Effect of gait speed on post-stroke step length and anterior ground reaction force relative to individualized predicted values

Background. Training at faster gait speeds is recommended to improve activity limitations in adults with stroke. Walking faster can also increase step length and anterior ground reaction force (AGRF) relative to post-stroke habitual walking patterns. Recent work has developed a prediction equation that utilizes individual characteristics to predict a neurotypical or pre-stroke step length and AGRF value for an individual. However, it is unclear how the predicted individualized step length and AGRF values from this prediction equation relate to actual step length and AGRF values, and how the relationship between these values may change with gait speed. We aimed to understand the effect of gait speed on post-stroke step length and AGRF relative to a predicted individualized step length and AGRF value and to understand whether step length asymmetry direction (i.e., whether the paretic or non-paretic step was longer) impacted this relationship. We hypothesized that the difference between the predicted individualized values and actual values would decrease as speed increased for both step length and AGRF. Methods. Participants post-stroke walked on a treadmill at two or three speeds. We recorded AGRF and step length values. We then calculated an individualized predicted step length and AGRF value using a previously developed prediction equation. We fit two linear mixed effects models, the first with step length as the outcome and the second with AGRF as the outcome. Both models included fixed effects for gait speed, limb, and speed by limb interaction. We then repeated the same analyses, splitting the participants by those who took longer paretic steps and those who took longer non-paretic steps. Results. We found that walking faster did not impact the difference between actual and predicted step lengths, regardless of which limb took a longer step. Walking faster increased the difference between paretic AGRF and the predicted AGRF value, specifically for individuals who took longer paretic steps at their self-selected gait speed. Conclusions. Fast walking alone does not improve AGRF to the level of a predicted individualized value for individuals who took longer paretic steps. Fast walking may need to be paired with another intervention to increase paretic AGRF.