Tobacco smoking is a major cause of morbidity and mortality, including during pregnancy. Although effective ways of promoting smoking cessation during pregnancy exist, the impact of these interventions has not been studied at a national level. We estimated the prevalence of smoking throughout pregnancy in the Netherlands and quantified associations of maternal smoking throughout pregnancy with socioeconomic, behavioural, and neonatal risk factors for infant health and development.
Data of five national surveys, containing records of 14,553 Dutch mothers and their offspring were analyzed. From 2001 to 2007, the overall rate of smoking throughout pregnancy fell by 42% (from 13.2% to 7.6%) mainly as a result of a decrease among highly educated women. In the lowest-educated group, the overall rate of smoking throughout pregnancy was six times as high as in the highest-educated group (18.7% versus 3.2%). Prenatal tobacco smoke exposure was associated with increased risk of extremely preterm (≤28 completed weeks) (OR 7.25; 95% CI 3.40 to 15.38) and small-for-gestational age (SGA) infants (OR 3.08; 95% CI 2.66 to 3.57). Smoking-attributable risk percents in the population (based on adjusted risk ratios) were estimated at 29% for extremely preterm births and at 17% for SGA outcomes. Infants of smokers were more likely to experience significant alcohol exposure in utero (OR 2.08; 95%CI 1.25 to 3.45) and formula feeding in early life (OR 1.91; 95% CI 1.69 to 2.16).
The rates of maternal smoking throughout pregnancy decreased significantly in the Netherlands from 2001 to 2007. If pregnant women were to cease tobacco use completely, an estimated 29% of extremely preterm births and 17% of SGA infants may be avoided annually.
Citation: Lanting CI, Buitendijk SE, Crone MR, Segaar D, Bennebroek Gravenhorst J, et al. (2009) Clustering of Socioeconomic, Behavioural, and Neonatal Risk Factors for Infant Health in Pregnant Smokers. PLoS ONE 4(12): e8363. doi:10.1371/journal.pone.0008363
Editor: Philippa Middleton, The University of Adelaide, Australia
Received: July 3, 2009; Accepted: November 19, 2009; Published: December 18, 2009
Copyright: © 2009 Lanting et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was financially supported by grants from the Netherlands Organization for Health Research and Development (ZonMW) (grant number 2500.0001), STIVORO - for a smokefree future (www.stivoro.nl), and Stichting Zorg voor Borstvoeding (www.zorgvoorborstvoeding.nl). The funders had no role in study design, data collection and analysis, interpretation of results or decision to publish.
Competing interests: The authors have declared that no competing interests exist.
Tobacco smoking is a major cause of morbidity and mortality, including during pregnancy. Smoking in pregnancy is associated with very preterm birth (less than 32 weeks), fetal growth restriction and low birth weight , . It increases the risk of sudden infant death syndrome, and it has been linked to an increased likelihood of asthma in childhood and a lower IQ in adulthood –. There are effective ways to help pregnant women to stop smoking and reduce preterm birth and low birth weight , but the impact of these interventions has not been studied at a national level in the Netherlands.
In most developed countries, there are marked social differences between women who smoke and those who do not, with continued smoking into pregnancy showing a strong association with socioeconomic disadvantage , . Also, across the literature, smoking has been associated with low rates of breastfeeding . In the present study, we report on the prevalence of smoking throughout pregnancy for 2001 to 2007 and quantify associations of maternal smoking throughout pregnancy with socioeconomic (low educational attainment, single parenthood), behavioural (alcohol consumption during pregnancy, initiation of formula-feeding), and neonatal (preterm birth and intra-uterine growth retardation) risk factors for infant health and development.
The data of five national surveys, containing anonymous records of 14,553 Dutch mothers and their offspring were analyzed.
The study design was approved by the Leiden University Medical Centre Medical Ethics Committee. The Committee did not require that informed consent was given for the surveys: return of the anonymous questionnaire was accepted as implied consent.
Between 2001 and 2007, five nationwide surveys were carried out among mothers of infants under six months of age. Data were collected by means of a two-step procedure. Firstly, for each survey, each of the 65 organizations executing the Mother and Child Health Care program in the Netherlands were asked to provide names and addresses of five Well-Baby Clinics that were willing to participate. In our invitation to the organizations it was asked that they identify Well-Baby Clinics whose clients vary in socioeconomic status. Next, each participating Well-Baby Clinic randomly distributed anonymous questionnaires with stamped, pre-addressed return envelopes to the first 20 visiting mothers of infants under six months of age. Participants completed the questionnaires at home.
