已发表的论文

Background
The Alexander technique probably helps back pain but it is unclear whether or not it can be combined with physiotherapy exercise classes, how many lessons are needed and what mechanisms might be involved.

Objectives
To undertake a feasibility trial of the Alexander technique and supervised exercise classes and perform exploratory biomechanical and neuromuscular physiological marker analyses to better understand mediators of recovery.

Design
Feasibility parallel-group randomised controlled trial.

Setting
General practices in southern England.

Participants
Patients with recurrent back pain (at least 3 weeks’ duration of a current episode).

Interventions
Participants were allocated by an external randomisation line to four groups: (1) normal care, (2) 10 Alexander technique lessons, (3) 12 physiotherapy exercise classes, (4) Alexander technique lessons plus exercise classes.

Main outcome measures
The feasibility outcomes were recruitment, acceptability and follow-up. The clinical outcomes were the Roland–Morris Disability Questionnaire (RMDQ), days in pain, the Von Korff pain and disability scale, overall improvement, fear of activity, enablement, the Oswestry Disability Index and the Aberdeen pain and function scale. Laboratory-based markers were axial muscle tone and flexibility, electrical activity, muscle tone and mechanical properties of elasticity and stiffness, trunk strength, and proprioception.

Results
In total, 83 patients consented, 69 were randomised and 56 (81%) were followed up at 6 months. Most patients had long-standing pain (median > 300 days of pain). The RMDQ and other instruments were sensitive to change and the preliminary evidence suggests that the Aberdeen scale could be a useful measure. Study procedures were feasible and three methods of recruitment were successfully piloted. At 6 months the control group had changed little (RMDQ 1 point lower than at baseline) and, compared with the control group, clinically important improvements in RMDQ were seen in all groups, albeit not significant [Alexander technique −3.0, 95% confidence interval (CI) −6.7 to 0.8]; exercise classes −2.9, 95% CI −6.5 to 0.8; combined Alexander technique + exercise classes −2.50, 9% CI −6.20 to 1.19]. Novel biomechanical variables strongly associated with RMDQ score at 6 months were muscle tone (0.94 increase in RMDQ per unit increase in Hz, 95% CI 0.48 to 1.40; p < 0.0001), lumbar proprioception (1.48 increase in RMDQ per degree, 95% CI 0.83 to 2.12; p < 0.0001) and muscle elasticity (−4.86 increase in RMDQ per unit log decrement, 95% CI −0.01 to −9.72; p < 0.05). At 3 months the Alexander technique improved proprioception and exercise classes improved trunk extension strength. At 6 months the Alexander technique improved the timing of multifidus muscle onset and the active straight leg raise test and exercise classes improved multifidus muscle thickness and the ability to contract. The combined effects of the Alexander technique and exercise classes were improvements in muscle tone, elasticity and thickness and contractile ability. These observations provide possible links between intervention, proprioception, muscle tone and elasticity and outcome. In terms of harms, one patient fell in the exercise class group.

The trial is feasible and the interventions may provide clinically important benefits. Exploratory analysis suggests that muscle tone, elasticity and proprioception are strongly associated with improved RMDQ score and are likely to be modified by the interventions.