Abstract

Objectives:

To compare the clinical effects of extracorporeal shock wave therapy (ESWT) combined with conventional physical therapy (CPT) versus monotherapy for chronic low back pain (CLBP).

Methods:

A frequentist network meta-analysis was conducted in Stata/MP 18 using the network package. PubMed, Web of Science, Embase, and the Cochrane Library were searched from inception to April 11, 2026. Randomized controlled trials (RCTs) evaluating ESWT with or without CPT in adults with CLBP were included. Two reviewers independently screened studies; disagreements were resolved by consensus with a third reviewer. Two reviewers independently extracted study characteristics, intervention details, and outcomes. Risk of bias was assessed using RoB 2.0, and certainty of evidence using GRADE. Treatment ranking probabilities were estimated using surface under the cumulative ranking curves (SUCRA).

Results:

Fourteen RCTs reported pain score (VAS/NRS) and eight reported functional score (ODI); lower scores indicate improvement. According to SUCRA ranking, combined therapy ranked first for both pain relief (96.0%) and functional improvement (99.3%), followed by ESWT, CPT, and sham ESWT in sequence. Combined therapy (ESWT with CPT) provided greater pain relief than sham ESWT (SMD = −1.70; 95% CI: −2.69 to −0.71) and CPT alone (SMD = −0.88; 95% CI: −1.29 to −0.47). ESWT was also superior to sham ESWT (SMD = −1.22; 95% CI: −1.83 to −0.61). For functional improvement, combined therapy achieved significantly greater reductions in ODI than ESWT (MD = -3.60; 95% CI -6.70 to -0.51), sham ESWT (MD = −7.60; 95% CI: −14.00 to −1.12), and CPT (MD = −5.29; 95% CI: −7.53 to −3.06).

Conclusion:

Combined therapy shows promise for alleviating pain and improving function in patients with CLBP. ESWT may help reduce pain, but no significant effects on functional outcomes have been observed. However, more high-quality randomized controlled trials are needed to confirm these findings.

Systematic review registration:

https://www.crd.york.ac.uk/prospero/, identifier 420251170304.

1 Introduction

Chronic low back pain (CLBP) is a clinical syndrome characterized by persistent low back pain lasting for 12 weeks or more (Nicol et al., 2023; Zheng et al., 2025). It can be broadly categorized into myofascial, discogenic, and facet joint-related subtypes (Zhai et al., 2024; Zheng et al., 2025). Epidemiological data indicates that approximately 619 million individuals worldwide were affected by low back pain in 2020, with projections suggesting this number will increase to 843 million by 2050 (Baran et al., 2022; GBD 2021 Low Back Pain Collaborators, 2023). The long-term pain associated with CLBP significantly impairs patients’ quality of life, making it a pressing global public health issue (Baran et al., 2022; GBD 2021 Low Back Pain Collaborators, 2023; Tieppo et al., 2024).

Extracorporeal shock wave therapy (ESWT), a non-invasive modality, has gained attention in CLBP management for its remarkable analgesic efficacy (Simplicio et al., 2020; Yue et al., 2021; Wang et al., 2024). The mechanism of ESWT involves the high-energy pressure wave to target tissues, inducing mechanical transduction effects that provide analgesia, modulate inflammation, promote angiogenesis, improve local microcirculation, and facilitate tissue repair (Tenforde et al., 2022; Zhang and Ma, 2023). However, studies have indicated that ESWT yields no significant long-term functional improvements in CLBP patients (Rajfur et al., 2022; Tan et al., 2024).

Conventional physical therapy (CPT) has long been regarded as one of the preferred interventions for the CLBP clinical management (Hayden et al., 2021; Rizzo et al., 2025). However, Hayden et al. noted that although CPT consistently improves long-term functional outcomes in CLBP management, the quality of evidence supporting its efficacy in short-term pain relief is relatively low (O’Keeffe et al., 2020; Hayden et al., 2021). Meanwhile, Gilanyi et al. pointed out that adherence to conventional rehabilitation regimens is poor, particularly in populations presenting with significant pain symptoms (Gilanyi et al., 2024; Nevelikova et al., 2025).

