Skip to main content
NCBI home page
Journal of Pain Research logoLink to Journal of Pain Research
. 2026 Sep 26;19:631208. doi: 10.2147/JPR.S631208

Comparative Effects of Acupuncture Combined with Tuina versus Acupuncture Alone for Lumbar Disc Herniation

Chao Sun 1, Hui Li 2, Xiaole Guo 3, Meng Guo 4,✉
PMCID: PMC13625979  PMID: 42820071

Abstract

Background

Whether adding tuina to acupuncture provides clinically important benefit for lumbar disc herniation remains uncertain. This review compared acupuncture combined with tuina with acupuncture alone.

Methods

PubMed, Embase, Web of Science, the Cochrane Library, CNKI, Wanfang, and VIP were searched without language restrictions from the earliest available records to January 31, 2026. Randomized controlled trials comparing acupuncture combined with tuina versus acupuncture alone were included. Outcomes were JOA score, VAS score, ODI, and overall therapeutic response. Random-effects models were used for clinically or statistically heterogeneous outcomes. Risk of bias was assessed using outcome-specific RoB 2, and certainty was assessed using GRADE. The review was retrospectively registered in PROSPERO in March 2026 (CRD420261297721).

Results

Seventeen trials involving 1555 participants were included. The combined intervention was associated with higher JOA scores (SMD 0.74, 95% CI 0.17 to 1.31; I2=94.9%), lower VAS scores (SMD −1.42, 95% CI −1.97 to −0.87; I2=95.1%), lower ODI scores (SMD −2.08, 95% CI −3.13 to −1.03; I2=96.9%), and a higher overall therapeutic response (RR 1.12, 95% CI 1.08 to 1.17; I2=33.0%). Most outcome-specific results raised some concerns or were at high risk of bias. Certainty was very low for JOA, VAS, and ODI and low for overall therapeutic response.

Conclusion

Limited-certainty evidence suggests that adding tuina to acupuncture may provide a possible short-term benefit for pain and function. Substantial heterogeneity, methodological limitations, variable outcome definitions, and possible small-study effects preclude firm conclusions about clinical superiority.

Keywords: lumbar disc herniation, acupuncture, tuina, systematic review, meta-analysis, pain, lumbar function

Background

Lumbar disc herniation is displacement of disc material beyond the normal intervertebral disc space and may produce low back pain, radicular leg pain, sensory symptoms, and functional limitation through mechanical and inflammatory effects on adjacent nerve roots.1–5 Clinical diagnosis integrates symptoms and examination findings and, when indicated, CT or MRI. Disease course, anatomical level and subtype, neurological involvement, and spontaneous resorption can influence prognosis and treatment decisions.4–6

Management includes conservative care, interventional procedures, and surgery. Urgent decompression may be required for cauda equina syndrome or progressive neurological deficit, whereas nonoperative care is commonly used for patients without urgent surgical indications. Available strategies include medication, exercise and rehabilitation, acupuncture, manual therapy, and selected intradiscal procedures; surgical approaches are reserved for appropriate patients.4,6–9

Acupuncture involves needle insertion and stimulation at selected points and has been evaluated for lumbar disc herniation and related pain.10 Tuina is a practitioner-delivered Traditional Chinese manual therapy using structured pressing, kneading, rolling, mobilization, and related techniques. Unlike acupuncture, tuina does not involve needle insertion and primarily applies manual mechanical stimulation to muscles, soft tissues, and joints.11 Comparative reviews have evaluated acupuncture, tuina, traction, Chinese herbal medicine, and combinations of these treatments, but intervention definitions and certainty have varied.12,13

Randomized trials have compared acupuncture combined with tuina against acupuncture alone, but differences in patient characteristics, intervention protocols, outcome instruments, and response definitions make the incremental effect of tuina uncertain. This systematic review therefore evaluated the comparative effects of acupuncture combined with tuina versus acupuncture alone on pain, lumbar function, disability, and overall therapeutic response in patients with lumbar disc herniation.

Methods

Study Design and Reporting Framework

This systematic review and meta-analysis was conducted and reported in accordance with the PRISMA 2020 statement.14 The review was retrospectively registered in PROSPERO (CRD420261297721) in March 2026, after the literature search and review process had commenced. Because registration was retrospective, the possibility that post hoc decisions concerning eligibility criteria, outcome selection, or statistical analysis influenced the review was considered when interpreting the findings.

Search Strategy

PubMed, Embase (Embase.com), Web of Science Core Collection, the Cochrane Library, China National Knowledge Infrastructure, Wanfang Data, and the VIP Chinese Science and Technology Journal Database were searched from the earliest records available in each database, with no lower date limit, to January 31, 2026. No language restrictions were applied. Search concepts covered lumbar disc herniation, acupuncture, tuina, and randomized trials. Reference lists of relevant reviews and eligible reports were screened. Records were exported to EndNote. Complete database-specific strategies, platforms, fields, Boolean operators, dates, language restrictions, and yields are provided in Supplementary Table S1.

