Smoking and quit attempts during pregnancy and postpartum: a longitudinal UK cohort

Objectives Pregnancy motivates women to try stopping smoking, but little is known about timing of their quit attempts and how quitting intentions change during pregnancy and postpartum. Using longitudinal data, this study aimed to document women’s smoking and quitting behaviour throughout pregnancy and after delivery. Design Longitudinal cohort survey with questionnaires at baseline (8–26 weeks’ gestation), late pregnancy (34–36 weeks) and 3 months after delivery. Setting Two maternity hospitals in one National Health Service hospital trust, Nottingham, England. Participants 850 pregnant women, aged 16 years or over, who were current smokers or had smoked in the 3 months before pregnancy, were recruited between August 2011 and August 2012. Outcome measures Self-reported smoking behaviour, quit attempts and quitting intentions. Results Smoking rates, adjusting for non-response at follow-up, were 57.4% (95% CI 54.1 to 60.7) at baseline, 59.1% (95% CI 54.9 to 63.4) in late pregnancy and 67.1% (95% CI 62.7 to 71.5) 3 months postpartum. At baseline, 272 of 488 current smokers had tried to quit since becoming pregnant (55.7%, 95% CI 51.3 to 60.1); 51.3% (95% CI 44.7 to 58.0) tried quitting between baseline and late pregnancy and 27.4% (95% CI 21.7 to 33.2) after childbirth. The percentage who intended to quit within the next month fell as pregnancy progressed, from 40.4% (95% CI 36.1 to 44.8) at baseline to 29.7% (95% CI 23.8 to 35.6) in late pregnancy and 14.2% (95% CI 10.0 to 18.3) postpartum. Postpartum relapse was lower among women who quit in the 3 months before pregnancy (17.8%, 95% CI 6.1 to 29.4) than those who stopped between baseline and late pregnancy (42.9%, 95% CI 24.6 to 61.3). Conclusions Many pregnant smokers make quit attempts throughout pregnancy and postpartum, but intention to quit decreases over time; there is no evidence that smoking rates fall during gestation.


Strengths and limitations of this study
• As far as we are aware, this is the only study to investigate timing of quit attempts and propensity to stop smoking during pregnancy and postpartum, and to quantify longitudinal changes.
• Smoking behaviour is self-reported rather than validated; misreporting due to recall bias may have been minimised by collecting data at three time points, and by there being no expectation that they should try to stop smoking.
• Later survey findings were adjusted using multiple imputation to help address nonresponse bias due to attrition.
• As the study was conducted in just one geographical area of the UK and participants were predominantly white British, findings might not be generalisable; however, the demographic profile of participants was similar to that of other UK cohorts of pregnant smokers.

INTRODUCTION
Smoking in pregnancy is associated with increased risks of miscarriage, stillbirth, prematurity, low birth weight, perinatal morbidity and mortality, neo-natal and sudden infant death, infant respiratory problems, poorer infant cognition, and adverse infant behavioural outcomes. 1 2 Internationally, large numbers of pregnant women smoke; with rates between 12% and 22% in high income countries [3][4][5][6] and rates increasing in emerging and developing economies. 7 Pregnancy is probably the event which most motivates female smokers to try quitting; for example, in the UK over 50 per cent of pregnant smokers try to stop 5 and pregnant women are, therefore, particularly likely to be interested in receiving cessation support. Some health systems systematically offer such support; in the UK this is largely done in early pregnancy, 8 although official guidance recommends that support is provided throughout gestation. 9 We are aware of no studies that have investigated when, in pregnancy, smokers have the greatest propensity to try stopping, the timing of any quit attempts and potential influences on this. Women who smoke before pregnancy have varied smoking patterns after conception, 5 10-15 and although it is logical to try to minimise fetal exposure to tobacco smoke by offering cessation support in early pregnancy, support may be welcomed at other times in gestation. Three prospective, longitudinal studies have reported smoking patterns in pregnancy; 10 11 15 however, two are over 20 years old 10 15 (including the only one to have been conducted in the UK 10 ), the third had a smoker response rate of only 25% 11 and none of these studies report when in pregnancy women have tried to quit. 10 11 15 To help focus smoking cessation interventions at the most effective leverage points, we need contemporary, longitudinal data on the smoking and quitting behaviours of pregnant women and also on their receptivity to cessation support. Consequently, we investigate the frequency of pregnant smokers' quit attempts and the factors associated with these. We also attempt to quantify individual-level changes in smoking behaviour during these times.  18 (scores range, 0 to 6; higher score indicates greater cigarette dependence). Women were also asked to rate their interest in receiving different forms of cessation support on a Likert scale ranging from 1 'not at all' to 5 'extremely'.

