Skip to content

The Evidence

How does acupuncture compare to alternatives?

TL;DR — What the Evidence Shows

Few studies directly compare acupuncture to physical therapy, massage, or chiropractic. When they do, results tend to be similar. No single treatment clearly wins for chronic pain.

Common treatments like NSAIDs (pain-relief drugs like ibuprofen) work in the short term but carry risks to the stomach and heart. Opioids carry risks of addiction. Surgery results vary widely.

We apply the same standard of scrutiny to every treatment on this page. The evidence for many common treatments is also limited.

Treatment comparison: efficacy and risk

Treatment Best evidence for Certainty Key risks
Acupuncture Chronic LBP, neck pain (vs. no treatment) Moderate Certainty 5–8 serious AEs per million treatments (underreporting caveat applies — see Safety page)
Physical therapy / exercise Chronic LBP, neck pain Moderate Certainty Low; some soreness; requires patient adherence
Massage Chronic LBP (short-term) Moderate Certainty Low; short duration of benefit; cost
Chiropractic Chronic LBP Moderate Certainty Rare serious AE (vertebral artery dissection, ~1 per million); manipulation risk
NSAIDs Acute and chronic LBP (short-term) High Certainty GI bleeding: 5,000–16,500 GI deaths/yr (US); cardiovascular risk; no safe option for patients with both CV and GI risk
Opioids Acute pain; not supported for chronic pain Low Certainty No RCT data beyond 16 weeks; addiction 8–12%; 42,245 US overdose deaths in 2016
Epidural steroids Radicular pain (sciatica); short-term Moderate Certainty Adding steroid to local anesthetic no better than local anesthetic alone in RCTs; risks increase with repeated use
Spinal fusion (non-specific LBP) Specific structural indications only — not non-specific LBP Low Certainty Non-significant 4.13-pt ODI difference vs. no surgery in RCTs; 14-fold regional variation in US surgery rates

Sources: Cochrane reviews; Vickers 2018; Qaseem 2017 (ACP); Chou 2007; Bhala 2013 (NSAIDs); Krebs 2018 (opioids); Fairbank 2005 (fusion). Methodology & sources.

What does the head-to-head evidence show?

Acupuncture vs. massage — Cherkin 2001

One of the few direct comparison RCTs. At 10 weeks, massage was superior to acupuncture for disability outcomes, with both superior to self-care education. This is an important finding but has important limitations: it was conducted in 2001, and results may not generalize to current practitioners or patient populations.

Cherkin DC et al. "A randomized trial comparing traditional Chinese medical acupuncture, therapeutic massage, and self-care education for chronic low back pain." Arch Intern Med. 2001;161(8):1081–1088. PMID 11322842.

Acupuncture vs. physical therapy, chiropractic

No adequate head-to-head RCTs exist for acupuncture vs. chiropractic or acupuncture vs. exercise therapy for chronic back or neck pain. This is a genuine evidence gap, not a finding favoring or opposing either treatment.

GERAC Trial — Acupuncture vs. conventional treatment

The German Acupuncture Trials (GERAC) found that real acupuncture (47.6% responders) and sham acupuncture (44.2% responders) both outperformed conventional care (27.4% responders) for chronic low back pain. The proximity of real and sham response rates is consistent with the sham debate described on the back pain evidence page.

Haake M et al. "German Acupuncture Trials (GERAC) for Chronic Low Back Pain." Arch Intern Med. 2007;167(17):1892–1898. PMID 17893311.

Why conventional treatments are not automatically preferable

NSAIDs: no safe option for high-risk patients

For patients with both cardiovascular and gastrointestinal risk, there is no NSAID option that avoids both risk categories. Traditional NSAIDs increase GI risk; COX-2 inhibitors increase CV risk. This is not a theoretical concern — NSAID-associated GI toxicity accounts for an estimated 5,000–16,500 deaths annually in the US.

Opioids: no chronic pain evidence base

There are no RCTs demonstrating opioid efficacy for chronic non-cancer pain beyond 16 weeks. The addiction rate in chronic pain populations is 8–12% (Vowles et al. 2015). Approximately 806,000 people died of opioid overdose in the US between 1999 and 2023 (CDC).

Surgery: high variation, limited evidence for non-specific pain

For non-specific low back pain, spinal fusion shows a non-significant 4.13-point difference on the Oswestry Disability Index vs. no surgery in RCT evidence. Spinal surgery rates vary up to 14-fold across US regions — variation unexplained by patient characteristics alone, suggesting non-clinical factors drive a substantial portion of surgical decisions.

What do we not know?