The PRISMA Statement for Reporting Systematic Reviews and Meta-Analyses was used to guide this review. We developed a search strategy to identify relevant studies with the assistance of a reference librarian. The strategy combined terms related to smoking (eg, tobacco smoke pollution, environmental tobacco smoke, passive smoke, secondhand smoke, involuntary smoke, smokefree) with terms related to pregnant women and their families (eg, pregnant women, pregnancy, mother, maternal, fetus, infant, newborn, prenatal, antenatal). The search strategy was kept broad to retrieve as many relevant articles as possible (see online supplementary appendix 1 for the MEDLINE search strategy).

Databases were searched for items published from 1990 to August 2012. A later series of searches updated the retrieval through January 2013. There were no language restrictions.

Systematic searches were conducted of the following databases: MEDLINE (Ovid), MEDLINE In-process & Other Non-Indexed Citations (Ovid), Embase (Ovid), PsycInfo (Ovid), Global Health (Ovid), CINAHL (EbscoHost), ASSIA (Applied Social Sciences Index and Abstracts) (ProQuest), Web of Knowledge and Cochrane Library. We searched and reviewed items from WHO Global Health Library, WHO International Clinical Trials Registry, Clinicaltrials.gov, Trip Database, Scirus and Google Scholar. We also reviewed the reference lists of selected articles retrieved by the searches to identify additional articles of interest.

An article was considered for inclusion in the literature review if it satisfied the Population, Intervention, Comparison, Outcome (PICO) criteria: (1) population: non-smoking pregnant women exposed to SHS, (2) intervention: clinical interventions to reduce exposure to SHS at home, (3) comparison: no intervention to reduce exposure to SHS at home and (4) outcome: either reduced exposure of non-smoking pregnant women to SHS at home during pregnancy or quit rates among smoking partners during the pregnancy of the woman. The PICO criteria were developed a priori based on consultation from a WHO guidelines development group and external stakeholders at two international meetings of tobacco control and reproductive health experts. Clinical interventions, again, included both psychosocial and pharmaceutical interventions delivered in an antenatal care setting. Studies were included if the comparison was reported as standard of care which may or may not include some advice about SHS, and information on whether the comparison group included SHS education or advice was noted in the review. All study designs were considered in this review if they met the eligibility criteria. If studies reported birth outcomes, we included that information in this report, but did not exclude studies that do not report on birth outcomes. Studies were excluded if the intervention was not focused on reducing SHS exposure among non-smoking women during pregnancy or if the outcomes were not measured during or at the end of pregnancy.

A total of six reviewers assisted in screening abstracts, but for each abstract, at least two individuals independently screened titles and abstracts to identify articles eligible for full-text review. English language abstracts were screened, but we did not restrict language of full papers. Discrepancies in selection were resolved via discussion with first and second authors of this review, and a final list of eligible studies was decided by consensus. Next, full articles were read by two reviewers to confirm inclusion of each article for the final study. One article was published in Chinese, and translation assistance was provided by a scientific collaborator who had expertise in tobacco control. Relevant information was abstracted from each study and synthesised in narrative form.

Two reviewers assessed the quality of each study by adapting a published set of criteria, based on study design and internal validity, developed by the US Preventive Services Task Force.¹⁷ Studies were given a grade for research design (I=randomised controlled trials (RCT); II-1=well-designed controlled trial without randomisation; and II-2=well-designed cohort or case-control study) and a separate grade for internal validity (good, fair or poor). For RCTs, internal validity was based on the following six criteria: adequate randomisation, low attrition and high adherence, low differential or total loss to follow-up, clear definition of intervention, high reliability and validity of exposure and outcome measures, and inclusion of an intent-to-treat analysis. ‘Good’ studies met ≥5 of the six criteria, and ‘fair’ studies met <5 of the criteria. In addition, given the high risk of bias of self-reported measures in cessation trials among pregnant women¹⁸¹⁹ and the low reliability of self-reported SHS exposure,²⁰ studies that used self-reported measures of SHS or partner quitting were automatically considered of ‘poor’ quality. A meta-analysis was not conducted because of the limited number of studies that met our inclusion criteria and the diversity of SHS exposure outcomes the studies assessed. Only one study assessed quitting among partners of pregnant women.

