Negative-pressure wound therapy versus conventional dressings following stoma reversal: a GRADE-based systematic review and meta-analysis of randomized controlled trials with predefined subgroup analyses by closure technique Authors: Wajahat Mirza, Mehak Ejaz Khan, Hania Iqbal, Alishbah Khan, Hadi Mohammad Khan, Muhammad Bilal Moeen-Ud-Din, Abdalla M Hadhoud Published: 2026 ABSTRACT AbstractBackgroundSurgical site infection (SSI) after stoma reversal remains common (15–40%), prolonging hospitalization and potentially delaying adjuvant therapy. Prophylactic negative-pressure wound therapy (NPWT) is increasingly used for closed incisions; however, its efficacy with different skin-closure techniques remains unclear. We conducted a systematic review and meta-analysis of randomized controlled trials comparing NPWT with conventional non-pressure dressings after stoma reversal, with prespecified subgroup analyses by skin-closure technique and certainty of evidence appraised using GRADE.MethodsThe review protocol was registered in the PROSPERO International Prospective Register of Systematic Reviews (CRD420251182767). We searched MEDLINE, Embase, Scopus, Web of Science, CENTRAL, ClinicalTrials.gov, and WHO ICTRP without language restrictions. Eligible randomized controlled trials enrolled adults undergoing elective ileostomy or colostomy reversal and compared negative-pressure wound therapy (NPWT) with conventional (non-pressure) dressings with a specified skin-closure technique (purse-string or primary). The primary outcome was incisional surgical site infection within 30–42 days of surgery. Secondary outcomes included length of stay, time to complete wound healing, operative duration, and estimated blood loss. Two reviewers independently screened the studies, extracted data, and assessed the risk of bias using RoB 2. Random-effects meta-analyses were conducted with sensitivity analyses and pre-specified subgroup comparisons based on closure techniques. The certainty of evidence was appraised using GRADE.ResultsSeven randomized trials (n = 429) met the inclusion criteria (NPWT, n = 220; conventional dressings, n = 209). The primary pooled analysis showed no significant reduction in incisional SSI with NPWT versus conventional dressings (RR 0.42, 95% CI 0.14–1.30; I²=46%). By closure technique, the effects were larger with primary closure (RR 0.20, 95% CI 0.06–0.67; I²=0%) than with purse-string closure (RR 0.68, 95% CI 0.14–3.35; I²=48%), although the interaction was not statistically significant (P = 0.23). The risk of bias was generally rated as “some concerns,” primarily due to open-label care, and heterogeneity was low after the sensitivity analysis.Prespecified leave-one-out analyses identified influential studies for SSI, length of stay, and wound healing; however, these findings were interpreted as exploratory influence diagnostics and did not replace the primary pooled analyses.ConclusionsIn this primary pooled analysis, prophylactic NPWT did not significantly reduce incisional surgical site infection after stoma reversal, although the effect favored NPWT. Exploratory sensitivity and subgroup analyses suggested possible benefits in selected contexts, although these findings should not supersede the nonsignificant primary pooled analyses. Larger, standardized randomized trials are needed to clarify which closure techniques and patient groups derive the greatest benefit.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12893-026-03957-2.Keywords: Negative-pressure wound therapy, Stoma reversal, Surgical site infection, Ileostomy closure, Colostomy closure, Wound healing, Meta-analysis, Randomized controlled trial, Purse-string closure, Linear primary closure Introduction Stoma reversal, although technically less complex than index colorectal surgery, carries substantial postoperative morbidity, with surgical site infection (SSI) rates of 15–40% in contemporary series [1–3]. Surgical site infections (SSIs) are defined as infections occurring up to 30 days after surgery (or up to one year after surgery in patients receiving implants) that affect either the incision or deep tissue at the surgical site [1]. Restoration of intestinal continuity predisposes the wound to bacterial contamination, subcutaneous dead space, and impaired perfusion [4–6]. These complications lead to longer hospitalizations, delayed adjuvant therapy, and reduced quality of life [5–7]. Despite refinements in techniques and perioperative care, SSI rates often exceed 30% even in specialized centers [7–12]. The persistently high burden underscores the need for effective and evidence-based prophylaxis [13]. Negative-pressure wound therapy (NPWT) is increasingly used as prophylaxis for closed surgical incisions, acting through wound-edge approximation, interstitial fluid evacuation, improved tissue perfusion, and a sealed microenvironment that limits bacterial contamination [1–3, 14, 15]. This rationale is particularly salient after stoma reversal, where bacterial burden, tissue edema, and seroma predispose patients to infection [6–8, 16]. Randomized controlled trials have reported mixed findings, with several showing directionally lower SSI reductions [6, 7, 9–11] and others reporting marginal or no benefit [8, 12, 17]. Contemporary meta-analyses have been limited by methodological heterogeneity, protocol variability, and non-standardized outcomes [1–3, 14–16]. These factors and possible effect modification by clinical context underscore the need for rigorous pre-specified subgroup analyses [16, 18]. Existing evidence on NPWT after stoma reversal remains difficult to interpret because prior syntheses included heterogeneous study designs, variable NPWT protocols, inconsistent SSI definitions, and limited stratification by skin-closure technique [1–3, 14–18]. Because purse-string and primary closure may have different baseline SSI risks and wound-healing profiles, closure-technique–specific analysis is clinically relevant [7, 8, 10, 11, 16, 19]. This systematic review synthesized randomized controlled trial