Background The UK has the highest rates of teenage motherhood (TM) in Western Europe and unintended pregnancy risks considerable adverse health problems for teenagers and their infants as well as generating considerable emotional, economic and social costs for adolescents and society globally. The relationship to individual social and material disadvantage is established but the effects of area of residence are unclear. We test the following theories: that TM is a natural response to poor employment prospects; that residence in deprived areas or cities conveys an additional risk; and that social norms within areas are important.

Methods The Northern Ireland Longitudinal Study was used to identify a cohort of females who would be 15–19 between 2001 and 2011. Risk of TM was measured using logistic regression, adjusting for established risk factors (health, religion, family structure and socio-economic status). Settlement bands and Index of Multiple Deprivation signified urban/rural split and area deprivation; joblessness levels denoted labour market conditions, and historical TM levels, social norms. We assessed the additional explanatory power of each area factor in multilevel models, individually and then in combination. Changes in model fit were characterised using Akaike’s Information Criterion and likelihood ratio tests.

Results All the individual and household attributes were related to TM in the expected direction, e.g. the adjusted odds of TM amongst social renting women was 6.50 higher (95% CIs 4.64, 9.11) than their peers in the most expensive housing. Neither area deprivation nor labour market was independently related to TM. An urban/rural gradient was evident with TM lower in rural dwellers than city dwellers (OR_adj 0.61 [0.53, 0.69]); this was attenuated by the concentration of teenagers but not that of other age groups. Women living in the deciles with the highest historical levels of TM were about one and a half times as likely as those in the lowest areas to become a TM (OR_adj 1.45 [1.09, 1.92]).

Conclusion Teenage motherhood is independently though weakly associated with area of residence. The higher risk in cities may be in part due to greater opportunity given the higher concentration of teenagers, though other factors such as access to alcohol may also be important. The persistence of historical patterns suggests ingrained social norms which may be difficult to overcome.