Ultrasound-Guided Hydrodissection

What Works

Cinematic anatomical rendering of a peripheral nerve within its surrounding tissue plane

Tier 1 — Strong evidence

None. No use of this technique reaches strong evidence. There is no clinical practice guideline, no Cochrane review, and one placebo-controlled trial in the entire field.

Tier 2 — Moderate evidence

Choose 5% dextrose over a corticosteroid MODERATE

Where a fluid is being chosen, dextrose is the better-supported option, and the safety difference is the clearest part.

10 mL dextrose · single session · meralgia paresthetica

Shi 2024 (PMID 39621981), randomized double-blind, N=56: better on pain and quality of life at 4 and 6 months, better clinical response at 6 months, and zero adverse effects against six in the steroid arm. Supported by the Lee 2025 network meta-analysis ranking dextrose first for function at 4, 12 and 24 weeks.

Insist on ultrasound guidance and a scan beforehand MODERATE

The needle is aimed just past the point where the nerve is most swollen, which has to be found on a scan first. Treat this as a safety and targeting standard.

Pre-procedure diagnostic scan · target just distal to maximum nerve swelling

Sveva 2024 (PMID 38392587): 11 side effects among 923 patients, about 1.2%. Neo 2022 (PMID 34261895): no safety-related adverse outcomes across 6 trials and 356 wrists. Stated honestly: guidance has not been shown to beat an unguided injection on outcomes.

Use it after conservative care fails, before surgery MODERATE

This includes scar tissue causing symptoms after a previous operation, such as a carpal tunnel release that did not hold.

Colorado 2025 (PMID 40766979) states this positioning explicitly, and it matches the patients enrolled in every trial retrieved.

Tier 3 — Emerging and contested (3 items)

Volume for carpal tunnel: ask for 10 ml, not 5 ml CONTESTED

10 ml · single session

Both direct randomized volume comparisons favour 10 ml: Huang 2024 (PMID 37873682, N=24) on both symptom and function scores (P = .024 and P = .011), and Eyvaz 2025 (PMID 39642354, N=80) where 10 ml dextrose was the only arm beating saline on nerve size at 12 weeks. Against this, the Lee 2025 network meta-analysis ranks 5 cc first. Both direct trials are small. Follow the direct comparisons and know the conflict exists.

Platelet-rich plasma for longer-term symptom relief EMERGING

Ranked first for symptoms at 12 and 24 weeks by Lee 2025 (PMID 39894044), behind dextrose on function. Rests on a small number of trials inside a nine-study network, needs a blood draw, and costs substantially more.

Any nerve other than the median nerve at the wrist or the lateral femoral cutaneous nerve NO DATA

Colorado 2025 names the ulnar nerve at the elbow, radial tunnel, saphenous, sciatic and fibular nerves as possibly amenable "in select cases". Individual case reports exist. No randomized data for any of them. Uncontrolled case series in conditions that often settle on their own do not establish that a treatment works.

Exercise Prescription

There is no home exercise for this procedure, and inventing one would misrepresent the evidence. What the research does support is that the conservative care people are told to do beforehand is not a waiting-room formality.

Night splint, before and alongside MODERATE

Worn overnight · every night · comfortable, never tight enough to wake you

In the one trial with a conservative comparator, Turkyolu 2026 (PMID 41642953, N=44), the splint alone produced significant improvement in pain and symptoms, and adding the injection failed to beat it on four of five measures.

Change the aggravating position, do not stop using the limb MODERATE

Ongoing · modify the position that reproduces symptoms

Every trial population reached the injection after activity modification. No trial in this literature prescribed general rest, and none measured training load.

What Doesn't Work

Hydrodissection as a way to beat a simpler injection. Ghorbanpour 2025 randomized 66 wrists across guided hydrodissection with steroid, guided hydrodissection without steroid, and a blind unguided steroid injection, and found no significant difference on any measure.
Hydrodissection for weakness or muscle wasting. No trial enrolled these patients, and no arm of any trial improved muscle strength.
Courses of repeat injections. Every randomized trial used a single injection. Multi-session schedules come from case reports only, and a failed injection is a reason for a surgical opinion rather than a second injection.
Using nerve size on ultrasound to tell treatments apart. It shrank in all four arms of Eyvaz 2025, including both plain saline arms.
Quoting "safe and effective" as a summary of this literature. Safe is well supported. Effective, meaning better than not doing it, rests on one trial of 34 people.

