---
title: "Evidence maps"
url: "https://acuiq.com/wiki/research/evidence-maps"
updated: 2026-10-09
last_verified: 2026-10-09
license: CC-BY-4.0
license_url: "https://creativecommons.org/licenses/by/4.0/"
section: "research"
basis: "Modern research"
audience: "Anyone"
published: "2026-10-09"
reading_minutes: 8
sources: 6
---

# Evidence maps

_Five overviews of acupuncture's systematic reviews, from 2011 to 2025: what each counted, how it graded the evidence, and which conditions it placed at the top._

Since 2010 at least five groups have tried to take stock of the whole acupuncture literature at once by collecting its systematic reviews, sorting them by condition, and placing each condition in a category of evidence. The US Department of Veterans Affairs called its version an [evidence map](https://www.hsrd.research.va.gov/publications/esp/acupuncture.cfm): a visual overview of where research exists, where it does not, and what it found, meant to show where the evidence lies rather than to settle any single question. The maps searched different years and databases, counted different things and graded them on different scales, so their lists of conditions overlap only in part. This page reports what each map counted and concluded, and where they agree.

## The maps at a glance

| Map | Reviews searched | What it counted | Top category |
| --- | --- | --- | --- |
| Ernst, Lee and Choi, _Pain_, 2011 | Systematic reviews of acupuncture for pain, 11 databases, no language limit | 57 reviews | Neck pain was the one condition with unanimously positive conclusions from more than one high-quality review |
| Hempel and colleagues, US Department of Veterans Affairs, 2014 | English-language reviews published January 2005 to March 2013 | 183 reviews | Chronic pain: positive effect, high confidence |
| McDonald and Janz, Acupuncture Evidence Project, 2017 | Reviews and meta-analyses from March 2013 to September 2016, three later reviews, and two earlier government reviews as baseline | 122 conditions | 8 conditions with a positive effect |
| Lu and colleagues, _BMJ Open_ and _BMJ_, 2022 | Reviews with meta-analyses of randomised trials, January 2015 to November 2020, including four Chinese databases | 120 reviews, 77 conditions, 1,402 trials, 138,995 participants | 9 outcomes with moderate or large effects at moderate or high certainty |
| Hempen and Hummelsberger, _Complementary Therapies in Medicine_, 2025 | Reviews and meta-analyses in PubMed, 2017 to 2022 | 862 reviews, 184 conditions | 10 conditions with a positive effect |

## A review of reviews of pain, 2011

Ernst, Lee and Choi, in Exeter and at the Korea Institute of Oriental Medicine, [assessed 57 systematic reviews](https://doi.org/10.1016/j.pain.2010.11.004) of acupuncture for pain and also collected case reports of serious harm published since 2000. Four of the 57 reviews were of excellent methodological quality, and the authors found “numerous contradictions and caveats” across them. They concluded that the reviews had produced little convincing evidence that acupuncture reduces pain. They also gathered 95 reports of severe adverse effects, 5 of them fatal, most often a collapsed lung or an infection.

## The Veterans Affairs map, 2014

Susanne Hempel, Paul Shekelle and colleagues at the VA’s Evidence-based Synthesis Program in Los Angeles found 1,223 citations and kept 183 systematic reviews. Pain had the most (65), then wellness (44), other conditions (48), mental health (20) and adverse events (6). For each of the three VA priority areas they drew a bubble plot: the number of trials in the largest review on a condition set its height, the estimated effect against passive controls set its position along the horizontal axis, and the map’s confidence set the bubble’s size. Passive controls meant no treatment, a waiting list, sham acupuncture, or acupuncture added to care both groups received.

Effects were placed in four categories: evidence of no effect, unclear evidence, a potential positive effect, and a positive effect, the last reserved for statistically significant effects where the best reviews recommended acupuncture without major concerns. Confidence was medium by default, raised to high only for a Cochrane review or an individual patient data meta-analysis, and lowered to low where more than half the trials behind an effect were not indexed in PubMed. Chronic pain was rated a positive effect with high confidence, on the strength of an individual patient data meta-analysis of 29 trials. The authors noted the tension in that result: back pain and neck pain make up most chronic pain, yet they rated the evidence for each of those, and for fibromyalgia and shoulder pain, as unclear. They found that migraine reviews showed a positive effect driven by trials against no acupuncture rather than against sham. They stated that the map was not designed to inform policy on the use of acupuncture.

