Osteonecrosis of the femoral head · Review · 2026
How does shockwave compare with other hip-saving treatments for osteonecrosis?
In this review of 29 studies, shockwave ranked well in some comparisons. Hip function scores were about 9 points higher than with placebo or conservative care and about 15 points higher than with core decompression surgery. Treatments did not clearly differ in how often a hip replacement was needed later.
Ranked well in some comparisons- studies (17 randomized trials)
- 29
- hips treated
- 2,177
- hip-saving treatments compared, alone or combined
- 4
Good to know: Mostly early-stage disease; results differed from one outcome to another.
Niu QY, Yang XQ, Liu XW, et al. Frontiers in Cell and Developmental Biology, 2026 · Read the paper ↗
How this study worked
What happened
- Compared with
- Core decompression surgery, cell therapy, hyperbaric oxygen, placebo or conservative care, and combinations
Hip function
Difference in Harris Hip Score (0–100, higher is better)
Show the numbers
| Difference (hip score points) | likely range | |
|---|---|---|
| Shockwave vs placebo or conservative care | +8.83 | 1.42 to 16.24 |
| Shockwave vs core decompression | +14.76 | 5.97 to 23.54 |
| Decompression + shockwave vs decompression alone | +18.61 | 4.5 to 32.72 |
What the trials used
- Interventions and ESWT protocol
- The network compared core decompression (CD), cell-based strategies grouped as BMAC, hyperbaric oxygen (HBO), ESWT, placebo/conservative care, and combinations including CD+ESWT, HBO+CD and ESWT+HBO. The Methods say ESWT energy flux density, shock number, session number and guidance method were summarized in unavailable Supplementary Material 2, and the Discussion says protocols, including focused versus radial technology, varied across studies; the main text gives no device or dose. BMAC pooled biologically heterogeneous products for network connectivity.
Who took part
Adults with osteonecrosis of the hip, mostly at an early or middle stage, in 29 studies (2,177 hips).
Side effects
Side effects were reported more often with shockwave than with placebo or conservative care, and less often than with decompression plus bone-marrow cells. Mild and serious events were counted together.
All the numbers and the full appraisal
The results, in numbers.
Adults with osteonecrosis of the femoral head, predominantly early-to-mid stage (Steinberg I–II, ARCO/Ficat I–III; small proportion stage IV). The review included 29 comparative studies (17 RCTs, 12 retrospective studies), 2,177 hips.
Visual analogue scale (VAS)1 reported outcome
Pain intensity (VAS)
Visual analogue scale (VAS) SMD (95% CI)
ESWT versus core decompression (CD); negative favors lower pain
The network estimate favored ESWT over CD. The article also reported significant effects for HBO and HBO+CD versus placebo and several other active comparisons; all comparison estimates were from the network.
Analysis details and source
SMD; article states lower values indicate greater pain relief. Supplementary league table S1 was unavailable, so only comparisons printed in the Results narrative are extracted.
Harris Hip Score1 reported outcome
Hip function (HHS)
Harris Hip Score MD (95% CI); scale details not stated here
ESWT versus placebo; positive favors higher HHS
ESWT favored HHS versus placebo and CD (MD 14.76, 95% CI 5.97 to 23.54). The HHS network had substantial inconsistency (loop inconsistency factor 16.82, 95% CI 2.26 to 31.37), limiting confidence in effects and rankings.
Analysis details and source
The article also reported CD+ESWT versus CD, MD 18.61 (95% CI 4.50 to 32.72). Values and direction are transcribed as reported; no clinical threshold is asserted.
Odds of structural/radiographic progression as defined variably in included studies1 reported outcome
Imaging progression
Odds of structural/radiographic progression as defined variably in included studies OR (95% CI)
ESWT versus placebo; OR below 1 favors less progression
Logarithmic ratio scale.
ESWT and CD+BMAC had lower odds of imaging progression versus placebo; CD+BMAC also had lower odds than CD. The review notes heterogeneous definitions and limited harmonization.
Analysis details and source
The source states OR 0.15 (95% CI 0.03 to 0.66). Structural outcome definitions included collapse, stage advancement or loss of sphericity.