Per survey, 222 to 263 out of the 325 Well-Baby Clinics approached participated. From 2001 to 2007, a total number of 25,120 anonymous questionnaires were distributed, of which 15,428 (61.4%) were completed and returned. Of the completed questionnaires, 518 (0.03%) were excluded because the inclusion criterion regarding the age of the infant (≤6 months) was not fulfilled. A total number of 14,910 (59.4%) questionnaires were found eligible for analyses, 99.1% of which showed to be completed by the mother. Median age of the infant at completion of the questionnaire was two months (range zero to six months).
The sample was found to be representative of the Dutch female population with respect to maternal age at child birth, gestational age of the child, number of previous children, place of child birth (at home or in hospital), and residential region . Differences did exist for level of formal education and immigrant status. In the standard Dutch population of 15- to 65-year-old women, 24% hold a degree on university or higher-professional level (2006) and 18% (2007) are born outside the Netherlands . In our sample, these proportions were 36% and 6%, respectively.
The questionnaire dealt with maternal smoking before, during, and after pregnancy, paternal or partner smoking, and daily number of cigarettes smoked.
Educational attainment of mothers was used as a measure of socioeconomic status. Formal education was categorized into low (primary or junior education), intermediate (secondary education), or high (higher-professional or university education) level. Mother's age (in years), whether she lived with a partner or not, and her country of birth (the Netherlands or elsewhere) were recorded.
Mothers recorded whether they initiated formula feeding or breastfeeding after the birth. In the 2007-survey, questions on alcohol consumption during pregnancy were added. Exposure during the first trimester, during the second or third trimester of pregnancy, and occasional abuse during pregnancy (more than six glasses per drinking occasion, also referred to as binge drinking)  were measured.
Mothers recorded parity (1/2/≥3), and gestational age (in completed weeks), birth weight (in grams), and gender of the infant. Gestational age was categorized into extremely preterm (≤28 completed weeks), very preterm (28 to 31 weeks), moderately preterm (32 to 36 weeks), and term (≥37 weeks) birth. In order to assess fetal growth, we used Dutch references curves . Infants whose birth weight was below the 10th percentile at the corresponding gestational age, adjusted for gender and parity, were classified as small for gestational age (SGA).
Frequency tables and 95%-confidence intervals (95%CIs) were calculated for smokers and non-smokers. Differences between groups were assessed by means of Student's t, Mann-Whitney or chi-square tests, wherever appropriate. Mantel-Haenszel linear-by-linear association tests were used to identify trends. Relationships with socioeconomic, behavioural and neonatal factors were studied by means of (multinomial) logistic regression techniques. Because effects of stopping smoking at different stages during pregnancy differ , women who smoked throughout pregnancy (smokers) were compared to women who did not smoke at all during pregnancy (non-smokers). During the analyses, adjustments were made for: maternal educational attainment, age, single motherhood, parity, maternal country of birth, and child's gender. Strengths of associations are presented as crude and adjusted odds ratios (ORs) and 95% confidence intervals (95%CIs). Smoking-attributable risk percents among smokers and for the population (smokers and non-smokers) were calculated based on adjusted ORs, according to the method by Cole and MacMahon . Population attributable risk percents may be used to judge priorities in public health, as they quantify the potential impact of risk-factor intervention programs . Presented p-values are two-sided; p-values ≤0.05 were classified as statistically significant. Statistical analyses were performed with SPSS 14.0.
Prevalence and Trends
Of 14,553 participating mothers, 9.7% (n = 1416) reported to have smoked throughout pregnancy. Median cigarette consumption was five per day (range 1 to 60). Of the participating mothers, 25.6% (n = 3166) indicated to have smoked during the six months prior to pregnancy. During the first six months after the birth, 14.0% (n = 2074) of mothers reported to smoke. Both before and after pregnancy, the median number of cigarettes smoked by the mother was 10 per day (range 1 to 60). Of fathers, 27.4% (n = 4025) smoked, with a median daily cigarette consumption equal to 10 (range 1 to 83). 31.6% (n = 4635) of all infants lived together with at least one smoker (mother and/or father). In smoking households, median consumption of cigarettes amounted to 24 per day (range 2 to 119).