Table 1 provides summary characteristics of the five included studies, and online supplementary table S1 provides detailed information abstracted from the studies. Four interventions focused on reducing SHS exposure at home among nonsmoking pregnant women, and one intervention focused on promoting quitting among partners of pregnant women, including both smokers and non-smokers. Study locations were Washington, DC, USA; Isfahan, Iran; Guangzhou and Sichuan, China (two studies); and Brisbane, Australia. All five were RCTs (one stratified RCT), and sample sizes ranged from 91 to 758. In all five studies, pregnant women or their husbands were recruited during antenatal care. The US study recruited African American women who reported SHS exposure,²³ and the studies in Iran and China recruited married and literate women whose husbands were smokers.²⁴²⁵²⁷ In Australia, after having received permission from the pregnant women during prenatal care, male partners who smoked ≥10 cigarettes a day were recruited.²⁶

In summary, four of the studies involved psychosocial interventions with various forms of counselling interventions delivered to pregnant women within the antenatal care setting, and the fifth study involved psychosocial intervention plus medication to partners of pregnant women. Among these studies, all provided information on the harms of SHS and made follow-up contact with participants. One intervention included negotiation skills for pregnant women,²³ and two encouraged implementing smoke-free home rules.²³²⁷ All interventions promoted partners or household members to quit smoking; however, only one provided direct assistance to partners²⁶ and one provided educational materials to pregnant women targeted for household members.²⁴ One was high intensity,²³ three were medium²⁴²⁶²⁷ and one was low.²⁵ Interventions are described in more detail.

In the US study, the intervention was delivered over eight sessions during prenatal care and two sessions during postpartum visits, each session lasting an average of 30–45 min.²³ Several publications were found based on this single trial, but one paper (included in this review) analysed a subsample of non-smoking pregnant women who self-reported SHS exposure. Cognitive behavioural strategies to eliminate or minimise exposure to SHS were employed by trained counsellors, including role play, skills practice, building negotiation skills with partners and household members who smoked, education about the risk of SHS during pregnancy, and encouragement of household smoking bans.

In Guangzhou, China, an obstetrician gave brief advice on health risks of exposure to tobacco smoke (2–3 min in duration) to pregnant women who reported SHS exposure in the home from their husbands.²⁵ The advice included a description of health risks related to SHS exposure and the importance of avoiding exposure. Women were also encouraged to help their husbands stop smoking. In addition, women received an educational booklet describing the risks of SHS exposure and strategies for helping their husbands to stop smoking. In subsequent prenatal visits, women received brief reminders (1.5 min on average) of the importance of avoiding SHS and were encouraged to help their husbands to quit smoking.

In Sichuan, China, study participants in the intervention group were provided with educational materials about the harms of SHS, the importance of establishing smoke-free families and ways to avoid SHS exposure.²⁷ Women participated in seminars, role playing, watching videos and entered in contests to test their understanding of the self-education materials. In addition, they were provided counselling by an obstetrician, followed by monthly phone calls, and were provided access to a telephone hotline for support.

In Isfahan, Iran, trained midwives delivered one-on-one education on SHS risks to women for 15–20 min at the first prenatal visit and for 5–10 min at the second prenatal visit.²⁴ Women were also given educational materials which provided information on the health risks of SHS for the fetus, a picture of a low birthweight infant and how the toxic substances from SHS can cross the placenta to the fetus, and a resource booklet to use at home which used simple terms and pictures to communicate knowledge.

In Australia, both counselling and use of a nicotine patch were used to help partners of pregnant women to quit smoking; thus, the outcome of this study was not measured in the pregnant women.²⁶ The intervention included one partner counselling session with a general practitioner by telephone, which included a history and assessment of smoking status and dependence, an explanation of the use of the nicotine patch and recommended a 1-week supply of patches plus a prescription for an additional 3-week supply. After the counselling session, the 1-week supply of nicotine patches and intervention materials were mailed to partners: (1) an 18 min video with a nationally-known athlete becoming a father and SHS risks for the newborn; (2) information on how to use the nicotine patch; (3) a booklet on quitting; and (4) a letter written for the participant’s general practitioner explaining the study. After 1 week, a newsletter was sent on tips on quitting and motivational anecdotes.

Results from all five studies showed positive findings based on study-defined outcome measures. However, it should be noted that the outcomes were measured in the partners of pregnant women in one study, and of the four studies that measured the outcomes among pregnant women, there were variations in how reduction in SHS exposure were defined and the length of time the outcomes were measured. Furthermore, three of the five studies were based on self-report of SHS exposure and were not biochemically-validated, and thus were judged as poor quality. A meta-analysis was not performed because of the diversity of SHS outcomes; SHS outcomes included women’s report of the quit status of the husbands, report of cigarettes the husbands smoked near the women, and exposure to cigarette smoking in home, same room, or car with a smoker.