Eligibility Criteria

Randomized controlled trials were eligible when they enrolled patients diagnosed with lumbar disc herniation; compared acupuncture combined with tuina against the same acupuncture regimen alone; and reported JOA, VAS, ODI, overall therapeutic response, or another relevant clinical outcome. Diagnosis based on clinical criteria with or without CT or MRI confirmation was accepted as reported by the original trial. No restrictions were imposed on lumbar level, anatomical subtype, symptom duration, acute or chronic status, radiculopathy, baseline severity, or Traditional Chinese medicine syndrome; these variables were extracted to evaluate clinical heterogeneity. Reviews, case reports, animal studies, conference abstracts, duplicate reports, nonrandomized studies, and trials with unbalanced co-interventions were excluded.

Study Selection and Data Extraction

After deduplication, two reviewers independently screened titles and abstracts and then assessed full texts. Disagreements were resolved by discussion or consultation with a third reviewer. Two reviewers independently extracted study identifiers, sample sizes, sex, age, diagnostic and imaging criteria, lumbar level and subtype, radiculopathy, symptom duration and severity, intervention procedures, session frequency and duration, total course, follow-up, outcome definitions, assessment time points, and numerical data. Unreported information was recorded as NR and was not imputed. The extracted study characteristics and intervention details are summarized in Table 1 and Supplementary Table S2.

Table 1.

Characteristics of the Included Randomized Controlled Trials

Study Year EG/CG Sex M/F Age EG Age CG Diagnosis and Clinical Characteristics Experimental Intervention Control Outcomes/Follow-Up
B Rong15 2017 49/49 56/42 49.28 ± 7.84 49.65 ± 8.73 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina, 30 min/session, once daily for 10 days Acupuncture; procedural details NR JOA, VAS, ODI, overall therapeutic response; follow-up NR
ZJ Yang16 2016 80/80 40/40 (as reported; requires source verification) 46.34 ± 4.64 46.25 ± 4.75 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina, 30 min/session, once daily for 10 days Acupuncture; procedural details NR JOA, VAS, overall therapeutic response; follow-up NR
JL Liu17 2023 32/35 35/32 38.89 ± 6.79 38.22 ± 7.55 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina; procedural details NR Acupuncture; procedural details NR VAS, ODI, overall therapeutic response; follow-up NR
XP Zhou18 2010 30/30 39/21 39.58 ± 15.26 37.72 ± 14.65 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina, 20 min/session, 3 times/week for 3 weeks Acupuncture; procedural details NR JOA, VAS; follow-up NR
X Hua19 2018 30/30 29/31 43.60 ± 10.40 40.90 ± 9.70 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina, 10 min/session, every other day, 5 sessions Acupuncture; procedural details NR JOA, VAS; follow-up NR
LL Zhang20 2024 47/47 55/39 42.58 ± 10.14 42.08 ± 9.60 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina; procedural details NR Acupuncture; procedural details NR JOA, VAS, ODI; follow-up NR
WC Zhou21 2023 36/36 42/30 45.20 ± 3.43 43.90 ± 8.43 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture 30 min/session, 5 times/week for 2 weeks; tuina 3 times/week for 2 weeks Acupuncture; procedural details NR JOA, VAS, ODI, overall therapeutic response; follow-up NR
JW Yang22 2025 57/57 70/44 45.64 ± 7.17 46.37 ± 6.24 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture 30 min/session, 5 times/week for 2 weeks; tuina 30 min/session, twice/week for 2 weeks Acupuncture; procedural details NR JOA, overall therapeutic response; follow-up NR
JT Li23 2020 50/50 57/43 37.28 ± 8.29 38.51 ± 8.91 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina, once every 3 days, 10 sessions/course, 2 courses Acupuncture; procedural details NR JOA, VAS, ODI, overall therapeutic response; follow-up NR
Y Chen24 2010 30/30 35/25 34-49 (range) 37-50 (range) LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina, 40 min/session, every other day, 5 sessions/course, 4 courses Acupuncture; procedural details NR JOA; follow-up NR
ZY Xu25 2018 43/43 57/29 56 ± 7 56 ± 6 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina, 30 min/session, once daily for 14 days Acupuncture; procedural details NR JOA, VAS, ODI, overall therapeutic response; follow-up NR
YD Feng26 2025 51/51 34/68 46.47 ± 11.08 45.35 ± 9.97 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina for 3 weeks; session details NR Acupuncture; procedural details NR JOA, VAS, ODI, overall therapeutic response; follow-up NR
Y Xu27 2020 30/30 29/31 42.43 ± 14.42 42.15 ± 10.33 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina, 20 min/session, once daily for 10 days Acupuncture; procedural details NR VAS, overall therapeutic response; follow-up NR
Q He28 2010 90/90 117/63 19-65 (range) 18-63 (range) LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina, once daily, 5 times/week for 2 weeks Acupuncture; procedural details NR VAS; follow-up NR
JQ Liu29 2016 36/36 37/35 37.69 ± 4.86 35.22 ± 5.55 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina; procedural details NR Acupuncture; procedural details NR VAS, overall therapeutic response; follow-up NR
MX Zhao30 2025 40/40 46/34 44.35 ± 4.34 44.22 ± 5.04 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina, 30 min/session, 3 times/week for 8 weeks Acupuncture; procedural details NR VAS; follow-up NR
L Wang31 2018 45/45 64/26 47.97 ± 14.51 48.25 ± 13.19 LDH diagnosis reported; imaging, level, subtype, radiculopathy, duration, severity, and TCM syndrome NR in the available extraction Acupuncture plus tuina, once daily, 10 sessions/course, 2 courses separated by 5 days Acupuncture; procedural details NR JOA, VAS, ODI, overall therapeutic response; follow-up NR
Open in a new tab

Notes: Sex counts refer to the total randomized study population unless otherwise specified. Ages are presented as mean ± SD or range exactly as available. The sex entry for ZJ Yang (2016) does not sum to the randomized sample and should be verified against the original report before submission.

