|Home | About | Journals | Submit | Contact Us | Français|
Smoking tobacco during pregnancy results in exposure to the fetal neuroteratogen nicotine. The current study evaluated if the offspring of smokers show abnormalities in maternal ratings of executive function, prevalence of Attention Deficit Hyperactivity Disorder (ADHD), and academic performance. A secondary objective was to determine the utility of online data collection.
Mothers (N = 357) completed the parent form of the Behavioral Rating Inventory of Executive Function (BRIEF) and provided information about smoking during pregnancy.
The internal consistency of the BRIEF when administered electronically was quite satisfactory (Cronbach’s α = .98). As anticipated, ADHD was more frequently diagnosed in the offspring of women that smoked at least 10 cigarettes/day (odds ratio [OR] = 2.64, 95% CI = 1.22–5.71). Higher (i.e., more problematic) ratings relative to unexposed children (p < .01) were only identified on the total BRIEF score, the Metacognition Index, and on the Initiate, Plan/Organize, and Monitor scales among children exposed to ≥10 cigarettes/day. Nicotine-exposed children were also more likely to perform less well than their classmates in math (OR = 2.78, 95% CI = 1.59–4.87) and reading (OR = 2.00, 95% CI = 1.10–3.63), and these academic effects were independent of maternal education levels.
This report provides preliminary evidence that the BRIEF has adequate psychometric properties when administered electronically and that mothers who smoke have offspring with lower executive function proficiency. These findings contribute to a larger literature that indicates that smoking during pregnancy results in adverse reproductive outcomes and, possibly, subtle but enduring deficits in prefrontal function.
The adverse consequences of cigarette smoking on reproductive outcomes and on the neonate are well established. Nicotine exposure causes dose-dependent increases in the risk of stillbirth (Strandberg-Larsen, Tinggaard, Nybo Andersen, Olsen, & Grønbaek, 2008, Wisborg, Kesmodel, Henriksen, Olsen, & Secher, 2001), low birth weight/decreased head circumference (Jaakkola, Jaakkola, & Zahlsen, 2001; Roza et al., 2007; Winzer-Serhan, 2008), and sudden infant death syndrome (Alm et al., 1998; Wisborg, Kesmodel, Henriksen, Olsen, & Secher, 2000). In contrast, relatively less is definitively known about the long-term sequelae of in utero nicotine on neurobehavioral function (Batty, Der, & Deary, 2006; Knopik, 2009; Linnet et al., 2003; Shea & Steiner, 2008). Many, but not all, case–control and cross-sectional studies have found that the rates of Attention Deficit Hyperactivity Disorder (ADHD) are elevated (odds ratio [OR] = 2 to 4) among the offspring of smokers, even after accounting for group differences in maternal education and socioeconomic status (Langley, Rice, van den Bree, & Thapar, 2005; Linnet et al., 2003; Schmitz et al., 2006). Hyperkinetic disorder, the International Classification of Diseases equivalent of ADHD, frequency was similarly increased in children with a history of prenatal nicotine (Obel et al., in press). However, other investigations using novel methodologies have challenged the perspective that early developmental nicotine causes ADHD symptomology (Thapar et al., 2009), deficits in intelligence (Gilman, Gardener, & Buka, 2008), or academic difficulties (Lambe, Hultman, Torrång, Maccabe, & Cnattingius, 2006).
Executive functions are responsible for guiding, directing, and managing cognitive and emotional behaviors, especially when solving novel problems, and are mediated by a network of forebrain structures including the prefrontal cortex (Gioia, Isquith, Guy, & Kenworthy, 2000). The behavioral endpoints that have been reported to be sensitive to in utero smoking include auditory sustained attention (Kristjansson, Fried, & Watkinson, 1989) and perseverative errors on the Wisconsin Card Sorting Test (Cornelius, Ryan, Day, Goldschmidt, & Willford, 2001). Although both attentional vigilance and the ability to override a previously learned rule are elements of executive function, some studies indicate that executive function is not a simple unitary process (Huijbregts, Warren, de Sonneville, & Swaab-Barneveld, 2008, Miyake et al., 2000). Furthermore, there may be distinct developmental trends for different executive function component processes and their integration (Pennequin, Sorel, & Fontaine, 2010; Piper, Li, Eowiz, Kobel, Benice, Chu, et al., 2011). Therefore, a benefit of the Behavioral Rating Inventory of Executive Function (BRIEF) is that this instrument can assess various nonoverlapping aspects of executive functioning (e.g., inhibition, self-monitoring, working memory, and emotional control) in a variety of ecologically valid contexts (home, school, play). Although this measure has been used previously with women that used illicit drugs during pregnancy (Piper, Acevedo, Kolchugina, Butler, Corbett, Honeycutt,et al., 2011), to our knowledge, no prior investigations have evaluated the offspring of tobacco users.
