Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Effects of a smoking cessation program for pregnant women and their partners attending a public hospital antenatal clinic

Effects of a smoking cessation program for pregnant women and their partners attending a public... This study aimed to assess the effect of a Abstract hospital-based smoking cessation intervention delivered by midwives during routine antenatal and postnatal care on the smoking habits of pregnant women and Warren Jones Department of Obstetrics and Gynaecology: Flinders Medical Centre, South Australia here is clear evidence that maternal smoking is hazardous to the development of the foetus and that cessation improves birthweight.’ In addition, studies indicate that exposure of infants to environmental tobacco smoke is associated with an excess risk of lower respiratory tract infections, so that sustained cessation postbirth is important.’ Smoking prevalence is highest among women of greater socio-economic disadvantage, who attend public hospital antenatal clinics for pregnancy care. Given the large numbers of women attending such clinics, considerable effort has been expended over the past decade to evaluate appropriate smoking cessation interventions for this setting. Late pregnancy quit rates of 6-8% using a minimal advice-giving procedure and 1215% using cognitive behavioural programs with repeated contact and reinforcement have been reported among women who present as smokers at the first antenatal visit, although most women relapse to smoking after deliver~.~-~ While higher pregnancy quit rates can potentially be achieved with a cognitivebehavioural program, the difficulty of incorporating such programs into busy clinic environments represents a major barrier to more widespread adoption. Programs that are minimal in nature may be a more achievable objective for routine application beyond the life of a research trial. Most intervention programs have emphasised the potential health hazards of maternal smoking on foetal development, particularly low birth-weight, as a primary source of motivation to quit. However, previous research has suggested that many women are easily able to discount the risks posed by smoking in pregnancy, due to their own experience or that of their friends and relatives VOL. who have had ‘normal babies’ after smoking throughout their pregnancies.’-” In addition, many women perceive that ‘small babies’ do not incur significant health An alternative strategy is to present more concrete and salient information about risk, using visual or auditory feedback, on the premise that such information is more likely to result in an increased sense of risk.I3-l5 Virtually all models of health behaviour include perceived risk as one factor that is likely to influence the decision to change risky behaviour. l6-Iy Although studies of feedback to women of carbon monoxide levels or visual information from ultrasound scans have produced mixed result^,^@^^ uncontrolled studies demonstrating the increase in foetal heart rate after the mother smokes a cigarette have produced changes in smoking habits worthy of further investigation. Kelly and O’Conor exposed women at 35 weeks gestation to changes in their foetus’ heart rate following the smoking of a cigarette; 19% quit for at least the remainder of their pregnancy and a further 50% at least halved their smoking rate.24Although personal feedback of foetal heart rate following maternal smoking would be likely to be more salient in enhancing perceived risk,13 time constraints and hospital policies which ban smoking may not make this option feasible. A recording which allows women to compare the sound of a normal foetal heart rate with that following one cigarette may provide concrete information about the effect of maternal smoking on foetal well being. Another aspect of smoking cessation programs which has remained undeveloped is the important role of partner’s smoking on maternal smoking and relapse. Smoking by partners of pregnant women is consistently and strongly related t o cessation in their partners. At the first antenatal visit, women in the intervention group (n=1 10) were given a demonstration of the immediate effects of smoking on foetal heart rate, brief smoking cessation advice and smoking cessation booklets for themselves and their partners; at delivery, they were given brief advice and a booklet about postpartum cessation. Compared with an historical control group who received usual care (n=1 10) and assuming those lost to follow-up continued to smoke, biochemically-verified maternal cessation sustained from at least 24 weeks gestation to late pregnancy was 6.4% in the intervention group and 1.8% in the comparison group. However, there was no difference between maternal quit rates at six months postpartum. Partners were more likely to try to quit in the intervention group, but quit rates did not differ. Exposure to the intervention was not associated with increased levels of psychological distress, as measured by the 12-item General Health Questionnaire. We conclude that this type of intervention, when implemented by staff during routine care, is probably associated with a small improvement in maternal cessation in pregnancy, similar to that produced by minimal advice to quit smoking. (Aust N Z J Public Health 1998;22:31 3-320) Correspondenceto: Dr Melanie Wakefield, Epidemiology Branch, South Australian Health Commission, PO’Box 6, Rundle Mall, South Australia, 5000, Fax (08) 8226 6316, e-mail wakefield.melanie @ health.sa.gov.au 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Wakefield and Jones pregnancy25,‘6and relapse p~stpartum.~’-*~ Women who receive encouragement from their partners to quit smoking are more likely to quit or reduce cigarette consumption.3” Despite this, few smoking cessation interventions in pregnancy have incorporated structured advice to quit, directed to partners. The present study sought to test the effectiveness of a novel minimal smoking cessation intervention delivered as part of routine antenatal and postnatal care. In the course of being given minimal advice to quit smoking by midwives, women were able to listen to the sound of a normal foetal heart beat and to compare this with the sound of a foetal heart beat five minutes after the mother had smoked a cigarette. An additional novel aspect of the intervention was that advice and information were provided to partners of pregnant women. The evaluation aimed to assess the effect of this advice on maternal and paternal smoking behaviour during pregnancy and up to six months postpartum. Many smoking interventions for pregnant women have been criticised for ignoring the problem of addiction, the meaning of the behaviour for the women involved, and the guilt and anxiety felt by those who continue to smoke in the face of exhortations to q ~ i t . ~ ’ Bryce & Enkin have questioned whether the guilt and anxiety that may be induced through maternal anti-smoking programs only causes more stress among pregnant women, particularly among Given this, the intervention trial also sought those least able to to determine whether there might be unintended adverse consequences of smoking cessation intervention, as manifested by increased levels of psychological distress. Subjects and methods Sample and design We chose not to conduct a randomised controlled trial as antenatal staff were to be trained in the delivery of stop-smoking advice and this would have influenced advice to the comparison group and also because such a design would have posed considerable organisational difficulties in the context of a busy antenatal clinic where recruitment of patients to the study was the responsibility of clinic staff. Rather, an historical comparison group was recruited from the hospital before the introduction of the intervention. As the comparison group was recruited at a different time, women may have been exposed to different types of influences on their smoking behaviour and the effect of a smoking cessation intervention could not be partitioned out using this design alone. For this reason, additional comparison groups were recruited concurrently from one other hospital (C1 and C2). By comparing C1 with C2, one could determine whether there was sufficient similarity in outcome measures to conclude that no extraneous influences on cessation, occurring concurrently with data collection for the comparison and intervention groups, were apparent (i.e. a ‘time’ or ‘history’effe~t).~~ If there was no evidence for such an influence, the comparison group could be directly compared with the intervention group at the index hospital. However, if a such a time effect was apparent, it would need to be taken into consideration when the comparison and intervention groups were compared. Women eligible to enter the trial were public patients who reported being smokers at the first antenatal appointment. Of these women, potential participants were excluded if they reported smoking one or fewer cigarettes per day, were ‘shared-care’ patients (those whose antenatal care was mainly managed by a general practitioner), if they had existing psychiatric disturbance including a history of severe puerperal depression, were non-English speaking (intervention materials had not been produced in other languages) or presented for their first antenatal visit later than 20 weeks gestation. Women were dropped from further follow-up if they had a spontaneous abortion or termination, a stillbirth, a delivery before 34 weeks gestation (making a late pregnancy assessment of smoking status impossible) or changed hospitals during the antenatal period, thus making their case notes inaccessible. Natural quit rates from other studies indicate that an expected quit rate in the comparison groups between the end of the first trimester and delivery would be 4%.4 Other intervention studies with public antenatal patients suggest that firm advice to quit, supported by patient materials can achieve a quit rate of approximately 15%.4 With a power of 90% and where a=0.05,a sample size of 108 would be required in each group.34 The study aimed to recruit an additional 17 subjects per group, making 125 in all, in order to allow for expected drop-out due to adverse pregnancy outcome. Women who met the eligibility criteria were asked by the midwife conducting the first antenatal visit to be involved in the study. Those who consented completed the 12-item General Health Questionnaire (GHQ) at the time of their visit.35Information pertaining to smoking status and level of cigarette consumption of pregnant women and their partners, as well as other relevant baseline information, was abstracted from the medical case notes that had been made up during the first antenatal appointment. Patients were followed up at 24-26 weeks gestation, 32-34 weeks gestation and six