Abbreviations: CG, control group; EG, experimental group; JOA, Japanese Orthopaedic Association; LDH, lumbar disc herniation; NR, not reported; ODI, Oswestry Disability Index; TCM, traditional Chinese medicine; VAS, visual analogue scale.

Outcome Definitions

The principal outcomes were end-of-treatment JOA score, VAS pain score, ODI, and overall therapeutic response. Higher JOA values were harmonized to represent better function; lower VAS and ODI values represented benefit. When different versions, ranges, or reporting formats of a continuous instrument were used, effects were expressed as SMDs. Overall therapeutic response was analyzed as the proportion classified as cured, markedly improved, or improved rather than ineffective, as available in the meta-analysis dataset. Trial-specific reporting of thresholds, category counts, time points, and assessors is summarized in Supplementary Table S3. Because response definitions were insufficiently detailed to establish an objective concordant subgroup, a definition-based exclusion analysis was not performed.

Risk-of-Bias Assessment

Two reviewers independently assessed risk of bias using RoB 232 separately for each result contributing to the meta-analyses of JOA, VAS, ODI, and overall therapeutic response at the end of treatment. The effect of interest was assignment to intervention. Domains were randomization, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. Judgments were low risk, some concerns, or high risk. Participant awareness, assessor blinding, instrument validity, deviations from intervention, missingness, and selective reporting were considered; subjectivity alone was not treated as determinative. Disagreements were resolved by consensus or a third reviewer. Domain-level rationales are provided in Supplementary Tables S4–S7.

Certainty of Evidence

Two reviewers independently assessed certainty for each principal outcome using GRADE.33 Randomized evidence started at high certainty and was considered for downgrading for risk of bias, inconsistency, indirectness, imprecision, and publication bias. The Summary of Findings is presented in Table 2.

Table 2.

GRADE Summary of Findings

Outcome Studies/Participants Relative or Standardized Effect (95% CI) Certainty Reasons for Downgrading
JOA score 12/1096 SMD 0.74 higher (0.17 to 1.31 higher) Very low Downgraded for risk of bias, very serious inconsistency (I2=94.9% and effects in opposing directions), and suspected small-study effects.
VAS score 15/1381 SMD 1.42 lower (1.97 to 0.87 lower) Very low Downgraded for risk of bias, very serious inconsistency (I2=95.1%), and suspected small-study effects.
ODI 8/709 SMD 2.08 lower (3.13 to 1.03 lower) Very low Downgraded for risk of bias, very serious inconsistency (I2=96.9%), and suspected small-study effects; only eight studies informed asymmetry analyses.
Overall therapeutic response 11/1021 RR 1.12 (1.08 to 1.17); approximately 95 more responders per 1000 (64 to 135 more), assuming 795 per 1000 in controls Low Downgraded for risk of bias and suspected publication bias. Definitions were heterogeneous or incompletely reported; the absolute effect is illustrative.
Open in a new tab

Notes: GRADE certainty categories: high, moderate, low, and very low. For continuous outcomes, an SMD does not represent points on the original instrument. The illustrative absolute effect for overall response was calculated by applying the pooled RR to the aggregate control-group event rate used in the meta-analysis.

Statistical Analysis

Meta-analysis was performed in Stata 15.0. Continuous outcomes were summarized using MDs when identical instruments and ranges were used and SMDs otherwise; dichotomous outcomes were summarized using RRs. All estimates are presented with 95% CIs. Statistical heterogeneity was assessed using Cochran’s Q and I2. A random-effects model was used when meaningful clinical or statistical heterogeneity was present; a fixed-effect model was used for overall therapeutic response because heterogeneity was relatively low. Leave-one-out sensitivity analyses were performed for JOA, VAS, and ODI. Funnel plots and Begg and Egger tests were explored when sufficient studies were available; results were interpreted cautiously when fewer than 10 studies contributed or heterogeneity was substantial. Trim-and-fill analyses were treated as exploratory because their assumptions may not hold under marked heterogeneity. Two-sided P<0.05 was considered statistically significant.

Results

Study Selection

The searches identified 7706 records. Before screening, 519 duplicate records, 748 records marked ineligible by automated tools, and 521 records removed for other reasons were excluded, leaving 5918 records for title and abstract screening. Of these, 5878 were excluded. All 40 reports sought for retrieval were obtained and assessed; 23 were excluded because of irrelevant outcomes (n=7), additional unbalanced interventions (n=8), or insufficient quantitative data (n=8). Seventeen randomized controlled trials were included15–31 (Figure 1).

Figure 1.