Between 2001 and 2007, the overall rate of smoking throughout pregnancy declined from 13.2% to 7.6% (42% reduction; p<0.001; fig. 1). Median cigarette consumption during pregnancy did not change over time. Sharp decreases in rates of smoking throughout pregnancy were seen between the surveys of 2001 and 2002 (25% reduction), and between 2003 and 2005 (32% reduction) (both p-values <0.001). In the highest-educated group, a statistically significant decrease in smoking rates over the years was observed (p<0.001); the proportion of smokers fell from 4.7% to 1.6% (66% reduction; fig. 2). The overall rate was 3.2%. Among intermediate-educated women, the proportion of smokers decreased from 12.6% to 8.5% (33% reduction), with an overall rate of 9.5%. This decrease did not reach significance (p = 0.06). In the lower-educated group, there was no statistically significant change over time. The overall rate of smoking throughout pregnancy for mothers with less education was 18.7%.
Figure 1. Prevalence of smoking throughout pregnancy in the Netherlands 2001–2007.doi:10.1371/journal.pone.0008363.g001
Figure 2. Prevalence of smoking throughout pregnancy by level of formal education (low, intermediate or high) in the Netherlands 2001–2007.doi:10.1371/journal.pone.0008363.g002
Pregnant smokers tended to belong to the lower-educated group (OR 6.31; 95% CI 5.31 to 7.50) and to be single mothers (OR 2.41; 95% CI 1.75 to 3.32) (table 1). As compared to non-smokers, smokers more often were born in the Netherlands (OR 1.62; 95% CI 1.23 to 2.15). They also tended to be somewhat younger at child birth (OR 1.02; 95% CI 1.01 to 1.04). Mean age at birth was 30 years for smokers versus 31 for non-smokers (table 1). And, if they had a partner, smokers were more likely to live with a partner who also was a current smoker (OR 7.81; 95% CI 6.90 to 8.92).
Table 1. Logistic regression analysis of several maternal characteristics on smoking status in pregnant women in the Netherlands 2001–2007.doi:10.1371/journal.pone.0008363.t001
Overall, 28.6% (n = 526) of mothers reported social drinking during the first trimester of pregnancy. Of drinking mothers, 43.7% (n = 222) had less than one glass per month, 28.8% (n = 146) drank one to three glasses per month, and 27.5% (n = 140) drank more than three glasses per month. With respect to the second and third trimester of pregnancy, 24.3% (n = 497) of mothers reported social drinking: 62.8% (n = 309) had less than one glass per month, 22.4% (n = 110) drank one to three glasses per month, and 14.8% (n = 73) drank more than three glasses per month. Proportions of social drinking did not differ for smokers and non-smokers. 9.4% (n = 173) of participants reported occasional alcohol abuse during pregnancy (more than six glasses per drinking occasion). For smokers, this was 17.9% versus 8.7% for non-smokers (OR 2.08; 95%CI 1.25 to 3.45; table 1).
In our study group, 21.1% (n = 3132) of mothers initiated formula feeding. 38.2% of smokers opted for formula feeding versus 19.2% of non-smokers. The difference remained after adjustment for maternal socioeconomic factors (OR 1.91; 95% CI 1.69 to 2.16) (table 1).
Of all infants 5.6% were born preterm: 0.2% extremely preterm (≤28 weeks), 0.5% very preterm (28 to 31 weeks), and 4.8% moderately preterm (32 to 36 weeks). Median gestational age was 40 weeks for infants of smokers as well as for infants of non-smokers (5th percentile 35 weeks, 95th percentile 42 weeks, for both groups). After adjustments for maternal characteristics, the odds for a smoking mother, as compared to a non-smoking mother, to have an extremely preterm baby (≤28 completed weeks) were 7.25 (95% CI 3.40 to 15.38) (table 2). Proportions of very preterm or moderately preterm deliveries were equally divided among smokers and non-smokers. Smoking-attributable risk percents were 0.81 among smokers and 0.29 in the population (smokers and non-smokers).
Table 2. (Multinomial) logistic regression analysis of maternal smoking status throughout pregnancy on preterm birth and intrauterine growth in the Netherlands 2001–2007.doi:10.1371/journal.pone.0008363.t002
Overall, 10.2% of all infants had birth weights below the 10th percentile for their respective gestational age. Among smokers, 23.5% of infants were SGA (birth weight <10th percentile for gestational age). Among non-smokers this was 8.8% (OR 3.08; 95% CI 2.66 to 3.57). Mean (± standard deviation) birth weight of infants born to smokers was 3240 g±566 g versus 3516 g±571 g for infants born to non-smokers. The smoking-attributable risk percents were 0.67 and 0.17, respectively among smokers and in the general population.