months post-delivery by mailed self-completion questionnaire, accompanied by a post-paid return envelope. If questionnaires were not returned within two weeks, a reminder letter was sent, followed by a telephone call from a research assistant if no reply had been received one to two weeks later. At 32-34 weeks gestation, those women who claimed that they had quit smoking were additionally asked to provide a urine sample for cotinine assay. Recruitment for the comparison group and C1 occurred between October 1990 and April 1991 and follow-up to six-month postpartum was completed in June 1992. Recruitment for the intervention condition and C2 occurred betweenAugust 1991 and February 1992 and final follow-up was completed in March 1993. Intervention During discussion of the issue of smoking in the first antenatal appointment, midwives used a hard moulded-plastic desk-top model of a foetus-in-utero. The model simulated the increase in foetal heart rate that follows the smoking of a cigarette by the mother, from approximately 130 to 150 beats per minute, an increase clearly perceptible to the human ear. It was explained to the pregnant woman that this effect was due to the nicotine and carbon monoxide in cigarettes making less oxygen available to the baby. The one-toone setting permitted patients to raise concerns and questions about the model with the midwife. Advice to quit was supported by a small booklet which further explained the immediate effects of maternal smoking on the foetus, detailed the benefits of quitting, presented counter-arguments to some of the bamers women most commonly identified to achieving cessation and provided practical VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 22 NO. 3 Effects of a smoking cessation program for pregnant women and their partners suggestions for quitting. A booklet for partners who smoked was also provided, whenever women reported that their partner was a smoker. Before leaving the hospital after delivery, further advice was given by a primary nurse to both continuing smokers and those who may have quit during pregnancy, about smoking and infant health and other associated issues such as breastfeeding and weight loss after pregnancy. This advice was reinforced by the provision of a booklet to each woman. The intervention materials were developed with the assistance of information and fee:dback from a series of focus group discussions of pregnant women and their partners.36 The booklets were written in a non-judgmental manner and raised issues that had been identified to be of concern to women in the focus group discussions. Each booklet had a Flesch grade level of 5 or 6, indicating the text was of an easy or fairly easy ~ t a n d a r d . ~ ' Women in comparison groups received usual care, which most often involved a verbal caution that smoking in pregnancy was not advisable. Brief pamphlets about smoking were available, although these were not pregnancy-specific. reductions in consumption were assessed by comparing the number of cigarettes per day reported at the 32-34 gestational week followup, after adjustment for pre-pregnancy cigarette consumption and consumption at the first antenatal appointment. To assess potentially adverse effects of the intervention, women were asked to complete the 12-item GHQ at baseline and each follow-up point. This enabled tracking of psychological functioning during and after pregnancy. The GHQ has previously been shown to be responsive to change in a pregnant population."OThose women who scored 3 or higher were designated as 'high scorer^'.^' Estimates of the recurrent costs of the intervention, based on staff time, cost of the desktop demonstration models and quit smoking resources, were made. Recurrent costs provide an indication of how much it would cost similar antenatal clinics to implement the usual care and intervention programs. Statisticalanalysis For each outcome measure, a comparison was undertaken between C 1 and C2, after adjustment for potential confounders. Judgments about whether or not a time effect had occurred were based on the size of the difference in outcome between C 1 and C2 and the consistency of the difference in outcome measures. Where there was no evidence of a time effect, it was assumed that the comparison and intervention groups could be directly compared. For dichotomous outcome variables (such as attempt to quit and quit status), univariate associations with independent variables were assessed by conventional chi-square tests and statistical adjustment was performed by using logistic regression analysis. Variables considered as candidates for inclusion in each logistic regression model were those which were associated at the univariate level with the outcome variable at a p value of <0.25,42or which were required to be included for a priori reasons, namely, baseline daily cigarette consumption. For each logistic regression, a check for multicollinearity was performed by reference to the eigenvalues generated in a factor analysis of candidate independent variables, resulting in highly inter-correlated variables being discarded. Eligible variables were then included, with group membership, in a series of logistic regression models, with independent variables being discarded if they failed to condition the risk estimate for group membership. Odds ratios and 95% confidence intervals were calculated from the final model. All continuous variables, where not normally distributed, were transformed by a square root transformation and re-assessed for normality. Univariate association of baseline variables with continuous outcome variables and group membership, was assessed by means of conventional t-tests. Analysis of covariance was used to adjust for the effect of potential confounders. Variables considered as candidates for inclusion in each analysis of covariance with a p value of <0.25, were subjected to a factor analysis to check for multi-collinearity, resulting in highly correlated variables being discarded. The remaining variables were added as covariates in an analysis of covariance, with group membership being the main independent variable. The adjusted means yielded in this analysis were then squared to transform them back to their index values. Initial analysis of each of the outcome measures was undertaken using only those patients who responded at the relevant follow-up point. Subsequent analyses for dichotomous outcome variables were Measures At each of the three follow-up points, women were asked whether they were smoking and, if so, how many cigarettes, on average, they smoked per day, whether they had tried to quit and, if so, how long ago their last attempt was. Women who reported that they were no longer smoking were asked how long it had been since their last cigarette. In addition, at each follow-up point, those women with a partner were asked identical questions about his smoking habits. A woman or her partner were considered as having tried to quit smoking during pregnancy if it was indicated that they tried to quit at any time between baseline and the 32-34 week follow-up. An attempt to quit postnatally was judged by those women who indicated that they or their partner tried to quit between delivery and the six-month postnatal follow-up. In assessing the outcome of intervention studies, a combination of point prevalence and continuous abstinence measures has been r e c ~ m r n e n d e d . ~ ~ measure of continuous abstinence, a woman For the was considered to have quit smoking during pregnancy if she indicated she had quit smoking at the 24-26 gestational week followup (and had not smoked any cigarettes in the preceding seven days) and was still abstinent, as confirmed by a urinary cotinine assay, at the 32-34 gestational week follow-up. Those women claiming abstinence with urinary cotinine levels above 100 ng/ml, indicating that they had most probably smoked in the 48 hours prior to the sample being taken, were considered to be smoker^.^' For the measure of point prevalencte cessation, the criterion was self-reported cessation at the 32-34 week follow-up (with no smoking during the previous week), confirmed by biochemical validation. For partners, the criterion was identical, except that biochemical assessment was not available. Maintenance of non-smoking postpartum was considered to have been achieved for those women who indicated they were not smoking at the six-month postnatal follow-up and who smoked their last cigarette before the 24-26 week follow-up. The same criteria were applied in the classification of partner's smoking status, except that a. cotinine assay was not available to validate self-reported smoking status at the 32-34 week follow-up. Among those who lwere not considered to have quit smoking, VOL. 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Wakefield and Jones Table 1: Participation rates, by group. c 1 Attended first antenatal visit Smoker Eligible smokers Recruited eligible smokers Available for follow-up c2 903 341 (37.8%) 201 (58.9%) 128 (63.7%) 110 (85.9%) Comparison 83 1 265 (32.6%) 208 (78.5%) 128 (61.5%) 110 (85.9%) intervention 961 303 (31.5%) 200 (66.0%) 127 (63.5%) 110 (86.6%) 1,000 355 (35.5%) 194 (54.6%) 121 (62.4%) 103 (85.1 Yo) undertaken assuming all non-respondents to follow-up questionnaires were continuing smokers. The research trial was approved by the ethics committees of both participating hospitals. Results Sample recruitment and follow-up Table 1 shows that, of all pregnant women smokers attending for their first antenatal visit during the study period, between 33% and 45% were ineligible. Exclusions occurred due to presentation after 20 weeks gestation (38.6%), being a shared-care patient (36.4%), presence of a psychiatric history of concern (1 2.4%), smoking fewer than 2 cigarettes per day (10.8%) and being non-English speaking (1.8%). In C1 and C2, the most common reason for exclusion was presentation after 20 weeks gestation, while in the comparison and intervention groups, the most common reason was shared care. Of smokers eligible to be approached for inclusion in the study, around 62% consented to participate; 20-25% of eligible smokers in each group declined to participate and the remainder were unfortunately not approached by midwives to be part of the study. A comparison of the characteristics of eligible smokers who consented to participate in the study with those who refused or were missed was undertaken for each of the four groups, revealing no systematic pattern of statistically significant differences between participants and non-participants. During the trial, 71 women were dropped from the study due to previously specified criteria. Most had experienced either an elective or spontaneous abortion (28.2%) or had cancelled subsequent visits due to abortion (36.6%). Most of the remaining cases were dropped