evidence comparing NPWT with conventional dressings after stoma reversal, with prespecified subgroup analyses by skin-closure technique and GRADE assessment of evidence certainty [1–18]. Methods Study design We conducted a protocol-driven, PRISMA-compliant systematic review and meta-analysis of randomized controlled trials (RCTs) comparing prophylactic negative-pressure wound therapy (NPWT) with conventional non-pressure dressings after elective stoma reversal. The review protocol, including the statistical analysis plan, was registered in the PROSPERO International Prospective Register of Systematic Reviews (CRD420251182767) before study selection and data extraction. The prespecified protocol included the eligibility criteria, primary and secondary outcomes, subgroup analysis according to skin closure technique, and leave-one-out sensitivity analyses. Literature search strategy A comprehensive search was performed from database inception to November 1, 2025, with no language or date restrictions. We queried MEDLINE/PubMed, Embase, Scopus, Web of Science Core Collection, Cochrane CENTRAL, ClinicalTrials.gov, and WHO ICTRP. Google Scholar supports forward/backward citation chasing (first 200 results screened). Search strings combined controlled vocabulary and keywords for the intervention (e.g., “negative-pressure wound therapy,” “NPWT,” “vacuum-assisted”) and context (e.g., “stoma closure,” “stoma reversal,” “ileostomy,” “colostomy,” “purse-string,” “primary closure”), using RCT filters where available and manual screening otherwise. The full search strings and hit counts are shown in Supplementary Table 1, which also records the number of results returned by each database. We also hand-searched the reference lists of eligible studies and pertinent reviews to identify additional trials. Inclusion criteria We included RCTs enrolling adults undergoing elective stoma reversal (ileostomy or colostomy), in which the intervention was NPWT applied to the skin incision after reversal, and the comparator was a conventional (non-pressure) dressing. Trials had to (i) report at least one early wound-related outcome, including incisional surgical site infection (SSI) assessed at 30 days or the closest available trial-defined follow-up within 30–42 days, or complete epithelialization/healing at a prespecified time point; (ii) specify the skin-closure technique (purse-string or primary); and (iii) provide sufficient arm-level data to compute the effect estimates. Only randomized parallel-group studies were eligible; multi-arm trials were included if at least one arm met the NPWT definition and at least one arm used a conventional nonabsorbent dressing. Exclusion criteria We excluded non-randomized studies, including cohort, case–control, before-and-after, single-arm series, and conference abstracts without extractable outcomes. We also excluded pediatric populations, emergency stoma closures, procedures involving concomitant laparotomy or other major operations that materially altered postoperative wound care, and studies in which the comparator involved any form of pressure- or suction-assisted system. Studies exclusively evaluating suture or skin closure techniques without a direct NPWT versus conventional non-pressure dressing comparison were excluded from the study. Studies were also excluded if they lacked sufficient outcome data after unsuccessful attempts to contact the corresponding authors of the studies. At the trial level, the included RCTs commonly excluded patients with active infection or dermatologic disease at the incision site and clinical factors that are likely to impair wound healing or adherence to the wound-care protocol, such as uncontrolled diabetes or immunosuppression. Data extraction and outcome measures Two reviewers (W.M. and A.M.H.) independently screened the titles/abstracts and full texts, extracted data using a piloted form, and cross-checked the data entries. Disagreements were resolved through discussion with a third reviewer (M.B.M.D.). The extracted items included trial identifiers (year, country, and setting), sample sizes, demographics (age, sex, and BMI), stoma type and indication, skin-closure technique (purse-string vs. primary closure), perioperative antibiotics, NPWT device/type, pressure setting, application details and duration, and comparator dressing protocols. For the primary outcome, incisional SSI was extracted preferentially after 30 days. When the 30-day SSI was not reported, the closest available SSI assessment within 30–42 days was used for the analysis. This window was selected because the included randomized trials assessed early SSI and wound outcomes at slightly different postoperative intervals, most commonly at 30 or 42 days. Restricting the analysis to a 30-day window would have excluded otherwise eligible randomized trials and reduced the available evidence. Composite wound outcomes were not used as SSI events unless the incisional SSI component was separately extractable or clearly included according to the trial definition. Where available, superficial and deep incisional SSIs were analyzed together as overall incisional SSI outcomes. The definitions and assessment methods for SSI used in the included studies are summarized in Supplementary Table 2. When a study reported both composite wound management complications and SSI, the SSI-specific data were prioritized for the primary analysis. Additional outcomes, when available, included complete wound healing/epithelialization by a fixed time point, time to wound healing, wound management complications, length of stay, operative duration, and estimated blood loss (EBL). Quality assessment The risk of bias was evaluated independently by W.M. and A.M.H. using the Cochrane RoB 2 tool for randomized trials (randomization process, deviations from intended interventions, missing outcome data, outcome measurement, and selection of reported results). Domain-level judgments informed the overall RoB ratings for each outcome measure. Disagreements were resolved through discussions with