Smoking behaviour in pregnancy and its impact on smoking cessation at various intervals during follow-up over 21 years: a prospective cohort study


Prof. SA Doi, University of Queensland, Clinical Epidemiology Unit, School of Population Health, Herston Road, Herston, Brisbane, QLD 4006, Australia. Email


Please cite this paper as: Rattan D, Mamun A, Najman J, Williams G, Doi S. Smoking behaviour in pregnancy and its impact on smoking cessation at various intervals during follow-up over 21 years: a prospective cohort study. BJOG 2013;120:288–296

Objective  To determine whether mothers who quit or reduce their level of smoking in pregnancy comprise a group of health-conscious women who are disproportionally likely to adopt a healthier smoking lifestyle in the medium to longer term, compared with women who continue to smoke during pregnancy.

Design  A prospective cohort study.

Setting  A public hospital in Australia.

Population  A cohort of 6703 individual mothers who completed both initial phases of data collection in 1981–1983; mothers who smoked daily (2992) before pregnancy were included in this study.

Methods  Mothers were interviewed at 3–5 days post-delivery, 6 months, 5 years, 14 years and 21 years to determine their smoking status. An inverse probability-weighted Poisson regression with a robust error variance was fitted to the data using a log-link function and a binary response variable for smoking outcome, and adjusting for several possible confounding factors.

Main outcome measure  Smoking cessation at several follow-up points, for up to 21 years.

Results  Of the mothers who smoked daily before pregnancy, 12, 23, 37 and 41% reported having ceased smoking at 6 months and at 5, 14 and 21 years, respectively. The decision to quit smoking during pregnancy was found to be independently associated with a higher rate ratio (RR) of smoking cessation at 6 months (RR 30.60, 95% CI 20.50–45.69), 5 years (RR 4.36; 95% CI 3.61–5.27), 14 years (RR 2.42, 95% CI 2.12–2.75) and 21 years (RR 1.86; 95% CI 1.60–2.15), after adjusting for several possible confounding factors.

Conclusions  Pregnancy appears to be an opportunity for successfully quitting smoking, regardless of socio-economic circumstances or demographic background.


Although the prevalence of smoking has declined considerably in the industrialised countries, including Australia, over recent decades, smoking still remains one of the major preventable risk factors causally associated with morbidity and mortality.1 This is of particular concern if it occurs in pregnancy because the health effects of maternal smoking are cumulative, and are substantial across a lifetime. Whereas evidence that changing smoking behaviour positively during pregnancy will have positive health benefits for offspring is well recognised,2,3 pregnancy can also have a positive impact on maternal smoking behaviour. This may occur in two ways: women tend to reduce or quit smoking during pregnancy, and exhibit an increased tendency to cease smoking over the longer term. There is the possibility that the experience of women over the longer term might differ if they do or do not successfully quit in pregnancy. Currently, it is unclear whether mothers that quit in pregnancy are at an advantage in terms of long-term smoking cessation compared with women who continue to smoke in pregnancy. If they are, then pregnancy would be a unique window of opportunity for healthcare professionals to advise patients to quit smoking.4,5

Although as yet there is no conclusive evidence about whether, or to what extent, smoking cessation in the long term is initiated by the decision taken during pregnancy, there is a lot of evidence suggesting that the effects of smoking cessation dissipate after birth, and that the majority of women resume smoking.5–12 One study that has looked at a longer term outcome of mothers who had quit smoking in pregnancy proposed that the same factors associated with the decision to quit in pregnancy were also associated with the decision not to smoke 9 months later.13 To our knowledge, only one previous study has examined directly whether, or to what degree, smoking status during pregnancy has had an impact on smoking trajectories 3 years after the initial postpartum period.14 However, this was a study that used non-pregnant women as a reference category, suggesting that pregnant women who quit in pregnancy were not more likely to quit in the long term than non-pregnant women. Again, this fails to answer the question of the extent to which quitting smoking in pregnancy impacts on the decision to cease smoking in the long term. Of greater interest is the possibility that pregnant quitters or pregnant non-quitters may differ in their longer term smoking behaviour. We are in a unique position to investigate this using the Mater-University of Queensland Study of Pregnancy (MUSP) cohort, because this study has prospectively collected data on maternal smoking status and a range of factors over 21 years. This study thus investigates the impact of maternal changes in pregnancy smoking behaviour on long-term smoking cessation by these mothers.