from further follow-up because they had changed hospitals during the antenatal period. Reasons for dropout did not differ between groups. Dropouts were compared with study participants and no differences were detected in study characteristics. Thus, 110 women were available for followup in each group, except C 2 with 103 women. Table 2 shows the cumulative response rates to follow-up questionnaires, by group. There was no significant difference in response rates between groups at the first and second follow-up points. However, at the final follow-up point, the second two groups of women recruited at each hospital (C2 and the intervention group) had higher response rates than those recruited earlier (C 1 and the comparison group) (~’‘8.5, df=l, p=0.003). Comparison of characteristics of women in C1 and C2 shows a good degree of similarity, with no significant differences evident (Table 3). The comparison and intervention groups were also similar in composition, except for a tendency for women in the intervention group to smoke fewer cigarettes per day at the first antenatal visit and to have older partners. Implementation of intervention In the intervention group, 90% of women recalled receiving advice about smoking at the first antenatal visit and 87% recalled receiving written materials about smoking, with only 3% recalling neither. In the comparison group, as well as C1 and C2, less than 75% of women recalled receiving advice to quit at the first visit and less than 58% remembered receiving written materials, with a minimum of 19%recalling neither. In contrast, only 29% of women in the intervention group recalled receiving postpartum advice to quit and 32% recalled receiving written materials, with 62% recalling neither. This was not significantly different from recall rates of women in the comparison group, C1 and C2. Thus, the antenatal component of the program was delivered to most women in the intervention group, but the postpartum component was poorly implemented. Outcome measures In the comparison of C1 and C2, there was no systematic pattern in the direction of differences in outcome measures and the magnitude of differences between groups was small.Together, these findings implied that a time effect was unlikely and that the comparison group could be directly compared with the intervention group. Figure 1 summarises outcome measures for the comparison and intervention groups. Women in the intervention group were slightly more likely to have tried to quit during pregnancy after the first antenatal visit (44.7%) than controls (34.6%) and more likely to have tried to quit in the postpartum period (intervention 48.6%; controls 33.3%), although neither of these differences reached statistical significance. Rates of biochemically verified sustained cessation in pregnancy to 32-34 weeks gestation were greater in the intervention group (9.3%) than among controls (2.8%) after adjustment for confounders (age and pre-pregnancy cigarette consumption) (OR=l.7; 95% CI=l .O-3.0; p=0.07). Point prevalence quit rates in late pregnancy were also greater in the intervention group (10.1%) than among controls (5.1%) (OR=1.4; 95% CI=O.9-2.2; p=O.1I), after adjustVOL. Table 2: Cumulative response rates to follow-up questionnaires, by group. ~~ ~ c 1 n=103 24-26weeks 32-34 weeks 6 months pfn 78.6% 68.9% 40.8% c 2 n=110 82.7% 68.2% 56.4% Comparison Intervention n=110 n=110 77.3% 65.5% 47.3% 73.6% 68.2% 60.9% AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 22 NO. 3 Effects of a smoking cessation program for pregnant women and their partners Table 3: Comparison of group characteristics. c1 (1-1403) Mean age (SD) Mean gestation (SD) Employment status (Yo unemployed) Occupational status, for employed only Postcode (Yolow) Gravidity (?/o >1) Parity (Yo >O) Mean pre-pregnancy cigarettes /day c2 (n=llO) P (C1 vs. C2) 0.45 0.57 0.45 0.62 0.35 0.36 0.41 0.23 0.35 0.58 0.45 0.45 0.48 0.26 0.49 0.30 Comparison (n=llO) Intervention P (n=110) (Cornp vs. Int) 24.5(4.7) 11.5(3.3) 25.0(5.4) 11.2(3.1) 25.5 47.1 49.1 78.2 60.9 21.8(7.9) 14.2(7.9) 27.2 96.4 28.7(6.7) 24.4(13.3) 84.4 42.3 45.0 24.3(5.6) 11.3(3.2) 20.9 40.0 3.6 64.5 47.3 20.9(8.3) 12.2(8.4) 34.7 92.7 26.5(5.6) 19.9(10.1) 74.7 19.6 55.0 24.9(5.1) 11.4(3.0) 19.1 52.1 1.8 69.1 52.7 23.0(9.1) 10.6(6.4) 35.5 90.0 27.9(5.5) 20.6(8.8) 72.9 21.2 50.0 (Yo high) 30.1 53.1 42.7 72.8 55.3 23.4(9.0) 13.4(8.4) 30.8 94.2 27.9(5.6) 22.7(11.6) 90.2 37.5 35.6 (SD) Mean cigarettes /day at first antenatal visit (SD) GHQ score (Yo >3) Partner ('/o yes) For those with partners: Mean partner age (SD) Mean partner cigarettes (SD) Smoker (% yes) Employment status (% unemployed) Occupational status, for employed only (Yo high) Favours Comparison Favours Intervention group group 0.1 0.5 1 5 10 WOMEN TRIED TO QUIT Tried to quit in pregnancy OR= 1.18 (0.94-1.47) Tried to quit post-partum OR= 1.16 (0.90-1.49) QUIT Quit at first and second f.u. OR = 1.70 (0.97-2.96) Quit at second f.u. only OR = 1.43 (0.92-2.23) Quit at first, second and third f.u. OR = 0.98 (0.52-1.65) Quit post-parturn only OR = 1.28 (0.80-2.06) PARTNERS TRIED TO QUIT Tried to quit in pregnancy OR = 2.94 (1.10-7.88) Tried to quit post-partum OR = 1.88 (0.68-5.17) QUIT Quit at second f.u. only OR = 0.87 (0.05-14.4) Quit post-partumonly OR = 0.87 (0.33-2.27) 4 - Figure 1:Adjusted odds ratios and 95% confidence intervalsfor dichotomous outcome variables. VOL. ing for age and pre-pregnancy cigarette consumption. However, cessation sustained to the six-month postpartum period did not differ between groups (controls 3.8%, intervention 4.5%; OR=l .O, 95% CI=0.5-1.9, p=0.95). Since rates of questionnaire non-response were high and may have led to potential selection bias, all non-respondents were assumed to be continuing smokers. When this assumption was made, the higher sustained pregnancy quit rate in the intervention group (6.4%) compared with the comparison group ( 1 3%) approached statistical significance (OR=l.5,95% CI=O.9-2.6, p=0.13). For late pregnancy point prevalence cessation, there was a modest trend towards increased cessation in the intervention group (8.2%) compared to the comparison group (3.6%) (OR=1.3, 95% CI=O.9-2.0, p=O. 18). These analyses suggest that even if all of those who were lost to follow-up were presumed to have continued smoking, the increased rate of cessation among intervention group women compared with controls would still have approached statistical significance. However, a similar series of analyses, where varying assumptions were made about non-responders, confirmed that the intervention was unlikely to have increased postpartum cessation. A significantly greater percentage of partners were reported as trying to quit in pregnancy in the intervention group (34.0%) than in the comparison group (14.9%) ( ~ ~ ' 4 . 8 , df=l, p=0.03). However, point prevalence quit rates did not differ between groups, either during pregnancy (2.1 % control, 1.8% intervention) or postpartum (2.8% control, 4.4% intervention). Analysis of data relating to cigarettes smoked per day suggested that the intervention was unlikely to have influenced cigarette consumption in pregnancy or postpartum, or paternal cigarette consumption at either follow-up point. While a greater percentage of women.had high GHQ scores at the first and second follow-up points in pregnancy in the intervention group than among women in the comparison group, after 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Wakefield and Jones adjustment for baseline scores and other confounders, there was no difference in GHQ scores between groups at the first follow-up (OR=l.l, 95% CI=O.9-1.5, p=0.35) second follow-up (OR=1.2, 95% CI=O.9-1.5, p=0.22) or postpartum (OR=0.88, 95% CI=0.61.2, p=0.37). The recurrent costs of the intervention condition ($2.75 per smoker) were greater than the comparison condition ($1.25 per smoker). However, the recurrent cost per validated abstainer (assuming sustained late pregnancy cessation rates in which all dropouts were presumed to be continuing smokers, i.e. 6.4% in the intervention group and 1.8% in the comparison group) were $35.50 in the intervention group and $69.40 in the comparison group. Thus, an approximate doubling of the cost of intervening with pregnant smokers would result in approximately 3.5 times the quit rate. Discussion Because this was not a randomised controlled trial, there were a number of sources of potential bias in the study that needed to be examined. Potential selection bias occurred at entry to the trial (sample loss through exclusions, refusals and failure to approach eligible women) and at follow-up (due to dropout for specified criteria and non-response to follow-up). Overall, 36.5% of pregnant women who smoked were ineligible to be recruited to the study. In C 1 and C2, there was no difference between groups in the extent of, or reasons for, exclusion. However, the intervention group was found to have a significantly greater percentage of exclusions than the comparison group (34.0% and 21.5% respectively), attributable to a greater number of shared-care patients presenting during the recruitment phase of the intervention group. This reflected the operation of a hospital policy over the course of the study which encouraged low-risk pregnancies to be referred back to their general practitioner for routine antenatal care. Such women would have been unlikely to have had complications in previous pregnancies or medical conditions which could complicate their current pregnancies, and would have had fewer unhealthy behaviour patterns with respect to drug use. These characteristics implied that such patients may have been more likely to quit smoking in pregnancy (since these same baseline characteristics in the study subjects were associated with several of the outcome variables). The differential exclusion of women in the intervention and comparison groups with respect to shared-care meant that it would be more difficult to show an improvement in cessation in the intervention group and therefore may have acted to underestimate the effect of the intervention. Overall, of the eligible women, 62% consented to participate in the study. This was lower than expected and can be explained by the fact that between 10% and 18% of eligible women were not approached by midwives to be part of the study. The administrative burden on midwives to recruit women to the study was designed to fit in as much as possible with the normal interview procedure. However, in the context of a busy antenatal clinic, where service provision outranks research in terms of priority, there were inevitably times when midwives forgot, or lacked sufficient time, to ask women to be involved in the study. Overall, the participation rates did not differ between groups. There were few differences and clearly no systematic pattern of differences