Stellate Ganglion Block: What the Evidence Actually Supports

The stellate ganglion block interrupts sympathetic outflow to the head, neck and upper limb, and it has accumulated a longer list of proposed indications than almost any other block in pain practice — complex regional pain syndrome, post-herpetic neuralgia, post-traumatic stress disorder, hot flushes, long COVID, olfactory loss, perioperative arrhythmia and postoperative pain among them.

That breadth is the problem. A technique offered for a dozen conditions is usually one where the evidence is thin in most of them, and that is the case here. This reference separates what the controlled evidence supports from what rests on uncontrolled series, and sets out the sign that tells you the block worked. It sits under our clinical reference to nerve blocks in outpatient pain practice.

What is being blocked, and the sign that it worked

The target is the cervicothoracic sympathetic chain. In most people the inferior cervical ganglion is fused with the first thoracic ganglion to form the stellate ganglion, and the block is placed at the C6 or C7 level, where the needle is kept away from the vertebral artery and the pleura by anatomy that ultrasound now makes visible.

Horner syndrome is the expected consequence of a successful block, not a complication: ptosis, miosis and anhidrosis on the blocked side, usually with conjunctival injection and nasal congestion. A block that produces no Horner sign has not reliably interrupted the sympathetic supply, and interpreting the patient’s pain response after such a block is unsafe reasoning.

This matters when reading the safety literature. A 2026 meta-analysis of 33 randomised trials in 2,231 surgical patients reported upper eyelid ptosis at a relative risk of 31.67 (95% CI 6.43 to 155.90) against control. Read as an adverse-event signal that number looks alarming; read correctly it is largely the block working as designed. The distinction is worth making explicitly to patients during consent, because an unexpected droopy eyelid is frightening and a predicted one is not.

Complex regional pain syndrome

CRPS is the historic indication and still the one most often cited, so the size of the effect deserves to be stated precisely.

A 2024 meta-analysis pooled twelve randomised controlled trials of sympathetic ganglion block for CRPS. Methodological quality was good, with a mean PEDro score of 7.0. The pooled result was a reduction in visual analogue pain score of 6.24 mm on a 100 mm scale (95% CI −11.45 to −1.03), and a numerical rating scale reduction of 1.17 points that did not reach statistical significance (95% CI −2.42 to 0.08).

Six millimetres on a hundred-millimetre scale is statistically detectable and clinically negligible; it sits well below any accepted minimal clinically important difference. The honest summary is that the best pooled randomised evidence for the block’s flagship indication shows a real but very small average effect, with the second pain measure not significant at all.

That does not make the block useless in CRPS. Averages conceal responders, and a diagnostic sympathetic block retains a role in establishing whether a patient’s pain has a sympathetically maintained component. It does mean that promising a patient with CRPS substantial relief is not supportable from the pooled data.

Post-traumatic stress disorder

This is the indication driving most current interest, and the evidence base is smaller than the attention suggests.

A 2025 systematic review and meta-analysis screened 394 records and found two randomised controlled trials and one case-control study. The pooled reduction on the Clinician-Administered PTSD Scale was 6.24 points (95% CI −10.71 to −1.78, P = 0.006), favouring the block.

The more useful context comes from the design paper for a multisite trial now under way: of the two published randomised trials, one was negative and the other had limitations. That trial is triple-blind and placebo-controlled with a described sham and a standardised anaesthetic dose, and it exists precisely because the current evidence cannot settle the question.

A pain practice offering this should say that plainly. It is a reasonable thing to study and an unreasonable thing to market.

Postoperative pain and perioperative effects

The perioperative literature is the largest and the most internally consistent, and it is also the least relevant to a chronic pain clinic.

The 2026 meta-analysis of 33 randomised trials found reduced pain scores at 6, 12 and 24 hours after surgery, no significant difference at 48 hours, and no significant reduction in total opioid consumption. It found a shorter hospital stay by roughly one day and a reduced incidence of postoperative nausea and vomiting (RR 0.55, 95% CI 0.40 to 0.77).

A second 2026 meta-analysis of eight trials in 562 patients graded the certainty of this evidence as very low, while making a methodologically interesting observation: separating trials by whether the surgical site lay in a directly or indirectly innervated zone resolved the heterogeneity completely, to I² of zero in both subgroups. That is a reminder that anatomy predicts response, and that pooled estimates across unrelated surgical sites are not telling you much.

The longer indication list

Several other uses appear in the literature at a lower evidential level, and should be described that way.

• Long COVID — a 2026 systematic review of seven studies found symptomatic improvement in every one, with response rates from 55.8% to 100%, and no control group in any of them. Uncontrolled improvement in a fluctuating condition is not evidence of effect.
• Olfactory dysfunction — a 2026 meta-analysis of nine studies in 441 participants found subjective and objective improvement rates of 68.9% and 63.4%, but the between-group differences against control were not statistically significant. No major adverse events were reported.
• Post-herpetic neuralgia — a widely used indication where, as of a 2026 protocol registered to address exactly this, the evidence base remains insufficient and existing results inconsistent. The systematic review has not yet reported.
• Perioperative arrhythmia — an active area with a 2026 systematic review in the anaesthesia literature, outside the scope of an outpatient pain practice.

Safety

The stellate ganglion sits in company that makes technique non-negotiable: the vertebral artery, the pleural dome, the recurrent laryngeal nerve and the brachial plexus are all within reach of a misplaced needle. Intravascular injection into the vertebral artery can produce seizure after a very small volume, which is why aspiration and incremental injection are not optional and why the block is performed where resuscitation equipment is present.

Expected effects that patients should be warned about in advance include Horner syndrome, hoarseness, a sensation of a lump in the throat, and nasal congestion on the blocked side. Bilateral blocks at the same sitting are avoided because of the airway risk if both recurrent laryngeal nerves are affected.

The general safety architecture for any block — local anaesthetic systemic toxicity, asepsis, and the anticoagulation guidance that is routinely confused between societies — is set out in the nerve blocks reference.

For clinicians

This is a block where ultrasound changed the risk profile, because the structures that must be avoided can be seen rather than inferred from surface landmarks. What guidance does and does not improve is covered in musculoskeletal ultrasound for pain practice: accuracy and safety, not pain scores.

This reference carries no needle depths or trajectories. Empire teaches the stellate ganglion block under ultrasound guidance within Advanced Musculoskeletal Ultrasound Guided Injections, alongside the other cervical and sympathetic blocks in the pain management academy.