Trends in Pregnancy-Related Smoking Rates in the United States, 1987-1996 FREE

The addition of smoking to the list of nationally notifiable health conditions in 1996 recognized the role of tobacco use as the leading preventable cause of death in the United States.^1- 2 For several reasons, the public health implications of smoking are potentially compounded for women. Tobacco is associated with many adverse reproductive health effects in women, including adverse effects on pregnancy.^3- 4 Exposure to tobacco during the prenatal period and to environmental tobacco smoke during the postnatal period are leading causes of mortality and morbidity, including problem behaviors, among children.^4- 6

Most women who quit smoking during pregnancy do so on their own.^3,7 Nevertheless, most smokers do not quit smoking during their pregnancy.^3,7 Prenatal smoking cessation programs have been shown to increase quitting rates; however, postpartum relapse rates are very high.^7- 9 Although the prevalence of smoking in the United States has declined since the 1980s, it has increased among adolescents and young adults.^2,10- 12 Therefore, it is not known whether progress has been made toward achieving the desired reduction in the national prevalence of smoking by year 2000 among pregnant women and women of childbearing age.¹³

To assess recent temporal trends in pregnancy-related differences in the prevalence of smoking, we compared smoking behaviors of pregnant women with those of nonpregnant women of reproductive age. This report is based on Behavioral Risk Factor Surveillance System (BRFSS)¹⁴ data from 33 states that collected information on smoking each year from 1987 through 1996. These states represent all of the major US geographical regions and contain about three fourths of the US population.¹⁵

The BRFSS¹⁴ is an ongoing, state-representative telephone survey that gathers information about modifiable risk behaviors related primarily to chronic diseases. Designed to produce risk factor estimates for the noninstitutionalized civilian population aged 18 years or older, it provides baseline data for setting national health promotion and disease prevention objectives.¹³ The sampling weights used in the BRFSS are based on the probability of selection and adjusted for nonresponses and disproportionate sampling of subgroups relative to the state's population. The median response rate to the survey (ratio of completed interviews to the sum of completed interviews and refusals) during the study period varied between 80% and 85%.

In this analysis, ever smokers (smoking initiation) were defined as those who responded "yes" to the opening question on smoking: "Have you smoked at least 100 cigarettes in your entire life?" Current smokers were defined as ever smokers who responded "yes" to the question "Do you smoke cigarettes now?" (which was asked each year from 1987 to 1995) or who responded "everyday" or "some days" to the question "Do you smoke cigarettes every day, some days, or not at all?" (which was asked during 1996). The median number of cigarettes currently smoked was calculated from participants' responses to the question "On the average, how many cigarettes a day do you smoke?" Women who were classified as having quit¹⁶ smoking were those who had initiated smoking but were not current smokers. As an indirect measure of motivation to quit smoking,¹⁶ we obtained the percentage of current smokers who had attempted to quit¹⁶ smoking, which was based on participants' response to the question "During the past 12 months, have you quit smoking for 1 day or longer?" asked from 1992 on. Heavy smoking was defined as smoking 20 or more cigarettes per day. A question on pregnancy status ("To your knowledge, are you now pregnant?") was asked after all questions on smoking and alcohol use.

We aggregated the weighted state-specific data from women aged 18 to 44 years from the 33 states selected for this analysis. From the total sample, we obtained the percentages of women who reported ever smoking (smoking initiation rate), were current smokers, and had quit smoking. We used 3 measures to assess annual differences in current smoking behaviors between pregnant women and nonpregnant women: the prevalence rate ratio (PRR), the median number of cigarettes smoked by current smokers, and the adjusted odds ratio; we used the latter measure to identify sociodemographic subgroups of women at differential risk for smoking. In the absence of information on each woman's smoking habits before and after planning or recognizing a pregnancy, the PRR provides an indirect measure of the magnitude of the reduction in smoking prevalence after women recognize they are pregnant (pregnancy-related reduction). A PRR close to zero indicates the greatest pregnancy-related reduction in smoking.

The overall rates of smoking initiation, current smoking, and quitting were age-adjusted to the age distribution of pregnant women in 1987 (Table 1). To examine variations in current smoking among sociodemographic subgroups and time trends in current smoking within those subgroups, we conducted analyses stratified by age, education, race, marital status, and employment status. Using these variables as independent variables, we created multiple logistic regression models to calculate adjusted odds ratios for current smoking separately for pregnant and nonpregnant women. Although a given odds ratio may overstate the strength of an association compared with the corresponding PRR, it is a valid measure for identifying sociodemographic subgroups at low or high risk for smoking. We used SUDAAN¹⁷ to obtain valid estimates of SEs. We created weighted least squares linear regression models (using the inverse of the variance estimates for weights) to determine the statistical significance of trends over time. A P value less than .05 was considered statistically significant for all analyses. Because most outcome measures showed either a linear relationship with calendar year or no significant secular variation, the results of the stratified analyses are reported only for the beginning and ending years.