between groups, in the characteristics of women who refused or were missed, compared with participants. Thus, there were no likely selection biases among those who were eligible to enter the trial which could have contributed to a spurious beneficial effect of the intervention. After recruitment, relatively few women were dropped from the study due to specified conditions (such as spontaneous abortion) and there was no differential rate of such dropout by group. During follow-up, however, response rates were disappointing, especially at the six-month postnatal follow-up. Cumulative response rates of 78.1% and 67.7% were obtained overall for the first two followups during pregnancy and, at this stage, response rates did not vary by group. At the six-month postnatal follow-up, response rates dropped to 52.2% overall, with a trend for the second two cohorts at each hospital achieving higher response rates at this time than the first two cohorts. Another study in Nottingham, England, which also followed women to six months after birth, reported similar response rates, commenting that one in seven women moved address during the pregnancy itself and that this pattern continued post-birth, restricting the effectiveness of postal follow-up from hospitaL2*Thus, during follow-up, there were potential biases introduced by the low response rates overall and by the differential response rates post-birth. However, re-analysis of the data based on a projection of likely behaviour change, given non-respondents’ baseline characteristics, did not markedly alter the direction or magnitude of group differences, compared with the initial analysis. This suggests that selection bias due to study attrition was unlikely to have accounted for observed differences in the study group outcome measures. Of all the sources of bias which may have potentially confounded the study results, there were none which would have artificially inflated the actual quit rate in the intervention group compared with the comparison group. As judged from the outcomes of C 1 and C2, there was no evidence to suggest that a time effect contributed to differences in outcome between the comparison and intervention groups. In addition, there were no sources of selection bias, either at entry to the trial or at follow-up, which would have accounted for improved outcome among women in the intervention group. The only source of likely selection bias (namely, an increased number of shared-care patients excluded from the intervention group compared with the comparison group) would, in fact, have resulted in an under estimate of the effect of the intervention. This leads to the conclusion that differences in antenatal outcome measures between the intervention and comparison groups were likely to be attributable to the intervention program, rather than underlying differences in the characteristics of the groups, or the result of other influences. In this trial, women who were ‘shared-care’ patients were excluded. In hindsight, it would have strengthened the study to include these patients. This is a practical proposition where intervention is focused on the first antenatal visit, as was the case in this trial, but may not be the case, in more intensive interventions which require multiple reinforcement and follow-up. Qualitative feedback from midwives suggested that most found the model to provide an interesting and helpful focus for discussing the smoking issue and were glad of new information to give to women about smoking. In their assessment, most women reacted 1in predictable ways to the model and most staff were comfortable with using the suggested ways of responding to the women provided during training. Coupled with the enthusiasm of the nurses VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 22 NO. 3 Effects of a smoking cessation program for pregnant women and their partners - to deal more effectively with the smoking issue and the fact that the store of patient booklets kept by the midwives needed regular restocking, this suggests that midwives did implement in a regular way, the antenatal components of the intervention that fell within their responsibility. At the six-month postnatal follow-up, recall of the postpartum intervention was low and this, together with the fact that the store of patient information booklets did not often require replenishing, suggests that the post-birth component of the intervention was not routinely carried out by nurses on the postnatal wards. It is likely that the extent and type of training provided to ward nurses - with frequent roster changes and high staff turnover, against a backdrop of organisational and structural change involving budget cuts and staff shortages - was insufficient to effect change in the behaviour of the nursing staff with respect to smoking cessation intervention. Further efforts to overcome the structural barriers preventing adequate training of midwives and the implementation of minimal smoking cessation intervention on the ward would need to be made to assist structured smoking cessation advice to become incorporated as a part of routine postpartum care. In future, it is likely that early discharge will be favoured for low-risk women after delivery and routine provision of advice and information about smoking cessation needs to take account of this trend. Hospital-based postpartum advice must be a priority, as relapse back to smoking occurs in most cases very soon after delivery and intervention, even at the time of the six-week postnatal follow-up, would be too late to prevent most relapse episodes.43 The rates of adjusted sustained cessation (6.4%) and point prevalence cessation (8.2%) in late pregnancy found in this trial are consistent with those reported in other studies which have used firm advice to quit and provision of written information about risk and tips for q ~ i t t i n g . ~ ~Windsor and~colleagues suggest that, among ,~~.~~, " antenatal patients attending public hospitals, strong advice to quit, readable risk information and reinforcement from the health care practitioner may increase the usual quit rate of 2-4% to 6-8%, as assessed in late pregnancy.44The inclusion of feedback is unlikely to improve upon these rates, at least when incorporated in the manner tested. Feedback of one's own foetus' heart rate after maternal smoking may be associated with greater motivation to quit smoking, compared with that simulated by a model, but this was not able to be tested in the non-smoking environment of the hospital. At the time this trial was conducted, no published intervention trial had specifically included a postpartum component to intervention in pregnancy. Although the trial attempted to include this, the implementation of the postpartum intervention was incomplete, with no more women in the intervention group recalling postpartum advice and written materials about smoking than controls and quit rates being not significantly different from controls. As a result, the present trial produced equivocal findings with regard to the question of whether the inclusion of a specific postpartum smoking cessation intervention incorporated into a minimal advice and feedback trial might influence postpartum cessation. Among males, partners in the intervention group were more likely to be reported as having tried to quit during pregnancy than controls. Despite this, there was no evidence for increased paternal cessation in pregnancy or postpartum. Very few other trials have attempted to influence the smoking behaviour of partners of pregVOL. nant women.J446,47 However, studies continue to show that partner cessation does influence maternal quit ratesZ5-27.30 that partners and do spontaneously make changes to their own smoking habits ~n It may be that partner support interventions might not increase partner quit rates per se, but might enable partners to be more supportive of women's cessation attempts, thereby increasing the likelihood of maternal cessation. Further efforts to develop interventions that might assist partners are worthy of attention. It was reassuring to note that, as judged by the GHQ, no adverse effects of exposure to the intervention on psychological stress were detected; scores in late pregnancy and postpartum did not differ between intervention and comparison groups. N o previous studies have attempted to measure unintended adverse effects of smoking cessation intervention, despite the fact that this is commonly raised as a potential barrier to giving smoking cessation advice and help during Further study is required to confirm these findings, as it is possible that other instruments may give a different picture. While minimal types of interventions have the appeal of being more easily integrated into antenatal care, cognitive-behavioural interventions involving more intensive skills-training and support to quit smoking are most likely to produce substantial changes in smoking behaviour in pregnancy, of the order of 10-15%cessation. However, antenatal clinics in public hospitals wishing to incorporate smoking cessation intervention into routine care need to assess to what extent they can service the requirements of programs using manuals that require additional time to introduce to patients and/or necessitate a system of follow-up support to provide additional counselling or written materials. In addition, many pregnant women may find extensive written materials cumbersome or unfamiliar to use Since many of the intervention trials have involved additional research funding for specific components of the intervention such as supportive follow-up visits, these types of interventions may no1 appeal to public hospital obstetric clinics which would need to find additional, ongoing funding for these components. In an environment of financial restraint and staff shortages, these types of strategies may not be preferred. In contrast, minimal intervention strategies that additionally appeal to the staff who will implement them may be more likely to be adopted routinely. Acknowledgment We wish to thank the staff and patients in the Department of Obstetrics and Gynaecology at Flinders Medical Centre and the Queen Elizabeth Hospital in Adelaide, project officers Jan Chorley and Susie Byrne and Professor Alistair Woodward for helpful comments on early drafts. The study was funded by the Research into Drug Abuse Grants program of the Commonwealth Department of Health and Family Services. Part of this work was undertaken while the first author was the recipient of an NHMRC Public HealthTravel Fellowship. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australian and New Zealand Journal of Public Health Wiley

Effects of a smoking cessation program for pregnant women and their partners attending a public hospital antenatal clinic

Download PDF
8 pages

Loading next page...
