Bulletin of the World Health Organization

Acupuncture-related adverse events: a systematic review of the Chinese literature

Junhua Zhang a, Hongcai Shang a, Xiumei Gao a & Edzard Ernst b

a. Research Centre of Traditional Chinese Medicine, Tianjin University of Traditional Chinese Medicine, 88 Yuquan Road, Tianjin, 300193, China.
b. Peninsula Medical School, University of Exeter, Exeter, England.

Correspondence to Junhua Zhang (e-mail: zjhtcm@gmail.com).

(Submitted: 09 February 2010 – Revised version received: 21 July 2010 – Accepted: 04 August 2010 – Published online: 27 August 2010.)

Bulletin of the World Health Organization 2010;88:915-921C. doi: 10.2471/BLT.10.076737


Acupuncture is popular in most countries, but nowhere more than in China. Because its use is so widespread, safety is an important issue that deserves close attention. Serious adverse events resulting from acupuncture, including pneumothorax, cardiac tamponade, spinal cord injury and viral hepatitis, have been identified in previous literature reviews.14 Prospective surveys to determine the frequency of acupuncture-related adverse events have been conducted in Germany,5,6 Norway7 and the United Kingdom of Great Britain and Northern Ireland.8,9 These studies have shown an incidence of mild, transient acupuncture-related adverse events that ranges from 6.71% to 15%. The most common adverse events of this type were local pain from needling (range: 1.1–2.9%) and slight bleeding or haematoma (range: 2.1–6.1%). In a prospective observational study of 190 924 patients, the incidence of serious adverse events (death, organ trauma or hospital admission) was about 0.024%.5 Another large-scale observational study showed a rate of adverse events requiring specific treatment of 2.2% (4963 incidents among 229 230 subjects).6 Studies such as these have shown that in extremely rare cases acupuncture can lead to serious, sometimes life-threatening complications, in addition to mild and transient adverse events.

Because most reports on the safety of acupuncture have been published outside China, the objective of this article was to summarize the Chinese literature on the subject of acupuncture-related adverse events and determine the possible reasons that such events occur.


Inclusion/exclusion criteria

In December 2009 we searched the following electronic databases: Chinese Biomedical Literature Database (1980–2009), Chinese Journal Full-Text Database (1980–2009) and Weipu Journal Database (1989–2009). The search terms were: (acupuncture OR needle) AND (induce OR cause OR adverse event OR adverse reaction OR side effect OR complication OR harm OR risk OR mistake OR infection OR injury OR fainting OR haemorrhage OR bleeding OR death OR pneumothorax OR pain). We searched for these terms (in Chinese) as free text in the title or abstract, and we also hand-searched the reference lists of all reports located through the electronic searches.

Case reports, case series, surveys and other observational studies were included in the review if they reported factual data on complications related to acupuncture. Review articles, translations and clinical trials were excluded. The search was limited to Chinese-language papers.

Different types of acupuncture can lead to different adverse events. To present clear results, we only included reports on traditional needle acupuncture, defined as a procedure in which stainless steel filiform needles are inserted into acupoints – acupuncture points located throughout the body that are associated with specific therapeutic effects – and manipulated in place. Other types of acupuncture, such as electroacupuncture, laser acupuncture and auricular acupuncture, were excluded.

Two authors (Zhang and Shang) independently examined the titles and abstracts of all papers found through the search to determine if they fulfilled the inclusion criteria outlined above. The full texts of potentially relevant articles were retrieved for detailed assessment. Disagreements between the two authors were resolved by discussion.

Information on author, patient, acupuncturist, acupuncture site, adverse event, treatment and outcome was extracted from the primary articles and entered into a pre-formulated spreadsheet. Acupoints were described by pinyin name (i.e. the Latinized spelling of traditional Mandarin Chinese names) and code according to a standard nomenclature developed by the World Health Organization.10,11 These data have been summarized in three tables according to the type of adverse event.


