The ATLET study: Can subjects with long-standing motor incomplete spinal cord injury learn to walk? A randomized clinical trial

Abstract

Body-weight supported locomotor training (BWSLT) has been used to improve walking function in subjects with incomplete Spinal Cord Injuries (SCI), but randomized controlled trials (RCTs) among subjects with long-standing (>2 years) incomplete SCI and poor walking function are scarce. This research aims to describe the methodology and report the results from two parallel, but independent, single-blinded RCTs.

Inclusion criteria were: incomplete SCI American Spinal Injury Association (ASIA) Impairment Scale grade C and D (AIS C and D), >2 years post injury, wheelchair dependent and BMI ≤30. Intervention groups received 60 days of BWSLT, manually assisted (study 1) or robot-assisted (study 2) with control groups receiving low-intensity usual care.

Compared to controls, the intervention groups experienced a significant increase in lower extremity muscle strength (LEMS), but no improvement in walking function, quality of life (QoL) or psychological outcomes.

Due to challenging recruitment, our study ended up recruiting persons who, at enrolment, were autonomously motivated subjects with high degree of physical activity and high expectations. This could possibly explain our null findings in most outcomes, possibly due to a “ceiling effect” where the participants already had attained close to their maximum potential.

Our study was underpowered as we were only able to recruit 44 (73%) of the planned 60 study subjects. Therefore, for training that starts several years post-injury, we can neither refute nor support the efficacy of BWSLT for subjects with long-standing incomplete SCI with poor or non-existing baseline walking function. However, our results do not exclude the possibility that such training can be useful in other subjects with SCI, i.e. those with SCI in subacute or early chronic phase, or among SCI subjects with some walking function at baseline. Such studies would be useful and, if initiated, we recommend that block randomization based on baseline function be utilized.