Prenatal Smoking Exposure, Low Birth Weight, and Disruptive Behavior Disorders

Prenatal Smoking Exposure, Low Birth Weight, and Disruptive Behavior Disorders JOEL T. NIGG, PH.D.,

AND

NAOMI BRESLAU, PH.D.

ABSTRACT Background: Prenatal problems are among theorized etiologies for child disruptive behavior problems. A key question concerns whether etiological contributors are shared across the broad range of disruptive psychopathology or are partially or largely distinct. Method: We examined prenatal smoking exposure and low birth weight as risk factors for attentiondeficit/hyperactivity disorder (ADHD), oppositional defiant disorder (ODD), and conduct disorder (CD) in a populationbased longitudinal design from ages 6 to 17 years. Multiple informants were used, with emphasis on parent and teacher report for ADHD, parent- and self-report interview for ODD, and self-report interview for CD, in keeping with evidence about the most valid sources of information for these respective syndromes. Results: The association of prenatal smoking exposure with ADHD was highly confounded by family variables. In contrast, low birth weight independently predicted ADHD, even with family variables statistically controlled. The opposite pattern appeared for ODD and CD. Prenatal smoking exposure but not low birth weight predicted ODD independent of potential confounding variables. Prenatal smoking exposure also predicted CD. The effect on CD was via its effect on ODD. Conclusion: Prenatal smoking exposure may contribute to ODD and via that route to later CD, but does not have a specific association with ADHD in this sample. Findings have implications for distinct etiological contributors to these often comorbid aspects of the disruptive behavior domain. J. Am. Acad. Child Adolesc. Psychiatry, 2007;46(3):362Y369. Key Words: attention-deficit/hyperactivity disorder, oppositional defiant, conduct disorder, prenatal nicotine, prenatal smoking, low birth weight.

Numerous studies have linked prenatal factors with subsequent disruptive behavior disorders, including attention-deficit/hyperactivity disorder (ADHD) and conduct problems, although few studies have examined oppositional defiant disorder (ODD). Among the most well-replicated early correlates of these disruptive behavior outcomes have been prenatal smoking exposure and low birth weight (LBW). Few studies have looked at these suspected causes together, in particular to evaluate whether LBW accounts for smoking effects, and few studies have looked at ADHD and other disruptive behaviors in the same sample to gauge relative Accepted August 22, 2006. Dr. Nigg is with the Department of Psychology and Dr. Breslau is with the Department of Epidemiology, Michigan State University, East Lansing. Correspondence to Dr. Joel T. Nigg, Psychology Department, Michigan State University, 115 Psychology Building, East Lansing, MI 48824-1116; e-mail: [email protected]. 0890-8567/07/4603-0362 Ó2007 by the American Academy of Child and Adolescent Psychiatry. DOI: 10.1097/01.chi.0000246054.76167.44

362

specificity of effects across disruptive behavior disorders. Clarifying these effects is of great importance because it can illuminate etiological theories of related but distinct forms of psychopathology. Although symptoms of ADHD are highly heritable, multiple etiologies are likely (Levy et al., 1997; Sherman et al., 1997), including a role of prenatal risk factors. Likewise, the etiology of aggression and conduct problems is likely multifactorial (Waldman et al., 2002). Some 11.4% of women in the United States reported that they smoked cigarettes during pregnancy in the year 2002 (Centers for Disease Control and Prevention, 2004). Prenatal smoking exposure has been associated with child externalizing problems in prospective population studies that included multiple methods of assessing child behaviors, after controlling for socioeconomic status, other demographic factors, maternal alcohol or drug use, and maternal depression (Brook et al., 2000; Williams et al., 1998). Findings, however, have not been entirely consistent. For example, Silberg et al. (2003) reported that after familial (genetic)