THA conversion1 reported outcome
Conversion to total hip arthroplasty
THA conversion No comparative estimate reported in narrative
No statistically significant differences in THA conversion were detected versus placebo or between active strategies. This is not evidence of equivalence; follow-up, disease stage and decision-making may affect the outcome.
Analysis details and source
The narrative supplies no OR/CI or absolute event counts; SUCRA rankings are not extracted as effects.
OHS1 reported outcome
Oxford Hip Score (OHS)
OHS Network MD (95% CI)
The narrative reports placebo comparisons for CD, HBO and HBO+CD but does not give an ESWT estimate. Only three studies contributed; a fixed-effect inconsistency model was used.
Analysis details and source
No ESWT effect is inferred from another intervention or from its ranking.
SF-361 reported outcome
Health-related quality of life (SF-36)
SF-36 Network MD (95% CI)
HBO+CD improved SF-36 versus CD alone (MD 4.01, 95% CI 1.63 to 6.38); no ESWT-specific comparison is reported in the narrative. Only three studies contributed and the authors identified disagreement for CD versus HBO.
Analysis details and source
The article notes a sparse fixed-effect network; no ESWT effect inferred.
Composite adverse events1 reported outcome
Treatment-related adverse events
Composite adverse events OR (95% CI)
ESWT versus placebo; OR above 1 indicates higher odds of reported events
Logarithmic ratio scale.
The estimate favored placebo for fewer reported events; ESWT had fewer reported events than CD+BMAC (OR 0.17, 95% CI 0.03 to 0.83). The composite combined events of markedly different severity.
Analysis details and source
Events included surgery complications, donor-site morbidity, post-procedure pain, ESWT local discomfort/petechiae, and HBO barotrauma/oxygen toxicity; event types were not stratified. Source inconsistency: Table 1 lists 10 studies for adverse events, but Table 2 codes adverse events (code 8) for only two studies (Jayankura 2023 and Gangji 2011, both CD+BMAC versus CD) and for no ESWT study; the main text (Tables 1–2; Results 3.4.1.7) does not identify which studies supplied the ESWT adverse-event data, and the supplementary by-study adverse-event table was not available.
WOMAC; pairwise treatment/control synthesis1 reported outcome
WOMAC
WOMAC; pairwise treatment/control synthesis MD (95% CI)
The four-study pooled estimate was MD 12.60 (95% CI 2.57 to 22.63), with I² 99.16% and τ² 102.59; significance was not retained when either of two named studies (Li M et al. or Ma YC et al.) was left out. The text does not identify an ESWT-specific estimate.
Analysis details and source
No direction/scale interpretation is imposed because the source narrative does not state the relevant WOMAC metric direction or ESWT contribution.
Harris Hip Score (HHS)2 reported outcomes
Hip function (HHS), ESWT vs CD
Harris Hip Score (HHS) MD (95% CI)
ESWT versus core decompression (CD); positive favors higher HHS
Source-reported MD ESWT versus core decompression (CD); positive favors higher HHS: 14.76 (95% CI 5.97 to 23.54).
Analysis details and source
Network estimate; the review reports substantial network inconsistency for HHS (loop inconsistency factor 16.82).
Hip function (HHS), CD+ESWT vs CD
Harris Hip Score (HHS) MD (95% CI)
Core decompression plus ESWT versus core decompression alone
Source-reported MD Core decompression plus ESWT versus core decompression alone: 18.61 (95% CI 4.5 to 32.72).
Analysis details and source
Network estimate; the review reports substantial network inconsistency for HHS (loop inconsistency factor 16.82).
Adverse-event odds1 reported outcome
Adverse events, ESWT vs CD+BMAC
Adverse-event odds Odds ratio
ESWT versus core decompression plus bone marrow aspirate concentrate (CD+BMAC); source-reported OR
Logarithmic ratio scale.
Source-reported OR ESWT versus core decompression plus bone marrow aspirate concentrate (CD+BMAC); source-reported OR: 0.17 (95% CI 0.03 to 0.83).