Our inclusion criteria were met by 115 articles (98 case reports and 17 case series) (Fig. 1). We noted no clear trend in the frequency of reports of acupuncture-related adverse events over the past 30 years.

Fig. 1. Flow diagram for systematic review of the Chinese-language literature on adverse events related to traditional needle acupuncture, 1980–2009
Fig. 1. Flow diagram for systematic review of the Chinese-language literature on adverse events related to traditional needle acupuncture, 1980–2009

In total, 479 cases of acupuncture-related adverse events were reported. Patients ranged in age from 2 to 73 years. The first authors of the papers were members of medical departments, court jurisdictions and police departments. Only 20% of these authors were the acupuncturists who performed the procedure that caused the adverse event. The reported acupuncture-related adverse events were classified into three categories: traumatic (Table 1, available at: http://www.who.int/bulletin/volumes/88/12/10-076737), infectious (Table 2) and “other” (Table 3).

Traumatic events

Traumatic injuries were reported in 87 articles (73 case reports and 14 case series) and totalled 296 cases. The events were classified into seven subgroups according to the type and site of the injury.

Arachnoid and spinal dura mater

Nine cases of spinal epidural haematoma (in the cervical, thoracic and lumbar spine) were reported. No further information was provided.

Subarachnoid haemorrhage was reported in 35 patients, 3 of whom died. The others recovered after 1 to 8 weeks of treatment. One of the deceased patients had a history of hypertension and cerebral haemorrhage and died 10 days after the acupuncture. The other two patients died within 30 minutes of having undergone the acupuncture, perhaps as a result of injury to the medulla oblongata.

The acupoints most frequently involved in cases of subarachnoid haemorrhage and spinal epidural haematoma were Fengchi (GB20), Yamen (GV15), Fengfu (GV16), Dazhui (GV14) and Tianzhu (BL10). In several cases, the needles were inserted to a depth of 4 to 5 cm below the skin’s surface, and such deep insertion is suspected to have led to injury.

Thoracic organs and tissues

With a total of 201 cases, pneumothorax was the most frequently reported acupuncture-related adverse event. Four patients died from it and the others recovered after 2 to 30 days of treatment. One patient was a 70-year-old woman with a history of chronic bronchitis, emphysema, cor pulmonale and heart failure who died from pulmonary infection, heart failure and pneumothorax. Two more women died after not receiving timely treatment for pneumothorax caused by needling at the Jianjing (GB21) and Tianding (LI17) acupoints. The fourth patient died from tension pneumothorax but no further information was provided.

The acupuncture sites in these cases were primarily in the shoulder and scapular regions (64%) and in the chest (24%). In two cases, the Tianding (LI17) point in the neck area had been needled. The most frequently used acupoints were Jianjing (GB21; 30%), Feishu (BL13; 15%), Quepen (ST12; 10%) and Tiantu (CV22; 10%). Other acupoints were Ganshu (BL18), Shenshu (BL23), Tianding (LI17), Jiuwei (CV15), Juque (CV14), Jianzhen (SI9), Quyuan (SI13) and Dingchuan (EX-B1).

Chylothorax was reported after needling at the Feishu (BL13) point in a 21-year-old man with a malformed thoracic duct. Right ventricular injury was reported in four cases, two of which recovered after surgical treatment. The other two patients died from right ventricular puncture complicated by cardiac tamponade and multiorgan dysfunction syndrome. One case of aortic artery rupture was reported after needling at the Qimen point (LR14) at a depth of 4 cm; the patient died within 15 minutes. Coronary artery injury with cardiac tamponade was reported in a man who treated himself for chronic bronchitis and lost the needle at the Zhongfu point (LU1).

Abdominal organs and tissues

Injuries of abdominal organs and tissues were reported in 16 patients, all of whom recovered after surgery. These instances included perforations of the gallbladder, of the bowels and of the stomach, frequently complicated by peritonitis. A 2-year-old boy suffered intestinal wall haematoma with intestinal obstruction after acupuncture treatment for diarrhoea.