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

PRENATAL SMOKING EXPOSURE AND ADHD

transmission of antisocial traits was controlled, there was no independent association between maternal prenatal smoking and offspring conduct problems in a sample of twins. Button et al. (2005) found that prenatal smoking effects on conduct problems and ADHD were independent of each other and of genetic influences. Effects on ODD have scarcely been examined and may account for the observed effects on conduct disorder (CD) because ODD often precedes CD developmentally. Many of the largest population-based studies did not assess ADHD. However, prenatal smoking exposure has been associated with ADHD in case-control designs (Milberger et al., 1996, 1998), a small number of large prospective cohort designs (Kotimaa et al., 2003; Linnet et al., 2005), and several smaller studies. Linnet et al. (2003) reviewed 24 studies and concluded that a link between maternal smoking in pregnancy and child ADHD could not be ruled out. Despite these emerging findings, a number of issues remain. First, few studies examined other disruptive behavior disorders along with ADHD to clarify specificity of effects to ADHD versus CD or ODD, which often co-occur with ADHD. Second, parental behavioral problems were controlled only in some studies, with inconsistent results. Thapar et al. (2003) found that maternal smoking during pregnancy still had a significant, environmentally mediated effect on ADHD symptoms after genetic factors, birth weight, and child CD were statistically controlled. However, in another twin study, Knopik et al. (2005) found that smoking was not related to ADHD after controlling for genetic effects and confounders. A second issue is that LBW itself is an established outcome of smoking during pregnancy (Kramer, 1987). However, LBW exerts effects on development independently of smoking exposure (Breslau et al., 2005). LBW (defined as <2,500 g) has been linked relatively specifically with ADHD and behavioral attention problems (Breslau et al., 1996; Pinto-Martin et al., 2004). Indeed, child problems with inattention and hyperactivity/impulsivity, as well as increased incidence of ADHD, are among the most reliable outcomes of LBW in prospective population-based designs (Foulder-Hughes and Cooke, 2003) as well as retrospective case-control studies (Mick et al., 2002). Breslau et al. (1996, 2000) examined neuropsychiatric outcomes of LBW in a large-scale, stratified, random

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 20 07

sample of children followed longitudinally. They reported associations between children_s verbal, spatial, fine motor, tactile, intellectual, and attentional abilities and birth weight up to 3,300 g (Breslau et al., 1996). At age 6, children with LBW had increased rates of ADHD but not anxiety or ODD. This relationship was stronger in inner-city versus suburban children (Breslau et al., 1996), which could suggest LBW by environment interactions. At age 11, LBW children had an excess of teacher-rated symptoms of inattention/hyperactivity in the urban setting (Breslau and Chilcoat, 2000), but this relationship was not seen in the suburban setting; these findings suggest that some environments interact with or are protective in relation to LBW effects on attention problems. It is important to clarify these etiological relationships for the DSM constructs of ODD, CD, and ADHD at later ages after children have passed through the period of onset risk. When ADHD, ODD, and CD have been examined together in twin studies, evidence of partially distinct as well as overlapping etiological influences has emerged (Waldman et al., 2002). Clarifying the degree of specificity of prenatal exposure to maternal smoking and of LBW to these overlapping outcomes may contribute nosologically relevant evidence. Indeed, as the DSM-V comes under discussion, there has been increasing recognition of the importance of potential unifying etiological mechanisms that may cut across related disorders and help explain their cooccurrence as well as their taxonomic relationships and distinctions (Krueger, 1999). We address a series of questions concerning whether prenatal smoking effects are specific to ADHD versus ODD or CD when LBW is well represented and can be examined simultaneously in a group of children followed to age 17; in other words, completely through the period of risk of onset of ADHD, ODD, and CD. We hypothesized, based on prior literature, that LBW would be related to ADHD but not to ODD or CD and examined the relationship of prenatal smoking exposure to ADHD, ODD, and CD. METHOD Sample Data are from a longitudinal study of LBW (e2,500 g) and normal birth weight children who were assessed at ages 6, 11, and 17. We targeted the 1983 to 1985 birth year cohorts of newborns

363

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

NIGG AND BRESLAU

who were 6 to 7 years of age in 1990 to 1992, the scheduled period of the initial fieldwork. Two major hospitals in southeast Michigan, one in the city of Detroit and the other in a middle-class suburb, were selected. In each hospital, for each year from 1983 through 1985, random samples of LBW and normal birth weight newborns were drawn from discharge records. Children with severe disabilities, identified at birth and at age 6, were excluded. Of the 1,095 children in the target sample, 823 (75%) participated. The second assessment was conducted in 1995 to 1997, with children assessed as they passed their 11th birthday. Of the 791 children remaining in the Detroit area, 717 (90.6%) were reassessed at age 11, 87.1% of the initial sample. In 2000 to 2002, when the children passed their 17th birthday, we assessed the sample a third time. A total of 713 children were assessed, including 56 children who were not assessed at age 11. More details are available in previous reports (Breslau and Chilcoat, 2000; Breslau et al., 1994; for the age 17 follow-up, see Breslau et al., 2005). Measurement of Key Variables Child ADHD, ODD, and CD. ADHD, ODD, and CD diagnoses were assessed by multiple informants and methods. For each disorder, we selected a primary method of identifying cases and then checked with the secondary source. In all of the analyses, we relied on only one data source at a time; we never substituted one source for another if a source was missing at a given wave. Instead, missingness was handled as explained in the data analysis below.