Online survey administration is becoming an increasingly common methodology in the substance abuse field with studies of adult alcohol (Collins, Logan, & Neighbors, 2010; Kypri, Paschall, Langley, Baxter, & Bourdeau, 2010), cannabis (Mullens, Young, Dunne, & Norton, 2010), methamphetamine (Hirshfield, Remien, Humberstone, Walavalkar, & Chiasson, 2004; Hirshfield, Remien, Walavalkar, & Chiasson, 2004), ecstasy (Gamma, Jerome, Liechti, & Sumnall, 2005; Rodgers et al., 2006), prescription stimulant (McCabe & Teter, 2007), and nicotine (Heffernan, Ling, Parrott, Buchanan, Scholey, & Rodgers, 2005) users. To our knowledge, no online investigations have been conducted in the neurotoxicology and teratology field. This is unfortunate for two reasons. First and foremost, with the appropriate safeguards and confidentiality protections, sensitive/illegal behaviors may be more readily disclosed in electronic surveys, which minimize the risk of interviewer judgments (Hirshfield, Remien, Humberstone, et al., 2004; Hirshfield, Remien, Walavalkar, et al., 2004). Second, the individual items on computerized questionnaires can be tailored automatically to each respondent based on prior responses. This could involve the administration of additional questions about the timing and extent of drug use only if the respondent reported lifetime use. Additionally, automated data collection and processing is faster and more efficient than more traditional (i.e., paper and pencil or computer-aided interviews) methods and is acceptable to various populations (Gamma, Jerome, Liechti, & Sumnall, 2005; Heffernan et al., 2005; Shakeshaft, Bowman, & Sanson-Fisher, 1998). The present online study examines maternally rated executive function using the BRIEF in children exposed to nicotine during pregnancy. As abnormalities in executive function may contribute to deficits in school performance as well as a diagnosis of ADHD, these endpoints were also evaluated. Since this is the first Internet-based study to use the BRIEF, some psychometric properties were also determined.
Mothers (N = 357) of children aged 5–18 years were recruited for a child behavior investigation. Flyers were posted on community boards throughout Doernbecher Children’s Hospital, Oregon Health and Science University (OHSU), the Portland metro area, western Oregon, and western Washington (e.g., grocery stores, libraries, coffee shops). Links to the study were also displayed on the community and volunteer sections of Craigslist (craigslist.org) as well as on message boards for parents (e.g., iVillage.com). The majority of respondents were from the northwest (68.3% from Oregon, Washington, or Idaho) with other participants mostly from adjacent states of California and Montana. This anonymous online survey was administered through Research Electronic Data Capture (REDCap), version 1.3.9, a web-based application for building and managing online databases with maximal security for sensitive information (Harris, Taylor, Thielke, Payne, Gonzalez, & Conde, et al., 2009). The Institutional Review Board at OHSU approved all procedures.
After providing an online consent to participate in this study, the parents began the survey, which typically took about 20 min to complete. The items on the first half were organized from less to more personal and included questions about maternal and child demographics (e.g., age, sex, ethnicity), academic performance (e.g., “Please rate your child’s performance in math with relation to their scores on the state’s standardized test.” with options of below, at, or above grade level), and child/maternal neurological or psychiatric conditions (e.g., Has your child been diagnosed with any of the following? with options of ADHD, fetal alcohol syndrome [FAS], and brain trauma). Items on maternal drug use (nicotine, alcohol, marijuana, cocaine, and the opiates) were organized into two periods: during pregnancy and specifically during the third trimester. If the respondent answered in the affirmative then additional item(s) about the extent of use (e.g., how many cigarettes did the biological mother smoke each day?) were displayed.