The acupoints associated with such adverse events were Tianshu (ST25), Zhongwan (CV12) and Qimen (LR14). Most of the patients underwent acupuncture for abdominal pain, attributable mainly to appendicitis or cholecystitis. Deep needling accounted for most of the abdominal injuries.

Neck area

Six cases of injuries in the neck region were reported, including neural injuries (4), a false aneurysm of the carotid artery (1) and thyroid haemorrhage (1). One patient died after acupuncture at the Tiantu point (CV22); the needle had been inserted to a depth of 6 cm.


Five articles reported injuries to the eyes, including orbital haemorrhage (3), traumatic cataract (1), injury of the oculomotor nerve (1) and retinal puncture (1). One case of optic atrophy accompanied by haemorrhage and traumatic cataract resulted in visual impairment.

The acupoints in the above cases were Jingming (SL1), Qiuhou (EX-HN7) and Chengqi (ST1). When needling acupoints in the area of the orbital cavity, bleeding is difficult to avoid, even for the experienced acupuncturist. Deep needling can also injure the oculomotor nerve, the retina and neighbouring tissues.

Peripheral nerves, vessels and other tissues

Three cases of haemorrhage were reported after acupuncture on the cheeks and the hypoglottis. One case of calf haematoma complicated by diabetic foot was caused by needling at the Tiaokou (ST38) and Chengshan (BL57) acupoints.

Four cases of peripheral motor nerve injuries and subsequent motor dysfunction were reported. Three children suffered adductor muscle fibrosis and adduction deformity of the thumb as a result of local vascular and muscular injuries from needling at the Hegu point (LI4).

The acupoints most frequently involved in the injuries were Taiyang (EX-HN5), Neiguan (PC6) and Hegu (LI4). Forceful needle manipulation at these points, which are quite superficial, can cause injury to peripheral nerves, capillaries and muscle fibres.

Needling site pain and broken needle

Four cases of pain at the needling site were reported in two articles. An intra-abdominal lump turned out to be caused by an acupuncture needle fragment that had broken off 15 years earlier.

Infectious events

Nine cases of bacterial infection and two cases of viral infection were reported. All patients recovered after appropriate treatment.

Infections were mainly due to poor sterilization of acupuncture needles. Acupoints on the head became infected most often, perhaps because hair makes it difficult to implement aseptic technique. Two cases of facial abscess may have been caused by acupuncture to relieve toothache.

Other adverse events

A total of 172 acupuncture-related adverse events that were neither due to trauma nor to infection were reported. Local allergic reactions occurred after acupuncture in four patients with an allergy to metal needles.

In our review, fainting was the most common adverse event associated with acupuncture, and it occurred primarily in patients receiving acupuncture for the first time. In total, 150 cases of fainting were reported. In one report of 82 cases, 60% (49) of the patients fainted during the first treatment. Of these 49 fainting spells, 83% occurred when acupuncture was being applied to the head or neck.

Stroke after acupuncture was reported in five patients (aged from 58 to 73 years). One case of stroke occurred in a 72-year-old woman who received acupuncture on her arm. The other four patients had a history of stroke and hypertension. Three patients died from cerebral haemorrhage that was considered to be causally related to the acupuncture.

Other adverse effects included cardiac arrest, pyknolepsy (epileptiform attacks resembling petit mal), shock, fever, cough, thirst, aphonia, leg numbness and sexual dysfunction. However, the existence of a causal link between acupuncture and these adverse events is uncertain.


Many types of acupuncture-related adverse events have been identified in the Chinese literature. Injuries and infections appear to be related to inappropriate technique, whereas other types of adverse events are not. Fainting is vasovagal in origin and minor bleeding is sometimes inevitable.