Mothers were interviewed using the Diagnostic Interview Schedule for Children (DISC; version 2.1) at ages 6 and 11, from which we derived DSM-III-R diagnoses for ODD, CD, and ADHD. These data were obtained before DSM-IV was published; however, convergence across DSM-III-R and DSM-IV ADHD combined type is acceptable (Biederman et al., 1997). We obtained diagnostic interviews of youths using the National Institute of Mental Health Diagnostic Interview Schedule (DIS) at age 17 for ADHD, ODD, and CD using DSM-IV criteria, but we did not interview parents at that age. Teachers completed the Achenbach Teacher Report Form at ages 6, 11, and 17, from which we used the Attention Problems Scale (which includes items for inattention, impulsivity, and overactivity). Because youth report of ADHD symptoms is of doubtful validity and tends to lead to underestimation (Barkley et al., 2002; Danckaerts et al., 1999), we used data from parent diagnostic interviews at ages 6 and 11 (primary analysis) and teacher ratings of attention problems at ages 6, 11, and 17 (secondary check on results) to examine effects on ADHD-related behavior problems. Youth interviews of conduct problems, in contrast, have been shown to have better validity than informant reports (Jewell et al., 2004), so we emphasized the youth interview for ODD and CD at age 17 (primary analysis), with parent report as a secondary check. Table 1 provides descriptive statistics on the sample broken down by maternal smoking status. Birth Weight. Information on children_s birth weight comes from medical records at baseline. Children were classified as LBW or normal birth weight using the standard World Health Organization cutoff of <2,500 g (Kramer, 1987).

TABLE 1 Description of Sample Classified by Maternal Smoking Status Smoked Outcome LBW, % Urban, % Maternal age, y Maternal substance use disorder, % Behavior problems at age 6 y Externalizing (teacher) Attention problem (teacher) Internalizing problem (teacher) % ADHD (DISC-parent) % ODD (DISC-parent) Behavior problems at age 11 y Externalizing (teacher) Attention problem (teacher) Internalizing (teacher) % ADHD (DISC-parent) % ODD (DISC-parent) Behavior problems at age 17 y Externalizing (teacher) Attention problem (teacher) Internalizing (teacher) ODD (DIS-Y), % CD (DIS-Y), %

Never (n = 353)

Not While Pregnant (n = 198)

While Pregnant (n = 247)

50.4 42.5 25.2 (5.6) 6.1

49.0 49.0 25.6 (5.4) 10.0

73.3 57.9 27.4 (5.4) 25.9

50.2 (9.3) 54.2 (6.5) 49.4 (9.8) 14.4 8.5

51.5 (8.9) 55.2 (7.2) 49.0 (10.3) 19.7 13.1

51.9 (9.9) 56.4 (8.7) 49.9 (10.4) 24.3 17.8

49.6 (9.2) 54.4 (7.0) 49.6 (10.2) 11.0 6.2

50.4 (9.4) 54.6 (6.8) 48.8 (9.3) 10.5 12.1

52.7 (11.0) 55.5 (7.5) 49.0 (9.6) 14.7 16.6

47.8 (7.9) 52.9 (5.2) 48.8 (9.5) 7.1 4.2

49.7 (9.3) 54.1 (7.0) 48.9 (10.0) 12.8 4.7

50.0 (8.0) 53.5 (5.0) 48.5 (9.7) 15.8 9.1

Note: LBW = low birth weight; ADHD = attention-deficit/hyperactivity disorder; DISC = Diagnostic Interview Schedule for Children; ODD = oppositional defiant disorder; DIS-Y = Diagnostic Interview Schedule for Youth; CD = conduct disorder.