The BRIEF accounted for the remaining 86 items. The BRIEF is a widely employed parental rating instrument for the clinical evaluation of children aged 5–18 years with inherited and acquired neurobehavioral conditions focusing on the child’s everyday activities at home and at school. Each behavior is rated as never, sometimes, or often a problem (1–3 points, respectively) in the last six months. The eight BRIEF scales form two measures of executive functioning (Metacognition and Behavioral Regulation), and these are summed to form an overall measure of executive functioning (the Global Executive Composite or GEC). The Metacognition Index comprises the following five scales: (a) Initiate, the capacity to act independently to produce ideas, responses, or problem-solving strategies; (b) Working Memory, the ability to hold information to complete a task; (c) Plan/Organize, the capability to anticipate future events, form goals, and construct the appropriate steps to complete an objective; (d) Organization of Materials, the degree of orderliness of work and play spaces; and (e) Monitor, self-monitoring habits scales. The Behavioral Regulation Index consists of three scales: (a) Inhibit, the ability to regulate one’s behavior at the appropriate time and not act on impulse; (b) Shift, the ability to switch attention and change focus; and (c) Emotional Control, the capacity to regulate emotional responses. Standardized T50 scores were determined from raw scores based on age/sex norms with higher scores indicating greater severity. The inconsistency scale is obtained by calculating the difference between 10 pairs of items (range = 0–20) with a score ≥9 interpreted as inconsistent. For example, a response of often (3 points) to “Has explosive, angry outbursts” and never (1 point) to “Has outbursts for little reason” would contribute two points to the inconsistency score.
The paper and pencil BRIEF has excellent internal consistency (Cronbach’s α = .97) and good test–retest reliability over two weeks (r = .86 for the GEC, .88 for Metacognition Index, and .84 for the Behavioral Regulation Index) in a normative sample. Importantly, Gioia et al. noted that the education level of the rater showed small, but significant, negative correlations with some BRIEF scales, that is, lower education was associated with more problematic ratings which accounted for as much as 5% of the variance. Good convergent and divergent validity with other standard parent report measures are described elsewhere (Gioia, Isquith, Guy, & Kenworthy, 2000).
Statistical analyses were completed with the Statistical Package for the Social Sciences, version 16.0, with data expressed as mean (±SEM) and p < .05 considered statistically significant. Exclusion criteria were child age (<5 or >18), incomplete/unfinished questionnaires (N = 138), responses from any other source besides the biological mother (father, grandparent, adoptive/foster parent, N = 80), or children with brain trauma (N = 3; Sesma, Slomine, Ding, & McCarthy, 2008) or FAS (N = 7; Chasnoff, Wells, Telford, Schmidt, & Messer, 2010). Categorical-level analyses were completed with the χ2, or, if the N/cell was < 5, likelihood ratios. As no online research with the BRIEF has been completed previously, three quality checks were conducted. First, the internal consistency (Cronbach, 1951) of the instrument was determined and compared with prior data from paper administration (Gioia et al., 2000). Second, the percentage of responses that met the inconsistent criteria was examined. Third, as the BRIEF is one of many instruments used in the diagnosis of ADHD (Gioia et al., 2000; McCandless & O’Laughlin, 2007), a comparison of unexposed ADHD+ versus ADHD− was completed to evaluate test validity. A BRIEF T50 ≥ 65, 1.5 SDs above the mean, is interpreted as a clinically significant (Gioia et al., 2000). The OR of clinically significant problems with the 95% CI was also determined for nicotine, education, and income (all coded dichotomously: exposed relative to unexposed, less than or equal to high school relative to above, <$10K/year during pregnancy relative to above). Analysis of covariance was also completed with covariates selected empirically based on variables in Table 1 that statistically (p < .05) differentiated women that did (NIC+) and did not (NIC−) use nicotine during pregnancy. Additional analyses were also conducted with the NIC+ divided into a low (1–9 cigarettes/day) and high (10+ cigarettes/day) groups. This categorization has been used previously by others (Huijbregts, Warren, Sonneville, & Swaab-Barneveld, 2008; Olds, Henderson, & Tatelbaum, 1994; Sexton, Fox, & Hebel, 1990; Thapar et al., 2009).
Cronbach’s α was .98 for GEC, .97 for the Metacognition Index, and .96 for the Behavioral Regulation Index (corresponding values for the paper and pencil version were .97, .96, and .94, see Supplementary Figure 1 for further details). Furthermore, only one respondent met the criteria for inconsistent responding (0.3%). As the Metacognition Index has been shown previously to differentiate ADHD+ and ADHD− children (McCandless & O’Laughlin, 2007), this pattern was evaluated and verified with online BRIEF administration, ADHD− = 54.8 ± 0.8, ADHD+ = 69.5 ± 1.6, t(271) = 7.91, p < .0005. Overall, these three internal checks for the reliability and validity of online BRIEF administration all indicated very satisfactory psychometric properties.