Red Flags

These mean this procedure is the wrong pathway, not that it needs adjusting. If any apply to you, get assessed before booking an injection.

Cinematic anatomical rendering of a compressed peripheral nerve

Weakness that is getting worse, or the muscle pad at the base of your thumb looking flatter than the other side. Refer to hand or orthopaedic surgery. No arm of any trial improved muscle strength.
Numbness getting rapidly worse, or numbness that has turned constant rather than coming and going. Refer to neurology or surgery.
Symptoms that are not confined to one nerve's territory, or that are the same on both sides. Refer to GP for blood tests, or to neurology. This points away from a single trapped nerve.
Constant pain that never lets up, night pain unrelated to position, fever, or unexplained weight loss. Refer to GP urgently.
A new numb patch or new weakness that appeared after a previous injection at the site. Refer back to the proceduralist and to neurology.
Redness, heat or suspected infection where the injection would go. Procedure not safe. Refer to GP or A&E.

Return to Training

For lifters and active people working back to full loading after symptoms settle.

No pins and needles at rest for two consecutive weeks
Grip strength within 10% of the other side, or of your own baseline if both sides are affected
Can hold the position that used to bring symptoms on, under load, for a full working set
No flare-up in the 24 hours after a full training session
Symptom scores improved and stable across two consecutive check-ins
No new or worsening weakness at reassessment

Stated plainly: no trial in this literature prescribed, measured or modified training load. These criteria are standard nerve-irritability load management applied to this population, not a finding from the hydrodissection evidence.

Hold both hands up in a mirror and compare the muscle pads at the base of each thumb.

If one looks flatter or more hollow than the other, book a doctor's appointment this week instead of an injection. Losing that muscle changes the plan entirely, and no study has shown this procedure brings it back.

Takes less than 2 minutes. No equipment needed.

The Verdict

Eleven trials asked which fluid to inject. One asked whether injecting helps.

A nerve is meant to slide through its tunnel the way a bike brake cable slides inside its housing. When the surrounding tissue grips the cable, every bend tugs on the nerve instead of letting it glide, and that tug is the burning and the pins and needles. Hydrodissection squirts fluid into the housing to float the cable free again. The honest catch is that nobody has shown the floating is what helps, rather than simply putting a needle in the right place.

Here's what's really happening: a doctor uses ultrasound to inject fluid around a trapped nerve to free it from the tissue gripping it, and in the biggest real-world review about three in ten people got a clearly meaningful improvement.
What most people get wrong: a lot of studies does not mean the question got answered, because nearly every trial compared one fluid against another fluid rather than against not injecting at all.
The one change that matters: if you are having it done, ask for 5% dextrose rather than a steroid, because in the one head-to-head comparison it worked better and caused zero side effects against six.

Best for

Mild to moderate carpal tunnel syndrome that has not settled with a night splint, or meralgia paresthetica (the numb patch on the outer thigh), when you want to try something before surgery.

Skip if

You are losing grip strength or thumb muscle bulk, or you have not yet given a night splint and activity changes a fair trial. Both point somewhere else first.

Want the full evidence? Keep scrolling

Conviction

LOW overall, and a single grade would misrepresent this. Scored per claim, because "no data" and "not supported" are different statements from "low".

Procedural safetyMODERATE-HIGH

Dextrose in preference to corticosteroidMODERATE

Positioning after failed conservative careMODERATE

Improves carpal tunnel versus no injectionLOW

Better than a simpler or unguided injectionNOT SUPPORTED

Best fluid volumeCONTESTED

Any nerve beyond median wrist / lateral femoral cutaneousNO DATA

How it actually works in humansNO DATA

Repeat or multi-session coursesNO DATA

Severe disease or established nerve damageNO DATA

What would change my mind: "it improves carpal tunnel versus no injection"

A three-arm double-blind trial in mild to moderate carpal tunnel syndrome, at least 150 people, comparing hydrodissection with 10 ml dextrose against a sham procedure (a needle and ultrasound contact but no injection around the nerve) and against splinting alone, with symptom and function scores at 12 and 24 weeks.

A difference of 0.5 or more on the symptom scale favouring hydrodissection over sham would move this from LOW to MODERATE. A null result against sham would move it to NOT SUPPORTED and retire the rationale for the technique rather than merely downgrade it.