364

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

PRENATAL SMOKING EXPOSURE AND ADHD

Prenatal Smoking Exposure. Maternal cigarette smoking during pregnancy was ascertained at baseline interview, when children were 6 years of age. As is standard on the NIMH-DIS (Shaffer et al., 1989), mothers were asked whether they smoked daily for 2 months or longer during pregnancy of the index child. A positive answer was followed by an inquiry about the number of cigarettes smoked daily. These questions were in a section of the interview that focused on pregnancy, delivery, and the child_s early health history (Shaffer et al., 1989). At baseline, a separate interview schedule, administered before the inquiry about perinatal history, included a section on tobacco use disorder, which opens with the question BDid you ever smoke daily for a month or more?[ At the 5-year reassessment (age 11), mothers were asked about daily smoking during the preceding 12 months. The availability of information on daily smoking at any time up to baseline and at reassessment allowed us to evaluate the effects of smoking during pregnancy relative to two reference groups: mothers who never smoked and mothers who smoked but not during pregnancy. Family Variables. Because we did not have formal assessment of maternal ADHD or antisocial behaviors to control specific familial transmission, we included in our models a measure of maternal lifetime substance use disorders assessed by diagnostic interview of the mother at baseline using the NIMH-DIS. It was intended to provide partial control for familial transmission of risk of deviance. Maternal education was used as an indicator of social class. Statistical Analysis Because of our interest in the nosological issues noted earlier, we focus on DSM categorical diagnoses. However, we also used dimensional symptom ratings to obtain convergent evidence of our conclusions about the ADHD-related domain. Multiple logistic regression was used to estimate the risk of specific disorders (ODD, CD), based on diagnostic interviews with youths at age 17. For analyses of data from repeated assessments over time, we applied generalized estimating equations (GEEs; Diggle et al., 1994) to test the relationship of maternal smoking during pregnancy to outcomes of interest. GEE analysis uses data on all cases, including those with incomplete data, and permits simultaneous modeling of specific risk factors in relation to outcome over multiple assessments, taking into account the correlations between scores across assessment waves.

The exchangeable correlation option was used as the working correlation. Interactions with age of assessment were evaluated in all models, but were never significant. In the absence of age interactions, the approach yields an estimate of the association of the risk factor with outcome across all assessment waves. When the outcome is dichotomous, GEE analysis yields coefficients from which estimated risks were derived. Analysis was conducted on 798 children with data on maternal smoking in pregnancy. (Informants for 25 children were caregivers other than mothers.) For diagnoses assessed by youth report at age 17 (one time point only), we used multiple logistic regression. Successive regression models are presented to address the sequence of questions regarding the association of prenatal smoking exposure with each outcome. In each model that follows the first unadjusted model, we first test whether LBW accounts for the initially observed association between maternal smoking and children_s behavioral outcome and then whether the association holds up when potential confounders are added. Finally, in each table, we present a model that includes potential confounders but does not include LBW, which cannot be considered a confounder. RESULTS

Table 1 provides descriptive data on the three subgroups of the sample compared in this analysis, indicating differences between mothers who smoked in pregnancy and mothers who did not on variables related to child development. It also displays mean behavior problems and percentages of disorders across subgroups and the association of prenatal smoking exposure and LBW as well as rates of maternal lifetime substance use disorder. ADHD

Table 2 shows the results of DSM-III-R ADHD diagnosis from maternal DISC interviews at ages 6 and

TABLE 2 Successive GEE Models of the Effects of Maternal Smoking in Pregnancy on ADHD (DISC-Parent; Ages 6 and 11) Model 1 Model 2 Model 3 Model 4 Model 5 Smoked during pregnancy Ever smoked daily LBW vs. NBW Urban vs. suburban Alcohol/drug Mother_s education
OR

95% CI

OR

95% CI

OR

1.69 1.27 V V V

1.16Y2.46* 0.84Y1.92

1.56 1.28 1.41

1.06Y2.31* 0.85Y1.94 0.99Y2.01

1.46 1.24 1.38 1.71 V

V V V

V V V V

95% CI 0.99Y2.16** 0.82Y1.89 0.96Y1.97 1.22Y2.38** V V V

OR

95% CI

OR

95% CI

1.19 1.12 1.35 1.51 1.38

0.79Y1.80 0.73Y1.72 0.94Y1.94 1.06Y2.15* 0.88Y2.16

1.27 1.11 V 1.52 1.39

0.85Y1.90 0.72Y1.71 1.07Y2.18* 0.89Y2.18

2.26 1.48 1.60

1.17Y4.37* 0.80Y2.70 0.90Y2.83

2.29 1.47 1.62

1.18Y4.41* 0.80Y2.70 0.92Y2.86

Note: Reference for smoking variables is never smoked; reference for maternal education is college or higher. GEE = generalized estimating equation; DISC = Diagnostic Interview Schedule for Children; OR = odds ratio; LBW = low birth weight; NBW = normal birth weight. * p < .05; **p < .005.