Accumulating evidence from clinical studies has supported that the combination of ESWT and CPT confers distinct advantages in pain alleviation and functional recovery (Liu et al., 2023; Cashin et al., 2025; Zhang et al., 2025). However, ESWT treatment protocols vary substantially across studies, and the efficacy of the combined therapy remains inconclusive (Elgendy et al., 2020; Burton, 2022). The present study employed a network meta-analysis (NMA) to explore the efficacy of combined therapy and provide evidence-based insights to guide its rational clinical application.

2 Methods

This NMA was reported in accordance with the relevant extension of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline (Page et al., 2021). This review was registered in PROSPERO (420251170304).

2.1 Search strategy

Four electronic databases, including PubMed, Web of Science, Embase, and the Cochrane library, were searched from inception to April 11, 2026. A supplementary search of the ClinicalTrials.gov trial registry was conducted to identify ongoing or unpublished studies. Search strategies combined free-text terms and controlled vocabulary where applicable for “extracorporeal shock wave therapy” and “low back pain.” In addition, the reference lists of included studies and relevant review articles were screened to identify potentially eligible studies. The detailed search strategies for each database are provided in Supplementary Material A.

2.2 Literature screening

Citations retrieved from databases were imported into EndNote version 20 (Clarivate Analytics), and duplicate citations were removed. Two authors independently reviewed the titles and abstracts of each record. The full text was retrieved for potentially eligible studies, and the same 2 authors independently reviewed the full texts, selected studies that met the eligibility criteria for inclusion, and conducted data extraction. Any disagreements were resolved by discussion or by consultation with other authors.

2.3 Inclusion and exclusion criteria

We identified eligible studies according to PICOS. Population: adults with CLBP (pain duration >3 months), including chronic nonspecific low back pain and closely related lumbopelvic pain syndromes reported as low back pain in the original trials. Intervention: ESWT alone or ESWT combined with CPT. Comparator: sham ESWT or CPT alone. Outcomes: pain intensity (VAS/NRS) and disability/function (ODI). Study design: RCTs. Sham ESWT was defined as zero-energy output or the use of an energy-blocking transmission medium. CPT included stretching, strengthening, and other routine rehabilitation modalities. When trials were multi-arm studies or included additional intervention arms outside the predefined network nodes (e.g., injections or pharmacological treatments), only arms that could be mapped to ESWT, sham ESWT, CPT, or combined therapy were included in the network, and non-mappable arms were excluded.

2.4 Data extraction

Two reviewers independently extracted the primary information from a pre-set standardized form. The following data were extracted (1): Basic information, including first author and year of publication. (2) Characteristics of participants, including sample size, age, and gender. (3) Intervention details include the treatment, dosage, duration, process, and follow-up period. (4) Outcomes information. When multiple follow-up time points were reported, the earliest post-intervention or nearest common follow-up time point across studies was used for quantitative synthesis. If the data were ambiguous, the authors of the individual studies were contacted by email or the data were obtained using the image−recognition software GetData v2.26.

2.5 Risk of bias within individual studies

Two reviewers independently assessed risk of bias using the Cochrane Risk of Bias tool, version 2.0 (RoB 2.0) (Sterne et al., 2019); disagreements were resolved by a third reviewer. RoB 2.0 evaluates five domains: bias arising from the randomization process; bias due to deviations from intended interventions; bias due to missing outcome data; bias in measurement of the outcome; and bias in selection of the reported result. For each study, domain-level judgments were low risk, some concerns, or high risk. Overall judgments followed RoB 2.0 guidance: low risk if all domains were low risk; some concerns if ≥1 domain had some concerns and none were high risk (also including cases with insufficient information or non-applicable items); and high risk if any domain was high risk.

2.6 Certainty of evidence

We assessed certainty of evidence using GRADE (Guyatt et al., 2011; Brozek et al., 2021; Qiao et al., 2025). Because the evidence base comprised randomized trials, comparisons started at high certainty and could be downgraded for risk of bias, inconsistency (including heterogeneity), indirectness (including potential intransitivity in networks), imprecision, and publication bias. Publication bias was examined using funnel plots; for network meta-analysis, comparison-adjusted funnel plots were used. For the network meta-analysis, we first graded direct estimates. We then graded indirect estimates by taking the lower rating of the two contributing direct comparisons in the dominant first-order loop and considering transitivity. For each network estimate, we adopted the higher of the direct and indirect ratings and then evaluated incoherence (network inconsistency) and imprecision to determine the final certainty rating.