M.B.M.D. Documentation of judgments and justifications was maintained to ensure transparency and to conduct subsequent sensitivity analyses. Statistical analysis All meta-analyses were performed using Review Manager (RevMan) version 5.4. Random-effects models were used for all pooled analyses. For dichotomous outcomes, risk ratios (RRs) with 95% confidence intervals (CIs) were pooled using the Mantel–Haenszel method to calculate the RR. For continuous outcomes, mean differences (MDs) with 95% CIs were pooled using the inverse variance method when the outcomes were reported on the same scale. When studies reported medians with measures of dispersion, values were converted to means and standard deviations using established methods described by Wan et al. [20] Luo et al. [21]. When only ranges were available, the method described by Hozo et al. was used. Zero-event cells were handled using a continuity correction of 0.5. Statistical heterogeneity was assessed using the I² statistic. Prespecified analyses included subgroup analysis by skin-closure technique and a leave-one-out sensitivity analysis to assess the influence of the individual studies. Publication bias was not formally assessed because fewer than 10 studies were available for each outcome. Sensitivity analysis, subgroup analysis, and GRADE certainty of evidence We prespecified a leave-one-out sensitivity analysis only, sequentially omitting each TRIAL to assess robustness and single-study influence. The primary subgroup analysis stratified the trials by skin closure technique (purse-string vs. primary) to explore the clinical heterogeneity attributable to the closure method. Between-subgroup differences were examined using Cochran’s Q-test for subgroup differences. The certainty of evidence for each critical outcome was assessed using GRADE across the domains of risk of bias, inconsistency, indirectness, imprecision, and publication bias, with summary judgments (high, moderate, low, and very low) and rationale provided in the Summary of Findings table. Results The database search identified 1,773 records, which were reduced to 1,011 unique records after the removal of 762 duplicates (Fig. 1). Title/abstract screening excluded 958 citations, leaving 53 reports for full-text review, all of which were retrieved. Of these, 46 were excluded, most commonly for non-randomized design (n = 27), followed by wrong intervention/comparator or no stoma reversal (n = 12), and insufficient data/duplicate populations (n = 7). Ultimately, seven randomized controlled trials met all the criteria and were included in the qualitative and quantitative syntheses (Fig. 1). PRISMA 2020 flow diagram for study selection in the review of negative-pressure wound therapy versus conventional non-pressure dressings after stoma reversal, showing database yield, screening, eligibility assessment, exclusion reasons, and the seven randomized trials included Characteristics of included studies Seven randomized trials of elective stoma reversal enrolled broadly comparable adult populations, with five using purse-string closure and two using primary closure, consistent with our pre-specified subgroups (Table 1). Eligibility frameworks were broadly similar across trials, although individual studies varied in their trial-level exclusions, including emergency closures or concomitant laparotomies, active dermatologic conditions at the incision site, and factors that could impair wound healing or adherence, such as uncontrolled diabetes or immunosuppression. The primary endpoints were early wound outcomes, including incisional SSI or complete epithelialization within 30–42 days. The secondary measures included healing quality, resource use, and complications. Table 1Core characteristics, eligibility criteria, and outcomes of randomized trials comparing negative-pressure wound therapy (NPWT) versus conventional dressings after stoma reversalStudyCountryNPWT nControl nInclusion criteriaExclusion criteriaPrimary endpoint(s)Secondary endpoint(s)Carrano 2021 [6]Italy50 randomized; 49 analyzed50 randomized; 45 analyzedAdults undergoing elective stoma reversal with purse-string skin closure after colorectal surgery.Pregnancy/breastfeeding, psychiatric or neurodegenerative disorders impairing consent/cooperation, and hypersensitivity to NPWT components.Wound complications within 30 days, including SSI, hematomas, and seromas, were recorded.Complete wound healing at 30 days, Vancouver Scar Scale scores on days 7 and 30, wound pain/itching, aesthetic satisfaction, and delayed wound healing were assessed.Kang 2023 [7]South Korea18 analyzed18 randomized; 16 analyzed Adults who underwent ileostomy or Hartmann reversal with purse-string closure were included. Dermatologic disease at the wound site, ASA physical status ≥ III, and inability to express informed intention/mental condition.Wound-healing period.SSI by CDC definition, length of stay, number of dressings, hospital visits for dressing, dressing cost, and POSAS at approximately 1 month postoperatively.Kojima 2021 [8]Japan20 combined NPWT arms: 10 one-week NPWT + 10 two-week NPWT10 Patients scheduled for ileostomy closure with a standardized purse-string suture at participating university hospitals were included in the study. Hemodialysis, poorly controlled diabetes, systemic steroids, cognitive/psychiatric impairment affecting wound care, and other comorbidities markedly affect healing.Wound-reduction rate.Superficial incisional SSI, length of hospital stay, and wound healing duration were also assessed.Tiang 2024 [9]Australia20 randomized; 19 analyzed20 randomized; 19 analyzed Adults who underwent elective loop ileostomy reversal with purse-string skin closure were included. Emergency operation, concomitant laparotomy, and situations in which wound-specific quality of life measures would be uninterpretable.Complete wound healing was observed on day 42, and percentage reduction in wound depth was observed on days 7 and 14. Patient-reported scar cosmesis using VAS on day 42 and SSI rate. Uchino 2016 [10]Japan28 enrolled; 24 