Statistical analysis
In order to quantify the proportion of quit attempts made after the first trimester of pregnancy, we aimed to recruit 850 participants. 16 Analyses were conducted using Stata version 14.0 (Stata Corp, College Station, TX, USA).
Descriptive statistics summarised participants' characteristics and smoking behaviour at each time point and we compared those responding to all three questionnaires with those who did not using chi-squared and t-tests for categorical and continuous variables respectively, with p values of <0.05 deemed significant. Characteristics found to be significantly associated with non-completion of later questionnaires, and hence absence of smoking data, were used with multiple imputation to adjust for attrition of smoking behaviour at later time points.
An exploratory analysis was performed to investigate the factors associated with reporting having made a quit attempt of any duration on the baseline questionnaire. For this analysis items were dichotomised; six self-efficacy items that had high internal consistency (Cronbach's α=0.95) 19 were combined into a single score out of 30 20 21 ethnicity, 22 qualifications held, 20 previous pregnancy, 20 21 23 24 number of cigarettes smoked per day, HSI, 21 23 25 partner smoking status, 21 occupation, 26 planned or surprise pregnancy, 20 27-29 depression, long term disability or mental illness, 30 smoking beliefs and self-efficacy 31 ). Variables which showed a significant association (p<0.05) in the univariable analysis were included in a multivariable logistic regression model. Variables that achieved significance (p<0.05) remained in the multivariable model and all nonsignificant variables identified from the univariable analysis were re-entered into the model consecutively to assess whether they became significant. The final multivariable model included only significant variables (p<0.05). A likelihood ratio test identified that age should be included in the multivariable analysis as a continuous variable. Where collinearity between variables was anticipated (for example, the number of cigarettes smoked per day and HSI), we included the variable that resulted in a better fitting model. As this analysis only included baseline data, we did not need to take account of attrition.
Multiple imputation for missing outcome data for smoking in late pregnancy (34)(35)(36) weeks gestation) and at 3 months after delivery was performed using Stata's mi command, based on 20 iterations. The outcomes were imputed using multivariable logistic regression models based on the following baseline variables: age, smoking status, gestation, general health, depression, previous pregnancy, smoking in previous pregnancy, smoking urges, qualifications, and ethnicity. All baseline variables were included in the analysis in dichotomised format. The percentage of women smoking at each outcome was obtained using Rubin's rule. 32 Where necessary, an augmented regression approach was used to overcome issues relating to perfect prediction during the multiple imputation. Participants had similar socio-demographic characteristics to those in previous pregnancy cohorts and have been reported elsewhere. 16 Just over half (488, 57.4%) were current smokers and 729 (85.7%) of the 850 women in the cohort reported their longer-term quitting intentions (data missing for 121 (14.2%)). Of these 729 women, 424 (58.2%) planned to stop smoking permanently, 21 (2.9%) intended to stop until their baby was born, and 181 (24.8%) were unsure; however, 103 (14.1%) did not plan to stop.
Responding to all three surveys was associated with being older, less cigarette dependant, primiparous, in a planned pregnancy and being a 'recent ex-smoker' at the outset of the study (Table 1).   Figure 2 shows a preliminary descriptive analysis of smoking behaviour across pregnancy within the 397 participants who returned all three questionnaires and illustrates variability in individual's smoking behaviour. Of note, 13.5% (5/37) of women who had stopped smoking in the 3 months before pregnancy were smoking again 3 months after childbirth, whereas 34.2% (55/161) of women who reported that they had quit after finding out they were pregnant had returned to smoking 3 months postpartum. As these data are not adjusted for non-response at follow-up, they may not be consistent with adjusted figures reported below.