2.7 Statistical analysis

The main characteristics of the included studies were qualitatively summarized. A frequentist network meta-analysis was conducted in Stata/MP 18 using the network package, whereas pairwise meta-analyses were performed using standard Stata commands. We evaluated clinical similarity and the transitivity assumption; statistical consistency was assessed using node-splitting (local) and the design-by-treatment interaction model (global). Network geometry was displayed as a four-node plot (ESWT, Sham ESWT, CPT, and Combined therapy) for each outcome; node size and edge thickness were proportional to the number of participants per intervention and the number of direct-comparison trials, respectively. Pain outcomes were summarized as standardized mean differences (SMD) with 95% confidence intervals (CIs), and function outcomes as mean differences (MD) with 95% CIs. Effects were coded so that negative SMDs/MDs favored the active intervention (greater improvement). Small-study effects (publication bias) were explored using funnel plots. Treatment ranking was summarized using the surface under the cumulative ranking curve (SUCRA; 0–100%, higher values indicating more effective interventions). Due to the limited number of included studies, sensitivity analyses were performed to assess the robustness of the study findings. All tests were two-sided with P < 0.05 considered statistically significant.

3 Results3.1 Literature screening process and results

The PRISMA flowchart was presented in Figure 1 (Page et al., 2021). The initial electronic search identified 290 potentially relevant publications. After removing 126 duplicate records, a total of 164 records were screened based on reading titles and abstracts, resulting in the exclusion of 149 records. Among the remaining 15 studies eligible for full-text review, 2 were excluded based on inclusion and exclusion criteria (Schneider, 2018; Kong et al., 2023). Additionally, one eligible study was retrieved by manually screening the reference lists. Ultimately, 14 RCTs (n = 738) were included.

Flow diagram of the studies screened and included according to the PRISMA.

3.2 Description of included studies

The meta-analysis included 14 studies (Lee et al., 2014; Han et al., 2015; Moon et al., 2017; Walewicz et al., 2019; Eftekharsadat et al., 2020; Elgendy et al., 2020; Guo et al., 2021; Kızıltaş et al., 2022; Rajfur et al., 2022; Wu et al., 2023; Back et al., 2024; Tan et al., 2024; Fu et al., 2026; Nedelka et al., 2025) with 738 participants (Table 1). A total of 8 countries contributed to the publication of these studies, with China being the most prolific (4 studies, 28.6%). The average age of the participants was 47.5 years. The intervention duration ranged from a single session to 6 weeks, with an intensity of 0.0298-0.35 mJ/mm². Among these interventions, ESWT was included in five studies, Sham ESWT in three studies, CPT in eight studies, and Combined therapy in ten studies. Fourteen studies included pain score as an outcome measure and eight studies reported ODI results (Figure 2A; Supplementary Figures 1, 2). The specific ESWT parameters can be found in Supplementary Materials (Supplementary Table 1).