The study

The MUSP began between 1981 and 1983 as a collaboration of the Mater Misericordiae Hospital, South Brisbane, and The University of Queensland. Some 8557 pregnant women were invited to join the study during their first clinic visit at the Mater Mothers Hospital in Brisbane, Australia. The main focus of the study has been the assessment of health and social outcomes for both mother and child. The various waves of the study throughout this period encompass the prenatal, postnatal, childhood, adolescent and young adult periods for the child. At enrollment, the pregnant women were on average at approximately 18 weeks of gestation. In total, 7223 (84% of the mothers invited to participate) mother–child pairs were enrolled, and 6703 individual mothers completed both of the initial phases of data collection. These mothers and their offspring form the MUSP cohort. Women were prospectively interviewed (via a questionnaire administered by research staff, who also assisted with queries) 3–5 days after delivery, and again after 6 months and 5, 14 and 21 years. Obstetricians, sociologists and a statistician developed the main research questions, and measurements related to these, in the early stages of the study. These included validated scales and new scales and questions, with the latter undergoing extensive piloting to assess reliability and validity. No active smoking cessation interventions were undertaken during the index pregnancies. The original study and subsequent follow-up received ethical approval from an ethics committee at the University of Queensland and Mater Hospital. Participants gave signed informed consent for their participation and for that of their children. The present study is limited to the 2992 mothers enrolled who smoked daily before pregnancy. Full details on the study participants and measurements have been reported previously.15–18


Smoking behaviour in pregnancy

Women were asked about whether and how much they smoked before they became pregnant. Only daily smokers were selected for this study. At 3–5 days after delivery, mothers were asked to recall their smoking level during the last trimester. The smoking question was ‘In the last 3 months of your pregnancy, how often did you smoke cigarettes?’, with response options of ‘every day’, ‘every few days’, ‘once or so, only’ or ‘did not smoke at all’. Women who responded ‘every day’ were classified as ‘unchanged’ and women who answered ‘every few days’ or ‘once or so, only’ were classified as ‘decreased’. Women that answered ‘did not smoke at all’ were classified as having ‘quit’. Thus we created three categories of smoking behaviour during pregnancy: unchanged, decreased and quit.

Postpartum smoking cessation

Smoking status was assessed at 6 months, and at 5, 14 and 21 years of follow-up. On each occasion, mothers were asked a question about the number of cigarettes smoked: ‘In the last week, how many cigarettes did you usually smoke per day?’ (with 50 or more, 30–49, 20–29, 10–19, 1–9, nil smoked, as possible responses). These responses were then dichotomised into those that had ceased smoking (responded none) versus those that had continued smoking (all other responses). It is important to note that smoking cessation was independently assessed at each time point, and this outcome was independent of smoking cessation or continuation at other time points. Thus a mother who ceased smoking at one time point could have resumed smoking at another, and vice versa.

Potential explanatory variables at first clinical visit (FCV)