A flowchart of study selection process showing identification, screening and inclusion steps. The flowchart illustrates the study selection process. It begins with identification, where records are identified from databases totaling 7706, including PubMed, Embase, Cochrane library, Web of science, CNKI, VIP and Wanfang. Records removed before screening include 519 duplicates, 748 marked as ineligible by automation tools and 521 removed for other reasons, leaving 5918 records for screening. During screening, 5878 records are excluded by reading the title and abstract. Reports sought for retrieval total 40, with none not retrieved. Reports assessed for eligibility also total 40, with 23 excluded due to no relevant outcome, combined interventions, or unavailable data. Finally, 17 studies are included in the review, with 17 reports of included studies.

Open in a new tab

PRISMA 2020 flow diagram of study selection.

Study Characteristics

The 17 trials were published from 2010 to 2025 and included 1555 participants: 776 assigned to acupuncture combined with tuina and 779 to acupuncture alone. Sample sizes ranged from 60 to 180. Reporting of diagnostic imaging, anatomical subtype, disease duration, baseline severity, radiculopathy, and follow-up was incomplete. Intervention intensity and duration also varied, and three reports provided no procedural detail beyond acupuncture plus tuina. Characteristics are presented in Table 1 and Supplementary Table S2.

Risk of Bias

Outcome-specific RoB 2 assessments are shown in Figure 2, with rationales in Supplementary Tables S4–S7. For JOA, none of 12 results was low risk, 11 raised some concerns, and one was high risk. For VAS, one of 15 results was low risk, 13 raised some concerns, and one was high risk. For ODI, none of eight results was low risk, seven raised some concerns, and one was high risk. For overall therapeutic response, none of 11 results was low risk, 10 raised some concerns, and one was high risk. Concerns most often involved insufficient reporting of randomization and allocation concealment, limited information about deviations from intended interventions, and absence of blinded outcome assessment. Missing outcome data were not a major concern.

Figure 2.

Four heatmaps of risk-of-bias by study, columns D1 to D5 and Overall, for JOA, VAS, ODI and response. Four heatmaps titled JOA, VAS, ODI and Overall therapeutic response. Each heatmap is a traffic light style grid with rows listing studies by author and year and columns labeled D1, D2, D3, D4, D5 and Overall. A legend below maps Low, Some concerns and High to three distinct fill categories; cells also contain symbols: plus, exclamation mark, or x. Column meaning text shown: D1, randomization process; D2, deviations from intended interventions; D3, missing outcome data; D4, measurement of the outcome; D5, selection of the reported result. Across all four heatmaps, D3 is consistently marked with plus for every listed study. D1, D2 and D4 are predominantly marked with exclamation marks across studies. D5 is mostly plus, with one study showing x in D5 and x in Overall in each heatmap. JOA rows: B Rong 2017; ZJ Yang 2016; XP Zhou 2010; X Hua 2018; LL Zhang 2024; WC Zhou 2023; JW Yang 2025; JT Li 2020; Y Chen 2010; ZY Xu 2018; YD Feng 2025; L Wang 2018. In JOA, L Wang 2018 has x in D5 and x in Overall; all other studies show plus in D5 and exclamation mark in Overall. VAS rows: B Rong 2017; ZJ Yang 2016; JL Liu 2023; XP Zhou 2010; X Hua 2018; LL Zhang 2024; WC Zhou 2023; JT Li 2020; ZY Xu 2018; YD Feng 2025; Y Xu 2020; Q He 2010; JQ Liu 2016; MZ Zhao 2025; L Wang 2018. In VAS, Q He 2010 shows plus in D2, plus in D4, plus in Overall; L Wang 2018 shows x in D5 and x in Overall; remaining studies show exclamation mark in D2 and D4 and exclamation mark in Overall. ODI rows: B Rong 2017; JL Liu 2023; LL Zhang 2024; WC Zhou 2023; JT Li 2020; ZY Xu 2018; YD Feng 2025; L Wang 2018. In ODI, L Wang 2018 shows x in D5 and x in Overall; all other studies show exclamation mark in Overall. Overall therapeutic response rows: B Rong 2017; ZJ Yang 2016; JL Liu 2023; WC Zhou 2023; JW Yang 2025; JT Li 2020; ZY Xu 2018; YD Feng 2025; Y Xu 2020; JQ Liu 2016; L Wang 2018. In Overall therapeutic response, L Wang 2018 shows x in D5 and x in Overall; all other studies show exclamation mark in Overall.

Open in a new tab

Outcome-specific risk-of-bias judgments for JOA, VAS, ODI, and overall therapeutic response. D1, randomization process; D2, deviations from intended interventions; D3, missing outcome data; D4, measurement of the outcome; D5, selection of the reported result.

Effect on JOA Score

Twelve studies reported JOA. The combined intervention was associated with a higher JOA score (SMD 0.74, 95% CI 0.17 to 1.31; P=0.011; I2=94.9%; Figure 3). The magnitude was highly heterogeneous.

Figure 3.