DISCUSSION
To our knowledge, this is the first UK study to use prospectively collected, longitudinal data to quantify changes in smoking behaviour during pregnancy and postnatally.
Despite over 50% of smokers reporting quit attempts across all three trimesters, there was no evidence that overall smoking rates changed between joining the study at around 8-24 weeks gestation and late pregnancy. Within 3 months of giving birth, around one third of women who achieved abstinence before or during early pregnancy had returned to smoking. However, we observed a trend, not previously reported in longitudinal data, whereby those who quit before pregnancy may be less likely to return to smoking postpartum than those that quit on learning of their pregnancy; those that only achieved abstinence in late pregnancy appeared to be most likely to return to smoking researchers emphasised that responses were of interest irrespective of smoking status and, as women completed questionnaires at three stages, they did not have to recollect their behaviour over long periods. Additionally, studies looking at both self-report and biochemically-validated smoking data suggest that self-reported smoking can be both accurate and reliable. 15 33 It is possible that pregnant women who were concerned about the stigma of smoking may have avoided participation; we do not know how this might have affected findings, but women who consented to join the cohort had similar characteristics to those who declined. 16 As the survey was conducted in just two Nottingham hospitals, it is hard to say how far findings can be generalised. To help assess generalisability, we included survey items that permitted comparison with previous studies; we found our participants who continued to smoke in early pregnancy were similar to pregnant smokers enrolled in other major UK cohorts. 16 This suggests that the principal findings may apply to pregnant smokers in the UK generally. Likewise, although absolute smoking rates and smoking cessation advice and treatment may vary, pregnant smokers from other high income countries generally have similar characteristics to those in the UK. 20 34 35 Therefore, it could be considered reasonable to extrapolate many of our findings to pregnant smokers in high income countries generally. Although we had very high rates of eligible women joining the cohort, a further limitation was that attrition was relatively high, with response rates to the two later questionnaires of 60% and 56%. This is a common problem with longitudinal studies, 36 and as young, pregnant smokers were likely to be a particularly difficult group to maintain contact with, we used a number of recommended methods to try to maximise response rates. 16 36 37 However, rather than simply relying on incomplete data, we have tried to address non-response bias by adjusting later surveys' findings using multiple imputation. In addition, differences in characteristics between the whole cohort and those that responded to all three surveys (Table 1) need to be considered when viewing the unadjusted smoking 'trajectory' analysis shown in Figure 2.
The originality of this study is a key strength. As previously mentioned, we could only find three previous observational studies in which pregnant women's smoking behaviours were longitudinally recorded at more than one time point in pregnancy and these have limitations. 10  The finding that most women in our cohort had quit in the early stages of pregnancy (before joining the study), and that smoking rates did not change between the second trimester and 36 weeks' gestation is consistent with cross-sectional estimates for smoking prevalence obtained in a large US study, which reported these by month of pregnancy. 38 In that study, smoking prevalence at 1 month gestation was 26%, then between the fourth and eighth month of gestation, smoking rates each month were 13-14%. 38 Other retrospective studies have found that most women who successfully quit are likely to achieve this soon after finding out they are pregnant, 5 39 often within the first few days. 39 Many quit spontaneously after discovering they are pregnant. 40 Therefore, it seems that after the early stages of pregnancy, despite still reporting quit attempts, women's smoking behaviour actually undergoes very little change.
One study found that 70% of pregnant women making their first quit attempt did so in their first trimester; however, these data were collected up to 5 years after delivery, and only considered first quit attempts. 14 We found that some women made multiple quit attempts throughout pregnancy and most reported cutting down since becoming pregnant. Far fewer women reported making quit attempts in the 3 months after childbirth than they did during pregnancy. Even in early pregnancy, around half of women had no intention to quit within the next 30 days; intention to quit in the short term was even lower in late pregnancy, and was lowest of all postpartum. This diminishing intention to quit has not been reported before and could be considered when designing and delivering cessation interventions.
We found that women who were primiparous, smoked fewer cigarettes per day, had a planned pregnancy and believed smoking during pregnancy could seriously harm their baby were more likely to have made a quit attempt during early pregnancy. These findings are comparable to previous literature examining the characteristics of pregnant smokers who successfully achieve cessation. Primiparous women have previously been found to be more likely to successfully quit smoking. 20 21 23 24 This may be because women who have smoked throughout a previous pregnancy without experiencing complications may view the risks of smoking during pregnancy differently to primiparous women, and therefore be less motivated to make a quit attempt. 21 Similarly, previous studies have found that heavier smoking is negatively associated with successfully quitting in pregnancy, 21 23 25 and heavier smokers are less likely to have high motivation to quit during pregnancy. 25 Women whose pregnancies are unintended have previously been found to be more likely to continue smoking during pregnancy, 20 27-29 and likewise pregnant smokers who do not report concern about the effect smoking might have on the health of their unborn baby were more likely to have low motivation to quit smoking. 25 These findings identify women who are most likely to make a quit attempt and will potentially benefit the most from NHS support. Heavier smokers and women in second or later and unplanned pregnancies who are less likely to try quitting may require different, more intensive or tailored forms of support.