 
/lp/wiley/effects-of-a-smoking-cessation-program-for-pregnant-women-and-their-6V0SbDteQb

References (56)

Publisher
Wiley
Copyright
Copyright © 1998 Wiley Subscription Services, Inc., A Wiley Company
ISSN
1326-0200
eISSN
1753-6405
DOI
10.1111/j.1467-842X.1998.tb01383.x
Publisher site
See Article on Publisher Site

Abstract

This study aimed to assess the effect of a Abstract hospital-based smoking cessation intervention delivered by midwives during routine antenatal and postnatal care on the smoking habits of pregnant women and Warren Jones Department of Obstetrics and Gynaecology: Flinders Medical Centre, South Australia here is clear evidence that maternal smoking is hazardous to the development of the foetus and that cessation improves birthweight.’ In addition, studies indicate that exposure of infants to environmental tobacco smoke is associated with an excess risk of lower respiratory tract infections, so that sustained cessation postbirth is important.’ Smoking prevalence is highest among women of greater socio-economic disadvantage, who attend public hospital antenatal clinics for pregnancy care. Given the large numbers of women attending such clinics, considerable effort has been expended over the past decade to evaluate appropriate smoking cessation interventions for this setting. Late pregnancy quit rates of 6-8% using a minimal advice-giving procedure and 1215% using cognitive behavioural programs with repeated contact and reinforcement have been reported among women who present as smokers at the first antenatal visit, although most women relapse to smoking after deliver~.~-~ While higher pregnancy quit rates can potentially be achieved with a cognitivebehavioural program, the difficulty of incorporating such programs into busy clinic environments represents a major barrier to more widespread adoption. Programs that are minimal in nature may be a more achievable objective for routine application beyond the life of a research trial. Most intervention programs have emphasised the potential health hazards of maternal smoking on foetal development, particularly low birth-weight, as a primary source of motivation to quit. However, previous research has suggested that many women are easily able to discount the risks posed by smoking in pregnancy, due to their own experience or that of their friends and relatives VOL. who have had ‘normal babies’ after smoking throughout their pregnancies.’-” In addition, many women perceive that ‘small babies’ do not incur significant health An alternative strategy is to present more concrete and salient information about risk, using visual or auditory feedback, on the premise that such information is more likely to result in an increased sense of risk.I3-l5 Virtually all models of health behaviour include perceived risk as one factor that is likely to influence the decision to change risky behaviour. l6-Iy Although studies of feedback to women of carbon monoxide levels or visual information from ultrasound scans have produced mixed result^,^@^^ uncontrolled studies demonstrating the increase in foetal heart rate after the mother smokes a cigarette have produced changes in smoking habits worthy of further investigation. Kelly and O’Conor exposed women at 35 weeks gestation to changes in their foetus’ heart rate following the smoking of a cigarette; 19% quit for at least the remainder of their pregnancy and a further 50% at least halved their smoking rate.24Although personal feedback of foetal heart rate following maternal smoking would be likely to be more salient in enhancing perceived risk,13 time constraints and hospital policies which ban smoking may not make this option feasible. A recording which allows women to compare the sound of a normal foetal heart rate with that following one cigarette may provide concrete information about the effect of maternal smoking on foetal well being. Another aspect of smoking cessation programs which has remained undeveloped is the important role of partner’s smoking on maternal smoking and relapse. Smoking by partners of pregnant women is consistently and strongly related t o cessation in their partners. At the first antenatal visit, women in the intervention group (n=1 10) were given a demonstration of the immediate effects of smoking on foetal heart rate, brief smoking cessation advice and smoking cessation booklets for themselves and their partners; at delivery, they were given brief advice and a booklet about postpartum cessation. Compared with an historical control group who received usual care (n=1 10) and assuming those lost to follow-up continued to smoke, biochemically-verified maternal cessation sustained from at least 24 weeks gestation to late pregnancy was 6.4% in the intervention group and 1.8% in the comparison group. However, there was no difference between maternal quit rates at six months postpartum. Partners were more likely to try to quit in the intervention group, but quit rates did not differ. Exposure to the intervention was not associated with increased levels of psychological distress, as measured by the 12-item General Health Questionnaire. We conclude that this type of intervention, when implemented by staff during routine care, is probably associated with a small improvement in maternal cessation in pregnancy, similar to that produced by minimal advice to quit smoking. (Aust N Z J Public Health 1998;22:31 3-320) Correspondenceto: Dr Melanie Wakefield, Epidemiology Branch, South Australian Health Commission, PO’Box 6, Rundle Mall, South Australia, 5000, Fax (08) 8226 6316, e-mail wakefield.melanie @ health.sa.gov.au 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Wakefield and Jones pregnancy25,‘6and relapse p~stpartum.~’-*~ Women who receive encouragement from their partners to quit smoking are more likely to quit or reduce cigarette consumption.3” Despite this, few smoking cessation interventions in pregnancy have incorporated structured advice to quit, directed to partners. The present study sought to test the effectiveness of a novel minimal smoking cessation intervention delivered as part of routine antenatal and postnatal care. In the course of being given minimal advice to quit smoking by midwives, women were able to listen to the sound of a normal foetal heart beat and to compare this with the sound of a foetal heart beat five minutes after the mother had smoked a cigarette. An additional novel aspect of the intervention was that advice and information were provided to partners of pregnant women. The evaluation aimed to assess the effect of this advice on maternal and paternal smoking behaviour during pregnancy and up to six months postpartum. Many smoking interventions for pregnant women have been criticised for ignoring the problem of addiction, the meaning of the behaviour for the women involved, and the guilt and anxiety felt by those who continue to smoke in the face of exhortations to q ~ i t . ~ ’ Bryce & Enkin have questioned whether the guilt and anxiety that may be induced through maternal anti-smoking programs only causes more stress among pregnant women, particularly among Given this, the intervention trial also sought those least able to to determine whether there might be unintended adverse consequences of smoking cessation intervention, as manifested by increased levels of psychological distress. Subjects and methods Sample and design We chose not to conduct a randomised controlled trial as antenatal staff were to be trained in the delivery of stop-smoking advice and this would have influenced advice to the comparison group and also because such a design would have posed considerable organisational difficulties in the context of a busy antenatal clinic where recruitment of patients to the study was the responsibility of clinic staff. Rather, an historical comparison group was recruited from the hospital before the introduction of the intervention. As the comparison group was recruited at a different time, women may have been exposed to different types of influences on their smoking behaviour and the effect of a smoking cessation intervention could not be partitioned out using this design alone. For this reason, additional comparison groups were recruited concurrently from one other hospital (C1 and C2). By comparing C1 with C2, one could determine whether there was sufficient similarity in outcome measures to conclude that no extraneous influences on cessation, occurring concurrently with data collection for the comparison and intervention groups, were apparent (i.e. a ‘time’ or ‘history’effe~t).~~ If there was no evidence for such an influence, the comparison group could be directly compared with the intervention group at the index hospital. However, if a such a time effect was apparent, it would need to be taken into consideration when the comparison and intervention groups were compared. Women eligible to enter the trial were public patients who reported being smokers at the first antenatal appointment. Of these women, potential participants were excluded if they reported smoking one or fewer cigarettes per day, were ‘shared-care’ patients (those whose antenatal care was mainly managed by a general practitioner), if they had existing psychiatric disturbance including a history of severe puerperal depression, were non-English speaking (intervention materials had not been produced in other languages) or presented for their first antenatal visit later than 20 weeks gestation. Women were dropped from further follow-up if they had a spontaneous abortion or termination, a stillbirth, a delivery before 34 weeks gestation (making a late pregnancy assessment of smoking status impossible) or changed hospitals during the antenatal period, thus making their case notes inaccessible. Natural quit rates from other studies indicate that an expected quit rate in the comparison groups between the end of the first trimester and delivery would be 4%.4 Other intervention studies with public antenatal patients suggest that firm advice to quit, supported by patient materials can achieve a quit rate of approximately 15%.4 With a power of 90% and where a=0.05,a sample size of 108 would be required in each group.34 The study aimed to recruit an additional 17 subjects per group, making 125 in all, in order to allow for expected drop-out due to adverse pregnancy outcome. Women who met the eligibility criteria were asked by the midwife conducting the first antenatal visit to be involved in the study. Those who consented completed the 12-item General Health Questionnaire (GHQ) at the time of their visit.35Information pertaining to smoking status and level of cigarette consumption of pregnant women and their partners, as well as other relevant baseline information, was abstracted from the medical case notes that had been made up during the first antenatal appointment. Patients