A forest plot of JOA score effects across studies, with mixed directions and an overall positive effect. Forest plot with columns labeled, Study, SMD (95 percent CI) and Weight. The x-axis is labeled, Standardized mean difference, with tick labels at minus 1, 0, 1, 2 and 3. The y-axis lists 12 studies plus an Overall row; the y-axis has no unit. Study effects shown as squares with horizontal confidence intervals and a vertical reference line at 0: B Rong (2017): 2.32 (1.80, 2.84), weight 8.23 percent. ZJ Yang (2016): minus 1.07 (minus 1.41, minus 0.74), weight 8.57 percent. XP Zhou (2010): 0.19 (minus 0.32, 0.69), weight 8.25 percent. X Hua (2018): 0.58 (0.06, 1.10), weight 8.22 percent. LL Zhang (2024): 1.18 (0.74, 1.62), weight 8.38 percent. WC Zhou (2023): 0.80 (0.32, 1.28), weight 8.30 percent. JW Yang (2025): 0.65 (0.27, 1.03), weight 8.49 percent. JT Li (2020): 0.76 (0.36, 1.17), weight 8.44 percent. Y Chen (2010): 2.06 (1.43, 2.69), weight 7.95 percent. ZY Xu (2018): 1.61 (1.12, 2.10), weight 8.28 percent. YD Feng (2025): 0.51 (0.11, 0.90), weight 8.46 percent. L Wang (2018): minus 0.58 (minus 1.00, minus 0.16), weight 8.41 percent. Overall row text reads, Overall (I2 equals 94.9 percent). The pooled effect is 0.74 (0.17, 1.31), weight 100.00 percent, shown by a diamond spanning 0.17 to 1.31 with center at 0.74.

Open in a new tab

Forest plot for JOA score. A random-effects model was used.

Effect on VAS Score

15 studies reported VAS. The combined intervention was associated with a lower VAS score (SMD −1.42, 95% CI −1.97 to −0.87; P<0.001; I2=95.1%; Figure 4). The magnitude was highly heterogeneous.

Figure 4.

A forest plot of VAS score showing mostly negative standardized mean differences and an overall reduction. A forest plot displays studies with columns labeled Study, SMD (95% CI) and Weight. The x-axis shows Standardized Mean Difference (SMD) from -5 to 0. Each study is represented by a square with a horizontal confidence interval line. Study details include: B Rong (2017), SMD -4.03 (-4.73, -3.33), 6.36%; ZJ Yang (2016), -2.61 (-3.03, -2.18), 6.80%; JL Liu (2023), -0.94 (-1.44, -0.43), 6.68%; XP Zhou (2010), -0.52 (-1.04, -0.01), 6.67%; X Hua (2018), -0.69 (-1.21, -0.16), 6.66%; LL Zhang (2024), -1.01 (-1.44, -0.58), 6.79%; WC Zhou (2023), -0.75 (-1.23, -0.27), 6.72%; JT Li (2020), -0.60 (-1.00, -0.20), 6.83%; ZY Xu (2018), -0.95 (-1.39, -0.50), 6.77%; YD Feng (2025), -0.31 (-0.70, 0.08), 6.84%; Y Xu (2020), -0.34 (-0.85, 0.17), 6.68%; Q He (2010), -0.47 (-0.78, -0.16), 6.92%; JQ Liu (2016), -0.86 (-1.34, -0.37), 6.71%; MX Zhao (2025), -3.85 (-4.61, -3.10), 6.25%; L Wang (2018), -3.88 (-4.59, -3.17), 6.33%. Overall, I2 = 95.1%, SMD -1.42 (-1.97, -0.87), weight 100%.

Open in a new tab

Forest plot for VAS score. A random-effects model was used.

Effect on ODI

8 studies reported ODI. The combined intervention was associated with a lower ODI score (SMD −2.08, 95% CI −3.13 to −1.03; P<0.001; I2=96.9%; Figure 5). The magnitude was highly heterogeneous.

Figure 5.

A forest plot of ODI standardized mean difference across studies, with an overall negative effect. Forest plot with columns: Study, SMD (95% CI), Weight. X-axis: Standardized mean difference, ticks at -5 to 0. Vertical line at 0, dashed at -2. Eight studies with markers and confidence intervals: B Rong (2017) SMD -4.72, CI -5.50 to -3.94, weight 12.12%. JL Liu (2023) -0.82, CI -1.32 to -0.32, weight 12.62%. LL Zhang (2024) -1.32, CI -1.77 to -0.87, weight 12.69%. WC Zhou (2023) -0.31, CI -0.77 to 0.16, weight 12.67%. JT Li (2020) -3.65, CI -4.30 to -3.01, weight 12.38%. ZY Xu (2018) -0.98, CI -1.43 to -0.53, weight 12.69%. YD Feng (2025) -0.51, CI -0.91 to -0.12, weight 12.76%. L Wang (2018) -4.61, CI -5.41 to -3.81, weight 12.07%. Overall: I2 = 96.9%, diamond at SMD -2.08, CI -3.13 to -1.03, weight 100%.

Open in a new tab

Forest plot for ODI. A random-effects model was used.

Effect on Overall Therapeutic Response

Eleven studies reported overall therapeutic response. The combined intervention was associated with a higher response rate (RR 1.12, 95% CI 1.08 to 1.17; P<0.001; I2=33.0%, heterogeneity P=0.135; Figure 6). Definitions and thresholds were incompletely reported, limiting interpretability.

Figure 6.