Implications for practice
Although our data suggest that motivation to quit may be strongest in early pregnancy, some women will be receptive to quitting at any time, as indicated by their multiple quit attempts throughout pregnancy, and this confirms that it is important to discuss smoking with women at every appointment and to refer them for stop smoking support. 9 Preventing resumption of smoking after pregnancy is a critical public health issue; if women restart their lifelong health is at risk, and their infants are more likely to be exposed to second-hand smoke 41 and to eventually become smokers. 42 Women often need help to resist returning to smoking after childbirth, but there are currently few effective interventions for this. 43 Women appear to be more inclined to consider quitting during pregnancy than in postpartum, and this is important when designing interventions. However, some postpartum women do make quit attempts, or may be planning to quit in the medium term, so engaging with them again after birth, to think about planning for this in the medium term, rather than immediately, might be a successful option. Previous studies have shown that women who quit spontaneously early in pregnancy are likely to be different and more successful than those who quit later, 40 and we found that women appear to be more likely to return to smoking after childbirth the later in pregnancy they quit. Therefore, exploring potential reasons for this, for example demographic factors or women's intentions, could help to identify if different women may benefit from alternative approaches to help prevent relapse, perhaps by developing more tailored interventions. Although quit attempts might suggest receptivity to quitting, what is not well understood is how interest in smoking cessation support may change during pregnancy.

Conclusions
Many pregnant women who smoke attempt quitting throughout their pregnancy, but this makes little difference to overall smoking rates. After giving birth, most smokers seem less inclined to make further quit attempts and many who quit in early pregnancy return to smoking. Women who quit in late pregnancy may be most likely to return to smoking after childbirth, whilst those who stopped prior to pregnancy may be least likely to relapse. It therefore imperative to discuss smoking with women, including recent exsmokers, throughout pregnancy and postpartum, and to continue to offer and provide specialist stop smoking support.

CONTRIBUTORS
SC helped conceive the study, made a substantial contribution to the development of the protocol and questionnaires, assisted with day-to-day troubleshooting during the datacollection phase, and drafted and revised this manuscript. SO helped design the data collection process, recruited participants into the cohort, managed the day-to-day running, assisted with data analysis, and contributed to the drafting of this manuscript.
JL-B contributed to the development of the study protocol and questionnaires, advised on analysis, and contributed to the preparation of this manuscript. EB undertook the analyses and interpretation of data in Table 2 as part of a BMedSci project and contributed to the preparation of this manuscript. LV assisted with data analysis and interpretation, and contributed to the preparation of this manuscript. KB helped design the data collection process, recruited participants into the cohort, managed the day-today running, and contributed to the preparation of this manuscript. FN, MU, KEP and SS all contributed to the development of the study protocol and questionnaires, contributing expertise from their own particular knowledge base, and to the preparation of this manuscript. TC conceived the study, and made substantial contributions to the development of the study protocol and questionnaires, and to the preparation of this manuscript. All authors read and approved the final manuscript. Rachel Whitemore assisted with study administration, telephone follow-ups, and data entry.

COMPETING INTERESTS
KP is a trustee of The Equality Trust (a registered charity), and receives occasional honoraria, all of which are donated to The Equality Trust or for student support at the University of York. In the last 5 years TC has been paid honoraria on 2 occasions for speaking at meetings or conferences organised by Pierre Fabre Laboratories (a nicotine replacement therapy manufacturer).
Other authors have no conflicts of interest to report.

