A multilevel analysis of the influence of neighborhood- and individual-level socioeconomic factors on smoking among South African adults

Abstract

Background: In addition to the influence of an individual’s socioeconomic status, the neighbourhoods in which people live may influence health-related behaviours including smoking. This study therefore sought to determine the influence of the socioeconomic context in which South African adults lived on their smoking behaviour, and explore the potential gender differences of contextual influences. Method: This study involved a representative sample of South African adults (≥16 years) who participated in the 2010 (n=3,112) and 2011 (3,003) South African Social Attitude Survey (SASAS). The 2009 General Household Survey (n =25,548 households) was used to obtain the socioeconomic characteristics of the neighbourhoods where SASAS participants lived, including proportion of households with access to tap water, access to flush toilets and the level of employment in the area (3-item deprivation index; α=0.84). Information obtained from SASAS included participants’ tobacco use status and socio-demographic characteristics, including participants’ self-rated socioeconomic position within the society. Data analysis included a multi-level Poisson regression analysis. Results: Of the respondents who participated in the 2010/2011 survey, 19.4% (n=1302) were current smokers (30% men and 9.8% women). Smoking was more prevalent among those living in areas in the upper-third socioeconomic status (SES) than in areas in the lower-third SES (22.9% vs. 13.5%; p= 0.01). The neighbourhood socioeconomic context had a greater influence on the prevalence of smoking among women than among men. In particular, the gender gap in smoking prevalence was higher among those living in areas in the lowest-third SES (24.6% men vs. 4.6% women) than among those in areas of highest-third SEP (31.5% men vs. 15% women). Overall, smoking was less likely among those with greater than high school education than among those with less than high school education (OR=0.68; 95%CI=0.56-0.82). Conclusion: The findings suggest a greater neighbourhood socioeconomic contextual influence on women than men and highlight the need for community-level interventions targeting the least educated living in areas of highest socioeconomic position in South Africa. Interventions and public health policies to decrease tobacco smoking should be developed with some neighbourhood-specific modifications and should also be actively implemented.