"Perceived neighborhood walkability" and physical activity in four urban settings in South Africa

Abstract

Introduction. In Africa, studies on the associations between the perceived neighbourhood walkability and physical activity, particularly, by socio-economic status (SES) remain scarce. This study explores these associations by validating the Neighbourhood Environmental Walkability Scale (NEWS-Africa) in an urban setting of South Africa to gain a better understanding of the construct of neighbourhood “walkability”. Methods. A convient sample of residents from four suburbs in urban metropole (n=52, 18-65yr, 81% women) in the Western Cape Province, South Africa (viz. Langa, Khayelitsha, Pinelands and Table View) were recruited through invitations following community gatherings and church services. Measures were obtained on perceived neighbourhood walkability, self-reported and measured physical activity and socio-economic status. Langa and Khayelitsha represented two primarily low-SES townships, whereas Pinelands and Table View represented suburbs of a higher-SES. Participants completed the 76-item (13 subscales) NEWS-Africa survey by structured interviews and reported weekly minutes of walking for transport and recreation using items from the International Physical Activity Questionnaire. Objective data on physical activity was collected using accelerometers, and ground-truthing was used to assess the neighbourhood environment using global information systems (GIS) in a 1000m buffer around each geocoded household. The research was carried out in three parts: 1) Evaluating the reliability and construct validity of the NEWS-Africa instrument between the two-SES groups. 2) Examining some of the walkability constructs and subscales of the NEWS-Africa instrument using GIS and ground-truthing, and the extent to which the SES of communities influenced these associations. 3) Examining the differences in self-reported physical activity (domains), measured physical activity (MVPA) when groups are divided according to SES, GIS walkability (1000m buffers) and if the data support the notion of utilitarian walking in low SES groups, irrespective of the built environment attributes. Results. For the combined-SES groups, the test-retest reliability indicated a good reliability with 10 out of the 13 scales of the NEWS-Africa being significantly and positively correlated. The Spearman’s correlations ranged from (rs = -0.43, p=0.00 to rs = 0.79, p=0.00). For construct validity of the NEWS -Africa instrument against self-reported physical activity, only three scales were related to walking for transport: Neighbourhood surroundings scale (rs= -0.34, p=0.01), Safety from Traffic scale (rs =0.34, p< 0.05) and people in the low-SES and combined SES perceived public bus/ train stops to be nearer than they actually were (rs =-0.50, P< 0.05). Of the 13 scales of the NEWS-Africa questionnaire, 6 were significantly correlated to GIS-measured walkability index parameters. The Roads and walking paths scale was positively associated with GIS-measured walkability (rs = 0.3), and the Stranger danger scale was negatively associated with GIS-measured walkability (rs = -0.4). When we considered GIS-measured Land use mix, 3 of the NEWS- Africa scales were correlated (For the entire sample, the scales including Places for walking, cycling and playing overall scale (rs = 0.3), and Neighbourhood surroundings scale (rs = 0.3), were positively associated respectively). Conversely, Stranger danger scale was inversely correlated (rs = -0.6). Intersection density measured with GIS was significantly and positively associated with the Roads and walking paths scale for all groups combined (rs = 0.3). For GIS-measured walkability, self-report physical and measured physical activity, there were no associations in any of the domains for self-reported physical activity within the 1000m buffer for all groups. However, for the objectively measured physical activity in the 1000m buffer, vigorous physical activity (rs = -0.39) was inversely associated with intersection density in the low-SES and moderate (rs = -0.29) and total MVPA (rs = -0.31) were inversely associated with Intersection density in the high SES. Conclusions: The overall results of the current study across all chapters generally show a mismatch between the perceived and objectively-assessed built environment, particularly in low-income communities. Furthermore, in low-SES communities, we failed to show the expected relationships between attributes of the built environment and physical activity, suggesting that physical activity in these communities is more utilitarian in nature, and as such, may not be as influenced by aspect of the built environment. In summary, the data suggest that the environment (including crime rates, poor access to physical activity facilities and public transportation predominantly made by buses) has less of an association with physical activity in LMICs and more disadvantaged communities, where physical activity is used for utilitarian, rather than recreational purposes. This study stemmed from the need to broaden research on the relationship between the built environment and physical activity, considering walkability constructs. These findings also suggest that the definition of the construct of walkability be re-examined, in relation to low SES settings.