Demographics, substance use patterns, and academic performance are depicted in Table 1. Women that did (NIC+) and did not (NIC−) use cigarettes during pregnancy did not differ in ethnicity, frequency of ADHD, or alcohol use. However, NIC+ women were younger and had lower income (both during pregnancy and currently), less education, were more likely to also use marijuana, cocaine, and methamphetamine during pregnancy and specifically in the third trimester. The NIC− and NIC+ children did not differ based on age or sex, but NIC+ were more likely to be diagnosed with ADHD and to be behind their peers in math and reading.
Further analysis of the NIC+ group divided into low (1–9 cigarettes/day) versus high (10–30 cigarettes/day during pregnancy) revealed that maternal age and income were significantly different in the both the low and high groups relative to NIC−. Notably, these groups did not differ from each other. Relative to unexposed children, reading difficulties were present in the low-NIC group (OR = 2.05, 95% CI = 1.02–4.14), but math difficulties (OR = 4.80, 95% CI = 2.19–10.52) and ADHD (OR = 2.64, 95% CI = 1.22–5.71) showed significant elevations only among high NIC.
The GEC was significantly elevated by 0.5 of a SD in NIC+ (63.1± 1.5) relative to NIC− (57.8 ± 0.8) children, F(1,356) = 9.76, p < .005. This difference was retained with maternal education, age at pregnancy, pregnancy income, child ADHD diagnosis, maternal ADHD diagnosis, prenatal marijuana, cocaine, methamphetamine, or alcohol exposure, prematurity, or current income entered as covariates. Additional analyses noted some evidence for exposed boys exhibiting a more pronounced profile than girls. For example, the Behavioral Regulation Index was increased by eight points in males, NIC− = 57.9 ± 1.3, NIC+ = 65.9 ± 2.6, t(173) = 2.85, p ≤ .005, but only three points in females, NIC− = 55.7 ± 1.2, NIC+ = 58.6 ± 1.8, t(181) = 1.32, p = 0.19. In contrast, younger (<9.5) and older (>9.5) children were similarly affected (data not shown).
Figure 1 shows a comparison of each NIC group. Both the low (p < .05) and high (p < .01) exposed children differed from the unexposed on the GEC. Similarly, the low- and high-NIC groups were elevated on the Behavioral Regulation Index. The low-, but not high-, NIC children exhibited increases on the Inhibit scale with the reverse pattern noted for Emotional Control. The low (p < .05) and high (p < .01) groups had difficulties with Metacognition. There was partial evidence for dose-dependent exposure effect with significant (p < .01) increases for the Initiate, Plan/Organize, and Monitor scales for only the high-NIC children. On the other hand, there were no significant differences between low- and high-NIC groups.
Table 2 shows the likelihood of having a clinically significant problem (T50 ≥ 65) on the BRIEF, having an ADHD diagnosis or being behind peers academically based on maternal education, income, or nicotine use during pregnancy. The prevalence of all these outcomes was uniformly increased in the offspring of women with no post-secondary education (note that ADHD did not fulfill the statistical cutoff with p = .053). Similarly, women with lower incomes during pregnancy generally showed a less generalized pattern of difficulties on these domains. Maternal nicotine use was associated with a more focused profile on the BRIEF including on the Inhibit, Emotional Control, Organization, and Monitor scales. Importantly, only math showed a nicotine effect, which was retained when the variance attributable to education as well as income was removed.
There are two key findings of this report. The first is that online survey administration verified and extended upon several findings that have been documented with other methodologies. Women that smoked during pregnancy were poorer, less educated, younger, and more likely to use other recreational drugs as well as give birth prematurely. All these demographic and perinatal findings are congruent with what has commonly been reported in earlier investigations of maternal smoking (Batty et al., 2006; Winzer-Serhan, 2008). Electronic BRIEF delivery resulted in equivalent internal consistency with the paper and pencil form (Gioia et al., 2000). Support for the validity of computerized administration of this executive function instrument was also identified with the anticipated (McCandless & O’Laughlin, 2007) elevation in the Behavioral Regulation Index among unexposed children with an ADHD diagnosis. Together, the methodological implications of these outcomes may be broadly relevant for others as we found that many mothers were quite willing to participate in online research. Instruments that have adequate psychometric properties when administered electronically could be employed for investigations focused on other drugs that are used less commonly than nicotine to examine the neurobehavioral profile of children exposed to nicotine cessation agents or to more efficiently recruit from rural populations. The web-based procedures could be incorporated in longitudinal investigations that include geographically mobile families or coupled with a medical record release forms to obtain more detailed perinatal information. Additional online studies with participants obtained from a national registry of research volunteers are also ongoing.