were followed up at 24-26 weeks gestation, 32-34 weeks gestation and six months post-delivery by mailed self-completion questionnaire, accompanied by a post-paid return envelope. If questionnaires were not returned within two weeks, a reminder letter was sent, followed by a telephone call from a research assistant if no reply had been received one to two weeks later. At 32-34 weeks gestation, those women who claimed that they had quit smoking were additionally asked to provide a urine sample for cotinine assay. Recruitment for the comparison group and C1 occurred between October 1990 and April 1991 and follow-up to six-month postpartum was completed in June 1992. Recruitment for the intervention condition and C2 occurred betweenAugust 1991 and February 1992 and final follow-up was completed in March 1993. Intervention During discussion of the issue of smoking in the first antenatal appointment, midwives used a hard moulded-plastic desk-top model of a foetus-in-utero. The model simulated the increase in foetal heart rate that follows the smoking of a cigarette by the mother, from approximately 130 to 150 beats per minute, an increase clearly perceptible to the human ear. It was explained to the pregnant woman that this effect was due to the nicotine and carbon monoxide in cigarettes making less oxygen available to the baby. The one-toone setting permitted patients to raise concerns and questions about the model with the midwife. Advice to quit was supported by a small booklet which further explained the immediate effects of maternal smoking on the foetus, detailed the benefits of quitting, presented counter-arguments to some of the bamers women most commonly identified to achieving cessation and provided practical VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 22 NO. 3 Effects of a smoking cessation program for pregnant women and their partners suggestions for quitting. A booklet for partners who smoked was also provided, whenever women reported that their partner was a smoker. Before leaving the hospital after delivery, further advice was given by a primary nurse to both continuing smokers and those who may have quit during pregnancy, about smoking and infant health and other associated issues such as breastfeeding and weight loss after pregnancy. This advice was reinforced by the provision of a booklet to each woman. The intervention materials were developed with the assistance of information and fee:dback from a series of focus group discussions of pregnant women and their partners.36 The booklets were written in a non-judgmental manner and raised issues that had been identified to be of concern to women in the focus group discussions. Each booklet had a Flesch grade level of 5 or 6, indicating the text was of an easy or fairly easy ~ t a n d a r d . ~ ' Women in comparison groups received usual care, which most often involved a verbal caution that smoking in pregnancy was not advisable. Brief pamphlets about smoking were available, although these were not pregnancy-specific. reductions in consumption were assessed by comparing the number of cigarettes per day reported at the 32-34 gestational week followup, after adjustment for pre-pregnancy cigarette consumption and consumption at the first antenatal appointment. To assess potentially adverse effects of the intervention, women were asked to complete the 12-item GHQ at baseline and each follow-up point. This enabled tracking of psychological functioning during and after pregnancy. The GHQ has previously been shown to be responsive to change in a pregnant population."OThose women who scored 3 or higher were designated as 'high scorer^'.^' Estimates of the recurrent costs of the intervention, based on staff time, cost of the desktop demonstration models and quit smoking resources, were made. Recurrent costs provide an indication of how much it would cost similar antenatal clinics to implement the usual care and intervention programs. Statisticalanalysis For each outcome measure, a comparison was undertaken between C 1 and C2, after adjustment for potential confounders. Judgments about whether or not a time effect had occurred were based on the size of the difference in outcome between C 1 and C2 and the consistency of the difference in outcome measures. Where there was no evidence of a time effect, it was assumed that the comparison and intervention groups could be directly compared. For dichotomous outcome variables (such as attempt to quit and quit status), univariate associations with independent variables were assessed by conventional chi-square tests and statistical adjustment was performed by using logistic regression analysis. Variables considered as candidates for inclusion in each logistic regression model were those which were associated at the univariate level with the outcome variable at a p value of <0.25,42or which were required to be included for a priori reasons, namely, baseline daily cigarette consumption. For each logistic regression, a check for multicollinearity was performed by reference to the eigenvalues generated in a factor analysis of candidate independent variables, resulting in highly inter-correlated variables being discarded. Eligible variables were then included, with group membership, in a series of logistic regression models, with independent variables being discarded if they failed to condition the risk estimate for group membership. Odds ratios and 95% confidence intervals were calculated from the final model. All continuous variables, where not normally distributed, were transformed by a square root transformation and re-assessed for normality. Univariate association of baseline variables with continuous outcome variables and group membership, was assessed by means of conventional t-tests. Analysis of covariance was used to adjust for the effect of potential confounders. Variables considered as candidates for inclusion in each analysis of covariance with a p value of <0.25, were subjected to a factor analysis to check for multi-collinearity, resulting in highly correlated variables being discarded. The remaining variables were added as covariates in an analysis of covariance, with group membership being the main independent variable. The adjusted means yielded in this analysis were then squared to transform them back to their index values. Initial analysis of each of the outcome measures was undertaken using only those patients who responded at the relevant follow-up point. Subsequent analyses for dichotomous outcome variables were Measures At each of the three follow-up points, women were asked whether they were smoking and, if so, how many cigarettes, on average, they smoked per day, whether they had tried to quit and, if so, how long ago their last attempt was. Women who reported that they were no longer smoking were asked how long it had been since their last cigarette. In addition, at each follow-up point, those women with a partner were asked identical questions about his smoking habits. A woman or her partner were considered as having tried to quit smoking during pregnancy if it was indicated that they tried to quit at any time between baseline and the 32-34 week follow-up. An attempt to quit postnatally was judged by those women who indicated that they or their partner tried to quit between delivery and the six-month postnatal follow-up. In assessing the outcome of intervention studies, a combination of point prevalence and continuous abstinence measures has been r e c ~ m r n e n d e d . ~ ~ measure of continuous abstinence, a woman For the was considered to have quit smoking during pregnancy if she indicated she had quit smoking at the 24-26 gestational week followup (and had not smoked any cigarettes in the preceding seven days) and was still abstinent, as confirmed by a urinary cotinine assay, at the 32-34 gestational week follow-up. Those women claiming abstinence with urinary cotinine levels above 100 ng/ml, indicating that they had most probably smoked in the 48 hours prior to the sample being taken, were considered to be smoker^.^' For the measure of point prevalencte cessation, the criterion was self-reported cessation at the 32-34 week follow-up (with no smoking during the previous week), confirmed by biochemical validation. For partners, the criterion was identical, except that biochemical assessment was not available. Maintenance of non-smoking postpartum was considered to have been achieved for those women who indicated they were not smoking at the six-month postnatal follow-up and who smoked their last cigarette before the 24-26 week follow-up. The same criteria were applied in the classification of partner's smoking status, except that a. cotinine assay was not available to validate self-reported smoking status at the 32-34 week follow-up. Among those who lwere not considered to have quit smoking, VOL. 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Wakefield and Jones Table 1: Participation rates, by group. c 1 Attended first antenatal visit Smoker Eligible smokers Recruited eligible smokers Available for follow-up c2 903 341 (37.8%) 201 (58.9%) 128 (63.7%) 110 (85.9%) Comparison 83 1 265 (32.6%) 208 (78.5%) 128 (61.5%) 110 (85.9%) intervention 961 303 (31.5%) 200 (66.0%) 127 (63.5%) 110 (86.6%) 1,000 355 (35.5%) 194 (54.6%) 121 (62.4%) 103 (85.1 Yo) undertaken assuming all non-respondents to follow-up questionnaires were continuing smokers. The research trial was approved by the ethics committees of both participating hospitals. Results Sample recruitment and follow-up Table 1 shows that, of all pregnant women smokers attending for their first antenatal visit during the study period, between 33% and 45% were ineligible. Exclusions occurred due to presentation after 20 weeks gestation (38.6%), being a shared-care patient (36.4%), presence of a psychiatric history of concern (1 2.4%), smoking fewer than 2 cigarettes per day (10.8%) and being non-English speaking (1.8%). In C1 and C2, the most common reason for exclusion was presentation after 20 weeks gestation, while in the comparison and intervention groups, the most common reason was shared care. Of smokers eligible to be approached for inclusion in the study, around 62% consented to participate; 20-25% of eligible smokers in each group declined to participate and the remainder were unfortunately not approached by midwives to be part of the study. A comparison of the characteristics of eligible smokers who consented to participate in the study with those who refused or were missed was undertaken for each of the four groups, revealing no systematic pattern of statistically significant differences between participants and non-participants. During the trial, 71 women were dropped from the study due to previously specified criteria. Most had experienced either an elective or spontaneous abortion (28.2%) or had