The following maternal characteristics during pregnancy, reported at FCV, were considered to be potential confounding factors on the basis of their potential association with maternal smoking during pregnancy, and possible impact later in life: maternal age (exact age); family income (low, AUS $10 399 or less; medium–high, AUS $10 400 or more); maternal education (did not complete secondary school, completed secondary school, completed further/higher education); maternal depression (depressed versus non-depressed, according to the Delusions Symptoms States Inventory19); quality of marital relationship (good relationship versus not good relationship, using Spanier’s Dyadic Adjustment Scale20); maternal alcohol consumption (abstainer, light, or one or more glasses per day); maternal race (white, Asian, aboriginal/islander); anxiety (anxious versus non-anxious, Delusions Symptoms States Inventory19); and parity (0, 1, 2, 3+). Four additional variables were recoded. Physical activity was measured by asking one question with five different subquestions, with each containing options such as ‘often’, ‘sometimes’ or ‘never’. The five activities recorded were: active sports; swimming or long walks; weekend trips in the car; working in the garden; doing physical exercise. For ease of analysis, we classified this further into little activity, some activity or very active based on the scores obtained. Total life events (life events scale for 6 months prior to FCV) were collected, and several items were used to construct this scale. The items used were: someone close to you died or has been seriously ill; big problem with own health; serious disagreements with your partner; serious disagreements with someone else close to you; serious financial problems; partner had major change in job situation; you had major change in job situation; serious problems with housing or accommodation; you or partner had problems with the law; you have had trouble at work; you have been divorced; you and your partner have separated; you have changed partners; you have moved house. Finally, for ease of analysis, the total number of adverse life events were classified into two categories: 0–3 versus 4+. Another additional variable was smoking intensity before pregnancy, which accounted for the total number of cigarettes smoked before pregnancy. This was recorded by asking the question: ‘In the last week how many cigarettes have you usually smoked per day?’ (with 50 or more, 30–49, 20–29, 10–19, 1–9 and nil smoked, as possible answers). For ease of analysis we constructed three mutually exclusive smoking categories by re-grouping smoking categories into high-intensity smokers (30 or more cigarettes daily), medium-intensity smokers (20–29 cigarettes daily) and low-intensity smokers (1–19 cigarettes daily).

Statistical analyses

We first compared the baseline characteristics of the women according to their decision to smoke or quit in pregnancy using the chi-square test for categorical variables and the F-test to compare mean age. We then cross-tabulated pregnancy quitting and postpartum smoking cessation to determine the rate of long-term smoking cessation at each time point. The sample numbers used in each analysis may differ according to the definition of different confounding factors, selection criteria and follow-up time. Finally, we fitted a series of regression models to assess the association between pregnancy smoking behaviour and long-term smoking cessation in mothers with complete follow-up at each time period.

As we had significant losses to complete follow-up over time, we adjusted for the drop-outs at each follow-up period by running an appropriately weighted regression.21 We did this by first computing the propensity (probability) to remain in the study separately for each time period by regressing (using logistic regression) an indicator variable on the baseline explanatory variables. The fitted model gives a predicted probability for each woman that a woman with her characteristics (i.e. values on the explanatory variables) would participate at that time point. Each woman who participated at the scheduled follow-up was then given weight equal to 1/p, where p is her fitted probability of participating from the logistic regression. The weighted analysis was reported, although it was not substantially different from the unweighted analysis. An inverse probability weighted Poisson regression with a robust error variance was then fitted to the data,22 using a log-link function and a binary response variable for smoking outcome. By applying this model, the rate ratios (RRs) and 95% confidence intervals (95% CIs) can be obtained. All analyses were performed using stata 11 StataCorp LP, College Station, TX, USA), and significance was considered at P < 0.05.


At the time of FCV to the study (Table 1) the women had an average age of 24 years (SD 5 years). Also, two-thirds had completed high school: most were predominantly white Australians, and two-thirds had a good income (recorded by 1981–1984 standards). The majority of women who smoked daily also drank alcohol (in varying quantities, of course). Interestingly, most of the women had good dyadic adjustment: few were depressed at FCV and few had reported anxiety disorders. Most engaged in some physical activity, and it is interesting to note that only half had planned the pregnancy and about half of the mothers were expecting their first child at FCV.

