Reliance on self-reporting underestimates pregnancy smoking rates in Scotland, with more than 2400 pregnant smokers estimated to be missed annually
- Correspondence to Geeta K Swamy
Department of Obstetrics and Gynecology, Duke University Medical Center, 2608 Erwin Road, Durham, NC 27705, USA;
Smoking during pregnancy is associated with numerous adverse outcomes, including spontaneous abortion, preterm birth, low birth weight, fetal growth restriction, placental abruption, perinatal mortality and neurocognitive disorders.1,–,3 A large retrospective Australian cohort study suggests that women who quit smoking by 15 weeks of gestation have similar perinatal outcomes to women who have never smoked.4 This is in contrast to women who quit later in gestation, who have outcomes similar to those among continued smokers. Medical and societal pressures may make women who smoke during pregnancy reluctant to disclose their smoking behaviour. Accurate identification of pregnant smokers is imperative to offer effective smoking cessation programmes early enough to affect pregnancy outcomes.5
Several studies have compared self-reported smoking during pregnancy with rates validated biochemically via cotinine, the main metabolite of nicotine.6 These studies have proved self-report to be inaccurate, with significant under-reporting of smoking. Maternal education level, a surrogate measure for socioeconomic status, has been linked to self-reporting accuracy.7 In addition to examining the accuracy of self-report, Shipton and colleagues were interested in determining whether area deprivation was associated with self-reported smoking accuracy during pregnancy. Maternal self-reported smoking (current, former, never, unknown/missing) was compared with cotinine concentrations for 3475 randomly selected serum samples obtained at the time of second trimester aneuploidy screening. The authors examined the contribution of socioeconomic status by using the Scottish Index of Multiple Deprivation (income, employment, education, housing, health, crime and geographical access), which is based on residential postcode.8
Using a serum cotinine threshold of >13.7 ng/ml to indicate current smoking, the authors identified 30% of women as smokers, compared with 24% by self-report. As expected, the prevalence of smoking was greater in the most deprived areas than in the least deprived areas. Furthermore, there was a striking difference in the discrepancy between self-report and cotinine-validated estimates by deprivation area. In the least deprived areas, 39% of women did not report their smoking, compared with 22% of women in the most deprived areas. However, owing to the difference in absolute prevalence rates (40% vs 14% in the most and least deprived areas, respectively), twice as many pregnant women living in the most deprived areas were undetected smokers as compared with those living in the least deprived areas.
Although this was a retrospective investigation, use of a large bio-repository provided a considerable sample size for analysis. Use of a deprivation index is an innovative approach to examining the contribution of psychosocial stress as it incorporates multiple measures rather than relying on a crude or simplistic measure such as maternal education. Although Shipton and colleagues provide useful information relevant to smoking cessation during pregnancy, several potential biases should be considered. Using serum samples obtained from women who choose to have prenatal serum screening for Down's syndrome and neural tube defects may introduce sampling bias, as such women may represent an older or more motivated population. Given that 70% of the population at large chose to undergo screening and that there was no difference in maternal age or self-reported smoking between those who did and did not undergo screening, it appears that such bias would be minimal. The timing of the study may be significant, as the prevalence of smoking during pregnancy in Scotland decreased from 24% in 2004 to 19% in 2008.9 Lower prevalence may lead to smaller discrepancies between self-report and cotinine validation. Furthermore, although smoking status was obtained at the booking appointment at 8–12 weeks, the process for ascertainment is unclear. For example, were women asked to complete a questionnaire, in which case they may have been less likely to respond accurately, or were they directly questioned by a midwife, which would have improved accuracy?
Smoking is clearly associated with poor pregnancy outcomes. Healthcare providers must be able to assess smoking accurately and be prepared to assist pregnant women with cessation. Nurses directly involved in prenatal care should consider asking more detailed questions to improve ascertainment. For example, a woman who is trying to cut down or quit smoking may find it difficult to answer whether she is a current or former smoker. Questions regarding second-hand smoke exposure, such as household or work exposure, may provide useful insight into a woman's risk of current smoking or failed cessation attempts. Although it is unclear whether it would be cost-effective, routine biochemical validation of smoking status during pregnancy, for example carbon monoxide breath tests or urine cotinine estimation, could aid in appropriately identifying and counselling women at high risk for adverse outcomes. Given the discrepancy in under-reporting by level of deprivation or socioeconomic status, a universal screening approach followed by effective smoking cessation programmes has the potential to affect the socioeconomic disparity in adverse pregnancy outcomes.
Competing interests None.