Smoking in pregnancy is a well documented and potentially avoidable risk factor for a multitude of conditions, including miscarriage, low birth weight, perinatal death, childhood asthma and atopic disease [1–3]. Despite evidence-based knowledge of the harmful effects, tobacco smoking is still prevalent during pregnancy.

Prevalence studies in the 1980s showed that one in three pregnant women in Norway smoked during pregnancy, at that time among the highest smoking prevalence in Europe [4–6]. Eriksson et al. showed that the point prevalence of smoking at 18 weeks of gestation in Trondheim was 34% in 1987 compared to 22% in 1994, a statistically significant reduction. No effect of the national campaign against smoking during pregnancy launched in 1989 was found [7]. Public health interventions and smoking bans have since then shown success in some Western countries [8]. Norway has a history of more than 40 years of regulation of tobacco advertising and tobacco smoking in public. The 1975 Tobacco Act involved an advertising ban, 16 years age limit for buying tobacco products and labelling of tobacco products. Restrictions on smoking in public restaurants, bars, cafes, pubs and discotheques came in 1993, but a total ban on smoking in restaurants and bars first took effect on June 1st 2004. The first national comprehensive mass media campaign on tobacco and health for many years was accomplished during the study period in 2003.

A review article from 2000 stated that pregnancy and the postpartum period provide a window of opportunity to promote smoking cessation[9]. A Cochrane review from 2004 concluded that smoking cessation programmes in pregnancy reduce the proportion of women who continue to smoke [10]. Further, a meta analysis from 2000 found that a brief cessation counselling session of 5–15 minutes, when delivered by a trained provider with the provision of pregnancy specific self help materials, significantly increased rates of cessation among pregnant smokers, and these evidence based procedures were recommended to be adopted by all prenatal health care providers [11].

In 1997 the Norwegian Government appointed Trondheim as a model city to try out a new public intervention to counteract the rising incidence of asthma and allergic diseases. It was a prerequisite that the intervention programme should be possible to implement in ordinary pre- and postnatal care, without extra cost, and within normal consultation timeframe. The PACT-study was initiated in collaboration between the Norwegian University of Science and Technology (NTNU) and the municipality of Trondheim. The PACT study is a still ongoing, controlled, prospective, intervention study that was started in 2000 [12]. The primary objectives of the PACT study were to investigate the effectiveness of the risk-factor intervention on behavioural changes among parents, secondly to investigate the efficacy on the incidence of allergic diseases in the offspring from increasing omega-3- fatty acid intake and reducing parental smoking and indoor dampness [13].

The objectives of this study were to investigate parental smoking behaviour during pregnancy after introduction of a prenatal, structured, multi-disciplinary smoking cessation programme in primary care, and to compare smoking behaviour among pregnant women in the city of Trondheim with Bergen and Norway.

The study was performed in the city of Trondheim, the capitol city in middle Norway with 160 000 inhabitants and approximately 2100 deliveries per year. The city holds a University with 20 000 students and 4500 employees. In all, 28 of 35 general practices (90 general practitioners), all eight community based midwifes and all 20 maternity health centres in Trondheim agreed to participate in the PACT study.

Cohorts and subjects

Sequential birth cohorts were established to evaluate the intervention programme. From September 1^st 2000 to May 30^th 2002 all pregnant women who consulted their GPs or community based midwifes for pregnancy care were eligible to participate in the control cohort of the PACT study. Of some 3600 eligible pregnant women in Trondheim during this period, 1788 (50%) women were included and completed the pregnancy questionnaire (Q1), and 1023 (57%) of the participating women completed the questionnaire (Q2) six weeks after delivery (Figure 1). Participating women in the control cohort received common, nationwide recommended, advice on life-style, including smoking behaviour, following the routines each health-worker was familiar with at that time.

From June 1^st 2002 to December 15^th 2004 women were invited and included to the intervention cohort of the study. Of some 5200 pregnant women eligible to participate in the intervention cohort during this period, 2051 (40%) women gave their consent and answered the pregnancy questionnaire, and 1109 (54%) of the participating women completed the questionnaire six weeks after delivery.

All pregnant women were eligible to the PACT study if they were able to understand and fill in a questionnaire in Norwegian language with no other inclusion or exclusion criteria for either cohort.

We found a low smoking prevalence at inclusion, 4.9% and 7.1% in the intervention cohort and the control cohort, respectively. Only a quarter of the pre-pregnancy smoking women still smoked at inclusion time with no difference between the cohorts. During the intervention period from inclusion until six weeks postnatal, 7 in 10 smokers still smoked six weeks postnatal with no significant difference between the cohorts. At inclusion 18% and 28% reported indoor smoking in the intervention cohort and the control cohort, respectively. At six weeks postnatal very few of the smokers reported indoor smoking, only one parent in the intervention cohort, and nine parents in the control cohort.

Data from MBR illustrating quitting rates in Trondheim, Bergen and Norway showed a seemingly higher proportion of women who stopped smoking at the beginning or during pregnancy in Trondheim than in Bergen in 2003 and 2004.

The study had a controlled design comparing sequential total and unselected cohorts of pregnant women from the beginning of pregnancy until six weeks postnatal. Choosing a controlled design including whole birth cohorts made it possible in a real life approach to test the intervention programme. The assessment of smoking behaviour was consistent through the observation period and across cohorts, and independent of clinical practice. Furthermore, the majority of care providers were trained and motivated to deliver the recommended intervention modalities on repeated occasions both to those who smoked and those who had quit smoking [18]. Finally, when health professionals take part in a scheduled and structured intervention, it may counteract any potential negative beliefs and attitudes against promoting smoking cessation [19]. The possibility to compare smoking cessation nationally and in the two comparable university cities of Trondheim and Bergen in the same period that the sequential cohorts in PACT were investigated was an additional strength of the study.

The one year time difference between the control cohort and intervention cohort might have biased the results towards a better effect of the intervention due to secular trends. However, this was the design of choice primarily because a public and community based intervention including the entire primary health care in the municipality would be impossible to implement without contaminating a co-existing control cohort. Secondly, comparing total birth cohorts also ensured high conformity between the cohorts regarding population size, race/ethnicity, maternal educational level, income, environment, urbanization and social characteristics [20]. The use of self reported questionnaires on smoking behaviour were adopted based on documentation indicating equal or better reliability compared to interviews using a structured questionnaire [21, 22]. Furthermore, a Norwegian validation study had already shown that Norwegian pregnant women generally reported their smoking habits correctly [23]. We used no biomarkers for tobacco smoking, as this is unfeasible in large epidemiologic studies, and earlier studies have demonstrated that such biomarkers give little or no additional accuracy to the registration of smoking behaviour when compared to self reported smoking in pregnancy [24, 25].

A new smoking intervention programme as part of a multiple health behaviour intervention did not reduce parental smoking prevalence during pregnancy in the intervention cohort compared to the control cohort. Most women were spontaneous quitters and gave up smoking early in pregnancy before the intervention took place. We found a low indoor smoking prevalence in both cohorts, which may reflect a high degree of awareness of the harmful effects of smoking during pregnancy. Data from the MBR showed a higher quitting rate in Trondheim compared to Bergen in 2003 and 2004 which may have been facilitated by the supplemental attention on smoking behaviour the PACT study initiated.