Comparative efficacy and safety of non-pharmacological nursing interventions for diabetic foot ulcers: a systematic review and network meta-analysis Authors: Juan Gong, Xiaohong Sun, Hong Fu, Yu Zhang, Ting Gao Published: 2025 ABSTRACT AbstractBackgroundNon-pharmacological nursing interventions are increasingly employed to enhance diabetic foot ulcer (DFU) healing and patient outcomes. However, the comparative effectiveness and safety profiles of various adjunctive nursing strategies remain unclear. This network meta-analysis systematically evaluated and ranked non-pharmacological nursing interventions for DFU management.MethodsA comprehensive systematic search was conducted in PubMed, Embase, PsycINFO, Cochrane Central Register of Controlled Trials, and Web of Science from inception to March 31, 2025. Randomized controlled trials (RCTs) comparing non-pharmacological adjunctive nursing interventions (e.g., hyperbaric oxygen therapy [HBOT], light therapy [LT], ultrasound therapy [US], extracorporeal shock-wave therapy [ESWT], exercise therapy [EXER], continuous diffusion of oxygen [CDO], negative pressure wound therapy [NPWT]) with standard care or alternative interventions for DFU were included. Primary outcomes assessed were healing rate, healing time, wound area reduction, recurrence rate, amputation rate, and adverse events. Bayesian network meta-analysis was conducted using random-effects models, with results expressed as odds ratios (OR), standardized mean differences (SMD), and surface under cumulative ranking curves (SUCRA).ResultsSixty-seven RCTs involving 5957 patients were included. LT significantly increased healing rates compared with standard care (OR = 7.62; 95% CI, 2.92–19.90; SUCRA = 88.8%). US produced the largest wound area reduction (SMD = −3.17; 95% CI, −4.81 to −1.53; SUCRA = 98.4%). ESWT was superior in shortening healing time (SMD = −2.02; 95% CI, −3.33 to −0.71; SUCRA = 94.4%) and reducing amputation risk (OR = 0.40; 95% CI, 0.23–0.70). EXER significantly reduced recurrence rates (OR = 0.07; 95% CI, 0.02–0.33; SUCRA = 97.4%). CDO demonstrated the fewest adverse events (OR = 0.27; 95% CI, 0.08–0.85; SUCRA = 89.2%).ConclusionThis analysis provides targeted evidence for selecting adjunctive nursing interventions based on specific clinical goals. LT and US are optimal for enhancing wound closure and reducing wound area, ESWT is most effective for rapid healing and limb preservation, EXER excels at recurrence prevention, and CDO offers superior safety. These findings facilitate tailored nurse-led care plans and inform clinical guidelines.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12912-025-04065-x.Keywords: Diabetic foot ulcer, Non-pharmacological interventions, Nursing care, Network meta-analysis, Wound healing Introduction Diabetic foot ulcer (DFU) is a severe complication of diabetes mellitus characterized by foot infections, ulcerations, and deep tissue damage, primarily due to peripheral neuropathy, peripheral arterial disease, and metabolic disturbances [1]. Patients with DFU typically exhibit chronic wounds, impaired healing capacities, and an increased risk of complications such as infections and amputations [2]. Currently, DFU is a growing global health issue, closely linked with the escalating prevalence of diabetes mellitus worldwide. Estimates suggest that by 2025, approximately 589 million adults aged 20–79 years will have diabetes, with this figure anticipated to rise to 853 million by 2050 [3]. The chronic and recurrent nature of DFU imposes substantial psychological and physical burdens on affected individuals and their families, significantly impacting patients’ quality of life. Additionally, the management of DFU incurs substantial economic costs for healthcare systems due to prolonged hospitalization, expensive wound care treatments, and increased disability rates [4]. Pharmacological approaches to DFU management, such as glycemic control, antibiotics, and vascular medications, are commonly applied; however, conventional treatments have important limitations and, when used in isolation, may be insufficient to achieve durable wound closure, partly due to antibiotic resistance, adverse drug reactions, and insufficient improvement in wound healing [5]. Consequently, non-pharmacological nursing interventions have emerged as critical adjunctive strategies in DFU care, offering significant advantages, including promoting wound healing, reducing complications, and enhancing overall patient recovery and quality of life through comprehensive and patient-centered nursing care [6]. Common non-pharmacological adjunctive interventions include hyperbaric oxygen therapy (HBOT), negative pressure wound therapy (NPWT), extracorporeal shock wave therapy (ESWT), ultrasound therapy (US), etc [7–9]. Clinical studies have explored various adjunctive non-pharmacological interventions for DFU, reporting significant benefits in enhancing healing rates, reducing healing time, minimizing wound recurrence, and decreasing the risk of amputation and adverse events [10, 11]. Previous meta-analyses have compared specific non-pharmacological treatments individually or in limited pairwise comparisons, confirming the general efficacy and safety of these interventions in managing DFU [12, 13]. Nevertheless, these analyses are constrained by their inability to comprehensively evaluate and rank multiple interventions simultaneously, leaving uncertainties regarding the comparative effectiveness of different adjunctive nursing strategies. Network meta-analysis (NMA) provides a robust statistical framework that integrates direct and indirect comparisons across multiple treatments, enabling comprehensive ranking and comparison of diverse interventions. Given the increasing clinical and economic burden associated with DFU and the critical role of adjunctive nursing interventions in its management, an NMA can systematically address the knowledge gaps and provide clearer guidance for clinical practice [14]. Therefore, this study aimed to systematically evaluate and rank the efficacy and safety of different non-pharmacological adjunctive nursing interventions for patients with diabetic foot ulcers. The findings of this analysis will offer