The overall rate of smoking throughout pregnancy fell by 42% from 2001 to 2007 in the Netherlands. This downward trend was mainly seen in the highest-educated group, and could not be detected in the lower-educated groups. Overall, in the Netherlands, 10% of women smoked throughout pregnancy.
At a European level, the percentages of smokers during pregnancy vary from 5% to 7% in Lithuania, the Czech Republic, Sweden, and Malta to 16% in Denmark and 21% in France . Studies on alcohol and cigarette consumption are prone to under-reporting. This might lead to underestimation of prevalences.
During recent decades, rates of smoking in pregnancy declined, in response to a general policy of discouraging smoking. In 1982, in the Netherlands, 46% of women smoked during pregnancy . In 1996, the proportion of smokers had fallen to 21% . We observed sharp decreases in 2001, and from 2003 to 2005. The 2003–2005 decrease coincided with the implementation of the brief intervention strategy V-MIS in the Netherlands. V-MIS is a 5 A's based smoking cessation protocol (Ask, Advise, Assess, Assist, Arrange) which has been proven to be effective in lowering smoking prevalence among pregnant women , .
From the present study it becomes clear that maternal smoking in pregnancy is not just a health risk in and of itself, but a proxy indicator for several other health risks as well. As compared to non-smokers, pregnant smokers are six times more likely to be of lower socio-economic status. By contrast, smokers are less likely to be immigrants, which probably reflect cultural differences in smoking habits in women. Throughout the literature, low socioeconomic status has been associated with poorer health and more frequent unhealthy behaviours such as smoking. The mechanism behind this association is complex and remains to be disentangled, but there are indications that psychological stress has a role. Compared with the more advantaged, individuals of low socio-economic status are believed to experience greater and more severe daily stressors. Smokers report that smoking helps them to deal with stress and anxiety . Pregnant women are vulnerable to pressures and report feeling constantly judged by others. . Also, findings show that significant psychological symptoms and strains are being reported amongst pregnant women who smoke , . Health risk perceptions may have a mediating role in the association between socio-economic status and smoking .
Occasional alcohol abuse during pregnancy was reported twice as much by smokers, as compared to non-smokers. This may well be related to elevated stress levels in the smoking pregnant population the more since proportions of social drinking did not differ for smokers and non-smokers. The highest levels of exposure occurred during the first trimester. Well known effects of fetal exposure to high levels of alcohol include facial anomalies, behavioural and CNS abnormalities, and growth restriction . Prenatal exposure to alcohol as well as to cigarette smoke increases the risk of preterm birth and low birth weight. Based on our data it can be estimated that in the Netherlands a yearly number of 3000 unborn babies are being exposed to the combined risks of smoking and (occasional) alcohol abuse during pregnancy.
In addition, pregnant women who smoked were twice as likely to choose formula feeding over breastfeeding. Breastfeeding is the preferred method of feeding to achieve optimal child health, growth, and development . Since both smoking and formula feeding are more common among the lower-educated, adjustments were in our analysis for educational differences between groups. Smokers not only initiate formula feeding more often, if they start breastfeeding, they tend to terminate early , . In the literature, several possible explanations for the adverse relation between cigarette smoking and breastfeeding have been offered . On the one hand, plausible physiological explanations exist. It has been suggested that nicotine in the maternal blood stream acts on human milk production by reducing the level of prolactin and that nicotine in human milk increases infant irritability and, hence, feeding difficulties . Psychosocial explanations involve smokers being more insecure about their ability to breastfeed and less health conscious in general . Until now the exact underlying mechanism remains unclear.
Almost all children of mothers who smoke during pregnancy will be exposed to environmental tobacco smoke in the postnatal period. And, even if smoking occurs outside the home and away from the infant, this does not seem to fully protect a smoker's infant from environmental tobacco smoke exposure due to contamination of dust and surfaces in homes of smokers. Environmental tobacco smoke exposure, as measured by cotinine levels in infant urine, was five to seven times higher in households of smokers trying to protect their infants by smoking outdoors, than in households of non-smokers . Between 2001 and 2007, 32% of all Dutch infants were exposed to at least one smoker (mother and/or father), with a median level of exposure to 24 cigarettes per household per day. In 1996, 44% of the Dutch six-months-olds were exposed to one or more smokers in the family although the level of exposure was lower: on average between five and 15 cigarettes per day .