wound-healing analysis31 enrolled; 29 wound-healing analysis A patient with ulcerative colitis scheduled for ileostomy closure with purse-string sutures. Protocol-defined medical exclusions and NPWT contraindications were also recorded. Complete wound healing.Wound complications, including incisional SSI, wound bleeding, enterocutaneous fistula, bowel obstruction, operative metrics, and postoperative glucose levels.Wierdak 2020 [11]Poland3536 Adults with colorectal cancer scheduled for protective loop ileostomy closure with primary wound closure were included in the study. Emergency/urgent operation; active infection; procedure other than ileostomy closure; parastomal hernioplasty; early reoperation/ICU transfer for noninfectious complications. Wound healing complications include:SSI, postoperative length of stay; complete wound healing duration.Yong Xu 2024 [12]China5051 Adults > 18 years of age with radical rectal tumor resection and protective loop ileostomy scheduled for elective ileostomy closure were included. Age < 18 years; loop ileostomy due to other surgeries; end ileostomy; emergency surgery; other surgical procedures during ileostomy period; and no informed consent. Incisional SSI.Intra-abdominal/surgical complications, operative time, blood loss, postoperative hospital stay, hospitalization cost, and SSI risk factor analyses.Abbreviations: NPWT Negative-pressure wound therapy (NPWT), PSC Purse-string (skin) closure, LPC linear/primary closure, SSI Surgical site infection, POSAS Patient and Observer Scar Assessment Scale, POD Postoperative day, ERAS Enhanced Recovery After Surgery, UC Ulcerative colitis, HbA1c Glycated hemoglobin, WMC Wound-management complications, VAS Visual analog scale, MPNPWT Micro-power NPWT Core characteristics, eligibility criteria, and outcomes of randomized trials comparing negative-pressure wound therapy (NPWT) versus conventional dressings after stoma reversal Abbreviations: NPWT Negative-pressure wound therapy (NPWT), PSC Purse-string (skin) closure, LPC linear/primary closure, SSI Surgical site infection, POSAS Patient and Observer Scar Assessment Scale, POD Postoperative day, ERAS Enhanced Recovery After Surgery, UC Ulcerative colitis, HbA1c Glycated hemoglobin, WMC Wound-management complications, VAS Visual analog scale, MPNPWT Micro-power NPWT The age and BMI distributions were balanced across the trials, and the sex ratios are explicitly shown, where applicable (Table 2). The stoma type was predominantly a looping ileostomy, and the indications ranged from colorectal cancer to ulcerative colitis, which aligned with the real-world case mix. NPWT technology and protocols varied: PICO at − 80 mmHg for 7–14 days, SNaP at − 125 mmHg for 7 days, and NANOVA with an early transition to standard dressings. Control arms used transparent films, gauze, or sterile dressings with defined change schedules. Perioperative antibiotic policies are broadly standardized (cefazolin-based or second-generation cephalosporins), thereby supporting fair outcome comparisons and robust subgroup analyses based on the closure technique (Table 2). Table 2Baseline demographics, stoma context, closure technique, antibiotic prophylaxis, and detailed wound care protocols for NPWT and comparator arms in randomized trials of stoma reversalStudyAge NPWTAge controlBMI NPWTBMI controlSex F/M NPWTSex F/M controlStoma type/indicationAntibioticsNPWT device/typePressureNPWT protocolControl dressing/protocolCarrano 2021 [6]56.32 ± 12.9255.08 ± 16.2523.81 ± 3.3823.45 ± 3.6615/3517/31 Ileostomy: 41 vs. 38; colostomy: 9 vs. 10. Oncological disease: 28 vs. 26; benign disease: 22 vs. 22. Cefazolin 20 mg/kgPICO single-use NPWT−80 mmHgThe technique was applied over purse-string closure, with a median duration of 7 days. Conventional wound dressing/packing protocol; wounds were assessed on days 7 and 30. Kang 2023 [7]72 (32–87)66.5 (33–78)23.7 (15.2–31.2)25.4 (18.7–31.9)9/98/8 Ileostomy reversal: 10 vs. 12; Hartmann’s reversal: 8 vs. 4. 2nd-generation cephalosporin 1 h before surgeryPICO portable NPWTNRDressing changes twice weekly; switch to conventional dressing if frequent inpatient changes are needed; follow-up twice weekly until healed.Simple transparent waterproof dressing, routine daily changes, and additional changes if wet/discharging.Kojima 2021 [8]Group B: 69 (48–75); Group C: 70 (43–84)Group A: 64 (48–82)Group B: 21.8 (16.5–23.8); Group C: 20.3 (17.0–26.2)Group A: 22.2 (16.4–30.1)Group B 2/8; Group C 6/4Group A 4/6 Ileostomy closure using purse-string sutures. NRPICO single-use NPWT + foam insert initially−80 mmHgNPWT was initiated on POD1 after wound cleaning was performed. Group B continued until POD7, and Group C continued until POD14, with dressing checks/changes on POD3 and later visits.Gauze exchange after standard purse-string closure until epithelialization.Tiang 2024 [9]55.67 ± 17.162.0 ± 15.127.62 ± 6.2126.12 ± 4.50Equal male/female numbers reported; exact arm-level counts NREqual male/female numbers reported; exact arm-level counts NRElective loop ileostomy reversal with purse-string closure: malignancy 11 vs. 14, IBD 8 vs. 5, diverticular disease 1 vs. 1.ERAS-based care; specific antibiotic not clearly reportedSNaP mechanically powered NPWT−125 mmHg The dressing was changed on POD3 and removed/discarded on POD7. A simple non-woven absorbent cotton dressing was changed daily or more often if exudate was significant until epithelialization. Uchino 2016 [10]48.1 ± 14.940.4 ± 15.919.8 ± 4.319.7 ± 3.811/178/23 Ileostomy closure in patients with ulcerative colitis. NR in extracted textPICO single-use NPWTNR in article text NPWT was applied after purse-string closure and used for wound management after ileostomy closure. Purse-string closure alone/simple dressing care.Wierdak 2020 [11]61.6 ± 11.362.4 ± 11.326.2 ± 4.526.2 ± 4.311/2416/20Protective loop ileostomy closure after colorectal cancer surgery.2nd-generation cephalosporin 20 mg/kg 30 min before incisionNANOVA negative-pressure dressingNRNANOVA over incision; removed at 72 h; Steri-Strips and standard sterile