A forest plot of overall therapeutic response showing most studies above the no effect line. Forest plot with columns: Study, RR (95% CI), Weight. X-axis: Risk ratio (0.6, 1.0, 1.7). Solid line at 1.0, dashed near 1.12. Studies as squares with horizontal CI lines. Values: B Rong (2017) RR 1.12, CI 1.00-1.25, 12.91%; ZJ Yang (2016) 1.28, 1.07-1.52, 5.28%; JL Liu (2023) 1.32, 1.03-1.70, 2.59%; WC Zhou (2023) 1.27, 1.01-1.59, 3.20%; JW Yang (2025) 1.12, 1.00-1.26, 12.07%; JT Li (2020) 1.31, 1.08-1.60, 4.10%; ZY Xu (2018) 1.14, 1.00-1.29, 9.77%; YD Feng (2025) 1.04, 0.96-1.13, 26.12%; Y Xu (2020) 1.04, 0.92-1.16, 11.97%; JQ Liu (2016) 1.13, 0.98-1.30, 8.01%; L Wang (2018) 1.31, 1.07-1.61, 3.98%. Overall (I2=33.0%) with diamond at RR 1.12, CI 1.08-1.17, 100% weight.

Open in a new tab

Forest plot for overall therapeutic response. A fixed-effect model was used.

Sensitivity Analyses

Leave-one-out analyses for JOA, VAS, and ODI did not materially alter the direction of the pooled estimates, but substantial heterogeneity persisted (Figures S1–S3). A definition-based sensitivity analysis for overall therapeutic response could not be performed because reporting of response criteria, thresholds, and assessment procedures was insufficient to support a reproducible exclusion rule. The response outcome was therefore interpreted as supportive rather than definitive evidence.

Small-Study Effects and Publication-Bias Analyses

For JOA, Begg’s test was not significant (P=0.055; continuity-corrected P=0.064), whereas Egger’s test suggested asymmetry (P=0.010). Begg’s tests were not significant for VAS (P=0.067), ODI (P=0.083), but Egger’s tests suggested asymmetry for VAS (P=0.005) and ODI (P<0.001). For overall therapeutic response, both Begg’s (P=0.006) and Egger’s (P<0.001) tests suggested asymmetry. These analyses were exploratory: ODI included only eight studies, and marked heterogeneity can produce funnel-plot asymmetry. Trim-and-fill imputed no studies for JOA or ODI, one for VAS, and four for overall response; the direction of the estimates was unchanged, but this did not exclude publication bias (Figures S4–S11).

Certainty of Evidence

Certainty was very low for JOA, VAS, and ODI because of risk of bias, very serious inconsistency, and suspected small-study effects. Certainty was low for overall therapeutic response because of risk of bias and suspected publication bias, with additional interpretive concern arising from incompletely reported response definitions (Table 2).

Discussion

This review found pooled associations favoring acupuncture combined with tuina over acupuncture alone for JOA, VAS, ODI, and overall therapeutic response. These findings suggest a possible short-term incremental benefit from adding tuina, but they do not establish the magnitude, durability, or clinical superiority of the combined intervention. Confidence is limited by very high heterogeneity for the continuous outcomes, predominantly concerning RoB 2 judgments, incomplete clinical reporting, and indications of small-study effects.

The JOA result suggests a possible functional benefit beyond pain relief, while the VAS and ODI results point in the same general direction for pain and disability. Acupuncture and tuina may plausibly address different symptom components, but mechanisms were not tested by the included trials and should not be inferred from these pooled clinical data. The observed effects must also be interpreted in light of variation in needling, manipulation, treatment intensity, practitioner experience, participant severity, anatomical subtype, and assessment timing.

The SMDs permitted synthesis across differing score versions or formats but cannot be interpreted as points on a particular scale. Heterogeneity was 94.9% to 96.9%, effects sometimes pointed in opposing directions, and several study-level estimates were unusually large. Consequently, the pooled SMDs indicate an average direction of association across heterogeneous settings rather than a precise effect applicable to a standardized treatment protocol. They also cannot establish whether changes exceeded a minimally important clinical difference.

Overall therapeutic response had lower statistical heterogeneity, but its apparent consistency should not be overinterpreted. The endpoint combined symptom changes, functional changes, and clinician judgment, while definitions and thresholds were incompletely reported. Knowledge of treatment assignment could also influence classification. It should therefore receive less evidentiary weight than validated continuous outcomes and be regarded as supportive evidence.

Risk-of-bias concerns arose mainly from insufficient reporting of randomization and allocation concealment, limited information about deviations from intended interventions, and lack of blinded outcome assessment. Inability to blind participants or practitioners does not automatically create high risk, but it increases vulnerability when outcomes are participant reported or assessor influenced. Missing data were generally not a major problem. The GRADE assessment nevertheless indicated very low certainty for the continuous outcomes and low certainty for overall response.

Funnel asymmetry and regression tests do not prove publication bias. Their reliability is limited with few studies and can be distorted by heterogeneity. This is especially important for ODI, with eight studies. Trim-and-fill results likewise depend on assumptions that are unlikely to be secure under extreme heterogeneity. These analyses support caution but cannot quantify the extent of unpublished evidence.

Clinically, the combined approach may be considered a possible conservative option for selected patients, but the evidence does not define an optimal protocol or identify who is most likely to benefit. It does not demonstrate superiority over other conservative treatments. Safety also remains uncertain because adverse events and treatment-related withdrawals were not reported consistently.

Limitations include small trials, incomplete methodological reporting, extreme heterogeneity, insufficiently standardized interventions, incomplete reporting of diagnostic and clinical characteristics, heterogeneous or unclear response definitions, limited follow-up, inadequate safety reporting, geographic concentration in China, and possible small-study effects. In addition, registration occurred retrospectively after the review had commenced, so post hoc decisions and selective reporting cannot be completely excluded.