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 20 07

365

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

NIGG AND BRESLAU

11. Prenatal smoking exposure was associated with increased risk of ADHD (model 1). LBW did not mediate the association; that is, the LBW-adjusted estimate of effects of smoking during pregnancy was only minimally reduced (model 2). The effect of maternal smoking did not hold up when maternal substance use disorders and education level were included (model 4), indicating that the association was highly confounded with familial factors. The final model (model 5) confirms this conclusion. GEE analysis of Teacher Report Form ratings of inattention/hyperactivity symptoms at ages 6, 11, and 17 replicated the results of the analysis based on DISC-III-R ADHD at ages 6 and 11 (results not shown). Prenatal smoking exposure predicted symptoms of inattention/hyperactivity, but this effect did not survive statistical control of maternal substance use problems or education level. Independent effects of LBW on teacher-reported attention problems (as a continuum) across the three assessments survived all controls (" = 1.07, SE = 0.03, p < .01). Note that, as displayed in Table 2, LBW was not associated with an increased odds ratio for ADHD according to maternal interview based DISC-III-R assessments at ages 6 and 11. In all analyses, no distinct effects were detected when inattentive versus hyperactive/ impulsive symptoms were examined separately (not shown). No effects of prenatal smoking exposure were found in relation to youth lifetime DSM-IV ADHD based on the DISC-IV interviews at age 17 (data not shown). ODD

Table 3 shows results for the logistic regression predicting child ODD assessed at age 17 by the youth interviews using the DIS-IV. The pattern of findings was in striking contrast to the findings on ADHD/ attention problems. The relative odds of ODD was more than twofold higher among children exposed to prenatal smoking ( p < .01; compared to offspring of mothers who never smoked) in all models, even with maternal behavioral and educational level, as well as LBW, controlled statistically. In contrast, LBW was unrelated to ODD. These results were replicated when ODD status was based on maternal interviews using DSM-III-R criteria at ages 6 and 11 (the relative risk of ODD was more than twofold; p < .05; data not shown), whereas LBW was unrelated to ODD.

366

CD

Table 4 shows results from logistic regressions of DSM-IV CD as assessed via the youth DIS-IV at age 17. It shows that prenatal smoking exposure was related to CD, whereas LBW was not, and this effect survived statistical control of confounders. Thus, the pattern was similar to that for ODD. Additional models showed that when CD is controlled in a model predicting ODD, the effect of smoking on ODD remained significant ( p < .05). However, when we set up the model to correspond to the developmental progression of ODD to CD, and thus entered age 6 and age 11 ODD (maternal report), the effect of prenatal smoking exposure on CD was no longer significant. Thus, prenatal smoking exposure may contribute to ODD and, via that route, indirectly, to later CD. DISCUSSION

Most of the extensive literature on prenatal smoking exposure and child disruptive behavior problems has looked at either ADHD or antisocial/disruptive behaviors in isolation. Moreover, most studies did not include simultaneous examination of LBW, which is an established outcome of prenatal smoking exposure (Kramer, 1987). We sought to add to this literature parallel examination of effects of prenatal smoking exposure on ADHD, ODD, and CD in a prospective, longitudinal design beginning at age 6 years and continuing through age 17 years, when all children had passed through the period of risk of onset for these disorders. We found that LBW was associated with teacher-rated ADHD, whereas the association of prenatal smoking exposure with ADHD appeared to be highly confounded with family characteristics, calling into question a causal interpretation. In striking contrast, ODD was predicted robustly by prenatal smoking exposure and did not have an independent association with LBW. Prenatal smoking exposure also predicted CD, apparently via its effect on ODD. This finding is consistent with the known developmental pathway from irritable temperament to ODD (Sanson and Prior, 1999) and then from ODD to CD (Lahey and Loeber, 1994). We focus on DSM-based categorical constructs and present collateral evidence from dimensional ratings using the Achenbach Attention Problems Scale. This addresses the need to evaluate etiological distinctions

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

PRENATAL SMOKING EXPOSURE AND ADHD

TABLE 3 Successive Logistic Regression Models of Maternal Smoking in Pregnancy and Self-Report of Lifetime ODD (DIS; Age 17) Model 1 Model 2 Model 3 Model 4 Model 5 Smoked during pregnancy Ever smoked daily LBW vs. NBW Urban vs. suburban Alcohol/drug Mother_s education
OR