ETHICS APPROVAL
Derbyshire Research Ethics Proportionate Review Sub-Committee gave ethical approval.

Results
Participants 13* (a) Report numbers of individuals at each stage of study-eg numbers potentially eligible, examined for eligibility, confirmed eligible, included in the study, completing follow-up, and analysed 9 & Figure 1 (b) Give reasons for non-participation at each stage Figure 1 (c) Consider use of a flow diagram Figure 1 Descriptive data 14* (a) Give characteristics of study participants (eg demographic, clinical, social) and information on exposures and potential confounders 9, Table 1, Appendix Pregnancy motivates women to try stopping smoking, but little is known about timing of their quit attempts and how quitting intentions change during pregnancy and postpartum. Using longitudinal data, this study aimed to document women's smoking and quitting behaviour throughout pregnancy and after delivery.

Setting
Two maternity hospitals in one National Health Service hospital trust, Nottingham,

England.
Participants 850 pregnant women, aged 16 or over, who were current smokers or had smoked in the 3 months before pregnancy, were recruited between August 2011 and August 2012.

Outcome measures
Self-reported smoking behaviour, quit attempts and quitting intentions.

Conclusions
Many pregnant smokers make quit attempts throughout pregnancy and postpartum, but intention to quit decreases over time; there is no evidence that smoking rates fall during gestation.

Keywords
Smoking cessation, pregnancy, longitudinal research, quit attempts, postpartum relapse, survey research

Strengths and limitations of this study
• As far as we are aware, this is the only study to investigate timing of quit attempts and propensity to stop smoking during pregnancy and postpartum, and to quantify longitudinal changes.
• Smoking behaviour is self-reported rather than validated; misreporting due to recall bias may have been minimised by collecting data at three time points, and by there being no expectation that they should try to stop smoking.
• Later survey findings were adjusted using multiple imputation to help address nonresponse bias due to attrition.
• As the study was conducted in just one geographical area of the UK and participants were predominantly white British, findings might not be generalisable; however, the demographic profile of participants was similar to that of other UK cohorts of pregnant smokers. rates between 12% and 22% in high income countries 3-6 and rates increasing in emerging and developing economies. 7 Pregnancy is probably the event which most motivates female smokers to try quitting; for example, in the UK over 50 per cent of pregnant smokers try to stop 5 and pregnant women are, therefore, particularly likely to be interested in receiving cessation support. Some health systems systematically offer such support; in the UK this is largely done in early pregnancy, 8 although official guidance recommends that support is provided throughout gestation. 9 We are not aware of any studies that have investigated when, in pregnancy, smokers have the greatest propensity to try stopping, the timing of any quit attempts and potential influences on this. Outside of pregnancy and postpartum, most adults tend to have fairly stable smoking behaviour. 10 Although overall smoking rates in pregnancy have declined, a significant proportion of women continue to smoke throughout pregnancy. 11 However, many women who smoke before pregnancy have varied smoking behaviour after conception, 5 11-16 and although it is logical to try to minimise fetal exposure to tobacco smoke by offering cessation support in early pregnancy, support may be welcomed at other times in gestation. In addition, of those that do stop, many relapse within the first few months postpartum. 17 18 Relatively few studies of prenatal smoking behaviour have been longitudinal, 12 13 15 16 19-23 with only two of these following up women postpartum, 12 13 and the only two studies to have been conducted in the UK are now over 20 years old. 12 19 Importantly, none of these studies asked about number of quit attempts or reported when in pregnancy women have tried to quit. To help focus smoking cessation interventions at the most effective leverage points, we need quit attempts, and the factors associated with these. We also attempt to quantify individual-level changes in smoking behaviour during these times. higher score indicates greater cigarette dependence). The questionnaires are available as an appendix to this paper.

