Effect of investigator observation on gait parameters in individuals with stroke.

Improvements in gait speed following various training paradigms applied to patients post-stroke does not always lead to changes in walking performance, defined as gains in daily stepping activity. We hypothesized that testing conditions, specifically the presence of an observer, influences patient behaviors and resultant outcomes may overestimate their true walking capacity. This potential Hawthorne effect on spatiotemporal and biomechanical measures of locomotor function in individuals post-stroke has not been assessed previously. Fifteen ambulatory individuals with chronic stroke wore instrumented insoles and performed two separate normal-pace walking assessments, including unobserved conditions during which participants were unattended and unaware of data collection, and observed conditions with an investigator present. Gait analysis was conducted outside of a laboratory setting using instrumented insoles equipped with a 3D accelerometer and pressure sensors which captured the spatiotemporal kinematics, vertical ground reaction forces and foot acceleration. Data were compared using paired comparisons, with subsequent correlation and stepwise regression analyses to explore potential associations between Hawthorne-induced changes in walking strategies, gait speed and locomotor performance (daily stepping). Except for cadence, other measures of spatiotemporal parameters and swing kinematics (acceleration) were not significantly different between observed vs unobserved conditions. However, analyses of ground reaction forces revealed significantly greater paretic limb loading (Δ1 peak = 1.5 ± 1.6 N/kg Δ2 peak = 1.4 ± 1.8 N/kg; p < 0.01) and increases in weight bearing symmetry (11-24%, p < 0.01) during observed vs unobserved conditions. This potential Hawthorne effect was greater in those with slower walking speeds and shorter stride lengths but was not related to daily stepping. The present findings suggest that biomechanical parameters of walking function may be related to the presence of an observer and highlight the need to separately measure locomotor capacity (gait speed) and performance (daily stepping).