Ground Truthing: Perceived Versus Objectively Measured Built Environment Attributes in South African Communities

Introduction :No ground truthing study of ‘neighbourhood walkability’ has been conducted in Africa that contrasts the lived experience of the neighbourhood-built environment against objectively measured attributes using GIS and subscales of the NEWS-Africa instrument. Methods: This is a cross-sectional survey and observation study. A convenient sample of residents (n=52) aged 18-65 years from four urban suburbs in low- and high-income settings, self-reported transport, leisure-time physical activity (PA) and device-measured moderate-to-vigorous PA (MVPA) were assessed. Built environment constructs was derived from individual-level street network measures (1000m buffer, using ArcGIS 10.51). We also assessed PA between the four groups based on income and GIS walkability. Results: There was no significant relationship between self-reported MVPA and GIS-measured walkability in the study sample. Only intersection density had a significant inversely association with moderate and MVPA (rho=-0.31, P<0.05) in the overall study sample. Self-report transport PA differed between groups (P<0.013). Device-measured vigorous activity differed between groups and was lower in both Low-SES groups compared to the High-SES/Low Walkable group (P<0.04). Total PA was not associated with GIS measured walkability, but inversely associated with positive perceived neighbourhood attributes. Conclusions: There is a modest inverse association between device-measured PA and GIS-measured intersection density. These results suggest that constructs of walkability, whether measured or perceived, may predict volitional and utilitarian PA differently in our study population.