Figure
(1-4) quit attempts were reported since childbirth (unadjusted data). The median number of quit attempts made by those who smoked across their pregnancy (smokers who completed both baseline and late pregnancy questionnaires, n=177) was 3 (IQR 1-6); these data were highly skewed with a range of 0-60 24-hour quit attempts reported. Table 3 shows data on smoking rates, quitting behaviour and quit intentions at the three time points adjusted for non-response, as appropriate, using multiple imputation; raw (unadjusted) data are included for reference in a supplementary online table (Appendix  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y   Table 3 Smoking behaviours reported in pregnancy and the postpartum adjusted for non-response at late pregnancy and 3 months postpartum using multiple imputation

DISCUSSION
To our knowledge, this is the first study to use prospectively collected, longitudinal data to quantify changes in smoking behaviour through the examination of multiple quit attempts and women's intention to quit during pregnancy and postnatally. Despite over 50% of smokers reporting quit attempts across all three trimesters, there was no evidence that overall smoking rates changed between joining the study at around 8-24 weeks gestation and late pregnancy. In smokers, intention to quit within the next month fell as the pregnancy progressed, and then fell further postpartum. Within 3 months of  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y giving birth, around one third of women who achieved abstinence before or during early pregnancy had returned to smoking. However, we observed a trend, not previously reported in longitudinal data, whereby those who quit before pregnancy may be less likely to return to smoking postpartum than those that quit on learning of their pregnancy; those that only achieved abstinence in late pregnancy appeared to be most likely to return to smoking postpartum. Women's motivation to try quitting was lowest in the first 3 months following childbirth; only around a quarter tried quitting during this time and far fewer reported intending to quit in the immediate future than had done so at either pregnancy time point.
The originality of this study is a key strength. As previously mentioned, we could find relatively few observational studies in which pregnant women's smoking behaviours were longitudinally recorded at more than one time point in pregnancy and . 12 13 15 16 19-23 Only two of these longitudinal studies followed women up postpartum, 12 13 only four reported any data on fluctuations or trajectories in smoking status, 12   researchers emphasised that responses were of interest irrespective of smoking status and, as women completed questionnaires at three stages, they did not have to recollect their behaviour over long periods. Additionally, studies looking at both self-report and biochemically-validated smoking data suggest that self-reported smoking can be both accurate and reliable. 16 41 It is possible that pregnant women who were concerned about the stigma of smoking may have avoided participation; we do not know how this might have affected findings, but women who consented to join the cohort had similar characteristics to those who declined. 24 As the survey was conducted in just two Nottingham hospitals, it is hard to say how far findings can be generalised. To help assess generalisability, we included survey items that permitted comparison with previous studies; we found our participants who continued to smoke in early pregnancy were similar to pregnant smokers enrolled in other major UK cohorts. 24 This suggests that the principal findings may apply to pregnant smokers in the UK generally. Likewise, although absolute smoking rates and smoking cessation advice and treatment may vary, pregnant smokers from other high income countries generally have similar characteristics to those in the UK. 28 42 43 Therefore, it could be considered reasonable to extrapolate many of our findings to pregnant smokers in high income countries generally. Although we had very high rates of eligible women joining the cohort, a further limitation was that attrition was relatively high, with response rates to the two later questionnaires of 60% and 56%. This is a common problem with longitudinal studies, 44 and as young, pregnant smokers were likely to be a particularly difficult group to maintain contact with, we used a number of recommended methods to try to  24 44 45 However, rather than simply relying on incomplete data, we have tried to address non-response bias by adjusting later surveys' findings using multiple imputation. In addition, differences in characteristics between the whole cohort and those that responded to all three surveys (Table 1) need to be considered when viewing the unadjusted smoking 'trajectory' analysis shown in Figure 2. Finally, we assessed smoking status at 3 months postpartum, and it is likely that some women who were abstinent at this point will have returned to smoking after this. 17 18 The finding that most women in our cohort had quit in the early stages of pregnancy (before joining the study), and that smoking rates did not change between the second trimester and 36 weeks' gestation is consistent with cross-sectional estimates for smoking prevalence obtained in a large US study, which reported these by month of pregnancy. 46 In that study, smoking prevalence at 1 month gestation was 26%, then between the fourth and eighth month of gestation, smoking rates each month were 13-14%. 46 Other retrospective studies have found that most women who successfully quit are likely to achieve this soon after finding out they are pregnant, 5 47 often within the first few days. 47 Many quit spontaneously after discovering they are pregnant. 48 Therefore, it seems that after the early stages of pregnancy, despite still reporting quit attempts, women's smoking behaviour actually undergoes very little change.
One study found that 70% of pregnant women making their first quit attempt did so in their first trimester; however, these data were collected up to 5 years after delivery, and only considered first quit attempts. 15 We found that some women made multiple quit attempts throughout pregnancy and we have previously reported that, at baseline, most reported cutting down or only smoking occasionally since becoming pregnant with less than 8% of our cohort saying that they smoked the same or more than before pregnancy. 24 Although self-reported, this reinforces findings from qualitative studies, which indicate that many persistent smokers report deliberate, and sometimes detailed, plans to cut down in their pregnancy, seeing this as a positive step and often as a route to quitting. 49 Far fewer women reported making quit attempts in the 3 months after childbirth than they did during pregnancy. Even in early pregnancy, around half of women had no intention to quit within the next 30 days; intention to quit in the short term was even lower in late pregnancy, and was lowest of all postpartum. This diminishing intention to quit has not been reported before and could be considered when designing and delivering cessation interventions; for example, earlier intervention may be more successful.
We found that women who were primiparous, smoked fewer cigarettes per day, had a planned pregnancy and believed smoking during pregnancy could seriously harm their baby were more likely to have made a quit attempt during early pregnancy. These findings are comparable to previous literature examining the characteristics of pregnant smokers who successfully achieve cessation. Primiparous women have previously been found to be more likely to successfully quit smoking. 28 29 31 32 This may be because women who have smoked throughout a previous pregnancy without experiencing complications may view the risks of smoking during pregnancy differently to primiparous women, and therefore be less motivated to make a quit attempt. 29 Similarly, previous studies have found that heavier smoking is negatively associated with successfully quitting in pregnancy, 29 31 33 and heavier smokers are less likely to have high motivation to quit during pregnancy. 33 Women whose pregnancies are unintended have previously been found to be more likely to continue smoking during pregnancy, 28 35-37 and likewise pregnant smokers who do not report concern about the effect smoking might have on the health of their unborn baby were more likely to have low motivation to quit smoking. 33 These findings identify women who are most likely to make a quit attempt and will potentially benefit the most from NHS support. Heavier smokers and women in second or later and unplanned pregnancies who are less likely to try quitting may require  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59