dressing afterward; then daily changes until suture removal on POD7. The sterile wound dressing was changed for the first time at 48 h and then daily until suture removal on POD7. Yong Xu 2024 [12]61.66 ± 14.2264.43 ± 9.0622.61 ± 3.0621.82 ± 2.9020/3017/34Protective loop ileostomy closure after rectal cancer surgery.NRMicro-power negative-pressure wound technique dressingMicro-power negative pressure; numeric setting NRIntermittent suturing with 2 − 0 Prolene; incision covered with MPNPWT dressing. Conventional dressing after intermittent suturing/standard postoperative care. Abbreviations.NPWT negative-pressure wound therapy, PSC Purse-string closure, LPC Linear/primary closure, BMI Body mass index, RPC Restorative proctocolectomy, IPAAIleal pouch–anal anastomosis, POD Postoperative day, NRNot reported, SNaP™ Smart Negative Pressure system, PICO Single-use portable NPWT, PREVENA Portable incisional NPWT system, MPNPWT Micro-power NPWT Baseline demographics, stoma context, closure technique, antibiotic prophylaxis, and detailed wound care protocols for NPWT and comparator arms in randomized trials of stoma reversal Abbreviations.NPWT negative-pressure wound therapy, PSC Purse-string closure, LPC Linear/primary closure, BMI Body mass index, RPC Restorative proctocolectomy, IPAAIleal pouch–anal anastomosis, POD Postoperative day, NRNot reported, SNaP™ Smart Negative Pressure system, PICO Single-use portable NPWT, PREVENA Portable incisional NPWT system, MPNPWT Micro-power NPWT Methodological quality and study integrity The methodological quality of the seven randomized trials was generally acceptable but not unequivocally low-risk. The randomization processes were clearly described and judged to be at low risk in three studies (Carrano 2021, Tiang 2024, Wierdak 2020), while the remaining trials provided insufficient detail on sequence generation and/or allocation concealment, yielding some concerns in D1. Because incisional NPWT cannot be blinded and peri-incisional care was delivered in open-label settings, every study had some concerns for D2 (deviations from intended interventions), despite protocolized dressing schedules. Missing outcome data were minimal and balanced, leading to a low risk of D3 in all trials. Outcomes (SSI within ~ 30–42 days and time/healing) were defined a priori, clinically objective, and assessed using standard criteria, indicating a low risk for D4. Selective reporting (D5) was low risk in six studies; Kojima 2021 had some concerns due to multi-arm reporting with limited outcome details. Consequently, the overall risk of bias was rated as “some concerns” for each study, driven primarily by unavoidable open-label care and incomplete descriptions of randomization, whereas other domains were consistently robust, supporting confidence in the direction and stability of the pooled effects (Fig. 2). Risk of bias (RoB 2) for included randomized trials. A Traffic-light plot showing domain-level judgments per study (D1–D5 and overall). B Summary bar plot displaying the proportion of studies rated low risk versus some concerns across domains Primary outcome Incisional surgical site infection (SSI) ≤ 30–42 days The pooled RR is 0.42 (95% CI 0.14–1.30) with I²=46%. This suggests a trend toward fewer SSIs with NPWT; however, the difference was not statistically significant (p = 0.13). Overall, Fig. 3 indicates no definitive difference, although heterogeneity was moderate. Forest plot comparing NPWT and non-pressure dressings for incisional SSI within 30–42 days (random-effects RR) Secondary outcomes Length of postoperative hospital stay (days) The pooled MD is − 0.33 days (95% CI − 1.30 to 0.64) with I²=61%. The direction favors NPWT (shorter stay), but the effect is not statistically significant (p = 0.51). Figure 4 shows moderate to substantial heterogeneity across the trials. Forest plot of length of hospital stay (days) for NPWT vs. nonpressure dressings (random-effects MD) Operative time (minutes) The pooled MD was 1.98 min (95% CI − 2.89 to 6.85) with I²=0%. There was no significant difference in operative time between the groups (p = 0.42), and the results were consistent. Figure 5 indicates neutrality with minimal heterogeneity in the funnel plot. Forest plot of operative time for NPWT vs. non-pressure dressings (random-effects MD) Estimated blood loss (mL) The pooled MD is − 3.78 mL (95% CI − 6.26 to − 1.31) with I²=27%. NPWT was associated with a statistically significant reduction in blood loss (p = 0.003) and low heterogeneity. Figure 6 supports the small but consistent perioperative benefit of NPWT. Forest plot of estimated blood loss for NPWT versus non-pressure dressing (random-effects MD) Time to complete wound healing (days) The pooled MD was 0.88 days (95% CI − 3.02 to 4.78) with I²=45%. This was not statistically significant (p = 0.66), and the direction was equivocal with moderate heterogeneity. Figure 7 shows no clear difference in the healing speed. Forest plot of time to complete wound healing for NPWT versus non-pressure dressings (random-effects MD) Postoperative patient-reported wound cosmesis Pooled across three RCTs, the random-effects mean difference in patient-reported wound cosmesis was 0.69 (95% CI − 1.23 to 2.62, p = 0.48), showing no statistically significant difference between NPWT and non-pressure dressings (Supplementary Fig. 1). Sensitivity analysis Leave-one-out analyses were performed as prespecified influence diagnostics to evaluate whether any single trial disproportionately affected the pooled estimates. Uchino 2016 was not excluded from the primary analysis; instead, it was sequentially omitted within a leave-one-out framework. This trial was the most influential study for the SSI estimate because it reported a directionally unfavorable SSI pattern for NPWT, with incisional SSI/wound infection occurring in 3/28 patients in the NPWT group versus 1/31 patients in the control group. After excluding Uchino 2016, heterogeneity decreased, and the pooled estimate became statistically significant in favor of NPWT (RR 0.31, 