## The Acupuncture Evidence Project, 2017

John McDonald and Stephen Janz [updated the VA map and a 2010 review](https://acupuncture.org.au/publicassets/1c94b136-802d-ed11-9115-00505687f2af/Acupuncture-Evidence-Project_Mcdonald-and-Janz_-Revised-Edition-07-02-17-Referenced.pdf) for the Australian Department of Veterans’ Affairs, adding reviews from March 2013 to September 2016. The project was commissioned and funded by the Australian Acupuncture and Chinese Medicine Association, a professional body, and its report discloses that McDonald co-authored three of the papers it cites. It graded evidence on the Australian National Health and Medical Research Council’s levels and, where possible, GRADE. Of 122 conditions in 14 clinical areas, it placed 8 in “positive effect”, 38 in “potential positive effect”, 71 in “unclear or insufficient evidence” and 5 in “no evidence of effect”, and found the level of evidence had risen for 24 conditions since the earlier reviews. Its five no-effect conditions were alcohol dependence, cocaine addiction, epilepsy, nausea in pregnancy and smoking cessation beyond six months. The authors wrote that sham controls are not inert and that this likely leads trials to underestimate the effect of acupuncture; [Sham acupuncture](/wiki/research/sham-acupuncture) sets out the evidence on that question.

## The Guangzhou map, 2022

Liming Lu, Nenggui Xu and colleagues at Guangzhou University of Chinese Medicine, with co-authors in Canada and Chile including Guyatt, one of the developers of GRADE, published the map in two parts. The [BMJ Open paper](https://doi.org/10.1136/bmjopen-2021-056803) covered 120 reviews with meta-analyses, re-ran meta-analyses where newer trials had appeared, and rated each of 77 chosen reviews’ outcomes with GRADE. It excluded trials in which acupuncture was added to another treatment. It rated 119 of the 120 reviews as low or critically low in methodological quality on the AMSTAR 2 checklist, most often because a review did not list the studies it excluded or explain changes to its protocol, and found that 97 had not used GRADE themselves.

The companion [BMJ analysis](https://doi.org/10.1136/bmj-2021-067475) counted 2,471 systematic reviews of acupuncture in Web of Science from 2000 to 2020, 64% of them reviews of randomised trials, and argued that conditions with moderate or high certainty of benefit are underused in practice and policy. It recommended that research funding go to conditions where large effects rest on low-certainty evidence. The map’s highest group was large effects at moderate certainty for post-stroke aphasia, myofascial pain and successful breastfeeding within 24 hours of birth; moderate effects at high certainty for neck pain and shoulder pain; and moderate effects at moderate certainty for vascular dementia, fibromyalgia pain, non-specific low back pain and nasal symptoms of allergic rhinitis.

## The 2025 update

Hempen and Hummelsberger [continued the series](https://doi.org/10.1016/j.ctim.2025.103149) from 2017 to 2022, using the same four categories as the VA map and the Evidence Project. From 862 reviews and meta-analyses covering 184 conditions, they placed 10 conditions in “positive effect”, 82 in “potential positive effect”, 86 in “unclear or insufficient” and 6 in “no evidence of effect”. They reported that more conditions sat in the upper two categories than in reviews up to 2017, and that a lack of high-quality trials behind some reviews limited how far reviews could be compared.

