Effectiveness of smoking cessation pharmacotherapies during pregnancy: A multi-national population-based study

Abstract

Aims: As clinical trial evidence on the effectiveness of smoking cessation pharmacotherapies during pregnancy is inconclusive, we conducted a large cohort study examining their effectiveness and comparative effectiveness during pregnancy. Design: Population-based cohort study. We used propensity score matching and conditional Poisson regression to compare pharmacotherapy-exposed with unexposed pregnancies, and to compare varenicline-exposed with nicotine replacement therapy (NRT) -exposed pregnancies. Setting: Birth records (2005–2020) from New South Wales (NSW) Australia, New Zealand (NZ) and Norway/Sweden linked to pharmacotherapy dispensing records. Participants/cases: Women with a birth record indicating smoking in early pregnancy [during first 20 weeks' gestation (NSW), at lead maternity registration (NZ) or during the first trimester (Norway/Sweden)]. Participants were dispensed prescription NRT, varenicline or bupropion in the first 18 weeks of gestation (NSW, Norway/Sweden) or between the first antenatal visit and childbirth (NZ). Measurements: We defined smoking cessation as not smoking after gestational week 20 (NSW), at gestational week 32–36 (Norway/Sweden) and at two weeks postpartum (NZ), identified via self-report and documented in the birth record. Findings: Our NRT analyses included 623, 7074 and 70 exposed and 6026, 68 161 and 700 propensity-score-matched unexposed pregnancies from NSW, NZ and Norway/Sweden, respectively. The associations between NRT and smoking cessation were mixed but tended toward reduced cessation compared with no pharmacotherapy. In NSW, NRT was associated with a reduction in cessation [relative risk (RR) = 0.68, 95% confidence interval (CI) = 0.48–0.97], while in NZ, the effect was smaller (RR = 0.93, 95% CI = 0.89–0.98) but inconclusive in Norway/Sweden (RR = 0.91, 95% CI = 0.61–1.35). Smoking cessation was also equally or less common among varenicline-exposed pregnancies (NSW: 308 exposed vs 3077 matched unexposed, RR = 0.89, 95% CI = 0.72–1.09, Norway/Sweden: 196 exposed vs 1960 matched unexposed, RR = 0.49, 95% CI = 0.34–0.70). Our comparison of varenicline-exposed with NRT-exposed pregnancies indicated increased smoking cessation among varenicline-exposed pregnancies (NSW: 108 vs 154 exposed, RR = 1.91, 95% CI = 1.10–3.22). There were too few bupropion-exposed pregnancies to support interpretation. Conclusions: Varenicline appears to be more effective at smoking cessation than nicotine replacement therapy during pregnancy. The uncertainty about the real-world effectiveness of nicotine replacement therapy and bupropion remains as this study's analyses were impacted by non-adherence and biased by unmeasured confounding.