Author, yearCountryDisease typeInterventionMean (SD) age, yGender M/FDurationFrequencyOutcomemeasuresElgendy et al., 2020EgyptNon-specific chronic low back painCombined therapy
CPT32.73(6.73)
33.26(5.48)5/10
5/106 weekstwice a weekVASHan et al., 2015Koreachronic low back painESWT
CPT49.7(8.3)
46.0(8.9)15
156 weeks.twice a weekVAS; ODIGuo et al., 2021Chinanonspecific low back painCombined therapy34.9 (8.7)25/224 weeks.twice a weekNRSTan et al., 2024Chinapostpartum patients with sacroiliac joint dysfunctionESWT
CPT
Combined therapy30.73(0.72)
30.56(0.73)
30.33(0.74)30
30
302 weekstwice a weekVAS; ODIKızıltaş et al., 2022Türkiyechronic low back painCombined therapy
CPT47.4(14.3)
45.3(12.2)23/13
14/201 weektwice a weekVAS; ODILee et al., 2014KoreaChronic Low Back PainCombined therapy
CPT53.92(10.38)
54.33(13.16)13
156 weekstwice a weekVASRajfur et al., 2022PolandChronic Low Back PainCombined therapy
CPT43.0(13.1)
45.4(14)10/10
10/105 weekstwice a weekVAS; ODIMoon et al., 2017Koreasacroiliac joint painESWT
Sham ESWT54.42(19.05)
59.18(15.30)3/11
1/10single sessionNRVAS; ODIWalewicz et al., 2019PolandChronic Low Back PainCombined therapy
CPT51.1(8.4)
55.8(9.3)6/14
5/156 weekstwice a weekVAS; ODIEftekharsadat et al., 2020IranLow Back PainESWT44.74(9.34)7/205 weeksonce a weekVASBack et al., 2024Brazilchronic non-specific low back painESWT
Convex tips
Sham ESWT56(11.6)
51(12.2)
48(1.5)10/18
6/21
8/19single sessionNRNRSFu et al., 2026Chinapiriformis syndromeCombined therapy60.4(12.7)12/23single sessionNRVAS; ODIWu et al., 2023Chinalow back painESWT
Thermomagnetic51.37(9.39)
52.62(9.91)26/25
31/254 sessionsOnce every 4–5 daysVASNedelka et al., 2025Czech Republiclumbar facet joint painESWT
Sham ESWT44
3933/31
35/295 weeksonce a weekVAS

Summary of studies characteristics.

ESWT indicates extracorporeal shock wave therapy; Sham ESWT, sham extracorporeal shock wave therapy; CPT, Conventional physical therapy; Combined therapy, ESWT with conventional physical therapy; VAS, Visual Analogue Scale; NRS, Numeric Rating Scale; ODI, Oswestry Disability Index.

Network Maps and Interval Plots. Interval plot and forest of pain (A, B), Interval plot and forest of function (C, D). Node size and edge thickness were proportional to the number of participants per intervention and the number of direct-comparison trials, respectively. Red confidence interval lines indicate a statistically significant difference. For GRADE assessment, two dark yellow blocks represent low certainty of evidence, and one dark yellow block represents very low certainty of evidence. ESWT, extracorporeal shock wave therapy; Sham ESWT, sham extracorporeal shock wave therapy; CPT, Conventional physical therapy; Combined therapy, ESWT with conventional physical therapy; SMD, standardized mean difference; MD, mean difference; CI, confidence interval.

3.3 Network effect estimates3.3.1 Pain outcome

In the network meta-analysis for pain (negative SMD indicates greater pain reduction), ESWT reduced pain compared with sham ESWT (SMD −1.22; 95% CI −1.83 to −0.61). In contrast, Combined therapy (SMD −0.48; 95% CI −1.25 to 0.30) and CPT (SMD 0.41; 95% CI −0.32 to 1.13) did not differ significantly from ESWT. Compared with Sham ESWT, CPT showed no significant difference (SMD −0.82; 95% CI −1.77 to 0.14), whereas Combined therapy showed significantly greater pain improvement (SMD −1.70; 95% CI −2.69 to −0.71). Compared with CPT, Combined therapy also produced significantly greater pain improvement (SMD −0.88; 95% CI −1.29 to −0.47) (Figures 2A, C; Supplementary Figure 3).

3.3.2 Functional outcome

For ODI (lower scores indicate better function), the network estimates suggested no significant differences between ESWT and Sham ESWT (MD −4.00; 95% CI −9.70 to 1.70) or CPT (MD −1.69; 95% CI −4.96 to 1.58). CPT did not differ significantly from Sham ESWT (MD −2.31; 95% CI −8.88 to 4.26). In contrast, Combined therapy showed significant advantages over ESWT (MD -3.60; 95% CI -6.70 to -0.51), Sham ESWT (MD -7.60; 95% CI -14.09 to -1.12) and over CPT (MD -5.29; 95% CI -7.53 to -3.06) (Figures 2B, D; Supplementary Figure 4).