evidence-based insights to optimize nursing care strategies, improve clinical outcomes, and inform healthcare policy decisions regarding DFU management. Methods The present systematic review and network meta-analysis (NMA) were performed and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and the PRISMA extension statement specifically developed for network meta-analyses of healthcare interventions [15, 16]. Data sources and search strategy A comprehensive systematic search was conducted in PubMed, Embase, PsycINFO, Cochrane Central Register of Controlled Trials (CENTRAL), and Web of Science from database inception to March 31, 2025. The search strategy included relevant Medical Subject Headings (MeSH) and keywords relating to diabetic foot (e.g., “diabetic foot ulcer”) and various non-pharmacological adjunctive care strategies (e.g., “Hyperbaric Oxygen Therapy”, “Light Therapy”, “Negative Pressure Wound Therapy”, “Extracorporeal Shock Wave Therapy”, “Exercise Therapy”). The detailed search strategy, including specific terms and combinations, is provided in Supplementary File 1. Reference lists from included studies and systematic reviews published in the past five years were additionally screened to ensure no eligible studies were overlooked. Two independent reviewers (J.G. and T.G.) conducted the screening of titles/abstracts and full texts. Any discrepancies were resolved through discussion or, if necessary, adjudication by a third reviewer (X.H.S.). Study selection Studies were included based on predefined inclusion criteria: (1) Population (P): Patients diagnosed with diabetic foot ulcers; (2) Intervention (I): Non-pharmacological adjunctive care strategies; (3) Comparator (C): Conventional wound care or another non-pharmacological intervention different from the experimental intervention for head-to-head comparisons; comparators had to receive conventional care comparable to the experimental arm, and trials with non-balanced pharmacologic co-interventions between arms were excluded; (4) Outcomes (O): Reported at least one of the following outcomes: healing rate, healing time, wound area change, recurrence rate, amputation rate, or adverse events; and (5) Studies published in English. During full-text screening, four otherwise eligible studies were excluded solely because they were reported in non-English languages. This restriction was applied to ensure accuracy in data extraction, correct interpretation of complex intervention protocols, and consistency in risk-of-bias assessment. Exclusion criteria comprised: (1) Studies combining pharmacological treatments; (2) Studies lacking clear descriptions of treatment modalities; (3) Studies without reported means and standard deviations for continuous outcomes or without response to requests for missing data. Two independent reviewers (J.G. and T.G.) assessed eligibility by reviewing titles, abstracts, and full texts based on these criteria. Data extraction Relevant data were managed using EndNote X9 software to exclude duplicates systematically. Two reviewers (J.G. and T.G.) independently extracted study characteristics, including authorship, publication year, study design, sample size, patient demographics (age, sex, duration of diabetes, baseline wound characteristics), coded in accordance with the TIDieR (Template for Intervention Description and Replication) and TIP (Treatment Implementation Protocol) frameworks, and reported outcome measures, as detailed in Supplementary File 2. To enhance transparency, a standardized summary of intervention definitions and examples for each network node has been provided in Supplementary File 5. Missing mean values and standard deviations were imputed according to methods outlined in the Cochrane Handbook [17]. Specifically, when SDs were unavailable, we derived them from SEs, 95% CIs, t statistics, or exact p values; when only medians with IQRs or ranges were reported, we approximated means and SDs using recommended formulas; and where studies mixed postintervention values and change scores, we combined them using standard approaches. For dichotomous outcomes with missing data, we used available-case/ITT-consistent denominators. For data unobtainable through these methods, authors were contacted at least four times within a six-week interval for clarification and data retrieval. If variance measures remained unavailable after author contact, we imputed SDs from pooled variability of clinically similar trials within the same node and assessed the impact in sensitivity analyses. Risk of bias assessment The revised Cochrane Risk of Bias tool (RoB 2) was employed independently by two reviewers (J.G. and T.G.) to evaluate study-level bias, focusing on domains including randomization, deviations from intended interventions, missing outcome data, outcome measurement, and selective reporting of outcomes [18]. Disagreements were resolved through consultation with a third reviewer (X.H.S.). Data coding Interventions identified across eligible studies were systematically coded as follows: CAP (Cold Atmospheric Plasma), CDO (Continuous Diffusion of Oxygen), ES (Electrical Stimulation), ESWT (Extracorporeal Shock Wave Therapy), EXER (Exercise Therapy), HBOT (Hyperbaric Oxygen Therapy), LT (Light Therapy), NPWT (Negative Pressure Wound Therapy), TMM (Telemedical Monitoring), and US (Ultrasound Therapy). Conventional care protocols across studies were uniformly coded as “CON.” For each intervention node, operational definitions and representative examples were standardized and are reported in Supplementary File 5 (“Definitions and examples of non-pharmacological interventions for diabetic foot ulcer management”). To ensure the validity of network comparisons, all alternative interventions included within a given node were carefully reviewed for clinical similarity in terms of therapeutic mechanism, delivery modality, and implementation context. Interventions were only pooled when deemed sufficiently homogeneous, and protocol-level details (e.g., treatment frequency, duration, and device specifications) are summarized in Supplementary File 2 and Supplementary File 5. Outcome measures The primary outcomes assessed included healing rate, defined as complete epithelialization