Open Access Open Badges Research article

Smoking in context – a multilevel approach to smoking among females in Helsinki

Sakari Karvonen1*, Petteri Sipilä2, Pekka Martikainen2, Ossi Rahkonen3 and Mikko Laaksonen3

Author Affiliations

1 STAKES, National Research and Development Centre for Welfare and Health, Box 220, 00531 Helsinki, Finland

2 Department of Sociology, Box 18, 00014 University of Helsinki, Finland

3 Department of Public Health, Box 41, 00014 University of Helsinki, Finland

For all author emails, please log on.

BMC Public Health 2008, 8:134  doi:10.1186/1471-2458-8-134

Published: 24 April 2008



Smoking is associated with disadvantage. As people with lower social status reside in less privileged areas, the extent of contextual influences for smoking remains unclear. The aims were to examine the spatial patterning of daily smoking within the city of Helsinki, to analyse whether contextual variation can be observed and which spatial factors associate with current daily smoking in the employed female population.


Data from a cross-sectional questionnaire were collected for municipal employees of Helsinki (aged 40–60 years). The response rate was 69%. As almost 4/5 of the employees are females, the analyses were restricted to women (n = 5028). Measures included smoking status, individual level socio-demographic characteristics (age, occupational social class, education, family type) and statistical data describing areas in terms of social structure (unemployment rate, proportion of manual workers) and social cohesion (proportions of single parents and single households). Logistic multilevel analysis was used to analyse data.


After adjusting for the individual-level composition, smoking was significantly more prevalent according to all social structural and social cohesion indicators apart from the proportion of manual workers. For example, high unemployment in the area of domicile increased the risk of smoking by almost a half. The largest observed area difference in smoking – 8 percentage points – was found according to the proportion of single households.


The large variation in smoking rates between areas appears mainly to result from variation in the characteristics of residents within areas. Yet, living in an area with a high level of unemployment appears to be an additional risk for smoking that cannot be fully accounted for by individual level characteristics even in a cohort of female municipal employees.