Future trials should be prospectively registered, adequately powered, and preferably multicenter. They should report allocation concealment, prespecified analyses, blinded assessment where feasible, detailed acupuncture and tuina protocols, diagnostic imaging and clinical phenotype, validated patient-centered outcomes, adverse events, and longer follow-up. Consensus on a core outcome set and transparent definitions of any composite response endpoint would improve comparability.

Conclusion

Limited-certainty evidence suggests that adding tuina to acupuncture may provide a possible short-term benefit for pain and lumbar function in patients with lumbar disc herniation. Substantial heterogeneity and methodological limitations preclude firm conclusions about clinical superiority. Better designed and transparently reported randomized trials are required.

Abbreviations

CI, confidence interval; CNKI, China National Knowledge Infrastructure; CT, computed tomography; GRADE, Grading of Recommendations Assessment, Development and Evaluation; I2, I-squared statistic; JOA, Japanese Orthopaedic Association; LDH, lumbar disc herniation; MD, mean difference; MRI, magnetic resonance imaging; NR, not reported; ODI, Oswestry Disability Index; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; PROSPERO, International Prospective Register of Systematic Reviews; RCT, randomized controlled trial; RoB 2, Risk of Bias 2; RR, risk ratio; SMD, standardized mean difference; VAS, visual analogue scale; VIP, VIP Chinese Science and Technology Journal Database.

Data Sharing Statement

All data analyzed in this systematic review are contained in the article and its Supplementary Material.

Ethics Statement

Ethics approval and informed consent were not required because this study analyzed data from previously published reports.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors report no conflicts of interest in this work.