of the ulcer without drainage as confirmed by clinical examination; healing time, measured as the duration (in weeks) from randomization until complete ulcer closure; wound area change, assessed using standardized planimetric methods such as digital photography with software-based planimetry, acetate tracings, or caliper measurement, and reported as the absolute reduction in wound area (cm2); recurrence rate, defined as the reappearance of a full-thickness ulcer at the same site during the follow-up period; amputation rate, defined as surgical removal of part of the foot or lower limb, further categorized as minor (toe/forefoot) or major (above the ankle); and adverse events, defined as all treatment-related complications (e.g., infection, bleeding, pain, or device-related harm) as reported in individual trial protocols. Each outcome was clearly defined, and assessment methodologies were consistently reviewed and documented. Data analysis Statistical analyses were performed using Stata software version 17.0 (StataCorp LLC, Texas, USA). Network meta-analysis was conducted to compare efficacy and safety outcomes across various non-pharmacological adjunctive interventions for diabetic foot ulcers. Network diagrams were generated to visualize direct and indirect comparisons, verifying the network’s structural appropriateness. Considering anticipated clinical heterogeneity, a random-effects model was applied to account for within- and between-study variance. Continuous outcomes (healing time and wound area change) were standardized using standardized mean differences (SMD) with 95% confidence intervals (CIs) due to variations in measurement methods and units. Dichotomous outcomes (healing rate, recurrence rate, amputation rate, and adverse events) were expressed as odds ratios (ORs) with 95% CIs. ORs were selected as the effect measure because they are the default and statistically robust metric within Stata’s mvmeta framework, ensuring consistency across studies with variable baseline risks and rare events such as recurrence and amputation. Although relative risks (RRs) were considered for their greater clinical interpretability, ORs were prioritized to maintain methodological coherence across the network and to avoid bias when event rates were low. All network meta-analyses were performed within a Frequentist framework using Stata’s “network” and “mvmeta” commands, which estimate treatment effects based on maximum likelihood with random effects to account for heterogeneity. The Frequentist framework was selected because it provides robust estimation of pooled effects, transparent calculation of confidence intervals, and reproducible SUCRA-based treatment rankings suitable for clinical interpretation. Treatment rankings were estimated based on surface under the cumulative ranking curves (SUCRA), with higher SUCRA values indicative of superior relative effectiveness. Comprehensive SUCRA-based ranking probabilities and cumulative ranking plots for all outcomes are presented in Supplementary File 9. Adjusted funnel plots and Egger’s regression tests were conducted to identify potential publication bias (p-value < 0.05 indicating potential bias) [19]. Prediction interval plots further explored heterogeneity, offering insights into variability and applicability of estimated treatment effects. To assess the plausibility of the transitivity assumption, potential effect modifiers (including baseline ulcer severity, participant characteristics such as age, sex, diabetes duration, and comorbidities, as well as variations in standard care protocols) were identified and their distributions compared across treatment nodes. Because the contributing studies differed by outcome, these assessments were conducted separately for each outcome network. Global network inconsistency was assessed using the design-by-treatment interaction model, and local inconsistency was evaluated with node-splitting and loop-specific approaches, each conducted separately for every outcome network; all tests yielded p > 0.05, indicating no statistically significant disagreement between direct and indirect evidence. Detailed outputs are provided in Supplementary File 6, and the complete inconsistency diagnostic results for each outcome network are summarized in Supplementary File 7. A prespecified sensitivity analysis excluding trials at high risk of bias (RoB 2) was conducted; results are presented in the Results section and Supplementary File 8. Finally, we assessed the certainty of evidence for each outcome using the CINeMA framework across six domains (within-study bias, reporting bias, indirectness, imprecision, heterogeneity, and incoherence), with full domain ratings and summary judgments provided in Supplementary File 10. Results Characteristics of included studies An initial electronic search identified 3033 records. After removing 1106 duplicates, 1927 articles were screened by title and abstract. Following the exclusion of 1025 articles based on title and abstract, 1584 articles underwent full-text eligibility screening. Ultimately, 67 RCTs involving 5957 patients with diabetic foot ulcers were included in the systematic review and network meta-analysis [10, 20–85], as depicted in the PRISMA flow diagram (Fig. 1). Prisma 2020 flow diagram of study selection. Numbers at each stage reflect records identified, screened, assessed for eligibility, and included in the network meta-analysis; reasons for exclusion at full-text are summarized in supplementary materials The included studies were published between 1992 and 2024, with a median publication year of 2017. Sample sizes ranged from 11 to 380 participants per study, with a median of 54 participants. The mean age of participants varied from 37.3 to 72.0 years, with a median of 60.2 years. The non-pharmacological nursing interventions evaluated were distributed as follows: HBOT in 15 studies, CDO and ESWT in 9 studies each, LT and NPWT in 8 studies each, ES in 6 studies, TMM in 5 studies, and CAP, EXER, and US in 3 studies each. Detailed characteristics of the included studies are provided in Supplementary File 2. Results of network meta-analysis Global design-by-treatment tests and local node-splitting/loop-specific