What would change my mind: "it works in nerves other than the median nerve"

A randomized trial of dextrose hydrodissection in any single non-carpal-tunnel nerve, at least 60 people, with 6-month follow-up.

That would be the first evidence the carpal tunnel result generalizes at all. Until one exists, every recommendation for another nerve is an analogy borrowed from the wrist, and should say so.

Go Deeper

Been offered an injection and not sure whether the evidence behind it is as strong as it sounds? The Verdict breaks down one treatment a week, free, with the numbers in plain English.

Join The Verdict — free

The Full Picture — Anatomy, Diagnosis & Evidence

What's Actually Going On

Cinematic cross-section of a peripheral nerve separated from surrounding connective tissue

Fluid is injected into the plane between the nerve and the tissue around it. The stated intent is mechanical: release the adhesions holding the nerve down, decompress it, reduce local swelling and let it glide normally again.

There is a competing explanation. Colorado 2025 suggests the benefit may come from restoring the nerve's own tiny blood and nerve supply rather than from physically separating tissue planes.

Both are hypotheses, and neither has been tested in a human study. That matters more than usual here, because the technique is named after the mechanism. Neo 2022 says directly that it is unclear whether the hydrodissection mechanism causes the improvement, and Shi 2024 lists the unclear mechanism among its own limitations.

The gap has a practical consequence. When Ghorbanpour 2025 compared guided hydrodissection against a blind steroid injection with no ultrasound at all, the two were indistinguishable at 3 months. If separating tissue planes were doing the work, the guided arm should have pulled ahead.

How to Identify It

Cinematic rendering of an ultrasound probe over a limb with the nerve visible beneath

There is no bedside test that predicts who responds to this procedure. No accuracy figures for selecting candidates appear anywhere in the retrieved literature, and rather than fill them in from general clinical knowledge, they are left empty here.

Diagnostic ultrasound of the nerve, to find focal swelling and its exact level Sn: no data | Sp: no data
Nerve conduction studies, to confirm and grade the problem and rule out a lesion higher up Sn: no data | Sp: no data
Diagnostic block at the suspected site, to check that site is actually the pain generator Sn: no data | Sp: no data

What these establish is a target, not a prediction. The procedure presupposes a correctly localized entrapment, so if nobody has imaged the nerve and found where it is trapped, there is no reason to prefer this over a simpler injection.

The Debate

There is no guideline row here, and that absence is the finding. A search for hydrodissection under guideline publication types returns zero records, while the same search mechanism run against carpal tunnel syndrome returns eighteen. A Cochrane Database search returns zero, against twenty-five Cochrane reviews with "carpal" in the title. The top two tiers of the evidence hierarchy are empty for this technique.

Do the reviews support it?

Buntragulpoontawee 2020, Sveva 2024, Lee 2025

Hydrodissection is "safe and effective" for peripheral nerve entrapment.

vs

Neo 2022, 6 RCTs, 356 wrists

"It is unclear whether the hydrodissection mechanism truly causes improvements in clinical outcomes."

All four reviews read substantially the same trials. The difference was not the data. Neo counted the placebo-controlled trials, found one, and reported the gap. Buntragulpoontawee even states the problem in its own results, that all studies compared different interventions with different comparisons, and concludes effective anyway.

How much fluid?

Lee 2025, network meta-analysis, 9 studies, N=458

5 cc ranks first for both symptoms and function.

vs

Huang 2024 (N=24) and Eyvaz 2025 (N=80)

10 ml beat 5 ml directly, on both symptom and function scores.

A ranking assembled across trials with different fluids and different follow-up windows contradicts the two studies that randomized volume as the only variable. Follow the direct comparisons: 10 ml. A ranking is not an effect size.

Is it already standard practice?

Wu 2021 narrative review

Dextrose perineural injection "has become the mainstream method for treating carpal tunnel syndrome", citing a textbook.

vs

PubMed, verified 2026-08-29

Zero guidelines and zero Cochrane reviews address the technique anywhere.

"Mainstream" is a claim no guideline body has made. Worth noting the same small group of authors recurs across the enthusiastic reviews, while the underlying trials come from Taiwanese, Turkish, Iranian, Chinese, Singaporean and Korean centres. The advocacy is concentrated. The evidence is not.