Table 1. Baseline characteristics and bivariate associations between socio-economic, demographic and other variables with smoking behaviour during pregnancy among mothers that smoked daily before pregnancy
VariablesUnchanged = 2090 (70.1%)Decreased n = 408 (13.7%)Stopped n = 484 (16.2%) P-value
Age at FCV
Mean (SD)24.13 (4.7)23.4 (4.3)24 (4.8)0.09
Education (n)
Incomplete high school471 (22.6)65 (15.9)81 (16.8)<0.001
Complete high school1365 (65.6)273 (66.9)322 (66.8) 
Post high school244 (11.7)70 (17.2)79 (16.4) 
Race of mother
White1900 (94.7)368 (93.9)438 (93.0)0.01
Asian12 (0.6)6 (1.5)11 (2.3) 
Aboriginal–Islander95 (4.7)18 (4.6)22 (4.7) 
Family Income
AUS $10 400 or more1110 (57.7)236 (64.0)295 (64.4)0.006
AUS $10 399 or less814 (42.3)133 (36.0)163 (35.6) 
Abstainer358 (17.2)57 (14)57 (11.8)<0.001
Light1385 (66.4)271 (66.4)365 (75.6) 
1+ per day344 (16.5)80 (19.6)61 (12.6) 
Dyadic adjustment
Good adjustment1837 (95.0)361 (97.3)438 (97.6)0.01
Conflict97 (5)10 (2.7)11 (2.4) 
Non-depressed1882 (91.6)377 (93.8)445 (94.1)0.09
Depressed172 (8.4)25 (6.2)28 (5.9) 
Non-anxious1684 (82)346 (86.3)407 (85.9)0.02
Anxious370 (18)55 (13.7)67 (14.1) 
Physical activity
Little physical activity161 (7.8)18 (4.5)38 (8.0)0.01
Some physical activity1664 (80.8)317 (78.9)385 (81.1) 
Very active234 (11.4)67 (16.7)52 (10.9) 
Planned, wanted954 (48.4)180 (47.2)234 (51.1)0.71
Unplanned, wanted620 (31.4)117 (30.7)132 (28.8) 
Unsure399 (20.2)84 (22)92 (20.1) 
Number of births/parity
0901 (43.1)248 (60.8)277 (57.2)<0.001
1591 (28.3)98 (24)110 (22.7) 
2352 (16.8)46 (11.3)70 (14.5) 
3 or more246 (11.8)16 (3.9)27 (5.6) 
Total life events
0–31766 (85.5)350 (86.6)416 (87.6)0.47
4+299 (14.5)54 (13.4)59 (12.4) 
Smoking intensity
Heavy159 (7.6)20 (5.0)17 (3.5)<0.001
Moderate892 (42.7)97 (24)130 (27) 
Light1037 (49.7)287 (71)334 (69.4) 

We found that 16% of the mothers who smoked daily before pregnancy (n = 484) actually stopped smoking during pregnancy, and a further 14% of the women (n = 408) decreased their frequency. Yet 70% of the women in our sample (n = 2090) did not change their smoking behaviour during pregnancy. Several variables, including race of the mothers, dyadic adjustments, anxiety, whether the pregnancy was wanted or unwanted and adverse life events, were not significantly associated with change of smoking behaviour in pregnancy. Women with higher income, with more than high school education, who did some physical exercise and who were nulliparous were most likely to quit smoking during pregnancy (Table 1).

Of the women with complete follow-up to the various time points, 12% reported smoking cessation at 6 months, 23% at 5 years follow-up, 37% at 14 years follow-up and 41% at 21 years follow-up (Table 2; assuming data missing at random, and consistent with the unadjusted but weighted results presented in Table 3). Comparable percentages for women with complete follow-up through to 21 years (= 1226) were 13.3, 24.7, 40 and 42.7% at these respective follow-up points. After adjusting for the potential confounding factors identified at FCV, and using inverse probability weights to address potential bias from losses to follow-up, quitting smoking during pregnancy was found to be independently associated with a greater rate of smoking cessation at 6 months (RR 30.60, 95% CI 20.50–45.69), 5 years (RR 4.36; 95% CI 3.61–5.27), 14 years (RR 2.42, 95% CI 2.12–2.75) and 21 years (RR 1.86; 95% CI 1.60–2.15) of follow-up (Table 3).

Table 2. Cross-tabulation of smoking status during pregnancy (n%) with long-term smoking status of women that smoked daily before pregnancy*
Decision in pregnancyStatus by follow-up time
6 months
Still SmokeCeased smokingTotal
  1. *Numbers vary across follow-up as a result of attrition that increased over time.