95% CI

OR

95% CI

OR

2.47 1.93 V V V

1.40Y4.38** 1.04Y3.60*

2.76 1.91 0.65 V V

1.53Y4.96** 1.02Y3.57* 0.39Y1.06

2.66 1.88 0.64 1.29 V

V V V

V V V

95% CI 1.47Y4.80** 1.01Y3.52* 0.39Y1.05 0.79Y2.10

V V V

OR

95% CI

OR

95% CI

2.31 1.74 0.63 1.17 1.08

1.24Y4.31* 0.92Y3.28 0.38Y1.04 0.71Y1.95 0.55Y2.14

2.07 1.77 V 1.14 1.07

1.13Y3.81* 0.94Y3.32

2.03 1.92 1.63

0.78Y5.33 0.81Y4.54 0.71Y3.75

1.99 1.90 1.62

0.76Y5.24 0.80Y4.51 0.70Y3.72

0.69Y1.89 0.54Y2.12

Note: Reference for smoking variables is never smoked; reference for maternal education is college or higher. ODD = oppositional defiant disorder; DIS = Diagnostic Interview Schedule; OR = odds ratio; LBW = low birth weight; NBW = normal birth weight. * p < .05; **p < .0051.

and overlaps in the DSM model, particularly in view of recent calls to identify higher-order dimensions that may account for comorbidity (Krueger, 1999). The results suggest potential etiological distinctions between ADHD and ODD/CD, while highlighting overlaps in the etiology of ODD and CD. They suggest that even though ADHD sometimes may be a precursor to ODD and CD, it also appears to remain on a partially distinct pathway, with partially distinct etiological determinants, in this case more strongly related to LBW and less clearly related to prenatal smoking exposure. This is consistent with the twin study by Waldman et al. (2002) in which behavioral genetic analyses indicated that certain environmental influences on ODD and CD are distinct from

environmental influences on ADHD, whereas there is considerable overlap in their genetic influences as well as in other environmental influences. Consistent with this picture, our results for ODD and CD contrasted with the results for ADHD with regard to environmental etiology. ODD and CD were not uniquely related to LBW but had robust relationships with prenatal smoking exposure. Because ODD develops into CD in about 25% of cases (Lahey and Loeber, 1994) and because ODD Bmediated[ the effect of LBW on CD in this study, we can suggest the following model. Prenatal smoking exposure amplifies irritable temperament in children (see Brook et al., 2000), laying the foundation for negative parentYchild interchanges that can emanate in CD. Theories of CD

TABLE 4 Successive Logistic Regression Models of the Effects of Maternal Smoking in Pregnancy on Self-Reports of Lifetime Conduct Disorder (DIS; Age 17) Model 1 Model 2 Model 3 Model 4 Model 5 Smoked during pregnancy Ever smoked daily LBW vs. NBW Urban vs. suburban Alcohol/drug Mother_s education
OR

95% CI

OR

95% CI

OR

95% CI

OR

95% CI

OR

95% CI

2.30 1.12 V V V

1.11Y4.77* 0.46Y2.76

2.39 1.12 0.86 V V

1.13Y5.04* 0.45Y2.75 0.44Y1.68

2.50 1.14 0.87 0.73 V

1.18Y5.32* 0.46Y2.81 0.44Y1.71 0.38Y1.40

2.32 1.10 0.87 0.71 1.08

1.05Y5.15* 0.44Y2.74 0.44Y1.71 0.36Y1.39 0.44Y2.64

2.25 1.11 V 0.70 1.07

1.03Y4.91* 0.45Y2.76

1.38 1.50 1.66

0.36Y5.21 0.49Y4.58 0.58Y4.74

1.38 1.50 1.66

0.36Y5.21 0.49Y4.59 0.58Y4.74

V V V

V V V

V V V

0.36Y1.38 0.44Y2.63

Note: Reference for smoking variables is never smoked; reference for maternal education is college or higher. DIS = Diagnostic Interview Schedule; OR = odds ratio; LBW = low birth weight; NBW = normal birth weight. * p < .05.