Of the 188,541 women aged 18 to 44 years who responded to the survey during the 10-year period, we excluded 793 (4.2%) who were not sure of their pregnancy status and 446 (2.4%) who chose not to answer the question on pregnancy. Between 1987 and 1996, the actual number of respondents increased from 712 (weighted estimate, 1.7 million) to 980 (weighted estimate, 1.7 million) among pregnant women and from 14,563 (weighted estimate, 32.1 million ) to 21,604 (weighted estimate, 35.7 million) among nonpregnant women. The percentage of women who reported that they were pregnant varied little over the study period (average weighted annual prevalence, 5%). The sociodemographic characteristics of the survey respondents (Table 1) are comparable with those of the population in the 33 states combined.¹⁵

Overall, the smoking initiation rate among women aged 18 to 44 years decreased significantly, from 44.1% in 1987 to 38.2% in 1996 (Figure 1). The largest decline in the smoking initiation rate occurred between 1987 and 1991 (Figure 1). However, among women aged 18 to 20 years, smoking initiation rates changed little between 1987 (26.1%) and 1996 (27.4%). There was no statistically significant difference in yearly smoking initiation rates between pregnant and nonpregnant women either in 1987 (41.7% vs 42.3%) or 1996 (36.4% vs 37.6%).

In the total sample, the percentage who had quit smoking changed little between 1987 and 1996 among both pregnant women (from 26.3% to 25.2%) and nonpregnant women (from 16.3% to 14.4%). Also, BRFSS data from 1992 through 1996 indicated that the percentage of currently smoking women of childbearing age who had attempted to quit smoking for at least 1 day in the past year actually declined significantly, from 57.8% to 50.3%.

Over time, the percentage of current smokers decreased significantly among both pregnant and nonpregnant women (Figure 1, Table 2). However, in each year, pregnant women were about half as likely as nonpregnant women to be current smokers, and the variation in the PRR over time was not statistically significant (average annual PRR, 0.54) (Figure 1, Table 2). Data available from BRFSS for 1994-1996 indicate that pregnant women who smoked were significantly more likely to be daily smokers than their nonpregnant counterparts (88.9% vs 79.2%). Among pregnant women who were current smokers, there was no statistically significant variation in the median number of cigarettes smoked per day during the 10-year period (average annual median number, 10), whereas among nonpregnant women, the median number of cigarettes smoked per day decreased significantly, from 19 in 1987 to 15 in 1996 (Figure 2).

For both pregnant and nonpregnant women, the secular decline in the percentage of current smokers occurred largely among women who had smoked 1 or more packs of cigarettes per day (Table 2). This subgroup of women also had the largest pregnancy-related reduction in smoking each year (indicated by the lowest PRR value of 0.3 in 1996).

Among both pregnant and nonpregnant women, the percentage of current smokers decreased between 1987 and 1996 within most sociodemographic subgroups studied, except for the youngest women and oldest pregnant women, among whom it increased slightly (Table 2). Among the subgroups of women in which current smoking rates decreased, the decrease was not significant among pregnant women who were married or were high school graduates or among nonpregnant women who had higher education or were white, unmarried, or unemployed (Table 2).

Smoking prevalence rates varied widely in 1987 among the sociodemographic subgroups studied (range, 9.4%-34.1% for pregnant and 20.0%-43.4% for nonpregnant), as did the PRRs (range, 0.4-1.2) (Table 2). By 1996, the variations in the PRRs had narrowed (range, 0.3-0.7), particularly among the subgroups that had the highest smoking prevalence rates in 1987, indicating that these subgroups had greater pregnancy-related reductions over time (Table 2).

Factors that were independently associated with a high risk for current smoking each calendar year were the same for pregnant and nonpregnant women, namely, having a high school education or less, being white, and being unmarried (Table 3). Among pregnant women, the variation between the odds of smoking in 1987 and the odds of smoking in 1996 was significant only within the 21- to 30-year-old age group, in which it decreased (Table 3). Among nonpregnant women, the odds of smoking increased significantly between 1987 and 1996 among white women and women who were unmarried (Table 3).