3.4 Pairwise meta-analyses

For VAS, ESWT did not differ significantly from CPT (MD −0.53; 95% CI −2.10 to 1.04), but showed a significant improvement compared with Sham ESWT (MD −1.48; 95% CI −2.45 to −0.52). In addition, the Combined therapy was significantly superior to ESWT (MD -0.88; 95% CI -1.40 to -0.35) and CPT (MD −0.90; 95% CI −1.12 to −0.67) (Figure 3).

Pairwise forest plots. VAS, Visual Analogue Scale; ODI, Oswestry Disability Index; ESWT indicates extracorporeal shock wave therapy; Sham ESWT, sham extracorporeal shock wave therapy; CPT, Conventional physical therapy; Combined therapy, ESWT with conventional physical therapy; SMD, standardized mean difference; CI, confidence interval.

For ODI, ESWT was significantly more effective than CPT (MD −4.51; 95% CI −8.80 to −0.22), but it did not show a significant benefit over Sham ESWT (MD −0.49; 95% CI −1.20 to 0.22). In addition, Combined therapy was significantly superior to ESWT (MD −4.47; 95% CI −7.88 to −1.06) and CPT (MD −4.21; 95% CI −6.97 to −1.46) (Figure 3).

3.5 Comparing efficacy of outcomes

The SUCRA plot of pain score showed that Combined therapy was ranked first (96.0%), followed by ESWT (66.3%), CPT (36.1%), and Sham ESWT (1.6%). The SUCRA plot of functional score showed that Combined therapy was ranked first (99.3%), followed by ESWT (59.0%), CPT (30.1%), and Sham ESWT (11.6%) (Figure 4).

Overall SUCRA rankings for efficacy outcomes. The SUCRA plot of pain (A), function (B), and pairs efficacy outcomes of pain and function (C). ESWT indicates extracorporeal shock wave therapy; Sham ESWT, sham extracorporeal shock wave therapy; CPT, Conventional physical therapy; Combined therapy, ESWT with conventional physical therapy.

3.6 Bias and quality of evidence ROB

Of the 14 included studies, 5 were judged to be at low overall risk of bias, 6 raised some concerns, and 3 were judged to be at high risk of bias. At the domain level, 9 studies (64.3%) were rated as low risk for bias arising from the randomization process; 10 (71.4%) for bias due to deviations from intended interventions; 13 (92.9%) for bias due to missing outcome data; 11 (78.6%) for bias in measurement of the outcome; and 5 (35.7%) for bias in selection of the reported result (Figure 5; Supplementary Figures 5, 6).

Study quality assessment by ROB 2.0.

3.7 Certainty of evidence

Certainty of evidence was assessed for 12 comparisons. Eight comparisons were informed by mixed evidence, and four were based on indirect evidence only. Among mixed-evidence comparisons, six were rated as low certainty and two as very low certainty; all indirect-only comparisons were rated as very low certainty.

3.8 Inconsistency test

No inconsistencies were identified in ODI. However, the direct pairwise estimate for ESWT versus CPT differed from the corresponding mixed network estimate, suggesting that the functional results may have been influenced by sparse network geometry and the contribution of indirect evidence rather than by statistical incoherence. No overall inconsistency was detected in pain scores, but local inconsistencies emerged for the Combined therapy and CPT, with the relevant comparisons downgraded in quality assessment. Sensitivity analyses for both direct pairwise and indirect comparisons (network meta-analysis) showed no evidence of heterogeneity. Results were consistent following exclusion of high-risk studies, supporting the robustness of our findings. While visualizations were provided for pairwise comparison analyses based on three or more studies (Supplementary Figure 7). The results from both analyses support the overall reliability of the treatment effects.

4 Discussion

This study conducted a network meta-analysis based on data from 14 randomized controlled trials involving a total of 738 CLBP patients. The analysis compared ESWT, Sham ESWT, CPT, and Combined therapy within a single evidence network. The results consistently showed that the Combined therapy held a significant advantage in both pain relief and functional improvement. ESWT demonstrated superior effectiveness in pain reduction compared to Sham ESWT. Furthermore, combined therapy outperformed CPT and sham ESWT in both pain relief and functional improvement, and was more effective than ESWT alone in enhancing functional outcomes. Notably, although combined therapy ranked highest in the SUCRA analysis, these ranking probabilities should not be interpreted as direct evidence of clinical superiority, because SUCRA reflects relative ranking rather than effect magnitude or clinical importance.

