Background: Stroke is one of the serious health problems worldwide with dramatic consequences in various domains. Stroke survivors experience impairments, activity limitations, and restrictions of participation, which associated with environmental and personal factors lead to an inactive lifestyle. Consequently, stroke survivors are exposed to a risk of recurrent stroke and other cardiovascular diseases, thus creating a vicious cycle. The use of physical activity (PA) as a therapeutic strategy to maximize functional recovery in the rehabilitation of stroke survivors has a growing evidence. It is therefore important to better understand factors that are associated with physical activity level in stroke survivors, in order to define interventions that may be more appropriate.

Purpose: To investigate how the model including self-efficacy, activity limitations, locomotion ability and social participation predict PA in stroke survivors.

Methods: 60 individuals with stroke (44 males; mean age: 56.7 ± 10.4 years old; median/interquartile time since stroke: 14.12 / 40.93 months) were recruited in Benin, a West-African francophone country. PA was measured by an African Francophone version of the International Physical Activity Questionnaire (IPAQ-AF) and by the Garmin Forerunner 15, a wrist-worn activity monitoring device. Self-efficacy, activity limitations, locomotion ability, and social participation were measured respectively by Stroke Self-Efficacy Questionnaire (SSEQ), ACTIVLIM-Stroke, ABILOCO-Benin, and the Reintegration to Normal Living Index (RNLI). SSEQ is a specific questionnaire that measures individual confidence in functional performance after stroke. ACTIVLIM-Stroke and ABILOCO-Benin are two Rasch-built questionnaires previously validated in Benin for stroke-related activity limitations and locomotion ability, respectively. RNLI is a generic questionnaire that measures the degree to which individuals with traumatic and neurological conditions achieve reintegration to normal social activities. We performed a standard multiple regression analysis using IPAQ-AF scores and step counts from Garmin Forerunner 15 as dependent variables. Independent variables were ACTIVLIM-Stroke, ABILOCO-Benin, SSEQ and RNLI scores.

Results: The model including self-efficacy, activity limitations, locomotion ability and social participation explained 46% of the variance in IPAQ-AF scores (R² = 0.46 p < 0.001) and 48% of the variance in step counts (R² = 0.48 p < 0.001). Only activity limitations presented a significant unique contribution to the prediction of PA scores (ß = 0.66, p < 0.001 for IPAQ-AF and ß = 0.46, p = 0.006 for step counts).

Conclusion: Self-efficacy, social participation, activity limitations and locomotion ability together explain 46 to 48% of the variance in PA scores in stroke survivors with 46 to 66% of a unique contribution of activity limitations dimension.

Implications: The findings suggest that self-efficacy, social participation, activity limitations, and locomotion ability are among the most important factors to consider in increasing physical activity level in stroke survivors. Particularly, the regression of activity limitations must be targeted during rehabilitation care.

Keywords: Stroke survivors; Physical activity; Predictors.

Ethics approval: N°2017-581 of the Research Ethics Committee of Centre Intégré Universitaire de Santé et de Services Sociaux de la Capitale Nationale (CIUSSS-CN) Quebec (Qc) Canada, and ethics approval from local participating rehabilitation centers.