References

  • 1.Gong G, Yan Z, Lai Q, et al. Inflammation preservation strategy: reconciling pain control and disc resorption in lumbar disc herniation. Front Immunol. 2025;16:1653681. doi: 10.3389/fimmu.2025.1653681 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Li S, Chen T, Shen X, Su W, Li X. Gentisic acid ameliorates lumbar disc herniation by regulating M1/M2 polarization via the MAPK14/S100A9/Rac1/2 pathway. Cell Biol Toxicol. 2025;41(1):161. doi: 10.1007/s10565-025-10117-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Wan H, Liu B, Zhang Q, et al. Gamma-delta T-cell infiltration predicts the prognosis of lumbar disc herniation. Immunology. 2026;177(1):199–15. doi: 10.1111/imm.70040 [DOI] [PubMed] [Google Scholar]
  • 4.Lai Q, Gong G, Lu Y, et al. Mechanisms and factors influencing resorption of herniated part of lumbar disc herniation: comprehensive review. J Inflamm Res. 2025;18:8553–8562. doi: 10.2147/JIR.S525233 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Cunha C, Yang H, Buser Z, et al. Lumbar disc herniation resorption: when and how does it occur? Neurospine. 2026;23(1):82–93. doi: 10.14245/ns.2551538.769 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Jin H, Lopez AM, Romero FG, Hoang R, Ramesh A, Bow HC. A systematic review of treatment guidelines for lumbar disc herniation. Neurospine. 2025;22(2):389–402. doi: 10.14245/ns.2550398.199 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Ambrosio L, Vadala G, de Rinaldis E, et al. Discectomy versus sequestrectomy in the treatment of lumbar disc herniation: a systematic review and meta-analysis. Spine J. 2025;25(2):211–226. doi: 10.1016/j.spinee.2024.09.007 [DOI] [PubMed] [Google Scholar]
  • 8.Yong JH, Wang E, Chin BZ, Hey HWD. Full-endoscopic versus microscopic lumbar discectomy for lumbar disc herniation: a systematic review and meta-analysis of 4186 cases. Spine J. 2026;26(5):981–995. doi: 10.1016/j.spinee.2025.12.016 [DOI] [PubMed] [Google Scholar]
  • 9.Takahashi T, Sato H, Takahashi Y, et al. Cost-effectiveness of intradiscal injection with condoliase versus microendoscopic discectomy for lumbar disc herniation: a retrospective multicenter study. Sci Rep. 2025;15(1):43913. doi: 10.1038/s41598-025-27802-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Mo Z, Tang S, Zhang R. Acupuncture for lumbar disc herniation: a systematic review and meta-analysis. Acupunct Med. 2018;36(2):62–70. doi: 10.1136/acupmed-2016-011332 [DOI] [PubMed] [Google Scholar]
  • 11.Liu ZF, Wang HR, Yu TY, Zhang YQ, Jiao Y, Wang XY. Tuina for peripherally induced neuropathic pain: a review of analgesic mechanisms. Front Neurosci. 2022;16:1096734. doi: 10.3389/fnins.2022.1096734 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Mo Z, Li D, Zhang R, Chang M, Yang B, Tang S. Comparisons of the effectiveness and safety of tuina, acupuncture, traction, and Chinese herbs for lumbar disc herniation: a systematic review and network meta-analysis. Evid Based Complement Alternat Med. 2019;2019:6821310. doi: 10.1155/2019/6821310 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Huang F, Xiao Z, Zhan X, et al. Tuina combined with adjuvant therapy for lumbar disc herniation: a network meta-analysis. Complement Ther Clin Pract. 2022;49:101627. doi: 10.1016/j.ctcp.2022.101627 [DOI] [PubMed] [Google Scholar]
  • 14.Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. doi: 10.1136/bmj.n71 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Rong B, Wang DL. Clinical efficacy of acupuncture at Baliao points combined with tuina for L5-S1 lumbar disc herniation. Liaoning J Trad Chin Med. 2017;44(3):587–590. Article in Chinese. [Google Scholar]
  • 16.Yang ZJ, Wang XM. Clinical observation of acupuncture combined with tuina for lumbar disc herniation. J New Chin Med. 2016;48(11):93–94. Article in Chinese. [Google Scholar]
  • 17.Liu JL, Wang M, Jing FQ, Zhao P, Zhou Y. Effects of acupuncture combined with manual tuina on plasma beta-endorphin levels and lumbar function in patients with lumbar disc herniation. Prog Mod Biomed. 2023;23(20):3869–3873. Article in Chinese. [Google Scholar]
  • 18.Zhou XP, Huang CJ, Huang GQ. Influence of acupuncture plus Long’s manual manipulations on functional improvement in lumbar intervertebral disc protrusion. J Acupunct Tuina Sci. 2010;8(6):375–379. doi: 10.1007/s11726-010-0454-4 [DOI] [Google Scholar]
  • 19.Hua X, Wang JW. Clinical observation of acupoint-pressing tuina combined with acupuncture for lumbar disc herniation. Shanghai J Trad Chin Med. 2018;52(9):51–53. Article in Chinese. [Google Scholar]
  • 20.Zhang LL, Yang TT, Wu YP, Wang X. Clinical effects of staged acupuncture combined with three-dimensional balanced spinal manipulation for lumbar disc herniation. J Cervicodynia Lumbodynia. 2024;45(5):976–980. Article in Chinese. [Google Scholar]
  • 21.Zhou WC, Wang HN. Clinical study of tendon-regulating tuina combined with conventional acupuncture for degenerative lumbar disc herniation. J New Chin Med. 2023;55(5):186–190. Article in Chinese. [Google Scholar]
  • 22.Yang JW, Liu WJ, Zheng W, Li LL, Yang R, Yan XQ. Effects of Tongdu Fuyang tuina combined with acupuncture on lumbar pain and function in patients with lumbar disc herniation. J Liaoning Univ Trad Chin Med. 2025;27(6):161–165. Article in Chinese. [Google Scholar]
  • 23.Li JT, Qi Y. Clinical efficacy of tuina combined with acupuncture for blood-stasis-type lumbar disc herniation. World Chin Med. 2020;16(6):920–924. Article in Chinese. [Google Scholar]
  • 24.Chen Y, Ran QF. Clinical observation of tuina combined with filiform-needle acupuncture for lumbar disc herniation. Beijing J Tradit Chin Med. 2010;29(7):543–544. Article in Chinese. [Google Scholar]
  • 25.Xu ZY. Clinical observation of acupuncture at Baliao points combined with tuina for L5-S1 lumbar disc herniation. Shanghai J Acupuncture Moxibustion. 2018;37(8):941–945. Article in Chinese. [Google Scholar]
  • 26.Feng YD, Zhang SM, Shi FD, et al. Clinical study of acupuncture combined with spinal tuina for lumbar disc herniation. Chin J Trad Med Traumatol Orthopedics. 2025;33(10):45–49. Article in Chinese. [Google Scholar]
  • 27.Xu Y, Kou JW, Zhao YJ, et al. Clinical study of acupuncture combined with acupoint tuina for lumbar disc herniation. Hebei J Trad Chin Med. 2020;42(7):1080–1083. Article in Chinese. [Google Scholar]
  • 28.He Q, Xu LH, Wu WF, et al. Clinical study of acupuncture combined with tuina for lumbar disc herniation. J Guangzhou Univers Trad Chin Med. 2010;27(3):242–245. Article in Chinese. [Google Scholar]
  • 29.Liu JQ. Clinical study of comprehensive traditional Chinese medicine treatment for lumbar disc herniation. Henan Trad Chin Med. 2016;36(2):297–299. Article in Chinese. [Google Scholar]
  • 30.Zhao MX. Effect of acupuncture combined with acupoint tuina in the treatment of sciatica caused by lumbar disc herniation. Med Front. 2025;(1). Article in Chinese. [Google Scholar]
  • 31.Wang L, Song YX. Clinical observation of acupuncture combined with tuina for lumbar disc herniation in middle-aged and older patients. Sichuan J Trad Chin Med. 2018;36(2):177–180. Article in Chinese. [Google Scholar]
  • 32.Sterne JAC, Savovic J, Page MJ, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ. 2019;366:l4898. doi: 10.1136/bmj.l4898 [DOI] [PubMed] [Google Scholar]
  • 33.Balshem H, Helfand M, Schunemann HJ, et al. GRADE guidelines: 3. Rating the quality of evidence. J Clin Epidemiol. 2011;64(4):401–406. doi: 10.1016/j.jclinepi.2010.07.015 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

All data analyzed in this systematic review are contained in the article and its Supplementary Material.

Articles from Journal of Pain Research are provided here courtesy of Dove Press

RESOURCES