CLBP is thought to involve a multi-level pathological cycle maintained by peripheral tissue injury, spinal-central sensitization, movement control dysfunction, and fear-avoidance behaviors (Ferdinandov, 2024). This pathological complexity suggests that long-term improvement in CLBP requires multi-dimensional interventions rather than mere analgesia or isolated motor function enhancement. Yue et al. (2021) found that ESWT significantly reduced pain at the 1-month follow-up (SMD ≈ −0.81), confirming its robust short-term analgesic efficacy. Existing research indicates that ESWT primarily exerts biological effects through mechanical transduction, including transient pain relief, enhanced local microcirculation, promotion of angiogenesis, activation of nitric oxide and various growth factor pathways, and modulation of inflammatory mediators and pain-related neuropeptides (Simplicio et al., 2020; Huang et al., 2024). Our network comparative analysis showed no significant improvement in patients’ ODI scores, a finding further corroborated by a systematic review by Wu et al (Wu et al., 2023), which noted that ESWT’s effects on disability outcomes are either minimal or statistically insignificant.

CPT is widely recognized for its long-term benefits in improving muscle instability and motor dysfunction (Goubert et al., 2017; Anthierens et al., 2024). Specifically, CPT mitigates central sensitization and activates deep stabilizing muscles (e.g., multifidus, transversus abdominis), thereby promoting motor control retraining, enhancing quality of life, and reducing recurrence risk (Nayel et al., 2025; Tomschi et al., 2025). Nevertheless, Gilanyi et al. (2024) pointed out that CPT is constrained by poor patient adherence—especially in those with severe pain—while Hayden et al. (2021) noted that despite its long-term functional benefits, CPT’s short-term analgesic effect is suboptimal. Notably, pairwise meta-analysis suggested that ESWT was superior to CPT for ODI, whereas the network meta-analysis showed no significant difference. As no inconsistency was detected in the ODI network, this discrepancy may be related to the sparse network structure, the contribution of indirect evidence, and between-study heterogeneity in treatment protocols and follow-up duration. The CPT effect typically manifests at 8–12 weeks, indicating that such discrepancies arise from short-term observation (Hayden et al., 2021). Consistently, O’Keeffe et al. (2020) reported that compared with group-based exercise and educational interventions, CPT reduced disability at 6 and 12 months but failed to alleviate pain—aligning with the earlier observation that CPT’s short-term analgesic effect is suboptimal. It should be noted that there is some heterogeneity among the intervention components of CPT. In addition, cautious interpretation is warranted for ODI findings, particularly in the comparison between ESWT and CPT, given the MCID threshold of 10 points (Ostelo and de Vet, 2005).

Against this backdrop, Jenkins et al. (2025) proposed a complementary therapeutic model for combined treatments: ESWT targets peripheral tissue injury to achieve rapid analgesia, while CPT exerts central regulatory effects to address sensitization and motor dysfunction. Some studies have demonstrated that combining ESWT with a structured exercise or rehabilitation program in patients with CLBP can not only achieve rapid pain relief but also contribute to improvements in functional outcomes. For example, in a randomized controlled trial by Taheri et al., the addition of ESWT to an exercise program and oral medication resulted in significantly greater short-term reductions in pain intensity (VAS) and disability (ODI) compared with sham ESWT combined with the same exercise and medication regimen (Taheri et al., 2021). Furthermore, other research suggests that, relative to conventional exercise or CPT alone, ESWT supplemented with exercise may yield more pronounced improvements in pain and dynamic balance, implying that the short-term biological effects of ESWT may provide a more favorable foundation for subsequent exercise-based rehabilitation (Lee et al., 2014).

This mechanism-driven combination directly addresses the multi-level pathology of CLBP, and accumulating evidence supports its feasibility (Karagiannopoulou e