## Where the maps agree

The three maps that list their top category in full overlap on a handful of conditions. A dash means the condition is not in that map’s top category: it may have been rated lower or not assessed.

| Condition | Evidence Project 2017 | Lu 2022 | Hempen 2025 |
| --- | --- | --- | --- |
| Chronic or non-specific low back pain | Yes | Yes | Yes |
| Knee osteoarthritis | Yes | – | Yes |
| Migraine prevention | Yes | – | Yes |
| Tension-type headache | Yes | – | Yes |
| Nausea and vomiting after surgery | Yes | – | Yes |
| Allergic rhinitis | Yes | Yes | – |
| Nausea and vomiting from chemotherapy | Yes | – | – |
| Pain after surgery | Yes | – | – |
| Neck pain | – | Yes | – |
| Shoulder pain | – | Yes | – |
| Fibromyalgia pain | – | Yes | – |
| Myofascial pain | – | Yes | – |
| Post-stroke aphasia | – | Yes | – |
| Vascular dementia | – | Yes | – |
| Breastfeeding within 24 hours of birth | – | Yes | – |
| Chronic pain | – | – | Yes |
| Cancer-related fatigue | – | – | Yes |
| Menopausal symptoms | – | – | Yes |
| Female infertility, alongside reproductive treatment | – | – | Yes |
| Chronic prostatitis or chronic pelvic pain in men | – | – | Yes |

Low back pain is the one condition all three place at the top; the VA map, from the years before, rated it unclear. Neck pain shows how far the methods diverge: Ernst and colleagues found it the only pain condition with consistent support among high-quality reviews, the VA rated it unclear, the Evidence Project placed it in its second category, and Lu and colleagues found a moderate effect at high certainty. The Lu map’s column measures something different from the other two, because it ranks outcomes by GRADE certainty and effect size rather than by the conclusions of review authors. The evidence pages in this wiki, such as [Low back pain](/wiki/evidence/low-back-pain), [Neck pain](/wiki/evidence/neck-pain) and [Knee osteoarthritis](/wiki/evidence/knee-osteoarthritis), report the individual reviews behind each row.

## What the maps cannot tell you

A map summarises reviews, and reviews summarise trials, so the same trial can be counted many times: the VA authors noted that reviews of the same topic included anywhere from one to 72 studies, and that primary studies contributed to several reviews at once, most of all in pain. No map re-examined the trials, so each inherits whatever the trials and reviews got wrong, including the risk-of-bias and reporting problems described on [Reading an acupuncture trial](/wiki/research/reading-an-acupuncture-trial) and the uneven geography of the literature set out on [Where the trials come from](/wiki/research/where-the-trials-come-from). The categories are also not interchangeable: a “positive effect” in one map and “moderate certainty” in another are judgements made by different rules. The VA map also pooled trials against sham with trials against no treatment, so a condition’s place on it does not say which comparison carried the effect. AcuiQ’s protocol pages record what individual studies prescribed rather than any map’s verdict; [Conformance](/wiki/research/conformance) reports how many of the conditions it indexes rest on a single source.

### Sources

1. Hempel S, Taylor SL, Solloway MR et al., Evidence Map of Acupuncture, US Department of Veterans Affairs Evidence-based Synthesis Program, January 2014 — https://www.hsrd.research.va.gov/publications/esp/acupuncture.cfm
2. Ernst E, Lee MS, Choi TY, Acupuncture: does it alleviate pain and are there serious risks? A review of reviews, Pain 2011 — https://doi.org/10.1016/j.pain.2010.11.004
3. McDonald J, Janz S, The Acupuncture Evidence Project: a comparative literature review (revised edition), Australian Acupuncture and Chinese Medicine Association, January 2017 — https://acupuncture.org.au/publicassets/1c94b136-802d-ed11-9115-00505687f2af/Acupuncture-Evidence-Project_Mcdonald-and-Janz_-Revised-Edition-07-02-17-Referenced.pdf
4. Lu L, Zhang Y, Ge S et al., Evidence mapping and overview of systematic reviews of the effects of acupuncture therapies, BMJ Open 2022 — https://doi.org/10.1136/bmjopen-2021-056803
5. Lu L, Zhang Y, Tang X et al., Evidence on acupuncture therapies is underused in clinical practice and health policy, BMJ 2022 — https://doi.org/10.1136/bmj-2021-067475
6. Hempen M, Hummelsberger J, The state of evidence in acupuncture: a review of metaanalyses and systematic reviews of acupuncture evidence (update 2017–2022), Complementary Therapies in Medicine 2025 — https://doi.org/10.1016/j.ctim.2025.103149