Implications for practice
Although our data suggest that motivation to quit may be strongest in early pregnancy, some women will be receptive to quitting at any time, as indicated by their multiple quit attempts throughout pregnancy, and this confirms that it is important to discuss smoking with women at every appointment and to refer them for stop smoking support. 9 One rather surprising finding was that in early to mid-pregnancy 44% (211/477) of smokers disagreed that smoking in pregnancy can harm their baby; as those who agreed with this statement were more likely to have made previous quit attempts at baseline, additional education on this issue should be considered by health professionals.
Preventing resumption of smoking after pregnancy is a critical public health issue; if women restart their lifelong health is at risk, and their infants are more likely to be exposed to second-hand smoke 50 and to eventually become smokers. 51 Women often need help to resist returning to smoking after childbirth, but there are currently few effective interventions for this. 52 Women appear to be more inclined to consider quitting during pregnancy than in postpartum, and this is important when designing interventions. A potential reason for restarting smoking and for making fewer quit attempts postpartum may be that women perceive that harm to the baby from smoking is much higher during pregnancy compared with after delivery. However, some postpartum women do make quit attempts, or may be planning to quit in the medium term, so engaging with them again after birth, to think about planning for this in the medium term, rather than immediately, might be a successful option. Previous studies have shown that women who quit spontaneously early in pregnancy are likely to be different and more successful than those who quit later, 48 and we found that women appear to be more likely to return to smoking after childbirth the later in pregnancy they quit. Therefore, exploring potential reasons for this, for example demographic factors or  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59

COMPETING INTERESTS
KP is a trustee of The Equality Trust (a registered charity), and receives occasional honoraria, all of which are donated to The Equality Trust or for student support at the University of York. In the last 5 years TC has been paid honoraria on 2 occasions for speaking at meetings or conferences organised by Pierre Fabre Laboratories (a nicotine replacement therapy manufacturer).
Other authors have no conflicts of interest to report.