Infections result primarily from poor aseptic procedure and insufficient knowledge on the part of acupuncturists, who often disinfect reusable acupuncture needles with alcohol instead of sterilizing them. The use of disposable sterile acupuncture needles and guide tubes is strongly recommended.12

Most traumatic events are caused by improper manipulation in high-risk acupoints. The depth of needle insertion is crucial. The lung surface is about 10 to 20 mm beneath the skin in the region of the medial scapular or midclavicular line.2 This may explain the high incidence of pneumothorax during needling in this area. Other traumatic complications, such as subarachnoid haemorrhage, cardiovascular injuries or perforation of the gallbladder, can also be caused by excessively deep needle insertion.

The patient’s condition also needs to be considered. Cardiovascular trauma occurred most frequently in patients with cardiomegaly. Patients with abdominal pain that has no clear diagnosis are at increased risk of trauma or infection from acupuncture at abdominal acupoints. Symptomatic treatment of abdominal pain with acupuncture can also delay effective therapy. During needling at peripheral acupoints on the legs, arms and face, manipulation should be carefully executed to avoid damaging nerves and blood vessels.

Some adverse events are inevitable but could be minimized through preventive measures. Fainting, which is a reflex caused by vagal excitation, is the most common adverse event during acupuncture.13 Its incidence can be reduced by preparing patients and positioning them properly; the patient should not be hungry or tired and should preferably be placed in the supine, lateral or prone position.

Of the 87 articles reporting traumatic events, 59 (70%) provided information about the acupuncturists. Of these 59 articles, 68% (40) indicated that the acupuncturists were practising in village clinics or rural hospitals when they performed the procedures that caused the traumatic events. All infections reported were caused by acupuncturists in rural areas. In China, acupuncturists in rural and urban hospitals have a great disparity in clinical skills. Acupuncturists practising in rural hospitals, township health centres or village clinics rarely receive formal education in medical colleges. It follows that training for the practice of acupuncture needs to be unified and improved.

Several serious adverse events were identified through a review of case reports,14 but very few were found in surveys79 or prospective observational studies.5,6 This suggests that serious acupuncture-related adverse events are rare. Bleeding and pain during needling are reported less often in the Chinese-language than in the English-language literature, perhaps because practitioners in China consider such events too trivial to report. Infections (primarily hepatitis) after acupuncture are reported frequently in the English-language literature1 but relatively rarely in the Chinese-language literature, even though non-disposable acupuncture needles are still used in China. It is possible that in China acupuncture-related infections are underreported.

Of the 87 articles reporting traumatic injuries, 72 (about 70%) were authored not by the acupuncturists themselves, but by the physicians who treated the adverse events. None of the articles reporting infections were authored by the acupuncturists, as opposed to 16 of the 20 (80%) reports of adverse events other than trauma or infection. Again, we suspect that underreporting of such events in the Chinese-language literature is much higher than in the English-language literature.

Our review has several limitations. Although our search strategy was comprehensive, we cannot guarantee that all relevant articles were identified. Many of the reports lacked detail, so that cause–effect relationships are often uncertain. In the absence of a denominator (i.e. the total number of acupuncture treatments practised over the study period), the reported adverse events do not lend themselves to generating incidence figures. There are 2688 hospitals of traditional Chinese medicine in China.15 If we assume, for instance, that each hospital receives 50 to 100 visits for acupuncture per day (a conservative figure), the annual number of acupuncture treatments would total from 50 to 100 million. This would suggest that the incidence of acupuncture-related adverse events is negligible. However, the true incidence remains unknown and cannot be accurately estimated. Collectively these factors limit the conclusiveness of our findings.


Various types of acupuncture-related adverse events have been reported in China. Similar events have been reported by other countries,19 usually as a result of inappropriate technique. Acupuncture can be considered inherently safe in the hands of well trained practitioners. However, there is a need to find effective ways to improve the practice of acupuncture and to monitor and minimize the health risks involved.


The authors thank YY Xu, X Zhang and WK Zheng for their help with the literature search.

Competing interests:

None declared.