Of the two studies that included biochemical validation, one tested nicotine in maternal hair samples²⁷ and the other tested for carbon monoxide (CO) in expired air of a subsample of partners who reported quitting.²⁶ However, both studies had additional risks of bias that should be considered when interpreting the outcomes measures.

In the Sichuan, China, study, the provision of educational materials about SHS and of counselling by an obstetrician was associated with decreased mean hair nicotine concentration in the intervention group compared with the control group: 0.3 log micro g/g at follow-up compared with 0.5 at baseline; and for control: 0.5 log micro g/g at follow-up compared with 0.4 at baseline.²⁷ This difference was reported as statistically significant at p<0.05. The outcome was measured approximately 1 month prior to delivery.

In the Australian study, 48 out of 291 (16.5%) men in the intervention group reported 6-month quitting compared with 25 out of 270 (9.3%) men in the control group (p=0.011, OR=0.52, 95% CI 0.31 to 0.86).²⁶ Of 24 self-reported quitters in the intervention group with CO testing, 23 (95.8%) were verified as having quit; of 21 self-reported quitters in control group, 14 (66.7%) were verified as having quit; seven refused CO testing and were assumed to be continuing smokers. Quit rate of partners of non-smoking pregnant women was not provided; however, the authors note that quit rates of the partners did not vary by pregnant women’s smoking status.

There were three studies that used self-reported SHS outcomes, and the definition varied by study. In the US study, which used the most intensive of all included interventions (counselling addressed multiple risk factors and lasted about 30–45 min), SHS exposure was defined from women’s report of exposure to cigarette smoking in the home, same room or car with a smoker during a typical week, and this outcome was measured during the second or third trimester of pregnancy. Women in the intervention group were less likely to self-report SHS exposure than women in the control group (OR 0.57, 95% CI 0.38 to 0.84).²³ No differences were found in low birth weight, neonatal hospitalisation or preterm delivery <37 weeks between infants of mothers in the intervention and control groups (p>0.05). They found only differences in preterm delivery at <34 weeks gestation were reported between treatment arms; however, pregnancy outcome data were not available for all participants. In a later published study, when missing pregnancy outcomes were imputed, no difference was detected in any pregnancy outcomes between treatment arms.³¹

In the briefest of interventions, conducted in Guangzhou, China, women in the intervention group were more likely than those in the control group to report their husbands not smoking in the past 7 days (8.4% vs 4.8%, p=0.04); however, no difference was found in their husbands not smoking in the last 30 days (6.1% vs 4.2%, p=0.26).²⁵ Women in the intervention group reported that their husbands were more likely to have attempted to quit (30.0% vs 22.2%, p=0.02) and to have decreased the number of cigarettes smoked daily (39.7% vs 17.7%, p<0.0001) compared with women in the control group.

In the study conducted in Isfahan, Iran, self-reported weekly SHS exposure was measured by the mean number of cigarettes smoked per week by the husband at home and near the pregnant woman. Self-reported weekly SHS exposure was statistically lower in the intervention group compared with the control group at third, fourth and fifth prenatal care sessions compared with the initial session, p<0.001 (eg, at fifth visit, 12.3 vs 25.4 weekly mean number of cigarettes the husband smoked near the pregnant woman).²⁴

The strengths of the research were that all studies had random assignment to intervention and control groups, and three of the five studies used an intent-to-treat analysis (it is assumed that all study participants lost to follow-up continued to be exposed to SHS or partner continued to smoke). Three of the five studies had outcome measures based only on self-report and so were judged to be of poor quality.²³²⁴²⁵

The remaining two studies were considered to be of fair quality: the Australian study and the Sichuan, China, study.²⁶²⁷ The Australian study did not report data on adherence to intervention and lacked sufficient detail on the randomisation process, and there were significant differences in baseline characteristics between intervention and control arms. Additionally, CO testing was only completed on a subsample of quitters, and CO is sensitive for only recent smoking within the past day.⁴⁵ Though the biochemical verification data suggest that quitters in the control group were more likely to incorrectly report their quit status, it was unclear how the subsample was identified and whether potential biases exist. The Yang et al study also did not report data on adherence to intervention and did not conduct an intention-to-treat analysis; loss to follow-up was approximately 14%. Nicotine accumulation in hair is less affected by daily variability of exposure, and 1 cm of hair near the scalp can indicate exposure in the past month; however, it is unknown what effect hair treatment, hair colour and growth rate has on nicotine levels in hair.⁴⁶