Background: Tobacco smoking in pregnancy causes serious health problems for the developing fetus and mother. When used by non-pregnant smokers, pharmacotherapies (nicotine replacement therapy (NRT), bupropion, and varenicline) are effective for increasing smoking cessation, however their efficacy and safety in pregnancy remains unknown. Electronic cigarettes (ECs) are becoming widely used, but their efficacy and safety when used for smoking cessation in pregnancy are also unknown.
Objectives: To determine the efficacy and safety of smoking cessation pharmacotherapies and ECs used during pregnancy for smoking cessation in later pregnancy and after childbirth, and to determine adherence to smoking cessation pharmacotherapies and ECs for smoking cessation during pregnancy.
Search methods: We searched the Cochrane Pregnancy and Childbirth Group's Trials Register (20 May 2019), trial registers, and grey literature, and checked references of retrieved studies.
Selection criteria: Randomised controlled trials (RCTs) conducted in pregnant women, comparing smoking cessation pharmacotherapy or EC use with either placebo or no pharmacotherapy/EC control. We excluded quasi-randomised, cross-over, and within-participant designs, and RCTs with additional intervention components not matched between trial arms.
Data collection and analysis: We followed standard Cochrane methods. The primary efficacy outcome was smoking cessation in later pregnancy; safety was assessed by 11 outcomes (principally birth outcomes) that indicated neonatal and infant well-being. We also collated data on adherence to trial treatments. We calculated the risk ratio (RR) or mean difference (MD) and the 95% confidence intervals (CI) for each outcome for each study, where possible. We grouped eligible studies according to the type of comparison. We carried out meta-analyses where appropriate.
Main results: We included 11 trials that enrolled a total of 2412 pregnant women who smoked at enrolment, nine trials of NRT and two trials of bupropion as adjuncts to behavioural support, with comparable behavioural support provided in the control arms. No trials investigated varenicline or ECs. We assessed four trials as at low risk of bias overall. The overall certainty of the evidence was low across outcomes and comparisons as assessed using GRADE, with reductions in confidence due to risk of bias, imprecision, and inconsistency. Compared to placebo and non-placebo (behavioural support only) controls, there was low-certainty evidence that NRT increased the likelihood of smoking abstinence in later pregnancy (RR 1.37, 95% CI 1.08 to 1.74; I² = 34%, 9 studies, 2336 women). However, in subgroup analysis by comparator type, there was a subgroup difference between placebo-controlled and non-placebo controlled RCTs (test for subgroup differences P = 0.008). There was unclear evidence of an effect in placebo-controlled RCTs (RR 1.21, 95% CI 0.95 to 1.55; I² = 0%, 6 studies, 2063 women), whereas non-placebo-controlled trials showed clearer evidence of a benefit (RR 8.55, 95% CI 2.05 to 35.71; I² = 0%, 3 studies, 273 women). An additional subgroup analysis in which studies were grouped by the type of NRT used found no difference in the effectiveness of NRT in those using patches or fast-acting NRT (test for subgroup differences P = 0.08). There was no evidence of a difference between NRT and control groups in rates of miscarriage, stillbirth, premature birth, birthweight, low birthweight, admissions to neonatal intensive care, caesarean section, congenital abnormalities, or neonatal death. In one study infants born to women who had been randomised to NRT had higher rates of 'survival without developmental impairment' at two years of age compared to the placebo group. Non-serious adverse effects observed with NRT included headache, nausea, and local reactions (e.g. skin irritation from patches or foul taste from gum), but data could not be pooled. Adherence to NRT treatment regimens was generally low. We identified low-certainty evidence that there was no difference in smoking abstinence rates observed in later pregnancy in women using bupropion when compared to placebo control (RR 0.74, 95% CI 0.21 to 2.64; I² = 0%, 2 studies, 76 women). Evidence investigating the safety outcomes of bupropion use was sparse, but the existing evidence showed no difference between the bupropion and control group.
Authors' conclusions: NRT used for smoking cessation in pregnancy may increase smoking cessation rates in late pregnancy. However, this evidence is of low certainty, as the effect was not evident when potentially biased, non-placebo-controlled RCTs were excluded from the analysis. Future studies may therefore change this conclusion. We found no evidence that NRT has either positive or negative impacts on birth outcomes; however, the evidence for some of these outcomes was also judged to be of low certainty due to imprecision and inconsistency. We found no evidence that bupropion may be an effective aid for smoking cessation during pregnancy, and there was little evidence evaluating its safety in this population. Further research evidence on the efficacy and safety of pharmacotherapy and EC use for smoking cessation in pregnancy is needed, ideally from placebo-controlled RCTs that achieve higher adherence rates and that monitor infants' outcomes into childhood. Future RCTs of NRT should investigate higher doses than those tested in the studies included in this review.
Copyright © 2020 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.