cancelled subsequent visits due to abortion (36.6%). Most of the remaining cases were dropped from further follow-up because they had changed hospitals during the antenatal period. Reasons for dropout did not differ between groups. Dropouts were compared with study participants and no differences were detected in study characteristics. Thus, 110 women were available for followup in each group, except C 2 with 103 women. Table 2 shows the cumulative response rates to follow-up questionnaires, by group. There was no significant difference in response rates between groups at the first and second follow-up points. However, at the final follow-up point, the second two groups of women recruited at each hospital (C2 and the intervention group) had higher response rates than those recruited earlier (C 1 and the comparison group) (~’‘8.5, df=l, p=0.003). Comparison of characteristics of women in C1 and C2 shows a good degree of similarity, with no significant differences evident (Table 3). The comparison and intervention groups were also similar in composition, except for a tendency for women in the intervention group to smoke fewer cigarettes per day at the first antenatal visit and to have older partners. Implementation of intervention In the intervention group, 90% of women recalled receiving advice about smoking at the first antenatal visit and 87% recalled receiving written materials about smoking, with only 3% recalling neither. In the comparison group, as well as C1 and C2, less than 75% of women recalled receiving advice to quit at the first visit and less than 58% remembered receiving written materials, with a minimum of 19%recalling neither. In contrast, only 29% of women in the intervention group recalled receiving postpartum advice to quit and 32% recalled receiving written materials, with 62% recalling neither. This was not significantly different from recall rates of women in the comparison group, C1 and C2. Thus, the antenatal component of the program was delivered to most women in the intervention group, but the postpartum component was poorly implemented. Outcome measures In the comparison of C1 and C2, there was no systematic pattern in the direction of differences in outcome measures and the magnitude of differences between groups was small.Together, these findings implied that a time effect was unlikely and that the comparison group could be directly compared with the intervention group. Figure 1 summarises outcome measures for the comparison and intervention groups. Women in the intervention group were slightly more likely to have tried to quit during pregnancy after the first antenatal visit (44.7%) than controls (34.6%) and more likely to have tried to quit in the postpartum period (intervention 48.6%; controls 33.3%), although neither of these differences reached statistical significance. Rates of biochemically verified sustained cessation in pregnancy to 32-34 weeks gestation were greater in the intervention group (9.3%) than among controls (2.8%) after adjustment for confounders (age and pre-pregnancy cigarette consumption) (OR=l.7; 95% CI=l .O-3.0; p=0.07). Point prevalence quit rates in late pregnancy were also greater in the intervention group (10.1%) than among controls (5.1%) (OR=1.4; 95% CI=O.9-2.2; p=O.1I), after adjustVOL. Table 2: Cumulative response rates to follow-up questionnaires, by group. ~~ ~ c 1 n=103 24-26weeks 32-34 weeks 6 months pfn 78.6% 68.9% 40.8% c 2 n=110 82.7% 68.2% 56.4% Comparison Intervention n=110 n=110 77.3% 65.5% 47.3% 73.6% 68.2% 60.9% AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 22 NO. 3 Effects of a smoking cessation program for pregnant women and their partners Table 3: Comparison of group characteristics. c1 (1-1403) Mean age (SD) Mean gestation (SD) Employment status (Yo unemployed) Occupational status, for employed only Postcode (Yolow) Gravidity (?/o >1) Parity (Yo >O) Mean pre-pregnancy cigarettes /day c2 (n=llO) P (C1 vs. C2) 0.45 0.57 0.45 0.62 0.35 0.36 0.41 0.23 0.35 0.58 0.45 0.45 0.48 0.26 0.49 0.30 Comparison (n=llO) Intervention P (n=110) (Cornp vs. Int) 24.5(4.7) 11.5(3.3) 25.0(5.4) 11.2(3.1) 25.5 47.1 49.1 78.2 60.9 21.8(7.9) 14.2(7.9) 27.2 96.4 28.7(6.7) 24.4(13.3) 84.4 42.3 45.0 24.3(5.6) 11.3(3.2) 20.9 40.0 3.6 64.5 47.3 20.9(8.3) 12.2(8.4) 34.7 92.7 26.5(5.6) 19.9(10.1) 74.7 19.6 55.0 24.9(5.1) 11.4(3.0) 19.1 52.1 1.8 69.1 52.7 23.0(9.1) 10.6(6.4) 35.5 90.0 27.9(5.5) 20.6(8.8) 72.9 21.2 50.0 (Yo high) 30.1 53.1 42.7 72.8 55.3 23.4(9.0) 13.4(8.4) 30.8 94.2 27.9(5.6) 22.7(11.6) 90.2 37.5 35.6 (SD) Mean cigarettes /day at first antenatal visit (SD) GHQ score (Yo >3) Partner ('/o yes) For those with partners: Mean partner age (SD) Mean partner cigarettes (SD) Smoker (% yes) Employment status (% unemployed) Occupational status, for employed only (Yo high) Favours Comparison Favours Intervention group group 0.1 0.5 1 5 10 WOMEN TRIED TO QUIT Tried to quit in pregnancy OR= 1.18 (0.94-1.47) Tried to quit post-partum OR= 1.16 (0.90-1.49) QUIT Quit at first and second f.u. OR = 1.70 (0.97-2.96) Quit at second f.u. only OR = 1.43 (0.92-2.23) Quit at first, second and third f.u. OR = 0.98 (0.52-1.65) Quit post-parturn only OR = 1.28 (0.80-2.06) PARTNERS TRIED TO QUIT Tried to quit in pregnancy OR = 2.94 (1.10-7.88) Tried to quit post-partum OR = 1.88 (0.68-5.17) QUIT Quit at second f.u. only OR = 0.87 (0.05-14.4) Quit post-partumonly OR = 0.87 (0.33-2.27) 4 - Figure 1:Adjusted odds ratios and 95% confidence intervalsfor dichotomous outcome variables. VOL. ing for age and pre-pregnancy cigarette consumption. However, cessation sustained to the six-month postpartum period did not differ between groups (controls 3.8%, intervention 4.5%; OR=l .O, 95% CI=0.5-1.9, p=0.95). Since rates of questionnaire non-response were high and may have led to potential selection bias, all non-respondents were assumed to be continuing smokers. When this assumption was made, the higher sustained pregnancy quit rate in the intervention group (6.4%) compared with the comparison group ( 1 3%) approached statistical significance (OR=l.5,95% CI=O.9-2.6, p=0.13). For late pregnancy point prevalence cessation, there was a modest trend towards increased cessation in the intervention group (8.2%) compared to the comparison group (3.6%) (OR=1.3, 95% CI=O.9-2.0, p=O. 18). These analyses suggest that even if all of those who were lost to follow-up were presumed to have continued smoking, the increased rate of cessation among intervention group women compared with controls would still have approached statistical significance. However, a similar series of analyses, where varying assumptions were made about non-responders, confirmed that the intervention was unlikely to have increased postpartum cessation. A significantly greater percentage of partners were reported as trying to quit in pregnancy in the intervention group (34.0%) than in the comparison group (14.9%) ( ~ ~ ' 4 . 8 , df=l, p=0.03). However, point prevalence quit rates did not differ between groups, either during pregnancy (2.1 % control, 1.8% intervention) or postpartum (2.8% control, 4.4% intervention). Analysis of data relating to cigarettes smoked per day suggested that the intervention was unlikely to have influenced cigarette consumption in pregnancy or postpartum, or paternal cigarette consumption at either follow-up point. While a greater percentage of women.had high GHQ scores at the first and second follow-up points in pregnancy in the intervention group than among women in the comparison group, after 22 NO. 3 AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH Wakefield and Jones adjustment for baseline scores and other confounders, there was no difference in GHQ scores between groups at the first follow-up (OR=l.l, 95% CI=O.9-1.5, p=0.35) second follow-up (OR=1.2, 95% CI=O.9-1.5, p=0.22) or postpartum (OR=0.88, 95% CI=0.61.2, p=0.37). The recurrent costs of the intervention condition ($2.75 per smoker) were greater than the comparison condition ($1.25 per smoker). However, the recurrent cost per validated abstainer (assuming sustained late pregnancy cessation rates in which all dropouts were presumed to be continuing smokers, i.e. 6.4% in the intervention group and 1.8% in the comparison group) were $35.50 in the intervention group and $69.40 in the comparison group. Thus, an approximate doubling of the cost of intervening with pregnant smokers would result in approximately 3.5 times the quit rate. Discussion Because this was not a randomised controlled trial, there were a number of sources of potential bias in the study that needed to be examined. Potential selection bias occurred at entry to the trial (sample loss through exclusions, refusals and failure to approach eligible women) and at follow-up (due to dropout for specified criteria and non-response to follow-up). Overall, 36.5% of pregnant women who smoked were ineligible to be recruited to the study. In C 1 and C2, there was no difference between groups in the extent of, or reasons for, exclusion. However, the intervention group was found to have a significantly greater percentage of exclusions than the comparison group (34.0% and 21.5% respectively), attributable to a greater number of shared-care patients presenting during the recruitment phase of the intervention group. This reflected the operation of a hospital policy over the course of the study which encouraged low-risk pregnancies to be referred back to their general practitioner for routine antenatal care. Such women would have been unlikely to have had complications in previous pregnancies or medical conditions which could complicate their current pregnancies, and would have had fewer unhealthy behaviour patterns with respect to drug use. These characteristics implied that such patients may have been more likely to quit smoking in pregnancy (since these same baseline characteristics in the study subjects were associated with several of the outcome variables). The differential exclusion of women in the intervention and comparison groups with respect to shared-care meant that it would be more difficult to show an improvement in cessation in the intervention group and therefore may have acted to underestimate the effect of the intervention. Overall, of the eligible women, 62% consented to participate in the study. This was lower than expected and can be explained by the fact that between 10% and 18% of eligible women were not approached by midwives to be part of the study. The administrative burden on midwives to recruit women to the study was designed to fit in as much as possible with the normal interview procedure. However, in the context of a busy antenatal clinic, where service provision outranks research in terms of priority, there were inevitably times when midwives forgot, or lacked sufficient time, to ask women to be involved in the study. Overall, the