Women who smoked were twice as likely to be single mothers. In addition, women with a partner who also was a current smoker were eight times as likely to smoke during pregnancy, as compared to women with a partner who did not smoke. In a Cochrane review by Park and co-workers (2004) it is concluded that there is a strong association between partner support and successful smoking cessation . In particular cooperative behaviours, such as talking the smoker out of smoking a cigarette, and reinforcement, were found to predict successful quitting. Negative behaviours, such as nagging and complaining, were found predictive of a relapse. Still, intervention programs that intend to improve partner support do not seem to increase long-term quitting rates, presumably because the interventions have not successfully changed the support provided .
We estimated that in women who, on average, smoke a daily number of five cigarettes throughout pregnancy, 81% of the total risk of extremely preterm birth and 67% of the total risk of SGA outcome can be attributed to tobacco smoke exposure. Other factors, such as socioeconomic status, maternal age, and parity, account for the remainder. We also estimated that at a population level (smokers and non-smokers), 29% of extremely preterm births and 17% of SGA outcomes can be prevented if pregnant women would refrain from smoking. In absolute terms this implies that, if pregnant smokers were to stop tobacco use completely, 12 extremely preterm and 170 SGA outcomes per 10,000 live births would be prevented annually in the Netherlands (181,336 live births per year , of which 0.4% extremely preterm (2004) ). Smoking-attributable risk percents were based on adjusted risk ratios. Factors we did not measure include psychosocial stress, multiple pregnancy, eating habits, and the level of health care during pregnancy. These may have caused residual confounding. The true preventive effect of smoking cessation may therefore be smaller.
In practice this gain will likely not be met soon without a much stronger concerted effort. Pooled data from 65 trials on effects of behavioural interventions for stopping smoking in pregnancy showed an absolute reduction of 6% in continued smoking . A comparable reduction was observed in the Netherlands between 2001 and 2007. Such an effort would be extremely worthwhile even just in monetary terms. Based on our data, the 6%-smoking reduction has lead to an estimated annual prevention of six extremely preterm and 93 SGA outcomes per 10,000 live births. The incremental hospitalization costs roughly are €50,000 for extremely preterm and €5,000 for SGA infants . Clearly, a stronger, more effective promotion of smoking cessation during pregnancy can have a significant effect on maternal and infant health, as well as on health care expenditure.
In conclusion, although the rates of maternal smoking throughout pregnancy decreased significantly in the Netherlands, one out of ten women still smoke throughout pregnancy. Children of smokers are more likely to face the consequences of preterm birth and fetal growth restriction. In addition, this group tends to be born into social disadvantage, is more likely to suffer significant alcohol exposure in utero, and is formula fed more often in early life. If pregnant women were to cease tobacco use completely, an estimated 29% of extremely preterm births and 17% of SGA infants may be avoided annually and at least 26 million euro would be saved on health care costs annually in the Netherlands alone.
Conceived and designed the experiments: CL MRC DS JPvW. Performed the experiments: CL. Analyzed the data: CL. Wrote the paper: CL SEB MRC DS JBG JPvW.
- 1. Simpson WJ (1957) A preliminary report on cigarette smoking and the incidence of prematurity. Am J Obstet Gynecol 73: 808–815.
- 2. Kramer MS (1987) Determinants of low birth weight: methodological assessment and meta-analysis. Bulletin of the World Health Organisation 65: 663–737.
- 3. Blair PS, Fleming PJ, Bensley D, Smith I, Bacon C, et al. (1996) Smoking and sudden infant death syndrome: results from 1993-5 case-control study for confidential inquiry into stillbirths and deaths in infancy. BMJ 313: 195–198.
- 4. DiFranza JR, Aligne CA, Weitzman M (2004) Prenatal and postnatal environmental tobacco smoke exposure and children's health. Pediatrics 113: 1007–1015.
- 5. Pattenden S, Antova T, Neuberger M, Nikiforov B, De Sario M, et al. (2006) Parental smoking and children's respiratory health: independent effects of prenatal and postnatal exposure. Tob Control 15: 294–301.
- 6. Mortensen EL, Michaelsen KF, Sanders SA, Reinisch JM (2005) A dose-response relationship between maternal smoking during late pregnancy and adult intelligence in male offspring. Paediatr Perinat Epidemiol 19: 4–11.
- 7. Rogers JM (2008) Tobacco and pregnancy: overview of exposures and effects. Birth Defects Res C Embryo Today 84: 1–15.
- 8. Lumley J, Chamberlain C, Dowswell T, Oliver S, Oakley L, et al. (2009) Interventions for promoting smoking cessation during pregnancy. Cochrane Database of Systematic Reviews 3: CD001055.