95% CI 0.11–0.89; I²=30%; p = 0.03). However, this result was interpreted strictly as a sensitivity-derived, hypothesis-generating finding and not as a replacement for primary pooled estimates. A strict 30-day-only SSI sensitivity analysis was also performed. Most trials reported SSI or wound infection within approximately 30 days, whereas Tiang 2024 assessed SSI during a 42-day follow-up period and reported no SSI events in either group. Because this was a double-zero study for SSI, its exclusion for a strict 30-day-only sensitivity check did not materially change the direction or interpretation of the pooled SSI effects. Therefore, the primary SSI analysis was retained as SSI within 30–42 days, with timing variation acknowledged as a limitation. For length of stay, exclusion of Carrano 2021 eliminated heterogeneity and showed a statistically significant reduction with NPWT (MD − 0.85 days, 95% CI − 1.51 to − 0.19; I²=0%; p = 0.01). For the time to complete wound healing, the exclusion of Xu 2024 resolved heterogeneity and showed faster healing with NPWT (MD − 2.80 days, 95% CI − 4.75 to − 0.85; I²=0%; p = 0.005). These secondary sensitivity findings were also interpreted as influencing diagnostics rather than definitive pooled effects (Table 3). Table 3Leave-one-out sensitivity analysis comparing negative-pressure wound therapy (NPWT) vs. non-pressure dressings (NPD) for outcomes with moderate heterogeneity at baseline. The “After” columns show the pooled effect and heterogeneity after removing the most influential study for that outcomeOutcomeEffect Estimate (Before)P-value (Before)I²% (Before)Study RemovedEffect Estimate (After)P-value (After)I²% (After)Incisional SSI ≤ 30–42 days (RR)0.42 [0.14, 1.30]0.1346Uchino 20160.31 [0.11, 0.89]0.0330Length of postoperative hospital stay, days (MD)−0.33 [− 1.30, 0.64]0.5161Carrano 2021−0.85 [− 1.51, − 0.19]0.010Time to complete wound healing, days (MD)0.88 [− 3.02, 4.78]0.6645Yong Xu 2024−2.80 [− 4.75, − 0.85]0.0050Estimated blood loss, mL (MD)−3.78 [− 6.26, − 1.31]0.00327Kang 20232.43 [0.05, 4.81]0.050Abbreviations: RR risk ratio, MD Mean difference, SSI Surgical site infection Leave-one-out sensitivity analysis comparing negative-pressure wound therapy (NPWT) vs. non-pressure dressings (NPD) for outcomes with moderate heterogeneity at baseline. The “After” columns show the pooled effect and heterogeneity after removing the most influential study for that outcome Abbreviations: RR risk ratio, MD Mean difference, SSI Surgical site infection Subgroup analysis The subgroup analysis by skin-closure technique suggested that the protective effect of NPWT against incisional SSI was concentrated in trials using primary closure (LPC), where the pooled risk ratio was 0.20 (95% CI 0.06–0.67) with I²=0%, indicating a statistically significant and consistent reduction in SSI. In contrast, among purse-string closure (PSC) trials, the pooled effect was 0.68 (0.14–3.35) with I²=48%, showing no clear benefit and moderate heterogeneity. The test for subgroup differences did not reach statistical significance (χ²=1.44, df = 1, P = 0.23; I²=30.4%); therefore, while LPC showed a stronger signal, a formal interaction could not be claimed. The overall pooled estimate across both subgroups was 0.42 (0.14–1.30) under a random-effects model, consistent with the main analysis and highlighting the technique-dependent variability in effect (Fig. 8). Subgroup meta-analysis of incisional surgical site infection (≤ 30–42 days) comparing NPWT versus conventional dressings stratified according to skin closure technique. Purse-string and primary closure studies were analyzed separately using random-effects models. Squares represent study-specific estimates weighted by inverse variance, and diamonds represent the pooled effects GRADE-certainty of evidence For the primary outcome of incisional SSI, certainty was based on the primary pooled analysis, which showed a directionally lower SSI risk with NPWT but did not reach statistical significance. The certainty was downgraded for risk of bias and imprecision because most trials had RoB 2 “some concerns” and the confidence interval crossed the null. Sensitivity analyses showed stronger effects after excluding influential studies; however, these sensitivity-derived estimates were not used as the basis for the GRADE certainty judgments. For secondary outcomes, the primary pooled analyses showed no statistically significant difference in length of stay, operative time, time to complete wound healing, or patient-reported wound cosmesis, while estimated blood loss was statistically lower with NPWT but of small clinical magnitude and sensitive to study exclusion. Overall, certainty was limited mainly by study-level risk of bias, imprecision, and inconsistency for selected outcomes (Table 4). Imprecision downgrading was primarily driven by wide confidence intervals crossing the null and limited event numbers. Inconsistency downgrades were applied when statistical heterogeneity was substantial or when the effect estimates varied meaningfully across studies. Table 4GRADE summary of findings for NPWT versus non-pressure dressings after stoma reversal across the pre-specified primary and secondary outcomesOutcomeRisk of BiasInconsistencyIndirectnessImprecisionPublication BiasEffect EstimateAbsolute EffectGRADE CertaintyIncisional surgical site infection ≤ 30–42 daysSeriousNot serious/moderate heterogeneity (I²=46%)Not seriousSeriousUndetected; not formally assessed because < 10 studiesRR 0.42 (95% CI 0.14–1.30)88 fewer per 1,000, ranging from 131 fewer to 45 more⭑⭑◯◯ LowLength of postoperative hospital staySeriousSerious (I²=61%)Not seriousSeriousUndetected; not formally assessed because < 10 studiesMD − 0.33 days (95% CI − 1.30 to 0.64)Continuous outcome⭑◯◯◯ Very lowOperative timeSeriousNot serious (I²=0%)Not seriousSeriousUndetected; not formally assessed because < 10 studiesMD 1.98 min (95% CI − 2.89 to 6.85)Continuous outcome⭑⭑◯◯ LowEstimated blood lossSeriousNot serious/low heterogeneity (I²=27%)Not seriousSerious*Undetected; not