assessments indicated no statistically significant inconsistency across the outcome-specific networks (all p > 0.05). In addition, certainty of evidence was appraised using the CINeMA framework; in brief, confidence was moderate for a few key contrasts (e.g., LT vs CON for healing rate; US vs CON for wound area; ES vs ESWT for healing time [high]) but low or very low for most other comparisons owing to imprecision, heterogeneity, and/or incoherence; full domain-level judgments are provided in Supplementary File 10. Healing rate The network meta-analysis for healing rate included 56 studies with 4640 patients. The network plot illustrating direct comparisons among interventions is presented in Fig. 2.1. SUCRA values (Fig. 3.1) ranked LT (SUCRA = 88.8%), US (78.8%), and EXER (76.2%) as the top three interventions for enhancing healing rate. Compared to CON, significant increases in healing rate were observed with LT (OR = 7.62, 95% CI: 2.92–19.90), US (OR = 5.94, 95% CI: 1.67–21.16), EXER (OR = 5.60, 95% CI: 1.43–21.86), ESWT (OR = 3.59, 95% CI: 1.82–7.07), HBOT (OR = 2.80, 95% CI: 1.49–5.24), NPWT (OR = 2.63, 95% CI: 1.28–5.41), and CDO (OR = 2.53, 95% CI: 1.40–4.58). Additionally, LT significantly outperformed TMM (OR = 4.77, 95% CI: 1.32–17.31). Results are summarized in Table 1. By CINeMA, the LT vs CON contrast was rated moderate confidence, whereas most other active-vs-CON comparisons were low and HBOT vs CON was very low, predominantly downgraded for heterogeneity and/or imprecision (Supplementary File 10). Network geometry by outcome. 2.1 healing rate; 2.2 healing time; 2.3 wound area change; 2.4 recurrence rate; 2.5 amputation rate; 2.6 adverse events. Node size is proportional to sample size and edge width to number of direct comparisons. Abbreviations: CAP, cold atmospheric plasma; CDO, continuous diffusion of oxygen; ES, electrical stimulation; ESWT, extracorporeal shock-wave therapy; exer, exercise therapy; HBOT, hyperbaric oxygen therapy; LT, light therapy; NPWT, negative pressure wound therapy; TMM, telemedical monitoring; US, ultrasound; CON, conventional care Treatment rankings by outcome (SUCRA). 3.1 healing rate; 3.2 healing time; 3.3 wound area change; 3.4 recurrence rate; 3.5 amputation rate; 3.6 adverse events. Higher SUCRA indicates higher probability of being among the best treatments for a given outcome Table 1Results of network meta-analysis for the effect of non-pharmacological nursing interventions on healing rate (odds ratio, 95% CI)LT1.28 (0.26,6.31)US1.36 (0.26,7.21)1.06 (0.16,6.84)EXER2.12 (0.65,6.89)1.65 (0.39,6.99)1.56 (0.34,7.13)ESWT2.72 (0.86,8.60)2.12 (0.51,8.76)2.00 (0.45,8.95)1.28 (0.60,2.76)HBOT2.89 (0.87,9.62)2.25 (0.52,9.71)2.12 (0.46,9.91)1.36 (0.52,3.59)1.06 (0.42,2.67)NPWT3.01 (0.97,9.31)2.34 (0.58,9.53)2.21 (0.50,9.77)1.42 (0.58,3.47)1.10 (0.47,2.59)1.04 (0.41,2.63)CDO3.41 (0.88,13.21)2.66 (0.54,13.02)2.51 (0.48,13.21)1.61 (0.50,5.13)1.25 (0.41,3.86)1.18 (0.36,3.86)1.13 (0.37,3.48)ES4.77 (1.32,17.31)3.72 (0.80,17.24)3.51 (0.70,17.54)2.25 (0.76,6.67)1.75 (0.61,5.02)1.65 (0.54,5.03)1.59 (0.56,4.49)1.40 (0.39,5.03)TMM7.62 (2.92,19.90)5.94 (1.67,21.16)5.60 (1.43,21.86)3.59 (1.82,7.07)2.80 (1.49,5.24)2.63 (1.28,5.41)2.53 (1.40,4.58)2.23 (0.86,5.79)1.60 (0.68,3.77)CONNote: Odds ratios (ORs) > 1 favor the row intervention over the column comparator. Abbreviations: CAP, cold atmospheric plasma; CDO, continuous diffusion of oxygen; ES, electrical stimulation; ESWT, extracorporeal shock-wave therapy; EXER, exercise therapy; HBOT, hyperbaric oxygen therapy; LT, light therapy; NPWT, negative-pressure wound therapy; TMM, telemedical monitoring; US, ultrasound therapy; CON, conventional care. CI denotes confidence interval Results of network meta-analysis for the effect of non-pharmacological nursing interventions on healing rate (odds ratio, 95% CI) Note: Odds ratios (ORs) > 1 favor the row intervention over the column comparator. Abbreviations: CAP, cold atmospheric plasma; CDO, continuous diffusion of oxygen; ES, electrical stimulation; ESWT, extracorporeal shock-wave therapy; EXER, exercise therapy; HBOT, hyperbaric oxygen therapy; LT, light therapy; NPWT, negative-pressure wound therapy; TMM, telemedical monitoring; US, ultrasound therapy; CON, conventional care. CI denotes confidence interval Healing time The analysis for healing time encompassed 17 studies with 2256 patients. The network plot is shown in Fig. 2.2. SUCRA rankings (Fig. 3.2) identified ESWT (SUCRA = 94.4%), NPWT (69.8%), and CDO (60.2%) as the leading interventions for reducing healing time. ESWT significantly shortened healing time compared to ES (SMD = −2.67, 95% CI: −4.81 to −0.54), CON (SMD = −2.02, 95% CI: −3.33 to −0.71), and HBOT (SMD = −1.64, 95% CI: −3.28 to −0.01). Results are detailed in Table 2. CINeMA indicated high confidence for ES vs ESWT, while ESWT vs CON and NPWT vs CON were generally low due to downgrades for heterogeneity or imprecision (Supplementary File 10). Table 2Results of network meta-analysis for the effect of non-pharmacological nursing interventions on healing Time (standardized mean difference, 95% CI)ESWT−0.99 (−2.68,0.70)NPWT−1.22 (−3.12,0.67)−0.23 (−1.97,1.51)CDO-1.64 (−3.28,-0.01)−0.65 (−2.45,1.15)−0.42 (−2.41,1.57)HBOT−1.90 (−4.59,0.80)−0.91 (−3.50,1.68)−0.68 (−3.40,2.05)−0.26 (−3.02,2.51)TMM-2.02 (−3.33,-0.71)−1.03 (−2.10,0.05)−0.80 (−2.17,0.57)−0.38 (−1.83,1.07)−0.12 (−2.48,2.24)CON-2.67 (−4.81,-0.54)−1.68 (−3.68,0.32)−1.45 (−3.62,0.72)−1.03 (−3.25,1.19)−0.77 (−3.67,2.12)−0.65 (−2.34,1.03)ESNote: Standardized mean differences (SMDs) < 0 favor the row intervention (shorter healing time). Abbreviations as in Table 1. SMD denotes standardized mean difference Results of network meta-analysis for the effect of non-pharmacological nursing interventions on healing Time (standardized mean difference, 95% CI) Note: Standardized mean differences (SMDs) < 0 favor the row intervention (shorter healing time). Abbreviations as in Table 1. SMD denotes standardized mean difference Wound area change The network meta-analysis for wound area change included 44 studies with 3680 patients. The network plot is displayed in Fig. 2.3. SUCRA values (Fig. 3.3) indicated that US (98.4%), NPWT (66.3%), and LT (57.2%) were the most effective