The Evolution of Stroke Rehabilitation Randomized Controlled Trials

Abstract

Background

In the interest of prioritizing resources and providing future direction for researchers, a complete overview of the landscape of stroke rehabilitation literature was conducted.

Aim

We aimed to examine the evolution of stroke rehabilitation randomized controlled trials, with respect to number, sample size, and methodological quality between 1970 and September 2012.

Methods

Using the Evidence-Based Review of Stroke Rehabilitation (http://www.ebrsr.com), all randomized controlled trials related to stroke rehabilitation interventions were eligible for inclusion and were divided into five groups based on the primary outcome (i.e., motor, cognitive, medical complications, psychosocial, and ‘other’).

Results

One thousand sixty-three randomized controlled trials met inclusion criteria, with motor studies accounting for 58·8% of the total. The total number of randomized controlled trials grew between 1970 and 2012, with 35·2% of all the studies published in the last five-years. Motor randomized controlled trials had the smallest median sample size compared with cognitive (P < 0·018), medical complications (P < 0·001), psychosocial (P < 0·001), and ‘other’ (P < 0·001) randomized controlled trials. Between 1973 and 1977 and 2008 and 2012, there was no statistically significant increase in median sample sizes (P = 0·845). Psychosocial randomized controlled trials had higher median Physiotherapy Evidence Database scores when compared with motor (P = 0·002), cognitive (P = 0·035), and ‘other’ randomized controlled trials (P = 0·036), but not medical complication randomized controlled trials (P = 0·591). Over time, median Physiotherapy Evidence Database scores for all randomized controlled trials significantly increased from 5 (interquartile range 0·5) in 1973–1977 to 7 (interquartile range 3) in 2008–2012 (P = 0·008).

Conclusions

Randomized controlled trials in stroke rehabilitation have increased over the past four decades, with an associated increase in methodological quality, but not sample size.