Type 2 diabetes is a long-term disorder in which the body's cells respond poorly to insulin, the hormone that moves glucose out of the blood, and blood glucose stays high. Prediabetes, which includes impaired glucose tolerance, is the stage in which glucose is raised but below the diabetic threshold. Cochrane has published no review of acupuncture for type 2 diabetes, prediabetes or insulin resistance. The reviews that exist were written by Chinese and Australian research groups and pool trials that mostly added acupuncture, electroacupuncture, acupressure or acupoint catgut embedding to diet, exercise or diabetes drugs, and compared the result with those treatments alone. They report lower blood glucose with the added acupuncture and rate the trials as low in quality. The one review that analysed sham-controlled trials separately, a 2026 meta-analysis by Lyu and colleagues, found a difference from sham in fasting glucose and none in HbA1c.
The reviews measure four things. Fasting blood glucose is the level after an overnight fast; 2-hour glucose is the level two hours after a meal or a glucose drink; HbA1c is the share of haemoglobin carrying glucose, which reflects the average over about three months; and HOMA-IR is an estimate of insulin resistance calculated from fasting glucose and fasting insulin.
What Cochrane has published
AcuiQ searched PubMed in October 2026 for Cochrane reviews of acupuncture, acupressure or electroacupuncture with diabetes, glucose or insulin, and found none on glycaemic control. Cochrane's review of acupuncture for symptomatic gastroparesis, a stomach complication of diabetes, is reported on Gastroparesis, and nerve damage in the feet and legs is on Diabetic neuropathy.
Systematic reviews
| Review | Population and methods | Trials and people | Comparison | What the authors report |
|---|---|---|---|---|
| Song 2018, searched to January 2018 | Insulin resistance in any condition; acupuncture alone or added | 9 trials, 562 people | Not stated in the abstract | HOMA-IR and fasting insulin improved; fasting glucose did not differ |
| Chen 2019, searched to June 2018 | Type 2 diabetes; acupuncture and its variants added to baseline treatment | 21 trials, 1,943 people | Sham acupuncture or no acupuncture, pooled | Fasting glucose 1.21 mmol/L lower, 2-hour glucose 2.13 mmol/L lower, HbA1c 1.12 percentage points lower |
| Li 2022, searched to July 2020 | Type 2 diabetes; acupuncture added to diabetes drugs | 21 trials, 1,188 people | Drugs alone | Fasting glucose 6.46 mg/dL lower (moderate certainty) and HOMA-IR lower (low); no difference in HbA1c (very low), 2-hour glucose or fasting insulin (low) |
| Zhang 2022, searched to December 2020 | Prediabetes; acupuncture-related therapies | 31 trials pooled | Usual care, sham and conventional medicine, pooled | Lower fasting glucose, 2-hour glucose and HbA1c; more people returned to normal glucose (risk ratio 1.43) |
| Yu 2023 | Prediabetes; catgut embedding added to conventional treatment | Not stated in the abstract | Conventional treatment alone | HbA1c 0.45 percentage points lower, fasting glucose 0.61 mmol/L lower; adverse event rates similar |
| Si 2025 | Type 2 diabetes; acupuncture as an adjunct | 21 trials, 2,117 people | Not stated in the abstract | HbA1c, 2-hour glucose, fasting glucose and fasting insulin improved; marked heterogeneity on some measures |
| Li 2026, searched to July 2025 | Type 2 diabetes; acupuncture | 20 trials, 1,479 people | No acupuncture | Lower fasting glucose, HbA1c, 2-hour glucose and HOMA-IR; no difference in insulin level |
| Lyu 2026, searched to April 2025 | Insulin resistance in diabetes; acupuncture | 16 trials, 1,087 people | Usual care and sham, kept apart | Against usual care: lower HOMA-IR, fasting glucose and HbA1c. Against sham: lower fasting glucose, no difference in HbA1c |
| Zhu 2026, network meta-analysis, searched to December 2024 | Type 2 diabetes; body, ear, laser and other acupuncture methods | 139 trials, 12,231 people | Conventional treatment and combinations | Body acupuncture with conventional treatment ranked first for fasting glucose and HbA1c (HbA1c 1.08 percentage points lower than conventional treatment) |
A network meta-analysis compares several treatments at once, including pairs no trial tested head to head. Chen and colleagues wrote that small samples and the poor methodological quality of the trials left the evidence not fully convincing. Li and colleagues in 2022, at research groups in Nanchang and Chengdu, found no data on diabetic complications and reported only mild adverse events. Li and colleagues in 2026, at Chengdu University, calculated prediction intervals, the range in which the result of a future trial would be expected to fall, and reported wide variation between trials. Zhu and colleagues, at Chengdu University of Traditional Chinese Medicine, rated the included studies as low in quality. Lyu and colleagues, at Capital Medical University in Beijing, graded their findings with the GRADE system, found that higher doses of acupuncture were linked to larger changes in HOMA-IR and 2-hour glucose, and could not analyse safety because the trials reported too little. Zhang and colleagues, at Edith Cowan University in Perth, warned that substantial clinical heterogeneity meant their estimates should be read with caution.