Analysis details and source
Network estimate. Adverse-event severity varied across trials, so do not infer uniform clinical seriousness. Source inconsistency: Table 1 lists 10 studies for adverse events, but Table 2 codes adverse events (code 8) for only two studies (Jayankura 2023 and Gangji 2011, both CD+BMAC versus CD) and for no ESWT study; the main text (Tables 1–2; Results 3.4.1.7) does not identify which studies supplied the ESWT adverse-event data, and the supplementary by-study adverse-event table was not available.
How treatment was delivered
The network compared core decompression (CD), cell-based strategies grouped as BMAC, hyperbaric oxygen (HBO), ESWT, placebo/conservative care, and combinations including CD+ESWT, HBO+CD and ESWT+HBO. The Methods say ESWT energy flux density, shock number, session number and guidance method were summarized in unavailable Supplementary Material 2, and the Discussion says protocols, including focused versus radial technology, varied across studies; the main text gives no device or dose. BMAC pooled biologically heterogeneous products for network connectivity.
The NMA combined adverse events of different severity. Compared with placebo, reported odds were higher for CD (OR 85.29, 95% CI 7.57–961.22), CD+BMAC (159.40, 12.21–2080.96), and ESWT (26.94, 3.59–202.26); ESWT had lower odds than CD+BMAC (0.17, 0.03–0.83). The source cautions that complications, donor-site morbidity, post-procedure pain, local discomfort/petechiae, barotrauma and oxygen toxicity were not stratified by severity. These sparse, very imprecise estimates do not establish overall comparative safety.
The network combined RCTs and retrospective studies; the NMA counted hips, and bilateral-hip clustering handling is not clear in the accessible main text. Stage, lesion features, follow-up, outcome definitions and protocols varied; transitivity is uncertain. HHS showed substantial inconsistency; SF-36 had a node-splitting disagreement. Several endpoints were sparse and estimates imprecise; WOMAC heterogeneity was extreme and leave-one-out sensitivity altered significance. ESWT protocol details and league tables were in unavailable supplements.
Source inconsistency (extractor-noted): the review counts 17 RCTs and 12 retrospective studies, but its own reference titles describe five of the listed RCTs as retrospective studies (Xu 2024, Sun 2026, Wang 2022, Moghamis 2021, Martinot 2020) and three of the listed retrospective studies as randomized trials (Hauzeur 2018, Ma 2014, Sen 2012); the 17/12 split is kept as printed but is uncertain.
Findings checked against the full source by a separate automated reviewer.
Limitations noted by the reviewers
- The network combined RCTs and retrospective studies; the NMA counted hips, and bilateral-hip clustering handling is not clear in the accessible main text. Stage, lesion features, follow-up, outcome definitions and protocols varied; transitivity is uncertain. HHS showed substantial inconsistency; SF-36 had a node-splitting disagreement. Several endpoints were sparse and estimates imprecise; WOMAC heterogeneity was extreme and leave-one-out sensitivity altered significance. ESWT protocol details and league tables were in unavailable supplements.
- Source inconsistency (extractor-noted): the review counts 17 RCTs and 12 retrospective studies, but its own reference titles describe five of the listed RCTs as retrospective studies (Xu 2024, Sun 2026, Wang 2022, Moghamis 2021, Martinot 2020) and three of the listed retrospective studies as randomized trials (Hauzeur 2018, Ma 2014, Sen 2012); the 17/12 split is kept as printed but is uncertain.
Source-linked critical reading
Systematic review with network meta-analysis
Technical source record
Read complete article XML, checked raw Tables 1–2 against parsed convenience extraction, all six figure captions, each outcome subsection and all sensitivity/inconsistency analyses, discussion, limitations, conclusion and end matter. Extracted only narrative quantitative comparisons; unavailable supplement tables were not reconstructed.
Some network comparisons favored ESWT for pain, hip function or imaging progression, but there was no significant difference in conversion to hip replacement. The estimates are uncertain because studies and treatments differed, some analyses were inconsistent or sparse, and the review does not identify a best ESWT protocol.
In adults with osteonecrosis of the femoral head, how do joint-preserving procedures and combinations compare for patient-reported symptoms, function, structural progression, conversion to arthroplasty and adverse events?