formally assessed because < 10 studiesMD − 3.78 mL (95% CI − 6.26 to − 1.31)Continuous outcome⭑⭑◯◯ LowTime to complete wound healingSeriousNot serious/moderate heterogeneity (I²=45%)Not seriousSeriousUndetected; not formally assessed because < 10 studiesMD 0.88 days (95% CI − 3.02 to 4.78)Continuous outcome⭑⭑◯◯ LowCertainty symbols: ⭑⭑⭑⭑ High, ⭑⭑⭑◯ Moderate, ⭑⭑◯◯ Low, ⭑◯◯◯ Very low GRADE summary of findings for NPWT versus non-pressure dressings after stoma reversal across the pre-specified primary and secondary outcomes Certainty symbols: ⭑⭑⭑⭑ High, ⭑⭑⭑◯ Moderate, ⭑⭑◯◯ Low, ⭑◯◯◯ Very low Risk of bias Downgraded one level because most trials were rated as having RoB 2 “some concerns,” mainly due to open-label care and incomplete reporting of randomization/allocation procedures. Inconsistency Downgraded for length of stay because of substantial heterogeneity in the primary pooled analysis (I²=61%). Other outcomes were not downgraded for inconsistency because the heterogeneity was low to moderate and did not preclude interpretation. Indirectness Not downgraded because the included populations, interventions, comparators, and outcome definitions matched the review PICO. Imprecision Downgraded when confidence intervals crossed the null or included both clinically meaningful benefits and no effect/harm. The estimated blood loss was downgraded because the absolute difference was clinically trivial and sensitive to study removal. Publication bias Formal testing was not performed because fewer than 10 studies were available for each outcome; therefore, publication bias was judged as undetected but could not be excluded. Discussion This GRADE-assessed systematic review and meta-analysis of seven randomized controlled trials found that prophylactic negative-pressure wound therapy (NPWT) was associated with a directionally lower risk of incisional surgical site infection (SSI) after stoma reversal, but the primary pooled estimate did not reach statistical significance (RR 0.42, 95% CI 0.14–1.30; I²=46%) [1–3]. Primary analyses of length of stay, operative time, time to complete wound healing, and patient-reported wound cosmesis also did not show statistically significant differences, while estimated blood loss was statistically lower with NPWT but of small absolute magnitude [1–3, 14–16]. Exploratory leave-one-out analyses demonstrated the sensitivity of pooled estimates to individual studies; however, these analyses were intended as influence diagnostics rather than confirmatory evidence. These findings suggest uncertainty around the pooled estimates and should not be interpreted as replacing the primary analysis. The absolute effect estimates from the primary pooled analysis corresponded to approximately 88 fewer SSI events per 1,000 patients, although the confidence interval ranged from 131 fewer to 45 more events per 1,000 patients. In the leave-one-out sensitivity analysis, the estimated NNT was approximately 10; however, this was derived after excluding Uchino 2016 and should, therefore, be interpreted as exploratory rather than definitive [1–3, 16, 18, 22, 23]. The potential clinical relevance should be considered in the context of SSI-related morbidity, delays in adjuvant therapy, and costs [1–3, 18, 22–26]. Future studies should clarify whether the benefits differ according to closure technique, wound characteristics, baseline SSI risk, and perioperative care protocols [7–12, 16, 19, 22, 23, 27–29]. Subgroup analyses by closure type showed primary closure RR 0.20 (95% CI 0.06–0.67; I²=0%) versus purse-string RR 0.68 (95% CI 0.14–3.35; I²=48%) [1–3, 6–8, 10, 11, 16, 22, 23]. Although the magnitude of the effect appeared greater in the primary closure cohorts, the interaction test was not statistically significant; therefore, these findings should be interpreted as exploratory rather than confirmatory [7, 8, 10, 11, 16, 19]. Residual heterogeneity may reflect device/pressure settings, duration, baseline SSI risk, antibiotic protocols, and outcome windows [1–3, 6–8, 10, 11, 16, 22, 23]. Clinical heterogeneity may have influenced the pooled estimates, including variation in NPWT systems, pressure settings, wear duration, antibiotic protocols, and SSI assessment windows. Compared with previous meta-analyses, the present review differs in that it restricted inclusion to randomized controlled trials, incorporated prespecified subgroup analyses according to closure technique, applied formal GRADE certainty assessment, and emphasized sensitivity analyses as influence diagnostics rather than confirmatory evidence. These methodological refinements were intended to improve interpretability and reduce the bias introduced by observational data. The methodological strengths included exclusive RCT inclusion, prespecified outcomes focused on early SSI and healing, random-effects modeling, prespecified subgroup analysis by closure technique, leave-one-out sensitivity analysis, RoB 2 assessment, and formal GRADE certainty appraisal. GRADE certainty was low for SSI, operative time, estimated blood loss, and wound-healing outcomes, and very low for length of stay, primarily because of study-level risk of bias, imprecision, and inconsistency [14–16, 18, 24–26]. Sensitivity analyses identified influential studies, particularly Uchino 2016 for SSI, reinforcing that sensitivity-derived findings should be interpreted cautiously rather than as replacements for the primary pooled analyses [10, 14–16]. Removal of individual studies also reduced heterogeneity for length of stay and wound healing [6, 12, 17, 27, 28]. The effects of estimated blood loss were directionally unstable and clinically negligible (< 5 mL), reinforcing their low certainty and limited clinical value [2, 14–16]. Several limitations should be acknowledged, including the modest sample size, device heterogeneity, variability in perioperative protocols, and unavoidable lack of blinding. These factors may have contributed to the imprecision and clinical heterogeneity across the pooled estimates. Variations in the SSI follow-up