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 20 07

367

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

NIGG AND BRESLAU

suggest that it has roots in an irritable temperament (Lahey and Waldman, 2003), and developmental studies indicate that ODD as well as CD may be preceded by early irritability in the toddler years (Sanson and Prior, 1999). Whereas that effect may in part be due to heritable temperamental factors, temperamental elements are now recognized to be partially formed by early experiences (Saudino, 2005). Prenatal exposures are likely to be among them. Based on these data, it could be hypothesized that prenatal smoking exposure may be one element in the irritable precursors of ODD, which in turn can contribute to the emergence of CD. Previous results on this issue of smoking and CD have been inconsistent. Several studies have linked smoking to conduct problems, particularly in case-control designs (Milberger et al., 1998). However, in a twin sample, Silberg et al. (2003) found that prenatal smoking effects on dimensionally assessed conduct problems were explained by genetic transmission of antisocial tendencies from mothers to offspring; however, another twin study found that there was a persistent unique effect on conduct problems (Button et al., 2005). Potentially these effects depend on which aspects of externalizing problems are involved. Most CD symptoms have a median age at onset of age 9 (Lahey and Loeber, 1994), whereas the onset of ODD symptoms occurs around age 6 years. It may be that early temperamental precursors are important and that effects on CD per se are less consistent. However, we cannot rule out the possibility of genetic transmission in our findings, even though we did control for family confounders. In contrast, like Knopik et al. (2005) and unlike Thapar et al. (2003), we found that prenatal smoking effects on ADHD failed to survive control of confounders. Overall, the effect of prenatal smoking exposure on ODD was substantiated in both parent and youth diagnostic interviews. The CD effect was also reliable in diagnostic interview data. These results extend earlier reports on this sample (Breslau and Chilcoat, 2000) to age 17 and to DSM-IV disorders, clarifying the pattern of findings in this sample. This pattern of effects by informant is consistent with evidence regarding validity of different informants for the disruptive behavior disorders; that is, youth self-reports of ADHD appear to have poor validity (Barkley et al., 2002). In contrast, youth reports of conduct problems have greater validity than parent reports or even clinician impressions (Jewell et al., 2004).

368

Limitations

Three key limitations bear comment. First, we did not have DSM-IV measurement of the child disorders at the early ages because the study began before DSM-IV was released. This could particularly affect results for ADHD, which in DSM-IV underwent substantial change with the addition of new subtypes. ADHD cases here probably corresponded most closely to the DSM-IV combined type (Biederman et al., 1997). Furthermore, we have no data on ADHD diagnosis at age 17 based on parent interviews. Instead we used teacher ratings of attention problems, which do not conform to DSM criteria of ADHD and may have less validity than parent-based diagnostic interview. Second, we did not assess the effects of smoking during pregnancy by either the time during pregnancy in which it had occurred or the duration. Additional information may emerge if these were to be assessed. Third, we did not have data on ADHD or antisocial personality in the biological parents; therefore, our controls for family confounders were incomplete. It is possible that the effects of smoking on ODD/CD would be accounted for by additional measures of psychopathology in the parents. However, controlling for maternal substance use disorders did account for the observed effects of prenatal smoking on ADHD but not ODD or CD, suggesting the possibility of a differential effect of maternal smoking during pregnancy on ADHD versus ODD or CD. Clinical Implications

Although it goes without saying that it remains valuable to advise women to avoid smoking during pregnancy, there may be implications here for early prevention of disruptive behavior disorders. In particular, preventive intervention in the form of referral for parenting guidance may be valuable for irritable youngsters with a history of maternal smoking. Prevention efforts may also benefit from recognizing potentially specific risks of ADHD in children with LBW. CONCLUSIONS

In conclusion, our findings suggest that observed associations between prenatal smoking exposure and ADHD in offspring are confounded by maternal education and maternal history of substance use disorder, whereas LBW has independent effects on