interventions for reducing wound area. US significantly reduced wound area compared to CON (SMD = −3.17, 95% CI: −4.81 to −1.53), EXER (SMD = −3.23, 95% CI: −5.48 to −0.98), CAP (SMD = −2.79, 95% CI: −5.04 to −0.54), ESWT (SMD = −2.60, 95% CI: −4.62 to −0.58), HBOT (SMD = −2.55, 95% CI: −4.36 to −0.73), ES (SMD = −2.32, 95% CI: −4.43 to −0.21), LT (SMD = −2.31, 95% CI: −4.26 to −0.37), and NPWT (SMD = −2.07, 95% CI: −4.04 to −0.11). NPWT also significantly reduced wound area compared to CON (SMD = −1.10, 95% CI: −1.53 to −0.67). Results are presented in Table 3. CINeMA rated US vs CON as moderate confidence and several US-anchored head-to-head contrasts (e.g., CAP vs US, ESWT vs US, EXER vs US, HBOT vs US) as moderate to high; most remaining comparisons were low, typically downgraded for imprecision or heterogeneity (Supplementary File 10). Table 3Results of network meta-analysis for the effect of non-pharmacological nursing interventions on wound area reduction (standardized mean difference, 95% ci)US-2.07 (−4.04,-0.11)NPWT-2.31 (−4.26,-0.37)−0.24 (−1.74,1.26)LT-2.32 (−4.43,-0.21)−0.25 (−1.96,1.47)−0.00 (−1.69,1.68)ES-2.55 (−4.36,-0.73)−0.47 (−1.80,0.86)−0.23 (−1.53,1.06)−0.23 (−1.77,1.31)HBOT−2.50 (−5.60,0.59)−0.43 (−3.27,2.41)−0.19 (−3.01,2.64)−0.18 (−3.12,2.76)0.05 (−2.69,2.78)CDO−2.59 (−5.68,0.50)−0.52 (−3.35,2.31)−0.28 (−3.09,2.54)−0.27 (−3.21,2.66)−0.04 (−2.77,2.68)−0.09 (−3.80,3.62)TMM-2.60 (−4.62,-0.58)−0.53 (−2.13,1.08)−0.29 (−1.86,1.29)−0.28 (−2.06,1.50)−0.05 (−1.38,1.27)−0.10 (−2.98,2.78)−0.01 (−2.88,2.86)ESWT-2.79 (−5.04,-0.54)−0.72 (−2.60,1.16)−0.48 (−2.34,1.38)−0.47 (−2.51,1.56)−0.25 (−1.97,1.48)−0.29 (−3.34,2.75)−0.20 (−3.24,2.83)−0.19 (−2.13,1.75)CAP-3.23 (−5.48,-0.98)−1.16 (−3.04,0.73)−0.92 (−2.78,0.94)−0.91 (−2.95,1.13)−0.68 (−2.41,1.04)−0.73 (−3.78,2.32)−0.64 (−3.68,2.40)−0.63 (−2.57,1.32)−0.44 (−2.62,1.74)EXER-3.17 (−4.81,-1.53)-1.10 (−2.18,-0.02)−0.86 (−1.90,0.18)−0.85 (−2.18,0.48)−0.63 (−1.40,0.15)−0.67 (−3.30,1.95)−0.58 (−3.20,2.03)−0.57 (−1.75,0.61)−0.38 (−1.92,1.16)0.06 (−1.49,1.60)CONNote: Standardized mean differences (SMDs) < 0 favor the row intervention (shorter healing time). Abbreviations as in Table 1. SMD denotes standardized mean difference Results of network meta-analysis for the effect of non-pharmacological nursing interventions on wound area reduction (standardized mean difference, 95% ci) Note: Standardized mean differences (SMDs) < 0 favor the row intervention (shorter healing time). Abbreviations as in Table 1. SMD denotes standardized mean difference Recurrence rate The analysis for recurrence rate involved 9 studies with 699 patients. The network plot is shown in Fig. 2.4. SUCRA rankings (Fig. 3.4) identified EXER (97.4%), CDO (68.7%), and ES (50.9%) as the top interventions for minimizing recurrence rate. EXER significantly reduced recurrence rate compared to US (OR = 0.03, 95% CI: 0.00–0.70), CON (OR = 0.07, 95% CI: 0.02–0.33), and TMM (OR = 0.14, 95% CI: 0.03–0.70). Results are summarized in Table 4. CINeMA judgments were predominantly very low for recurrence due to sparse data, imprecision, heterogeneity, and incoherence; only the CON vs EXER comparison achieved low confidence, warranting cautious interpretation (Supplementary File 10). Table 4Results of network meta-analysis for the effect of non-pharmacological nursing interventions on recurrence rate(odds ratio, 95% CI)EXER0.25 (0.04,1.69)CDO0.15 (0.02,1.12)0.60 (0.10,3.73)ES0.14 (0.03,0.70)0.55 (0.14,2.18)0.92 (0.20,4.25)TMM0.07 (0.02,0.33)0.30 (0.09,1.00)0.50 (0.13,1.98)0.54 (0.28,1.05)CON0.03 (0.00,0.70)0.12 (0.01,2.50)0.20 (0.01,4.50)0.22 (0.01,3.81)0.40 (0.02,6.53)USNote: Odds ratios (ORs) < 1 favor the row intervention (lower recurrence rate). Abbreviations as in Table 1 Results of network meta-analysis for the effect of non-pharmacological nursing interventions on recurrence rate(odds ratio, 95% CI) Note: Odds ratios (ORs) < 1 favor the row intervention (lower recurrence rate). Abbreviations as in Table 1 Amputation rate The network meta-analysis for amputation rate included 25 studies with 2529 patients. The network plot is presented in Fig. 2.5. SUCRA values (Fig. 3.5) ranked ESWT (68.4%), LT (62.6%), and HBOT (58.2%) as the most effective interventions for reducing amputation rate. HBOT significantly lowered the amputation rate compared to CON (OR = 0.40, 95% CI: 0.23 - 0.70). Results are detailed in Table 5. CINeMA rated the certainty for nearly all amputation contrasts as very low, primarily due to imprecision and incoherence, thereby limiting the strength of inference for this endpoint (Supplementary File 10). Table 5Results of network meta-analysis for the effect of non-pharmacological nursing interventions on amputation rate (odds ratio, 95% CI)ESWT0.62 (0.01,30.20)LT0.45 (0.02,13.40)0.73 (0.09,5.79)HBOT0.57 (0.00,64.51)0.92 (0.02,45.36)1.26 (0.04,37.62)ES0.56 (0.00,63.35)0.91 (0.02,44.47)1.24 (0.04,36.83)0.99 (0.01,112.47)EXER0.36 (0.01,11.06)0.58 (0.07,4.91)0.79 (0.30,2.06)0.63 (0.02,19.67)0.64 (0.02,19.64)TMM0.35 (0.01,24.02)0.57 (0.02,14.84)0.78 (0.06,10.92)0.62 (0.01,42.66)0.63 (0.01,42.70)0.99 (0.07,14.63)CDO0.31 (0.01,9.46)0.51 (0.06,4.14)0.69 (0.29,1.64)0.55 (0.02,16.82)0.56 (0.02,16.80)0.88 (0.33,2.36)0.89 (0.06,12.69)NPWT0.18 (0.01,5.09)0.29 (0.04,2.13)0.40 (0.23,0.70)0.32 (0.01,9.06)0.32 (0.01,9.04)0.50 (0.23,1.10)0.51 (0.04,6.70)0.57 (0.30,1.12)CONNote: Odds ratios (ORs) < 1 favor the row intervention (lower risk of amputation). Abbreviations as in Table 1 Results of network meta-analysis for the effect of non-pharmacological nursing interventions on amputation rate (odds ratio, 95% CI) Note: Odds ratios (ORs) < 1 favor the row intervention (lower risk of amputation). Abbreviations as in Table 1 Adverse events The analysis for adverse events encompassed 28 studies with 3108 patients. The network plot is displayed in Fig. 2.6. SUCRA rankings (Fig. 3.6) indicated that CDO (89.2%), LT (74.1%), and NPWT (55.0%) were associated with the fewest adverse events. CDO significantly reduced adverse events compared to TMM (OR = 0.19, 95% CI: 0.04–0.90), ES (OR = 0.19, 95% CI: 0.04–0.94), and CON (OR = 0.27, 95% CI: 0.08–0.85). Results are presented in Table 6. CINeMA generally indicated low confidence across adverse-event