The catgut embedding trials Yu and colleagues pooled, at Hunan University of Chinese Medicine, used BL20 (Pishu), Weiwanxiashu (EX-B3), ST36 (Zusanli), BL23 (Shenshu), SP06 (Sanyinjiao), BL21 (Weishu) and KI03 (Taixi) most often. Where the trials come from discusses what the country of a trial tells a reader.
Trials with a sham
- Electroacupuncture with metformin, 2016. Firouzjaei and colleagues, at Nanjing University of Chinese Medicine, randomised 39 overweight people with type 2 diabetes taking metformin to electroacupuncture and ear acupuncture or to a sham, superficial needling beside the points with the stimulator switched off and ear tape without seeds, in 10 sessions over 3 weeks. Points included
CV12(Zhongwan),ST25(Tianshu),ST36,SP06,LI04(Hegu) andLI11(Quchi). Fasting glucose fell from 6.65 to 6.12 mmol/L and HOMA-IR from 4.25 to 2.67 in the needle group, and the authors report significant differences from sham in both, and in weight. - Self-acupressure, 2021. Salmani Mood and colleagues randomised 66 people with type 2 diabetes to a month of self-acupressure or to lightly touching the same points, and analysed 60. Fasting glucose and stress scores fell in the acupressure group; HbA1c, measured at 3 months, did not change significantly.
- Ear acupressure, 2025. Lee and colleagues, in South Korea, randomised 51 older adults with type 2 diabetes to 8 weeks of ear acupressure at points chosen for diabetes, sleep and stress (25) or at unrelated points (26). Blood glucose and HbA1c differed between groups; glycated albumin and fructosamine, two shorter-term glucose markers, did not. The authors described the effects as not strong.
Each trial is single-centre with fewer than 70 people and follow-up of 3 months or less. Sham acupuncture sets out what superficial needling and light touch can and cannot rule out, and Acupressure describes the method.
Trials in prediabetes
A 2023 trial by Li and colleagues at Guangzhou University of Chinese Medicine gave 55 people with prediabetes a 12-week lifestyle programme, with or without two 50-minute sessions a week of a warming acupuncture style named after Zuo Chuanbo, one of the authors. Fasting glucose after treatment was 5.73 mmol/L with acupuncture and 6.04 without, and only the acupuncture group's HbA1c fell significantly; no serious adverse events were reported. The control group had no sham.
What guidelines say
NICE guideline NG28 on type 2 diabetes in adults, published in December 2015 and last updated in February 2026, does not mention acupuncture or complementary therapy. The National Chinese medicine guideline for the prevention and treatment of diabetes in primary care (2022) suggests acupuncture, ear therapy, catgut embedding and point massage as adjuvant treatments for diabetes and prediabetes, according to a 2024 commentary by Jia and colleagues at Xiamen University. The World Federation of Chinese Medicine Societies' 2020 international guideline on traditional Chinese medicine for diabetes recommends therapies by level of evidence and states that its authors found few high-quality clinical studies. Clinical guidelines lists the guidelines AcuiQ tracks.
What the reviews could not settle
Most pooled trials added acupuncture to drugs or lifestyle advice and compared the result with those treatments alone, without a sham. Against sham, Lyu's review found a difference in fasting glucose and none in HbA1c, and the three sham trials above ran for 3 weeks to 3 months. Li's 2022 review found no trial data on diabetic complications, and Lyu's found too few safety reports to analyse. The 2026 reviews also differ on method: Zhu's network meta-analysis ranked body acupuncture first for fasting glucose and HbA1c, while Lyu's ranked electroacupuncture as the most promising type for insulin resistance.
The trials in people with obesity or polycystic ovary syndrome that measured insulin resistance are reported on Obesity and Polycystic ovary syndrome. AcuiQ's pages on type 2 diabetes, prediabetes, impaired glucose tolerance, insulin resistance, hyperglycaemia and metabolic syndrome list the protocols clinical texts and individual studies prescribed, each with its citation.