Second, nicotine-exposed children differ from their unexposed counterparts on several overlapping areas including psychiatric diagnoses, academic performance, and maternally rated executive function. Although our current understanding of the causes versus risk factors for ADHD is incomplete, identification of the extent that in utero nicotine is involved has been a highly active research area (Ball et al., 2010; Biederman, Monuteaux, Faraone, & Mick, 2009; Knopik, 2009; Kotimaa et al., 2003; Linnet et al., 2003; Milberger, Biederman, Faraone, Chen, & Jones, 1996; Obel et al., in press; Schmitz et al., 2006; Thapar et al., 2009).
Children whose mother smoked at least a half-pack per day were over twice as likely as unexposed children to have been diagnosed with ADHD. One confound of many of the prior investigations that have identified an increased frequency of ADHD among NIC-exposed children is that the rates of parental ADHD were also elevated (Biederman et al., 2009; Knopik, 2009; Milberger et al., 1996; Schmitz et al., 2006), which raises the possibility that a vulnerability to develop ADHD was inherited. Therefore, it is important to emphasize that the rates of maternal ADHD in this sample were equivalent in tobacco abstaining and tobacco using women. Another interesting finding from this investigation is that the offspring of women from lower educational backgrounds exhibited a nonsignificant tendency to more commonly be diagnosed with ADHD. If this finding is replicated in other community samples, future investigations should continue to pay close attention to this key potential confound (Langley et al., 2005). Although animal (Heath & Picciotto, 2009; Thomas, Garrison, Slawecki, Ehlers, & Riley, 2000) and, perhaps, human (Berlin et al., 2009) studies indicate that it is mechanistically plausible that early developmental smoking causes ADHD, as there was no nicotine-associated difference in ADHD frequency when adjusting for other variables, this would support the view that other socioeconomic and environmental factors have a larger contribution to the etiology of ADHD than does prenatal nicotine.
The current dataset documented that nicotine-exposed children were approximately two times as likely to be behind their peers on mathematics and reading, and these effects were not attributable to differences in maternal education or income, findings which are broadly concordant with Batstra, Hadders-Algra, and Neeleman (2003). Perhaps, the most definitive study to date to examine scholastic performance was conducted with a large (N = 50,000) sample of Swiss teenagers. Academic difficulties showed dose-dependent nicotine increases (OR = 2) within each of five maternal education levels. On the other hand, examination of siblings pairs where the mother smoked during one pregnancy but not the other also identified an elevation in school problems for both children which indicated that nonsmoking factors were responsible (Lambe et al., 2006). The same general pattern indicative of unmeasured genetic or environmental variables underlying intellectual performance deficits was subsequently replicated in an older all-male sample (Lundberg et al., 2010), indicating that the relationship between in utero nicotine exposure and school performance may be dependent on the sample characteristics.
There is currently no consensus whether prenatal nicotine causes or is only correlated with long-term reductions in academic success (Batstra et al., 2003; Knopik, 2009). Many cross-sectional and longitudinal investigations have identified significant decreases among the offspring of smokers on various measures of intelligence (Fried, Watkinson, & Gray, 2003; Gilman et al., 2008; Julvez et al., 2007; Mortensen, Michaelsen, Sanders, & Reinisch, 2004). A deficit in intellectual function on the Stanford–Binet was noted (Olds et al., 1994), and the pre-school–aged offspring of women that quit smoking during pregnancy, relative to those that persisted, had more difficulties on the verbal scale of the McCarthy assessment even after controlling for other prenatal and postnatal variables (Sexton et al., 1990). In contrast, nicotine-associated decrements in mathematics and reading on the Peabody Individual Achievement Test in children from the United States were nonsignificant when accounting for maternal education (Batty et al., 2006). Socioeconomic factors appear to be responsible for the nicotine group differences in reading but not mathematics or spelling in Dutch adolescents (Batstra et al., 2003).
