Twenty post-stroke patients were assigned to one of three treatment conditions or to a control group to test whether exercises, determined in a previous study to recruit maximal extensor digitorum participation, would improve finger extension function over time. The exercises were resisted grasp, resisted extension, and ballistic extension. Improved function was defined as increased active range of motion, speed of reversal of movement, and ability to grasp and release cylinders. Significantly more subjects assigned to ballistic or resisted extension conditions improved in their ability to rapidly reverse movement over the course of treatment as opposed to those assigned to resisted grasp or control conditions. However, Kruskal-Wallis nonparametric analyses of variance indicated that no exercise improved all three components of function significantly more than another or the control condition.

Since no clear difference was found between the control and treatment conditions, it was concluded that motor unit recruitment as an attribute of activity is insufficient to improve function in post-stroke patients. The strength of this conclusion is limited by a chance imbalance of patient assignment in which significantly more patients assigned to resisted and ballistic extension conditions were at a higher level of recovery of motor control.