participation rates did not differ between groups. There were few differences and clearly no systematic pattern of differences between groups, in the characteristics of women who refused or were missed, compared with participants. Thus, there were no likely selection biases among those who were eligible to enter the trial which could have contributed to a spurious beneficial effect of the intervention. After recruitment, relatively few women were dropped from the study due to specified conditions (such as spontaneous abortion) and there was no differential rate of such dropout by group. During follow-up, however, response rates were disappointing, especially at the six-month postnatal follow-up. Cumulative response rates of 78.1% and 67.7% were obtained overall for the first two followups during pregnancy and, at this stage, response rates did not vary by group. At the six-month postnatal follow-up, response rates dropped to 52.2% overall, with a trend for the second two cohorts at each hospital achieving higher response rates at this time than the first two cohorts. Another study in Nottingham, England, which also followed women to six months after birth, reported similar response rates, commenting that one in seven women moved address during the pregnancy itself and that this pattern continued post-birth, restricting the effectiveness of postal follow-up from hospitaL2*Thus, during follow-up, there were potential biases introduced by the low response rates overall and by the differential response rates post-birth. However, re-analysis of the data based on a projection of likely behaviour change, given non-respondents’ baseline characteristics, did not markedly alter the direction or magnitude of group differences, compared with the initial analysis. This suggests that selection bias due to study attrition was unlikely to have accounted for observed differences in the study group outcome measures. Of all the sources of bias which may have potentially confounded the study results, there were none which would have artificially inflated the actual quit rate in the intervention group compared with the comparison group. As judged from the outcomes of C 1 and C2, there was no evidence to suggest that a time effect contributed to differences in outcome between the comparison and intervention groups. In addition, there were no sources of selection bias, either at entry to the trial or at follow-up, which would have accounted for improved outcome among women in the intervention group. The only source of likely selection bias (namely, an increased number of shared-care patients excluded from the intervention group compared with the comparison group) would, in fact, have resulted in an under estimate of the effect of the intervention. This leads to the conclusion that differences in antenatal outcome measures between the intervention and comparison groups were likely to be attributable to the intervention program, rather than underlying differences in the characteristics of the groups, or the result of other influences. In this trial, women who were ‘shared-care’ patients were excluded. In hindsight, it would have strengthened the study to include these patients. This is a practical proposition where intervention is focused on the first antenatal visit, as was the case in this trial, but may not be the case, in more intensive interventions which require multiple reinforcement and follow-up. Qualitative feedback from midwives suggested that most found the model to provide an interesting and helpful focus for discussing the smoking issue and were glad of new information to give to women about smoking. In their assessment, most women reacted 1in predictable ways to the model and most staff were comfortable with using the suggested ways of responding to the women provided during training. Coupled with the enthusiasm of the nurses VOL. AUSTRALIAN AND NEW ZEALAND JOURNAL OF PUBLIC HEALTH 22 NO. 3 Effects of a smoking cessation program for pregnant women and their partners - to deal more effectively with the smoking issue and the fact that the store of patient booklets kept by the midwives needed regular restocking, this suggests that midwives did implement in a regular way, the antenatal components of the intervention that fell within their responsibility. At the six-month postnatal follow-up, recall of the postpartum intervention was low and this, together with the fact that the store of patient information booklets did not often require replenishing, suggests that the post-birth component of the intervention was not routinely carried out by nurses on the postnatal wards. It is likely that the extent and type of training provided to ward nurses - with frequent roster changes and high staff turnover, against a backdrop of organisational and structural change involving budget cuts and staff shortages - was insufficient to effect change in the behaviour of the nursing staff with respect to smoking cessation intervention. Further efforts to overcome the structural barriers preventing adequate training of midwives and the implementation of minimal smoking cessation intervention on the ward would need to be made to assist structured smoking cessation advice to become incorporated as a part of routine postpartum care. In future, it is likely that early discharge will be favoured for low-risk women after delivery and routine provision of advice and information about smoking cessation needs to take account of this trend. Hospital-based postpartum advice must be a priority, as relapse back to smoking occurs in most cases very soon after delivery and intervention, even at the time of the six-week postnatal follow-up, would be too late to prevent most relapse episodes.43 The rates of adjusted sustained cessation (6.4%) and point prevalence cessation (8.2%) in late pregnancy found in this trial are consistent with those reported in other studies which have used firm advice to quit and provision of written information about risk and tips for q ~ i t t i n g . ~ ~Windsor and~colleagues suggest that, among ,~~.~~, " antenatal patients attending public hospitals, strong advice to quit, readable risk information and reinforcement from the health care practitioner may increase the usual quit rate of 2-4% to 6-8%, as assessed in late pregnancy.44The inclusion of feedback is unlikely to improve upon these rates, at least when incorporated in the manner tested. Feedback of one's own foetus' heart rate after maternal smoking may be associated with greater motivation to quit smoking, compared with that simulated by a model, but this was not able to be tested in the non-smoking environment of the hospital. At the time this trial was conducted, no published intervention trial had specifically included a postpartum component to intervention in pregnancy. Although the trial attempted to include this, the implementation of the postpartum intervention was incomplete, with no more women in the intervention group recalling postpartum advice and written materials about smoking than controls and quit rates being not significantly different from controls. As a result, the present trial produced equivocal findings with regard to the question of whether the inclusion of a specific postpartum smoking cessation intervention incorporated into a minimal advice and feedback trial might influence postpartum cessation. Among males, partners in the intervention group were more likely to be reported as having tried to quit during pregnancy than controls. Despite this, there was no evidence for increased paternal cessation in pregnancy or postpartum. Very few other trials have attempted to influence the smoking behaviour of partners of pregVOL. nant women.J446,47 However, studies continue to show that partner cessation does influence maternal quit ratesZ5-27.30 that partners and do spontaneously make changes to their own smoking habits ~n It may be that partner support interventions might not increase partner quit rates per se, but might enable partners to be more supportive of women's cessation attempts, thereby increasing the likelihood of maternal cessation. Further efforts to develop interventions that might assist partners are worthy of attention. It was reassuring to note that, as judged by the GHQ, no adverse effects of exposure to the intervention on psychological stress were detected; scores in late pregnancy and postpartum did not differ between intervention and comparison groups. N o previous studies have attempted to measure unintended adverse effects of smoking cessation intervention, despite the fact that this is commonly raised as a potential barrier to giving smoking cessation advice and help during Further study is required to confirm these findings, as it is possible that other instruments may give a different picture. While minimal types of interventions have the appeal of being more easily integrated into antenatal care, cognitive-behavioural interventions involving more intensive skills-training and support to quit smoking are most likely to produce substantial changes in smoking behaviour in pregnancy, of the order of 10-15%cessation. However, antenatal clinics in public hospitals wishing to incorporate smoking cessation intervention into routine care need to assess to what extent they can service the requirements of programs using manuals that require additional time to introduce to patients and/or necessitate a system of follow-up support to provide additional counselling or written materials. In addition, many pregnant women may find extensive written materials cumbersome or unfamiliar to use Since many of the intervention trials have involved additional research funding for specific components of the intervention such as supportive follow-up visits, these types of interventions may no1 appeal to public hospital obstetric clinics which would need to find additional, ongoing funding for these components. In an environment of financial restraint and staff shortages, these types of strategies may not be preferred. In contrast, minimal intervention strategies that additionally appeal to the staff who will implement them may be more likely to be adopted routinely. Acknowledgment We wish to thank the staff and patients in the Department of Obstetrics and Gynaecology at Flinders Medical Centre and the Queen Elizabeth Hospital in Adelaide, project officers Jan Chorley and Susie Byrne and Professor Alistair Woodward for helpful comments on early drafts. The study was funded by the Research into Drug Abuse Grants program of the Commonwealth Department of Health and Family Services. Part of this work was undertaken while the first author was the recipient of an NHMRC Public HealthTravel Fellowship.

Journal

Australian and New Zealand Journal of Public Health – Wiley

Published: Jun 1, 1998

There are no references for this article.