Quit207 (45.4)249 (54.6)456
Decreased332 (88.5) 43 (11.5)375
No change1829 (98.4) 30 (1.6)1859
Total2368 (88)322 (12)2690
  5 years   
  Smokers Quitters Total
Quit157 (43.5)204 (56.5)361
Decreased208 (68.7)102 (32.9)310
No change1181 (87.7)165 (12.3)1346
Total1546 (76.6)471 (23.4)2017
  14 years   
  Smokers Quitters Total
Quit128 (35.5)233 (64.5)361
Decreased141 (48.8)148 (51.2)289
No change1010 (73.9)357 (26.1)1367
Total1279 (63.4)738 (36.6)2017
  21 years   
  Smokers Quitters Total
Quit100 (35.8)179 (64.2)279
Decreased110 (48.2)118 (51.8)228
No change680 (67.5)328 (32.5)1008
Total890 (58.7)625 (41.3)1515
Table 3. Unadjusted and adjusted rate ratios of long-term smoking cessation according to smoking status in pregnancy
Decision in pregnancy6 months5 years14 years21 years
RR95% CIRR95% CIRR95% CIRR95% CI
  1. *Reference category.

  2. **Adjusting variables entered via forward stepwise selection (P < 0.3 to enter): age, education, race, income, physical activity, parity, alcohol, smoking intensity, dyadic adjustment, depression, anxiety, pregnancy planned and total life events.

Continued*1 1 1 1 
Continued*1 1 1 1 


Using a community based longitudinal cohort study, we report for the first time that quitting smoking during pregnancy has an effect on long-term smoking behaviour. In mothers that quit during pregnancy, the rate of long-term smoking cessation was between two and four times greater compared with mothers who continued to smoke in pregnancy at each time point evaluated (5, 14 and 21 years postpartum), with a much greater influence at 6 months. Our study thus suggests that quitting smoking in pregnancy is independently associated with smoking cessation up to 21 years after delivery. This association was retained even after adjustment for various socio-economic and demographic factors in pregnancy, and the strength of the relationship of quitting in pregnancy with long-term smoking remained striking, even after adjustment for these potential confounding factors.

The strength of the association declined with follow-up time, however, and this could be because the proportion of women who smoke has fallen steadily over the past decade (for a variety of reasons), in both pregnant and non-pregnant women of reproductive age at similar rates.23 From this, we infer that there has been a consistent decline in smoking among women of reproductive age that would lead to a decline in smoking over time independent of the impact of the pregnancy decision. In this study we found that out of the total 6703 women, 2992 were smokers (44%) at the time of recruitment into the study. This figure was relatively high, given that reports from the mid-eighties suggest that women of this age group in Australia had a smoking prevalence of around 30%.24–26 Being young women, presenting at public hospital, with two-thirds of them also drinking might explain these figures. Currently this figure is estimated to be around 18%.27,28

It has always been thought that health risks like smoking are created and maintained by social systems, and the magnitude of those risks are largely a function of socio-economic disparities and psychosocial gradients.29 We show that the impact of the decision to quit in pregnancy overshadows these factors, and this concurs with some studies that have hypothesised (based on indirect evidence) that women who quit smoking during pregnancy may be more likely to be successful in quitting over their lifetime,30–32 and this would not be unexpected because they might be health conscious women. Others, however, did not think this to be likely based on surveys that had suggested that the relapse rates of pregnant smokers is very high soon after the delivery of the child,8,33 with estimates suggesting that up to 50–60% of pregnant quitters return to smoking within the first 6 months postpartum, increasing to 80–90% by 12 months.12,34 Although this is true, and we found 45% of pregnancy quitters had resumed smoking at 6 months, our study suggests that those that do cease smoking at 6 months are almost exclusively pregnancy quitters, and as time goes on there is also a contribution to smoking cessation from pregnancy smokers, thus explaining the decreasing risk ratios over long-term follow-up. This is therefore the first study confirming that pregnancy quitting retains a long-term impact on mothers with respect to smoking cessation over time.

