Functional strength training versus movement performance therapy for upper limb motor recovery early after stroke: a RCT

Abstract

Background
Not all stroke survivors respond to the same form of physical therapy in the same way early after stroke. The response is variable and a detailed understanding of the interaction between specific physical therapies and neural structure and function is needed.

Objectives
To determine if upper limb recovery is enhanced more by functional strength training (FST) than by movement performance therapy (MPT), to identify the differences in the neural correlates of response to (1) FST and (2) MPT and to determine whether or not pretreatment neural characteristics can predict recovery in response to (1) FST and (2) MPT.

Design
Randomised, controlled, observer-blind, multicentre trial with embedded explanatory investigations. An independent facility used computer-generated randomisation for participants’ group allocation.

Setting
In-patient rehabilitation, participants’ homes, university movement analysis facilities and NHS or university neuroimaging departments in the UK.

Participants
People who were between 2 and 60 days after stroke in the territory of the anterior cerebral circulation, with some voluntary muscle contraction in the more affected upper limb but not full function.

Interventions
Routine rehabilitation [conventional physical therapy (CPT)] plus either MPT or FST in equal doses during a 6-week intervention phase. FST was progressive resistive exercise provided during training of functional tasks. MPT was therapist ‘hands-on’ sensory input and guidance for production of smooth and accurate movement.

Main outcomes
Action Research Arm Test (ARAT) score for clinical efficacy. Neural measures were made of corticocortical [fractional anisotropy (FA) from corpus callosum midline], corticospinal connectivity (asymmetry of corticospinal tracts FA) and resting motor threshold of paretic biceps brachii (pBB) and extensor carpi radialis muscles (derived from transcranial magnetic stimulation).

Analysis
Change in ARAT scores were analysed using analysis of covariance models adjusted for baseline variables and randomisation strata. Correlation coefficients were calculated between change in neural measures and change in ARAT score per group and for the whole sample. An interaction term was calculated for each baseline neural measure and ARAT score change from baseline to outcome.

Results
A total of 288 participants were randomised [mean age 72.2 (standard deviation 12.5) years; mean ARAT score of 25.5 (18.2); n?=?283]. For the 240 participants with ARAT measurements at baseline and outcome, the mean change scores were FST?+?CPT?=?9.70 (11.72) and MPT?+?CPT?=?7.90 (9.18). The group difference did not reach statistical significance (least squares mean difference 1.35, 95% confidence interval –1.20 to 3.90; p?=?0.298). Correlations between ARAT change scores and baseline neural values ranged from –0.147 (p?=?0.385) for whole-sample corticospinal connectivity (n?=?37) to 0.199 (p?=?0.320) for MPT?+?CPT resting motor threshold pBB (n?=?27). No statistically significant interaction effects were found between baseline neural variables and change in ARAT score. There were no differences between groups in adverse events.

Limitations
The number of participants in the embedded explanatory investigation was lower than expected.

Conclusions
The small difference in upper limb improvement in response to FST and MPT did not reach statistical significance. Baseline neural measures neither correlated with upper limb recovery nor predicted therapy response.