Carlsson 2001.

Study characteristics
Methods	Three‐arm RCT Number analyzed/randomized: 27/50 Statistical analysis: Student's t test, Chi² test, Fisher exact test, Mann‐Whitney test, and Wilcoxon signed ranks test; power analysis not conducted Funding source: NR Ethical approval and informed consent obtained
Participants	Participant recruitment: by referral Setting: Swedish university hospital Inclusion criteria: 1) lumbar or lumbosacral pain experienced for 6 months or longer; 2) no radiation of pain below knee; 3) normal neurological examination findings for lumbosacral nerve function, including deep tendon reflexes, plantar response, voluntary muscle activation, straight leg raising, and sensory function Exclusion criteria: 1) major trauma or systemic disease; 2) ongoing pregnancy; 3) history of acupuncture treatment Age (mean ± SD): 49.8 ± 15.4 years Gender (female): 66% Pain duration (mean ± SD): 9.5 ± 7.0 years Pain intensity: no significant difference in baseline pain intensity (VAS scale)
Interventions	1) GROUP 1: manual acupuncture (semi‐standardized) Acupuncture points: some points on lower back (local points) and other points on lower limbs and forearms or hands (distal points; bilateral LI4, LI11, BL24, BL25, BL26, BL40, BL57, BL60, Yaoyan, and Jiaji) Depth: NR (30 to 70 mm needles) De Qi: stimulated three times during each treatment, usually at a depth of 2 to 3 cm Sessions: 10 (20 min; 1/week for 8 weeks; 9th at 6 months and 10th at 8 months). All three groups received same treatment schedule. Acupuncturist experience: board‐certified anesthesiologist with considerable acupuncture experience (> 10,000 treatments) 2) GROUP 2: electroacupuncture Two or three sessions of manual acupuncture administered, followed by treatments with electrical stimulation delivered by four needles (one pair on each side) to lower back. Stimulation frequency: 2 Hz every 2.5 seconds, interrupted by 15 Hz for 2.5 seconds (dense disperse) at perceptible but not painful stimulation intensity. Similar number of needles as used with the manual acupuncture group. Same schedule as Group 1. 3) GROUP 3: placebo Mock TENS administered by a disconnected stimulator attached to two large TENS electrodes. Electrodes placed on skin over most intensely painful area of low back. Flashing lamps visible to participant during stimulation. Same treatment schedule as Group 1 Co‐intervention: all participants permitted to take NSAIDs or weak opioids as needed Duration of treatment: 2 months for most of participants Duration of follow‐up: 6 months after the 8th session
Outcomes	1) Pain intensity: VAS (0 to 100) recorded twice daily (morning and evening) 2) Global assessment of therapy effectiveness: classified as improved, unchanged, or worse, based on the results of clinical interviews on pain‐related disability and physical examination 3) Pain‐related disability: level of activity (at work or at home) recorded weekly. Change in workplace activity (at 6 months) Assessment times: 1, 3, and 6 months after the end of sessions Costs: NR Adverse effects: NR
Notes	Conclusion: Compared with placebo, acupuncture had long‐term pain‐relieving effects for some people with chronic nonspecific LBP. Language: English Study not included in quantitative analysis for intermediate and long‐term follow‐up due to high dropout rates (44% at 3 months and 54% at 6 months)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The randomisation process was produced from a previously developed computer‐generated list.
Allocation concealment (selection bias)	Low risk	The randomized numbers were kept by a secretarial assistant who was not otherwise involved in the study.
Blinding of participants (performance bias) All outcomes	Low risk	Participants were told that they would receive three forms of stimulation, and were asked not to reveal to the independent observer what they were feeling. Blinding method not clearly described, but verification testing suggested some degree of successful blinding (P = 0.04)
Blinding of personnel /care providers (performance bias) All outcomes	High risk	Not possible
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Some outcomes measured by blinded assessor; other subjective outcomes recorded by participants, who may have been blinded
Incomplete outcome data (attrition bias) All outcomes	High risk	High dropout rate (44%) at second follow‐up due to subjects' failure to respond to treatment
Intention‐to‐treat‐analysis (attrition bias)	Unclear risk	Unclear whether ITT analysis was used; 44% (high) dropout rate
Selective reporting (reporting bias)	High risk	The outcomes were pooled from two acupuncture groups, which broke the randomisation and was unacceptable
Group similarity at baseline (selection bias)	Unclear risk	Doses of analgesic drug seemed not similar between the combined acupuncture groups and the placebo group.
Co‐interventions (performance bias)	Unclear risk	Could not be assessed because drugs were reported as pooled data of acupuncture groups
Compliance bias (performance bias)	Unclear risk	Most of the participants finished 8 of the 10 treatments; unclear if it was acceptable or comparable.
Timing of outcome assessments (detection bias)	Unclear risk	Global improvement at the third follow‐up was assessed between 6 and 48 months; others were assessed at the same time.
Other bias	Low risk	Not identified