Honest Limitations

1. The trials rank fluids. The clinic asks whether to inject at all.

What the research shows: eleven-plus randomized trials in carpal tunnel syndrome, three systematic reviews and a network meta-analysis, most reporting positive results.

The real-world gap: almost every one of those trials randomized between two active fluids or two volumes, so both groups received an injection. Exactly one trial controlled for where the injection went, and exactly one controlled for the technique itself, and the second was null.

What to do with that: when someone asks "does this work", answer from the two trials that asked, not the eleven that did not.

2. The evidence is one nerve wearing the name of all of them.

What the research shows: reviews titled for "peripheral nerve entrapment syndromes" report consistent benefit.

The real-world gap: Buntragulpoontawee 2020 found 9 of its 10 studies were carpal tunnel and the tenth was the ulnar nerve at the elbow. The only randomized trial outside carpal tunnel in the entire field is Shi 2024 in meralgia paresthetica.

What to do with that: a protocol quoted for a nerve in the foot, shin or knee is a carpal tunnel protocol with the nerve name swapped. Say so, rather than borrowing the confidence.

3. The unselected caseload does worse than the trial populations.

What the research shows: trial groups improve significantly on symptom and function scores at 12 and 24 weeks.

The real-world gap: Sabbineni 2025 reviewed 204 patients across a five-year clinical service and several nerves. At 6 months: 56% minimal improvement, 14% moderate, 30% significant. It is retrospective and uncontrolled, so natural recovery is not separated out.

What to do with that: set expectations from the 204-patient service data, not from the trial rankings.

The Nuance

Cinematic anatomical study of a nerve pathway through surrounding soft tissue

No trial has ever compared this procedure against the surgery it claims to defer. That is the most clinically important missing comparison in the field, and it is missing precisely where the technique makes its strongest claim.

The one trial with a conservative comparator points the other way. In Turkyolu 2026, a wrist splint alone produced significant improvement in pain, symptoms and function, and adding the injection did not clearly beat it on four of five measures.

None of that makes the procedure a bad choice. It is genuinely low risk, roughly 1.2% side effects across 923 patients, with no serious adverse events reported in any review. That safety record is real, and it is the strongest reason to consider it.

What has not been shown is that it does more than a cheaper, simpler injection, and it has never been tested against surgery. Someone choosing it is choosing a low-risk option with a real chance of meaningful benefit and a genuinely uncertain mechanism. That is a defensible choice. It should just be made with those words rather than with the word "effective".

Sources

Lee K et al. (2025), Yonsei Med J, PMID 39894044 — systematic review and network meta-analysis, 9 studies, N=458. Ranked dextrose first for function; 5 cc first for volume.
Neo EJR et al. (2022), Am J Phys Med Rehabil, PMID 34261895 — systematic review, 6 RCTs, 356 wrists. Only one included trial placebo-controlled; mechanism attribution unclear.
Ghorbanpour S et al. (2025), Muscle Nerve, PMID 40007483 — double-blind RCT, 66 wrists, 3 arms including blind unguided steroid. No significant difference between groups on any measure.
Wu YT et al. (2019), Muscle Nerve, PMID 30339737 — randomized double-blind controlled trial, N=34. The only trial controlling for injection site.
Shi X et al. (2024), Pain Physician, PMID 39621981 — randomized double-blind trial, N=56, meralgia paresthetica. The only randomized trial outside carpal tunnel syndrome.
Sveva V et al. (2024), J Pers Med, PMID 38392587 — systematic review. 11 side effects among 923 patients.
Buntragulpoontawee M et al. (2020), Front Pharmacol, PMID 33746745 — systematic review of injectates, 10 studies, 9 of them carpal tunnel.
Eyvaz N et al. (2025), Am J Phys Med Rehabil, PMID 39642354 — randomized double-blind trial, N=80, four arms comparing volume and fluid.
Huang CY et al. (2024), J Ultrasound Med, PMID 37873682 — randomized single-blind trial, N=24. 10 ml beat 5 ml on both subscales.
Turkyolu E et al. (2026), J Back Musculoskelet Rehabil, PMID 41642953 — randomized controlled trial, N=44. Null on four of five measures.
Sabbineni M et al. (2025), Clin Radiol, PMID 40203607 — retrospective review, N=204 across multiple nerves, 6-month follow-up.
Colorado B et al. (2025), Muscle Nerve, PMID 40766979 — expert clinical perspective on technique, target site and injectate choice.