comparisons (e.g., CDO vs CON downgraded for heterogeneity), with no contrast reaching moderate certainty (Supplementary File 10). Table 6Results of network meta-analysis for the effect of non-pharmacological nursing interventions on adverse events (odds ratio, 95% CI)CDO0.66 (0.09,4.81)LT0.37 (0.09,1.50)0.56 (0.09,3.57)NPWT0.36 (0.08,1.56)0.55 (0.08,3.62)0.98 (0.27,3.56)ESWT0.34 (0.07,1.65)0.52 (0.07,3.73)0.93 (0.23,3.80)0.95 (0.26,3.52)HBOT0.36 (0.04,3.34)0.54 (0.04,6.64)0.97 (0.12,8.04)0.99 (0.12,8.45)1.04 (0.11,9.56)CAP0.27 (0.03,2.32)0.41 (0.04,4.66)0.72 (0.09,5.57)0.74 (0.09,5.87)0.78 (0.09,6.65)0.74 (0.05,10.54)US0.27 (0.08,0.85)0.41 (0.08,2.03)0.72 (0.29,1.78)0.74 (0.28,1.95)0.78 (0.25,2.38)0.74 (0.11,5.05)1.00 (0.16,6.26)CON0.19 (0.04,0.94)0.29 (0.04,2.06)0.52 (0.13,2.15)0.53 (0.12,2.30)0.56 (0.12,2.68)0.54 (0.06,4.88)0.72 (0.09,6.12)0.72 (0.24,2.16)ES0.19 (0.04,0.90)0.29 (0.04,1.97)0.52 (0.13,2.05)0.53 (0.13,2.19)0.55 (0.12,2.56)0.53 (0.06,4.69)0.71 (0.09,5.88)0.71 (0.25,2.02)0.99 (0.22,4.49)TMMNote: Odds ratios (ORs) < 1 favor the row intervention (fewer adverse events). Abbreviations as in Table 1 Results of network meta-analysis for the effect of non-pharmacological nursing interventions on adverse events (odds ratio, 95% CI) Note: Odds ratios (ORs) < 1 favor the row intervention (fewer adverse events). Abbreviations as in Table 1 Risk of bias and publication bias The risk of bias in the 67 included randomized controlled trials was evaluated using the RoB 2. Assessments were conducted independently by two reviewers across five domains: bias arising from the randomization process, bias due to deviations from intended interventions, bias due to missing outcome data, bias in the measurement of the outcome, and bias in the selection of the reported result. Discrepancies were resolved through discussion or consultation with a third reviewer. Overall, 36 studies were judged to have a low risk of bias, 25 were classified as having some concerns, and 6 were deemed to have a high risk of bias. In the randomization process domain, 54 studies were assessed as low risk, 12 had some concerns, and 1 were at high risk. For deviations from intended interventions, 56 studies were rated as low risk, 7 had some concerns, and 4 were at high risk. Regarding missing outcome data, 51 studies were considered low risk, 12 had some concerns, and 4 were at high risk. In the measurement of the outcome domain, 59 studies were assessed as low risk, 7 had some concerns, and 1 were at high risk. For the selection of the reported result, all 67 studies were judged to have a low risk of bias. Detailed risk of bias assessments for each study are provided in Supplementary File 3. Publication bias was evaluated using funnel plots (Supplementary File 4: Figure S1). Funnel plots for healing rate, healing time, wound area change, recurrence rate, amputation rate, and adverse events showed varying degrees of symmetry. Figures S1.4 and S1.5 exhibited relatively symmetrical distributions, suggesting minimal bias, whereas Figures S1.1, S1.2, S1.3 and S1.6 displayed some asymmetry. Egger’s test revealed significant publication bias for healing rate and amputation rate (p < 0.05), necessitating cautious interpretation of these outcomes. For the remaining outcomes, Egger’s test p-values were greater than 0.05, suggesting no substantial evidence of publication bias. After exclusion of the six high-risk trials, repeat Egger’s tests for healing rate and amputation rate yielded p-values > 0.05 and the corresponding funnel plots showed improved symmetry, indicating attenuation of small-study effects and supporting the robustness of our primary inferences. Discussion This network meta-analysis integrated data from 67 RCTs involving 5957 patients with diabetic foot ulcers to evaluate and compare the efficacy and safety of various non-pharmacological adjunctive nursing interventions. Several key findings emerged from the analysis. LT and US demonstrated superior efficacy in promoting ulcer healing and reducing wound area. LT ranked highest in enhancing the healing rate, with a SUCRA value of 88.8%, significantly improving healing rates compared to CON. US was the most effective intervention in reducing wound area (SUCRA = 98.4%), significantly outperforming CON. These results indicate that LT and US offer distinct advantages in wound closure and tissue repair, providing critical intervention options for clinical nursing practice. Additionally, ESWT showed notable efficacy in reducing healing time, achieving the highest SUCRA value of 94.4% and significantly shortening healing time compared to conventional care. ESWT also significantly reduced amputation rates, underscoring its dual potential in accelerating wound healing and preventing severe complications. Finally, EXER and CDO effectively improved long-term outcomes; EXER significantly reduced ulcer recurrence rates (SUCRA = 97.4%), while CDO demonstrated the fewest adverse events (SUCRA = 89.2%), highlighting their respective roles in recurrence prevention and treatment safety enhancement. Non-pharmacological nursing interventions are critically important for the management and care of patients with diabetic foot ulcers, given their potential to improve patient outcomes and reduce treatment-associated risks. Previous studies have emphasized that such interventions can minimize the reliance on pharmacological approaches, thus reducing adverse effects, antibiotic resistance, and incomplete wound healing commonly associated with drug therapies [86]. For instance, interventions like negative-pressure wound therapy and hyperbaric oxygen therapy have previously shown significant benefits in improving healing outcomes and patient quality of life [87]. Despite the growing availability of diverse non-pharmacological options, clinical nurses often face uncertainty regarding the optimal selection of interventions due to limited comparative evidence. Consequently, this study was conducted to address this knowledge gap by providing robust comparative data to guide nursing decisions and enhance evidence-based practice. Healing rate and wound area change are considered the critical outcome measures for patients with diabetic foot ulcers, as they