Our findings support the suggestion that different health trajectories are the product of cumulative risk, protective factors, and other influences that are programmed into bio-behavioural regulatory systems during critical and sensitive periods.35 One such sensitive period for mothers and critical period for the fetus is pregnancy. Mothers who continue smoking during pregnancy have offspring who are at high risk of developing ill health, including cardiovascular disease, asthma, attention deficit-hyperactivity disorder and attention-deficit problems,36 and mothers who are given this information are more likely to change their smoking behaviour. This is evident, as nearly two-thirds to about half of women who smoke before pregnancy reduce their smoking or quit while pregnant.37 Socio-economic and demographic factors, however, moderate this quitting behaviour,37,38 and it is well known that education is a key factor. In our study, women who had education beyond high school were more likely to quit during pregnancy. Also, low income, greater parity, no intention to breastfeed and presence of other smokers in the household have all been independently associated with a lower likelihood of quitting during pregnancy. Our study confirmed these findings. In addition, we found that dyadic adjustment, being non-anxious and doing some physical exercise increases the chances of quitting during pregnancy. Furthermore, other studies have found that quitters were more likely to be light smokers, were less likely to have another smoker in their household, had a stronger belief in the harmful effect of maternal smoking, had a history of fewer miscarriages and entered prenatal care earlier.38 Factors that seem to have less impact in previous studies, however, include the number of cigarettes smoked before delivery, prenatal alcohol consumption and pregnancy weight gain.39 In addition, we found that the age of mothers at FCV, depression, whether the pregnancy was wanted or unwanted and adverse life events seemed to have less impact.

It is notable that in the present study we did not adjust for all possible confounding factors. In the future, research could address other theoretically relevant confounding factors, such as those derived from social cognitive models of behaviour change (e.g. the transtheoretical model).40–42 This model highlights the importance of various psychological processes such as motivation to quit, self-efficacy in quitting and beliefs about (internal and external) resources available to help quit behaviour. With the inclusion of such a model and the study of cognitive processes in future work, more integrated theoretical models of smoking cessation and pregnancy, and interventions to help pregnant women and new mothers quit smoking,43,44 are likely to be developed.

The strengths of our study compared with previous studies are that ours is a longitudinal cohort study. A valid concern related to the cross-sectional designs of previous studies is that data about cessation attempts have been collected retrospectively.45 Thus, the accuracy and richness of the information provided about past attempts might have been limited by recall problems. We do not have this problem as our study presents data from a large cohort of women presenting for public antenatal care at a large tertiary maternity hospital in Brisbane, Australia, over a 21-year period. However, the downside to our data was that they were derived from self-reports, for which there was no objective validation. It can be argued that self-reported measures of smoking during pregnancy are subject to problems of poor instrument design, recall bias and under-reporting. Also there might be reluctance by mothers to reveal their smoking habits, especially when they were pregnant, given shame, guilt and recall problems.23 An alternative and more reliable measurement is laboratory assessment of the smoking by using salivary cotinine levels, which was not feasible at that time. Finally, our sample, although representative of public obstetrical patients, differs from private obstetrical patients in a number of characteristics. Public patients are predominantly from middle to lower socio-economic status (SES) groups (lower family income and education levels), they are much more likely to be tobacco smokers, to be unmarried at the time they were recruited and more frequently exhibited a range of risk characteristics generally associated with poorer health outcomes. Initial attempts to recruit a subsample of private patients were of limited success, and a decision was therefore made to obtain as complete a sample of public patients as possible.


Given our findings, we conclude that pregnancy offers an opportunity for quitting smoking that has the potential to improve long-term quit rates and thus health outcomes for mothers. This implies that we need to focus on new approaches to achieving sustained cessation of smoking during pregnancy. As we achieve a better understanding of the multitude of factors that impact this important decision, we can better target these to more effectively promote smoking cessation during pregnancy. Our study makes clear that the dye is not irrevocably cast by social disadvantage: pregnancy appears to be an opportunity for successful quitting, regardless of social background. Investing in smoking cessation services for women during pregnancy is an important way of promoting the health of all mothers and their children.

