, 25, 26]. This pattern is particularly relevant for the PREVENT/Brandsma program, in which longer-term follow-up did not demonstrate a sustained significant reduction in parastomal hernia, and more broadly suggests that the four largest trials do not provide consistent evidence of a clinically important benefit [24]. Accordingly, the pooled results should not be interpreted as proof of a uniform treatment effect but rather as an average across studies with materially different designs, follow-up durations, and ascertainment methods. Sensitivity analysis demonstrating the loss of statistical significance after the removal of a single moderate-sized trial further underscores the fragility of the evidence base [27, 28].

Current practice patterns reveal significant variability in the adoption of prophylactic mesh. Surveys indicate that fewer than half of North American colorectal surgeons routinely use preventive meshes despite accumulating evidence [29, 30]. Concerns regarding mesh-related complications, technical complexity, and cost are barriers to adoption. The standardized retromuscular technique described in the included trials provides a reproducible approach for centers with appropriate surgical expertise [31–33]. The European Hernia Society guidelines emphasize the use of standardized techniques and evidence-based practices for parastomal hernia prevention [34].

The perioperative outcomes showed no statistically significant differences, although the evidence was characterized by substantial precision. Operative time demonstrated a non-significant mean increase of 8.52 min, and hospital length of stay showed no statistically significant difference, although confidence intervals included potentially meaningful increases, and heterogeneity was high [1–3]. External cost-effectiveness analyses suggest potentially favorable long-term economic profiles for prophylactic meshes [4–6]; however, such models depend on assumptions that may not be generalizable across different healthcare systems [7–9].

Long-term outcomes beyond the primary follow-up period remain important areas for future investigation. Data on mesh durability, late complications, sustained hernia prevention efficacy, and patient-reported outcomes are limited [10, 13, 14]. Available QoL data suggest no consistent adverse impact or potential benefits related to reduced hernia development and functional limitations [15–17].

The clinical implications should be interpreted cautiously, given the low certainty of the evidence, substantial heterogeneity, and fragility of the primary effect. While prophylactic mesh placement may reduce parastomal hernia rates, the magnitude of this benefit remains uncertain. These findings support selective implementation rather than routine universal adoption, emphasizing the need for shared decision-making and systematic outcome monitoring [21–23]. Additional large randomized trials using mandated CT-based outcome assessments and longer follow-ups are required before routine implementation can be confidently recommended [34–37].

Strengths and limitations

This systematic review had several methodological strengths that enhanced the reliability and clinical applicability of our findings. The exclusive focus on randomized controlled trials provides high-quality evidence with reduced selection bias, while the specific restriction to retromuscular mesh placement eliminates confounding from diverse anatomical approaches that have been limited in previous studies. Our comprehensive GRADE assessment provides a transparent evaluation of the certainty of evidence across all outcomes, facilitating informed clinical decision-making. The rigorous search strategy, which included multiple databases and trial registries, minimized the risk of missing studies, while duplicate screening and data extraction reduced errors. Predefined sensitivity and subgroup analyses explored the sources of heterogeneity and tested the robustness of the findings, while focusing on clinically relevant outcomes to ensure the practical applicability of the findings to surgical practice.

Several limitations warrant consideration when interpreting these results. The substantial heterogeneity in the primary outcome (I² = 73%) reflects the methodological differences across trials, which may limit the precision of the pooled estimates. Variations in parastomal hernia detection methods, ranging from mandated CT to clinical assessment, introduce measurement inconsistencies that could influence apparent treatment effects. The limited number of included trials (n = 7) restricts the power of subgroup analyses and prevents a comprehensive exploration of patient-level factors that might modify treatment effects. The follow-up periods varied across studies, with most reporting outcomes at 12 months, potentially missing late complications or hernias. The geographic concentration of European healthcare systems may limit the generalizability of our findings to other settings with different surgical practices and patient populations. Finally, the absence of a standardized QoL assessment across trials limits our ability to comprehensively evaluate patient-centered outcomes beyond clinical hernia rates.

Conclusion

This GRADE-assessed systematic review and meta-analysis suggests that prophylactic retromuscular synthetic mesh may reduce clinically detected parastomal hernia after end-colostomy creation, but the certainty of this benefit remains low because the pooled signal was heterogeneous, attenuated in trials using standardized CT-based assessment, and sensitive to individual-study removal. Although clear excess early harm has not been demonstrated, the current evidence does not support routine universal implementation. At present, selective use in appropriately chosen patients and experienced centers appears more justified than blanket adoption of the technique. Future research should focus not only on additional randomized comparisons where feasible, but also on longer-term follow-up of existing randomized cohorts, standardized reporting of imaging-defined and symptomatic outcomes, pooled analyses of patient-reported quality of life and functional burden, assessment of late mesh-related events, and evaluation of newer reinforcement technologies or configurations that may offer more durable benefits.

Abbreviations

PSH

Parastomal hernia

RCT

Randomized controlled trial

CT

Computed tomography

OR

Odds ratio

CI

Confidence interval

MD

Mean difference

QoL

Quality of life

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

PROSPERO

International Prospective Register of Systematic Reviews

GRADE

Grading of Recommendations Assessment, Development and Evaluation

SSI

Surgical site infection

ITT

Intention-to-treat

Authors’ contributions

WM conceived and designed the study, developed the protocol, led the literature search, screening, data extraction, and meta-analysis, prepared figures and tables, and drafted the manuscript. MEK and HI co-designed the methods, conducted screening and risk-of-bias appraisal, curated the dataset, and revised the Methods and Results sections. AK, MBMU, and HMK provided senior supervision, resolved disagreements, ensured clinical integrity and feasibility, and critically reviewed the manuscript. WM and MBMU performed database searches and reference management, managed PRISMA/RevMan files, prepared supplementary materials, and formatted the manuscript for submission. All authors read and approved the final manuscript and agreed to be accountable for all aspects of the work.

Funding

This systematic review received no specific funding from any commercial, public, or not-for-profit funding agency.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary materials. The extracted study-level dataset, data extraction forms, and full search strategy are available from the corresponding author, Wajahat Mirza, upon reasonable request at wajahatmirza.clinicalresearch@gmail.com.

Declarations

Ethics approval and consent to participate

Not applicable. This study was a systematic review and meta-analysis of published anonymized data and did not involve direct recruitment of human participants or collection of identifiable personal data.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.