Children with a history of in utero smoking exposure had more problems with maternally rated executive function. The BRIEF findings were quite robust with significant mean elevations in the NIC groups on the GEC, both indices, and six of the eight scales. Most notably, the difference between unexposed and nicotine-exposed children on the GEC were retained after removal of the variance attributable to several other potential confounds. The proportion of children scoring in the clinical range was also more frequent among NIC-exposed children on the Inhibit, Emotional Control, Organization, and Monitor scales. It should be reiterated that executive function is conceptualized by the BRIEF developers as a broad construct mediated by the frontal cortex with its associated cortical and subcortical connections that is responsible for intentional, goal-directed, problem-solving behaviors (Gioia et al., 2000). A fundamental strength of this instrument, unlike single laboratory executive function tests, is that this measure can be completed relatively quickly and can assess nonoverlapping aspects of executive functioning. Furthermore, although elevations in BRIEF ratings are well known among children with ADHD (Gioia et al., 2000; McCandless & O’Laughlin, 2007), abnormalities in executive function are certainly not unique to this condition and have also been identified in extremely low birth weight (Anderson & Doyle, 2004), FAS (Chasnoff et al., 2010), and children that experienced a traumatic brain injury (Sesma et al., 2008). While this is the first report to examine maternally assessed executive function in the offspring of smokers, there are prior studies with laboratory-based measures (Fried et al., 2003; Huizink & Mulder, 2006; Kristjansson et al., 1989), and the present finding of abnormalities in the offspring of smokers are generally concordant. Notably, the fact that no nicotine group differences on the clinically significant executive function measures (Table 2) survived after inclusion of prenatal income levels into the statistical models indicates that socioeconomic or other lifestyle factors are integral for the proportion of children meeting this criteria.
A potential limitation of this investigation is the reliance on retrospectively determined maternal drug use patterns. Although it is generally recognized that retrospective maternal drug use information is inferior to that which has been prospectively obtained (Huizink & Mulder, 2006), empirical examination of this issue with smokers does not lead to simplistic conclusions. The veracity of recall was repeatedly determined over a twenty-year period and found to be accurate for smoking (+ versus −) for the vast majority (94%) of women but correct classification of the number of packs smoked per day was lower (80%; Krall, Valadian, Dwyer, & Gardner, 1989). Furthermore, (Pickett, Kasza, Biesecker, Wright, & Wakschlag 2009) repeatedly evaluated urine cotinine during pregnancy and determined the correspondence with self-reported smoking measured during the second trimester as well as with smoking habits during pregnancy which were obtained over a decade later when the offspring were between the ages of 11 and 18. Among women whose urine tested positive for cotinine, the preponderance was classified correctly as smokers by both prospective (98.1%) and retrospective (95.6%) methods. Additionally, among women who prospectively denied smoking, approximately one quarter (22.7%) retrospectively reported nicotine use during pregnancy. (Pickett et al., 2009) concluded that retrospective measures may even be more informative than prospective ones for determining some smoking behaviors (e.g., packs per day during the first trimester). Additional prospective studies could incorporate a quantitative biomarker of smoking (Florescu et al., 2009) in conjunction with paternal, teacher, or self-ratings to further evaluate the generalizability, persistence, and potential strategies for remediation of the observed abnormalities in executive function and scholastic performance.
In conclusion, there is evidence possibly indicative of both a causal (Figure 1 and academic data of Table 1) and correlative (clinically significant problems in Table 2) relationship between in utero nicotine and subtle neurocognitive deficits. We suspect that future investigations will clarify that the type and strength of relationship depends not only on the domain measured but also upon on the extent and trimesters of maternal nicotine use, individual genetic differences (maternal and fetal), the ages participants are assessed, and the degree that confounds are present in different populations (Knopik, 2009). Ideally, an increased awareness of the reproductive and neurodevelopmental risks of nicotine will encourage more women to quit smoking prior to pregnancy.
This work was supported by the National Institute of Environmental Health Sciences (T32 ES007060-31A1), the Oregon Clinical Translational Research Institute (UL1 RR024140), the National Institute of Drug Abuse (L30 DA027582-01), the N.L. Tartar Trust to BJP, and the University of Puget Sound Richard Bangs Collier scholarship to SMC.
Special thanks to Jacob Raber, Ph.D., as well as Reid Olsen, Julie Mitchell, M.S., and Ruth Rowland for their technical support; Jean O’Malley, M.P.H., for statistical advice; Joel T. Nigg, Ph.D., for reading an earlier version of this manuscript; and the mothers that took part in this study.