windows, including assessments at both 30 and 42 days, may also have contributed to modest clinical heterogeneity. As fewer than 10 studies were available for each outcome, formal publication bias testing was not feasible. Therefore, small-study effects and publication bias cannot be excluded. Taken together, these findings do not establish a definitive clinical benefit but suggest that NPWT may warrant further evaluation in selected higher-risk patients, particularly when primary closure is planned. Future large multicenter RCTs employing standardized NPWT protocols, economic evaluations, and patient-reported outcome measures are essential for refining clinical indications and facilitating the integration of NPWT into enhanced recovery pathways [30–33]. Strengths and limitations To our knowledge, this review is the first GRADE-assessed synthesis restricted to randomized trials comparing NPWT with conventional dressings after stoma reversal, reducing the selection bias inherent in observational studies. A prespecified protocol guided comprehensive, language-unrestricted searches, duplicate screening and extraction, and standardized RoB2 appraisal across all domains. Technique-stratified subgroup analyses (purse-string vs. primary closure) offer clinically relevant insights into potential effect modifications. Leave-one-out analyses demonstrated that Uchino 2016 had a notable influence on the SSI estimate, although the overall direction of effect continued to favor NPWT. GRADE summaries provide transparent certainty ratings for the risk of bias, inconsistency, indirectness, imprecision, and publication bias in the studies. The inclusion of trials from multiple health systems supports external validity across different perioperative practices and NPWT availability. Limitations include a modest overall sample (429 patients), which constrains the power for definitive subgroup conclusions, especially for the purse-string versus primary closure interaction, where the difference was not statistically significant, despite a larger apparent effect with primary closure. Device heterogeneity (e.g., portable systems with varying pressures, interfaces, and wear times) precludes the provision of device-specific recommendations. Open-label care and incomplete reporting of randomization procedures contributed to “some concerns” regarding the risk of bias and GRADE downgrades. Clinical protocols varied, including antibiotic regimens, and blinding of outcome assessment was impractical due to visible devices, introducing potential ascertainment bias. Because fewer than ten studies were available per outcome, formal publication bias testing was not feasible. Therefore, small-study effects and publication bias cannot be excluded. Unmeasured factors, such as the timing of stoma closure, postoperative wound care practices, nutritional status, glycemic control, and smoking, may also have influenced outcomes. The generalizability of our findings beyond elective ileostomy and colostomy reversals is limited; emergency closures, pediatric populations, and procedures performed concomitantly with laparotomy were not represented. Conclusions In this meta-analysis, prophylactic NPWT did not significantly reduce incisional SSI in the primary pooled analysis. Exploratory subgroup and sensitivity analyses suggested possible benefits under selected conditions; however, these findings require confirmation in larger, adequately powered randomized trials. Although the current evidence does not definitively establish clinical benefits, NPWT may represent a reasonable adjunct in selected high-risk patients, particularly when primary closure is used. Supplementary Information Supplementary Material 1. Abbreviations Negative-pressure wound therapy non-pressure dressings Purse-string (skin) closure Linear/primary closure Surgical site infection Length of stay Body mass index Ulcerative colitis Restorative proctocolectomy Ileal pouch–anal anastomosis Postoperative day Enhanced Recovery After Surgery Single-use portable NPWT system Smart Negative Pressure system Portable incisional NPWT system Negative-pressure wound dressing system Micro-power negative-pressure wound therapy Patient and Observer Scar Assessment Scale Glycated hemoglobin Wound-management complications Visual analog scale Randomized controlled trial Preferred Reporting Items for Systematic Reviews and Meta-Analyses Cochrane Central Register of Controlled Trials World Health Organization International Clinical Trials Registry Platform Cochrane Risk of Bias 2 tool Risk ratio Mean difference Standardized mean difference Confidence interval Higgins’ inconsistency statistic Between-study variance Restricted maximum likelihood Hartung–Knapp–Sidik–Jonkman Number needed to treat Standard deviation Authors’ contributions W.M. conceived the study, designed the protocol, performed the literature search, screening, data extraction, statistical analysis, and drafted the manuscript. M.E.K. and H.I. performed duplicate screening and data extraction, and contributed to the critical revision. A.K., M.B.M.U., and A.M.H contributed to data interpretation and manuscript revision. H.M.K. provided clinical expertise, supervised the analysis, and contributed to critical revision. All authors approved the final manuscript. Funding No external funding was received for this systematic review and meta-analysis. Data availability All data extracted from included studies are presented within the manuscript tables and forest plots. The full search strategy and PRISMA checklist are available in the supplementary materials. Declarations Ethics approval and consent to participate Not applicable. This systematic review synthesized data from previously published randomized controlled trials; no primary data collection or patient involvement occurred. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement All data extracted from included studies are presented within the manuscript tables and forest plots. The full search strategy and PRISMA checklist are available in the supplementary materials.