directly reflect treatment efficacy and tissue repair capacity. The present study demonstrated that LT and US were the among most effective interventions for these outcomes. Specifically, LT achieved the highest SUCRA value for healing rate, while US ranked second in healing rate and first in reducing wound area. These findings align partially with previous studies that have shown the efficacy of physical energy-based therapies in accelerating wound healing; however, earlier studies have often reported individual interventions without comprehensive comparative analyses [88]. The superiority of LT and US can be explained by their mechanisms of action: LT promotes wound healing through photobiomodulation, enhancing cellular metabolism, collagen synthesis, and angiogenesis, while US facilitates wound repair by increasing tissue perfusion, stimulating fibroblast activity, and reducing inflammation [89, 90]. Notably, data on LT for wound area change were limited, highlighting the need for further studies to conclusively evaluate its efficacy in reducing wound size. Beyond photic and acoustic modalities, ESWT warrants focused consideration. Originally developed for lithotripsy, ESWT has been increasingly deployed in chronic wound care over the past decade because it delivers high-energy acoustic pulses that trigger mechanotransductive cascades within ischaemic tissues [91]. Consistent with this biological rationale, the present NMA identified ESWT as the premier strategy for accelerating ulcer closure, attaining the highest SUCRA value for healing time (94.4%) and conferring a two-standard-deviation reduction in time to healing versus conventional care. Notably, ESWT also achieved the best ranking for amputation avoidance, highlighting its dual clinical utility in expediting repair while preventing limb-threatening progression. These findings extend but also refine earlier pairwise meta-analyses that reported modest gains in granulation tissue formation yet failed to confirm amputation benefit, likely because those reviews pooled heterogeneous control groups and lacked head-to-head contrasts [92, 93]. Preclinical studies provide mechanistic support: shock-wave exposure up-regulates angiogenic mediators (VEGF, eNOS), mobilises endothelial progenitor cells, modulates neuro-inflammatory pathways, and enhances microvascular perfusion—processes that collectively accelerate the proliferative phase and re-establish tissue oxygenation [94, 95]. By hastening these early reparative events, ESWT may shorten the overall duration of ulcer care and reduce the window during which infection and ischaemia precipitate major amputation. Conversely, ESWT did not rank among the top performers for final healing rate or wound-area reduction. This apparent discrepancy may stem from protocol variability (energy flux density, session frequency), the requirement for intact peri-wound tissue to transduce mechanical signals, and the fact that healing rate metrics capture endpoint closure rather than temporal dynamics [93]. Thus, while ESWT is less influential on the ultimate proportion of ulcers that achieve complete epithelialisation, it meaningfully compresses the healing timeline and mitigates catastrophic sequelae, supporting its selective use as a time-critical limb-salvage adjunct in nursing practice. Nevertheless, CINeMA generally graded ESWT-related time-to-healing contrasts as low to moderate certainty because of between-study variability and imprecision, and amputation comparisons were typically very low certainty; these findings should be viewed as hypothesis-strengthening rather than definitive. Beyond acute wound resolution, long-term durability and safety constitute pivotal goals of nursing management for diabetic foot ulcers. The current NMA revealed that EXER provided the most robust protection against ulcer recurrence, achieving a SUCRA value of 97.4% and significantly reducing recurrence risk relative to conventional care. This superiority corroborates cohort evidence indicating that structured weight-bearing and resistance regimens improve plantar pressure distribution, enhance glycaemic control, and strengthen intrinsic foot musculature—factors that collectively reduce mechanical stress and reinjury of healed tissue [96]. In parallel, CDO exhibited the lowest rate of adverse events (SUCRA = 89.2%), outperforming both device-based and energy-based modalities. CDO delivers low-flow oxygen directly to the wound bed, sustaining a moist, hypoxic-relieved microenvironment without the barotrauma or shear forces inherent to other oxygenation strategies [97]. The absence of mechanical contact and systemic pharmacologic exposure likely underpins its favourable safety profile. Together, these findings emphasise that integrating patient-activated behavioural interventions such as EXER with low-risk oxygenation adjuncts like CDO may optimise long-term limb integrity while minimising treatment-related complications—an imperative for holistic, nurse-led chronic wound care. However, CINeMA certainty for recurrence and adverse-event endpoints was predominantly low to very low due to sparse data and heterogeneity, warranting cautious clinical interpretation of these rankings. Our findings both corroborate and extend intervention-specific meta-analyses. For NPWT, the superiority for wound-area reduction and favourable healing profile are consistent with prior pairwise syntheses, while the network framework clarifies NPWTs relative position against energy-based therapies (e.g., US, LT) across multiple outcomes [8, 13, 87]. For HBOT, our estimate of reduced amputation risk aligns with earlier reviews [7] and situates HBOT alongside ESWT, which our analysis ranks highest for shortening healing time and suggests benefit for amputation within the network—extending earlier ESWT reviews that focused on granulation/time-to-closure but were inconclusive for limb outcomes [98, 99]. Our ranking of LT for complete healing and US for wound-area reduction is concordant with prior light-/ultrasound-therapy syntheses [9, 36], and the NMA contributes a comparative hierarchy that is difficult to achieve with pairwise evidence. Finally, by evaluating CDO’ alongside device- and energy-based