Background:
International evidence suggests local tobacco retailing is associated with individual smoking behaviours. However, most evidence is cross-sectional. This longitudinal study examines the effects of exposure to high densities of tobacco retailers on maternal smoking during pregnancy.

Methods:
Scottish maternity records (2000-2015) were extracted. Information included maternal smoking and postcode of residence at delivery which enabled linkage to a GIS-based measure of residential tobacco retail density. Because the full population of births was available (n=c.750,000), it was possible to capture multiple births to the same mother and hence to examine the extent to which changes in smoking behaviour between pregnancies was associated with changes in tobacco retail density exposure through residential moves. A neighbourhood measure of “social norms” (maternal smoking rates) was also linked. A case crossover approach was adopted which treats each mother as their own control and removes all confounding by time-invariant characteristics of the mother.

Results:
Fixed-effects analysis suggests the risks of smoking during different pregnancies to the same mother are around 40% higher (OR 1.37, CI 1.15, 1.63) for those living in the highest density tobacco retailing areas compared to zero density areas. Including the local social norms variable slightly strengthened the observed effects with an excess risk of 39% in the highest density quintile (OR 1.39, CI 1.17,1.66).

Conclusions:
This longitudinal study show the significant role of the tobacco retailer environment in influencing smoking during pregnancy. Findings emphasise the need to target the supply of tobacco products as a means of reducing smoking rates.