Thomas 1991

Methods	RCT cross‐over Number analysed/randomised: 132/132 Power analysis: Post hoc analysis suggested 70 participants per group were needed Intention‐to‐treat analysis: NA Funding source: NR
Participants	Chronic mechanical neck disorder, cervical osteoarthritis Participant recruitment: Sweden
Interventions	INDEX TREATMENT Acupuncture (Acu): LI3 and GB20 bilaterally, DU14, DU16, DU20 with manual stimulation for 10s for Deqi every 5 minutes, 40 minute session COMPARISON TREATMENT Sham acupuncture: needles inserted superficially, not manually stimulated CO‐INTERVENTION Instructed to take no pain medication 24 hours before trials Duration of treatment: 1 session Duration of follow‐up: 2 hours
Outcomes	PAIN INTENSITY (VAS 0 to 10) Baseline: acupuncture 2.5, placebo acupuncture 2.0, placebo diazepam 1.9 End of study mean: acupuncture 1.8, placebo 1.6, placebo diazepam 1.7 Absolute benefit: acupuncture 0.7, placebo 0.4, placebo diazepam 0.2 Reported results: not significant when acupuncture was compared with placebo acupuncture but statistically significant when acupuncture was compared with placebo diazepam vs placebo acupuncture: SMD 0.17 (95% CI random ‐0.25 to 0.59) immediate post treatment vs placebo diazepam: SMD 0.09 (95% CI random ‐0.33 to 0.51) immediate post treatment Adverse effects: NR Costs of care: NR
Notes	‐‐
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Only described as randomised to order
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding (performance bias and detection bias) All outcomes ‐ patient?	High risk	Interventions perceivably different
Blinding (performance bias and detection bias) All outcomes ‐ care provider?	High risk	Not possible
Blinding (performance bias and detection bias) All outcomes ‐ outcome assessor?	High risk	Participant not blinded as outcome assessor
Incomplete outcome data (attrition bias) All outcomes ‐ Drop out rate acceptable?	Low risk	No drop‐outs
Incomplete outcome data (attrition bias) All outcomes ‐ Analyzed in the group to which they were allocated?	Low risk	All randomised participants were analysed
Selective reporting (reporting bias)	Unclear risk	No protocol
Similarity of baseline characteristics	Low risk	Each participant received all treatments
Co‐interventions avoided or similar?	Low risk	Avoided
Compliance acceptable?	Low risk	All participants received all treatments
Similar timing of outcome assessment?	Low risk	Immediate post treatment
Fatal Flaw	High risk	Improper error terms for analysis (SD should not be used for within‐patient analysis); randomisation and concealment not described; no adjustments made for order; balancing not apparent