- 9. Schneider S, Schutz J (2008) Who smokes during pregnancy? A systematic literature review of population-based surveys conducted in developed countries between 1997 and 2006. Eur J Contracept Reprod Health Care 13: 138–147.
- 10. Cavelaars AEMJ, Kunst AE, Geurts JJM, Crialesi R, Grötvedt L, et al. (2000) Educational differences in smoking: international comparison. BMJ 320: 1102–7.
- 11. Amir LH, Donath SM (2002) Does maternal smoking have a negative physiological effect on breastfeeding? The epidemiological evidence. Birth 29: 112–23.
- 12. Lanting CI, Van Wouwe JP, Reijneveld SA (2005) Infant milk feeding practices in the Netherlands and associated factors. Acta Paediatr 94: 935–42.
- 13. Statistics Netherlands, the Hague, the Netherlands. Available: www.cbs.nl. Accessed 2009 jan 19.
- 14. Hannigan JH, Armant DR (2000) Alcohol in pregnancy and neonatal outcome. Semin Neonatol 5: 243–254.
- 15. The Netherlands Perinatal Registry. Utrecht, the Netherlands. Available: www.perinatreg.nl. Accessed 2009 jan 19.
- 16. MacArthur C, Knox EG (1988) Smoking in pregnancy: effects of stopping at different stages. Br J Obstet Gynaecol 95: 551–5.
- 17. Cole P, MacMahon B (1971) Attributable risk percent in case-control studies. Brit J Prev Soc Med 25: 242–244.
- 18. Northridge ME (1995) Public health methods: attributable risk as a link between causality and public health action. Am J Public Health 85: 1202–1203.
- 19. EURO-PERISTAT project, with SCPE, EUROCAT, EURONEOSTAT. 2008. European perinatal health report.Available www.europeristat.com. Accessed 2009 jan 19.
- 20. de Jonge GA, van der Klaauw MM (1982) The incidence of smoking before during and after pregnancy. Ned Tijdschr Geneeskd 126: 1537–43.
- 21. Crone MR, Hirasing RA, Burgmeijer RJF (2000) Prevalence of passive smoking in infancy in the Netherlands. Patient Educ Counseling 39: 149–153.
- 22. Segaar D, Bolman C, Willemsen MC, de VriesH (2007) Identifying determinants of protocol adoption by midwives: a comprehensive approach. Health Educ Res 22: 14–26.
- 23. Vries H de, Bakker M, Mullen PD, van BreukelenG (2006) The effects of smoking cessation counseling by midwives on Dutch pregnant women and their partners. Patient Educ Couns 63: 177–187.
- 24. Ebert LM, Fahy K (2007) Why do women continue to smoke in pregnancy? Women Birth 20: 161–168.
- 25. Blalock JA, Fouladi RT, Wetter DW, Cinciripini PM (2005) Depression in pregnant women seeking smoking cessation treatment. Addict Behav 30: 1195–1208.
- 26. Goedhart G, van der WalMF, Cuijpers P, Bonsel GJ (2009) Psychosocial problems and continued smoking during pregnancy. Addict Behav 34: 403–406.
- 27. Lee JE, Lemyre L, Turner MC, Orpana HM, Krewski D (2008) Health risk perceptions as mediators of socioeconomic differentials in health behaviour. J Health Psychol 13: 1082–91.
- 28. Horta BL, Bahl R, Martines JC, Victora CG (2007) Evidence of the long-term effects of breastfeeding: systematic reviews and meta-analyses. Geneva: WHO. 52 p.
- 29. Horta BL, Kramer MS, Platt RW (2001) Maternal smoking and the risk of early weaning: a meta-analysis. Am J Public Health 91: 304–7.
- 30. Amir LH (2001) Maternal smoking and reduced duration of breastfeeding: a review of possible mechanisms. Early Hum Dev 64: 45–67.
- 31. Matt GE, Quintana PJ, Hovell MF, Bernert JT, Song S, et al. (2004) Households contaminated by environmental tobacco smoke: sources of infant exposures. Tob Control 13: 29–37.
- 32. Park EW, Schultz JK Tudiver FG, Campbell T, Becker LA (2008) Enhancing partner support to improve smoking cessation. Cochrane Database of Systematic Reviews 1: CD002928.
- 33. Russel RB, Green NS, Steiner CA, Meikle S, Howse JL, et al. (2007) Cost of hospitalization for preterm and low birth weight infants in the United States. Pediatrics 120: e1–9.