This analysis of population-based survey data indicates that the decline in rates of current smoking among pregnant women in the United States between 1987 and 1996 reflects a decline in smoking initiation among women of childbearing age rather than an increase in quitting rates related to pregnancy. Over time, pregnant women have continued to be about half as likely as nonpregnant women to smoke. Among pregnant women, we found no significant change during the 10-year study period in the median number of cigarettes smoked. Between 1987 and 1996, among young women, prevalence rates of smoking initiation and current smoking increased slightly.

Because this survey excluded people without telephones, those younger than 18 years, and those who did not live in private residences, and because we used self-reported information on pregnancy, these data may have underestimated the actual prevalence of smoking. Furthermore, because the survey produced point-prevalence data, they cannot be used to determine whether a woman is truly a nonsmoker. Even women who had quit smoking for a day (on the day of the interview) may have been considered nonsmokers in this survey. However, point-prevalence estimates of current smoking reported by other studies for corresponding years are generally consistent with our findings.^18- 19 Because the data were from a cross-sectional survey, we could not assess whether the substantial decline between 1987 and 1996 in the percentage of heavy smokers among both pregnant and nonpregnant women was due to heavy smokers stopping altogether, heavy smokers reducing the number of cigarettes they smoked, or fewer women who started smoking reaching a heavy level. Pregnant smokers were more likely to be daily smokers than their nonpregnant counterparts, probably because proportionately more of the less-than-daily smokers may have quit smoking when they learned they were pregnant.

For US women of reproductive age, the year 2000 national objective of a 12% prevalence rate of current smoking¹³ is unlikely to be achieved given that the 1996 prevalence rate of current smoking is about twice that of the year 2000 target. For pregnant women, the 1996 rate is only 2% higher than the year 2000 objective of 10%¹³; however, because of the reported recent increases in smoking initiation among adolescents and young women^10- 12 and no concurrent increase in quitting rates by current smokers shown in this analysis, the year 2000 objective for pregnant women is also unlikely to be met. Furthermore, most adult smokers in the United States become regular smokers before age 18 years,¹⁰ and for such smokers quitting is difficult; only 2.5% of the 34% who attempt to quit smoking each year are successful in quitting permanently.¹⁶ The fact that most smokers (54%) continue to smoke after they learn that they are pregnant underscores the need for more extensive efforts to reduce smoking by women over their entire lives and to reduce the number of women who begin smoking.

We found that the risk factors for smoking are similar for pregnant and nonpregnant women and that many such women are likely to rely on publicly funded health care providers for care. Clinicians who provide this care should extend the "ask, advise, assist, and arrange" strategy suggested in the clinical practice guidelines on smoking cessation^20- 21 to each woman. Efforts to curb smoking during pregnancy should begin in the preconception period. Clinically proven prenatal smoking cessation interventions that can be delivered in primary care settings are now available.^20- 24 "Minimal contact" primary care clinic-based smoking prevention interventions have been found to be effective.²⁴ Nicotine replacement products can improve the effectiveness of comprehensive behavioral interventions to reduce smoking.²⁵ For pregnant women, nicotine replacement products should be recommended only if the increased likelihood of smoking cessation outweighs the harmful effects of nicotine and potential concomitant smoking on the fetus.²⁰

Pregnancy may represent a period when smoking cessation activities are most effective. If prenatal smoking cessation is not feasible, a reduction in the number of cigarettes smoked could benefit the pregnancy, including a reduction in risk of low birth weight.⁸ To reduce postpartum relapses in smoking, interventions should continue beyond the perinatal period. Of note, in our study declines in smoking rates were not as pronounced among married pregnant women as they were among their unmarried counterparts. Because smokers may have partners whose smoking habits match their own, and because the presence of such a partner may negatively affect a woman's attempts to quit smoking,²⁶ screening for tobacco use should be extended to partners of women who smoke. For women who are unable to quit smoking and who are at risk of becoming pregnant, the feasibility of pregnancy postponement until smoking cessation is achieved should be explored.²⁷

Although the importance of reducing tobacco use by current smokers is clear, the findings from our study, including the unchanging trends in current smoking rates among young women, and reports from England,²⁸ which showed that the prevalence of smoking among pregnant women remained unchanged between 1992 and 1999 (average annual prevalence, 28%), signify the need for multiple approaches to reduce smoking initiation. The appeal of smoking and access and exposure to tobacco can be reduced through a broad-based approach that includes restriction of tobacco advertising and the promotion of counteradvertising, as well as economic deterrents such as a higher tobacco excise tax, stronger enforcement of age-at-sale policies, and elimination of smoking in the home, workplace, and schools.^10,29- 31 Our findings should alert other countries, especially developing nations in which tobacco marketing efforts and the prevalence of smoking are increasing,³² about the complexities associated with combating the epidemic of tobacco addiction and the need to stem smoking initiation by young people.