Seven studies (1,386 participants, range 30 to 250) were included in the review: five RCTs, one quasi-RCT and one prospective controlled observational study. All five RCTs scored 3 or more on the Jadad quality scale and the two non-randomised studies scored zero.
The mean abstinence rate at late pregnancy was 13% (95% CI 10.9 to 15.25) for intervention and control groups.
Pharmacotherapy had a significant effect on smoking cessation at the longest follow-up (RR 1.80, 95% CI 1.32 to 2.44; seven studies; Ι²=41.5%).
Results remained significant when analysed by type of study design. Non-RCTs (RR 3.25, 95% CI 1.65 to 6.39; two studies) reported a larger effect than RCTs (RR 1.48, 95% CI 1.04 to 2.09; five RCTs). Results remained significant for studies that used nicotine patches or bupropion, studies that used validated measures, studies with short-term follow-up (<12 weeks) and those without placebo groups. No significant treatment effects were found in studies that used nicotine gum, studies that used self reported measures, studies with long-term follow-up (>24 weeks) and those that compared treatment with placebo. There was evidence of statistical heterogeneity (Ι²=52.9%) for the analysis of self reported abstinence.
Some minor adverse events in both intervention and control groups were reported in three studies. There were no significant differences between groups for serious adverse events including maternal hospitalisation, low birthweight, spontaneous abortion, preterm birth, neonatal intensive care unit admission, placental abruption and foetal demise (two studies). Birth outcome data was reported for individual studies. Three of five studies found no significant between-group differences for mean birthweight rate, low birthweight rate, mean gestational age and preterm delivery. One study found lower preterm delivery rates and low birthweight rate for intervention group (nicotine gum).
There was evidence of publication bias.