Systematic review and network meta-analysis registered as PROSPERO CRD420261379078. It included 29 comparative studies: 17 RCTs and 12 retrospective studies, totaling 2,177 hips. Searches covered four databases through 20 April 2026. The evidence combines randomized and nonrandomized comparisons.
Adults with predominantly early-to-mid-stage disease were eligible; most studies involved Steinberg I–II or ARCO/Ficat I–III, with a small stage-IV proportion. The unit reported is hips, not necessarily unique people; handling of bilateral-hip clustering is not clear in the accessible report.
The network included ESWT alone and combinations such as CD+ESWT and ESWT+HBO. Included ESWT studies varied in energy, shock count, sessions, guidance and focused/radial modality; the energy, shock-count, session and guidance details were assigned to unavailable Supplementary Material 2. The paper grouped biologically distinct cellular products under BMAC for network connectivity.
Comparators included core decompression, cell-based approaches, hyperbaric oxygen, placebo or conservative care, and combinations. Network estimates may be indirect; the main narrative does not provide the full league tables.
Primary outcomes were VAS, HHS and imaging progression; secondary outcomes included THA conversion, OHS, SF-36, WOMAC and adverse events. The authors report that ESWT was favored over CD for VAS (SMD −2.60, 95% CI −4.93 to −0.27), over placebo for HHS (MD 8.83, 1.42 to 16.24), and over placebo for imaging progression (OR 0.15, 0.03 to 0.66). Timepoints and network-specific contribution counts were not stated alongside these narrative estimates. Additional structured network estimates: HHS ESWT versus core decompression MD 14.76 (95% CI 5.97–23.54); core decompression+ESWT versus core decompression MD 18.61 (4.50–32.72); and adverse-event OR ESWT versus CD+BMAC 0.17 (0.03–0.83). HHS network inconsistency and heterogeneous AE severity remain important limitations.
Results differed by outcome. No intervention significantly reduced THA conversion. HHS had substantial loop inconsistency (16.82, 95% CI 2.26 to 31.37), while SF-36 had a node-splitting disagreement. Rankings by SUCRA do not establish clinical superiority, especially in sparse or inconsistent networks.
The adverse-event network reported higher odds versus placebo for ESWT (OR 26.94, 95% CI 3.59 to 202.26) and lower odds for ESWT than CD+BMAC (0.17, 0.03 to 0.83). The pooled endpoint combined events of different types and severity, and the intervals are very wide.
The synthesis mixes RCTs and retrospective cohorts, with variable disease stages, protocols, follow-up and structural definitions. Transitivity is uncertain; some networks were sparse, HHS inconsistent, WOMAC extremely heterogeneous (I² 99.16%), and supplement-level estimates/protocols unavailable. The review reports 2,177 hips, so independent participant counts cannot be assumed. The design labels also conflict with the review's own reference titles: Xu 2024, Sun 2026, Wang 2022, Moghamis 2021 and Martinot 2020 are listed among the RCTs but titled retrospective studies, while Hauzeur 2018, Ma 2014 and Sen 2012 are listed as retrospective but titled randomized trials, so the 17/12 split is uncertain.
The findings are hypothesis-generating for joint-preserving treatment selection in early-to-mid-stage osteonecrosis, not a protocol recommendation. ESWT device, dose, guidance and course cannot be selected from the main text; the evidence does not show prevention of arthroplasty.
Future multicenter trials should stratify by stage and lesion characteristics, describe ESWT device and dose, standardize imaging progression and adverse-event definitions, account for bilateral hips, and follow patients long enough to assess collapse and arthroplasty. Individual-study contributions and direct versus indirect evidence should be transparent.
Read the original article.
Comparative efficacy and safety of core decompression, cell-based therapy, hyperbaric oxygen therapy, extracorporeal shock wave therapy, and combined regimens for osteonecrosis of the femoral head: a network meta-analysis.
Niu QY, Yang XQ, Liu XW, Yang F, He W. · Frontiers in cell and developmental biology · 2026
Open the publication record or available full-text files.
Checked against the full paper · Clinical review: Dr. Marcus Yu Bin Pai, · How we review →