Objective Most women who stop smoking during their pregnancy will relapse within the first 6 months after birth, with health risks to the mother and family. There have been few UK-specific studies identifying factors associated with relapse. Such information would help health care providers identify those most at risk so that extra support and resources can be targeted. We examined whether a small set of factors that are potentially readily measured or observable in routine health care settings were associated with women at risk for relapse postpartum.

Design and setting Analysis of the first wave of the Millennium Cohort Study. Variables in the final model were selected using backwards logistic regression weighted to account for the complex survey design.

Participants 2353 natural mothers who reported quitting smoking during their pregnancy.

Main outcome measure Postpartum smoking relapse.

Results 58% of mothers who quit during pregnancy were smoking again at 9 months. Only 16.2% of smoking couples quit together. The strongest risk factors were related to postnatal partner smoking status and single parenthood. Compared to mothers with a non-smoking partner, the risk of relapse was higher for those with a partner who smoked at 9 months (OR 3.2, 95% CI 2.6 to 3.9), a partner who smoked antenatally (1.2, 1.4 to 2.1), and those who had no live-in partner postnatally (3.2, 2.2 to 4.8) or antenatally (2.3, 1.5 to 3.4). Other significant risk factors remaining in the models were the mother having no-one to share her feelings with, drinking 1–2 times a month or 1–2 times a week, having fewer educational qualifications and more children in the household. Factors that did not appear to be influential in the final models were age, ethnicity, change in marital status from birth to 9 months, depression history, financial status, timing of entry into ante-natal care, feelings about the pregnancy or breastfeeding.

Conclusion Pragmatic variables can be used to identify women at higher risk for relapse. The validity of these findings are potentially limited by the inability to measure antenatal intent to relapse, the smoking cessation and relapse prevention interventions actually received, and timing/sequence of relapse for partner-mother pairs in this data set. Both partner smoking habits and single marital status are strongly associated with relapse and can be readily identified both antenatally and postnatally by healthcare providers.