Disclosure of interests

The authors declare that they have no conflicts of interest.

Contribution to authorship

D.R. designed the study, planned and executed the analysis, and drafted the article. S.D. supervised the conception, design and analysis of the study, and revised the article for intellectual content. A.M. contributed to data extraction, contributed to the analysis and revised the article for intellectual content. G.W. contributed to the analysis and revised the article for intellectual content. J.N. contributed to data extraction and revised the article for intellectual content.

Details of ethics approval

All phases of the MUSP have been approved by the ethics committees of The University of Queensland and Mater Hospital.


The study was funded by the National Health and Medical Research Council (NHMRC) of Australia, but the views expressed in the article are those of the authors, and not necessarily those of any funding body.


We thank all participants in the study, the MUSP data collection team and Greg Shuttlewood, data manager, MUSP, who has helped to manage the data for the MUSP.

Commentary on ‘Smoking behaviour in pregnancy and its impact on smoking cessation at various intervals during follow-up over 21 years: a prospective cohort study’

Pregnancy is often described as a ‘window of opportunity’ for addressing health needs and promoting positive behavioural change. Yet, this expression may be so hackneyed that it is like a window: we look right through it without thought about the window itself. The characteristics of this window matter: the support of friends and family may be amplified; some choices, like smoking or illicit drug use, are more stigmatised; health messages from professionals are frequent; and usual activities and routines are altered in fundamental ways. Multiple facets of life may change: increased attention to habits and choices in view of fetal wellbeing; modifications of work schedules; a move to a larger home or apartment; and a period of maternity leave or transition to full-time motherhood (

This research by Rattan and colleagues allows us to look 21 years into the distance to see something fascinating about women who chose to make a difficult change—stopping smoking—and succeeded during pregnancy. Over time, women who quit during pregnancy remained more likely to be nonsmokers. At first glance, this seems unremarkable. Yes, people who quit smoking are more likely to remain quit. And those who quit during pregnancy are often lighter smokers, with fewer co-existing challenges, as was the case in this cohort: those who achieved cessation were better educated, less likely to drink alcohol daily and were more likely to engage in some physical activity. But look closer. After taking into account key differences, including age, income, education, parity, relationship quality, alcohol use, anxiety and depression, intendedness of the pregnancy and baseline smoking, as well as potential bias in retention of cohort members over time, the effect is enduring and strong. The trend, in Australia and other developed countries, has been for smokers to give up smoking as they age. But that trend did not catch up with the value of having experience with cessation during pregnancy. At each follow-up period, those who stopped or reduced smoking during pregnancy were more likely to be non-smokers. How might this frame future research?

In behavioural time, pregnancy is long, and for the roughly half of women who planned for pregnancy the window is open even wider to ‘practise’ new behaviours. Time is available to consider, plan for, try and engraft new behaviours (Prochaska et al., J Consult Clin Psych 1988;56:520–8). Women describe undertaking changes, from avoiding certain products to being more consistent with meals or even dental hygiene (Right from the Start, KE Hartmann, unpubl. data). Pregnancy is a window in which we experience as mothers and observe as healthcare providers that there are real opportunities to ‘do life’ differently, in many cases with considerable joy attached to creating change. Experiencing success during crucial life transitions builds self-efficacy, even when the transition is stressful and profound. This study should inspire researchers to understand the life-course influence for mothers and their families of the choices they make when given support to address behaviours that they are motivated to change (Halfon, Soc Science Med, 2012;74:671–3; Horton, Lancet 2002;360:186). If evidence were to suggest similar enduring effects, imagine the benefit of learning skills to manage anxiety, improve sleep, understand and make new dietary choices, or enhance interpersonal relationships or parenting skills. Holistic care coordinated to support experiencing such changes in pregnancy would then be an investment not only in pregnancy wellbeing, but also in the entire future of a woman and her family.

Disclosure of interests

The author declares there are no conflicts of interest.

KE Hartmann

Vanderbilt University Medical Center, Nashville, TN, USA