comparators, we complement efficacy-focused topical oxygen reports and clarify that CDOs safety profile ranks favourably while efficacy remains competitive for selected outcomes [36]. Importantly, these cross-study comparisons were interpreted in light of CINeMA judgments, with stronger inferences reserved for contrasts graded moderate certainty and cautious language applied where certainty was low/very low. This study offers two principal strengths. First, by synthesising evidence from sixty-seven randomised controlled trials addressing ten non-pharmacological interventions and six clinically relevant outcomes, it provides a highly comprehensive comparative overview currently available and offers nurses with pragmatic guidance for tailoring care. Second, the application of dual independent reviewers and advanced Bayesian network modelling strengthened the reliability of effect estimates by concurrently accounting for both direct and indirect evidence. Several limitations merit consideration. Clinical and methodological heterogeneity remained despite random-effects modelling, because individual trials differed in ulcer severity, off-loading protocols, and treatment dosimetry. The evidence base for certain outcomes, particularly recurrence rate and amputation rate, relied on a relatively small number of trials with modest sample sizes, which resulted in wide confidence intervals and reduced statistical power. Publication bias cannot be fully excluded, as contour-enhanced funnel plots and Egger tests indicated possible small-study effects for healing and amputation outcomes. Approximately one quarter of the included trials were judged to have some concerns or high risk of bias, which may attenuate the certainty of the pooled findings. Finally, all included studies were published in English, and therefore language bias might have limited the comprehensiveness of the literature search. Indeed, four otherwise eligible studies were excluded solely due to non-English language. While modern AI-assisted translation tools may facilitate broader inclusion, we restricted to English to ensure consistent data extraction, accurate interpretation of complex intervention protocols, and robust risk-of-bias assessment. This methodological choice, however, introduces potential selection bias that should be considered when interpreting the findings. In addition, cost-effectiveness data and pragmatic real-world evaluations were largely absent, restricting the ability to translate these findings into diverse healthcare systems. Long-term endpoints such as durability of ulcer closure, functional mobility, and patient-reported quality of life were inconsistently assessed, precluding firm conclusions regarding sustained benefits. Moreover, variation in intervention protocols (e.g., treatment duration, intensity, and device specifications) limits direct comparability across trials and underscores the need for standardized regimens in future research. Although subgroup analyses and meta-regression for prespecified effect modifiers (e.g., ulcer severity/phenotype, baseline wound size and duration, HbA1c, off-loading regimen, and region) were planned, the small number of studies within several node–outcome combinations and inconsistent reporting of key covariates precluded credible exploration of heterogeneity; consequently, residual unexplained heterogeneity may persist and estimates should be interpreted with appropriate caution. Although interventions within the same node were carefully reviewed for clinical similarity and coded as adjuncts over shared conventional care before pooling, residual heterogeneity in protocol details and implementation settings cannot be entirely excluded, and this should be considered when interpreting indirect comparisons. Conclusion This network meta-analysis of sixty-seven randomised controlled trials delineates an evidence-based hierarchy for adjunctive nursing care in diabetic foot ulcers. Light therapy achieved the notable absolute improvement in complete healing; ultrasound therapy produced the considerable reduction in wound area. Extracorporeal shock-wave therapy markedly shortened healing time and simultaneously lowered major amputation risk, indicating value when urgent tissue salvage is required. Exercise programmes highly effectively prevented ulcer recurrence, while continuous diffusion of oxygen exhibited the safest profile with the fewest adverse events. These modality-specific advantages permit targeted selection of interventions that align with distinct clinical goals, whether rapid closure, limb preservation, long-term durability, or safety. Although outcome data for recurrence and amputation remain limited and study heterogeneity persists, the present findings provide concrete guidance for nurse-led treatment planning and resource allocation. Future trials should refine dosing protocols, explore combined regimens, and evaluate cost-effectiveness to strengthen guidelines for comprehensive diabetic foot care. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements The authors thank the financial support of the Research Funding for Doctoral Talents of the First Affiliated Hospital of Anhui Medical University (Grant No.1952). Author contributions J.G. and X.H.S. designed the study and developed the retrieve strategy. J.G. and T.G. executed the systematic evaluation as the first and second reviewers, searching and screening the summaries and titles, assessing the inclusion and exclusion criteria, generating data collection forms and extracting data, and evaluating the quality of the study. Y.Z. and J.G. performed meta-analysis. J.G. drafted the article, which was reviewed and revised by H.F and T.G. All authors reviewed the manuscript. Funding The research of our article was funded by Research Funding for Doctoral Talents of the First Affiliated Hospital of Anhui Medical University (Grant No.1952). Data availability The data of this study can be obtained from the corresponding author according to reasonable requirements. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. 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 The data of this study can be obtained from the corresponding author according to reasonable requirements.