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 2007

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

PRENATAL SMOKING EXPOSURE AND ADHD

ADHD. In contrast, the effects of prenatal smoking exposure on ODD and CD in childhood are independent of LBW as well as of maternal education and history of substance use disorder. These findings are consistent with a model of partially distinct etiological contributors to ADHD versus ODD and CD. Disclosure: The authors have no financial relationships to disclose. REFERENCES Barkley RA, Fischer M, Smallish L, Fletcher K (2002), The persistence of attention-deficit/hyperactivity disorder into young adulthood as a function of reporting source and definition of disorder. J Abnorm Psychol 111:279Y289 Biederman J, Faraone SV, Weber W, Russell RL, Rater M, Park KS (1997), Correspondence between DSM-III-R and DSM-IV attention-deficit/ hyperactivity disorder. J Am Acad Child Adolesc Psychiatry 36:1682Y1687 Breslau N, Brown GG, DelDotto JE et al. (1996), Psychiatric sequelae of low birth weight at 6 years of age. J Abnorm Child Psychol 24:385Y400 Breslau N, Chilcoat HD (2000), Psychiatric sequelae of low birth weight at 11 years of age. Biol Psychiatry 47:1005Y1011 Breslau N, DelDotto JE, Brown GG et al. (1994), A gradient relationship between low birth weight and IQ at age 6 years. Arch Pediatr Adolesc Med 148:377Y383 Breslau N, Paneth N, Lucia VC, Paneth-Pollak R (2005), Maternal smoking during pregnancy and offspring IQ. Int J Epidemiol 34:1047Y1053 Brook JS, Brook DW, Whiteman M (2000), The influence of maternal smoking during pregnancy on the toddler_s negativity. Arch Pediatr Adolesc Med 154:381Y385 Button TM, Thapar A, McGuffin P (2005), Relationship between antisocial behaviour, attention-deficit disorder and maternal prenatal smoking. Br J Psychiatry 187:155Y160 Centers for Disease Control and Prevention (2004), Smoking during pregnancyVUnited States, 1990Y2002. MMWR Morb Mortal Wkly Rep 53:911Y912 Danckaerts M, Heptinstall E, Chadwick O, Taylor E (1999), Self-report of attention deficit and hyperactivity disorder in adolescents. Psychopathology 32:81Y92 Diggle PJ, Liang KY, Zeger SL (1994), Analysis of Longitudinal Data. New York: Oxford University Press Foulder-Hughes LA, Cooke RWI (2003), Motor, cognitive, and behavioral disorders in children born very preterm. Dev Med Child Neurol 45:97Y103 Jewell J, Handwerk M, Almquist J, Lucas C (2004), Comparing the validity of clinician-generated diagnosis of conduct disorder to the diagnostic interview schedule for children. J Clin Child Adolesc Psychol 33:536Y546 Knopik VS, Sparrow EP, Madden PAF et al. (2005), Contributions of parental alcoholism, prenatal substance exposure, and genetic transmission to child ADHD risk: a female twin study. Psychol Med 35:625Y635 Kotimaa AJ, Moilanen I, Taanila A et al. (2003), Maternal smoking and hyperactivity in 8-year-old children. J Am Acad Child Adolesc Psychiatry 42:826Y833 Kramer MS (1987), Determinants of low birth weight: methodological assessment and meta-analysis. Bull WHO 65:663Y737

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 46:3, MARCH 20 07

Krueger RF (1999), The structure of common mental disorders. Arch Gen Psychiatry 56:921Y926 Lahey BB, Loeber R (1994), Framework for a developmental model of oppositional defiant disorder and conduct disorder. In: Disruptive Behavior Disorders in Childhood, Routh DK, ed. New York: Plenum, pp 139Y180 Lahey BB, Waldman ID (2003), A developmental propensity model of the origins of conduct problems during childhood and adolescence. In: Causes of Conduct Disorder and Juvenile Delinquency, Lahey BB, Moffitt TE, Caspi A, eds. New York: The Guilford Press, pp 76Y117 Levy F, Hay DA, McStephen M, Wood C, Waldman ID (1997), Attentiondeficit hyperactivity disorder: a category or a continuum? Genetic analysis of a large-scale twin study. J Am Acad Child Adolesc Psychiatry 36:737Y744 Linnet KM, Dalsgaard S, Obel C et al. (2003), Maternal lifestyle factors in pregnancy risk of attention deficit hyperactivity disorder and associated behaviors: review of the current evidence. Am J Psychiatry 160: 1028Y1040 Linnet KM, Wisborg K, Obel C et al. (2005), Smoking during pregnancy and the risk for hyperkinetic disorder in offspring. Pediatrics 116: 462Y467 Mick E, Biederman J, Prince J, Fischer MJ, Faraone SV (2002), Impact of low birth weight on attention-deficit hyperactivity disorder. J Dev Behav Pediatr 23:16Y22 Milberger S, Biederman J, Faraone SV, Chen L, Jones J (1996), Is maternal smoking during pregnancy a risk factor for attention deficit hyperactivity disorder in children? Am J Psychiatry 153:1138Y1142 Milberger S, Biederman J, Faraone SV, Jones J (1998), Further evidence of an association between maternal smoking during pregnancy and attention deficit hyperactivity disorder: findings from a high-risk sample of siblings. J Clin Child Psychol 27:352Y358 Pinto-Martin J, Whitaker A, Feldman J et al. (2004), Special education services and school performance in a regional cohort of low-birthweight infants at age nine. Paediatr Perinat Epidemiol 18:120Y129 Sanson A, Prior M (1999), Temperament and behavioral precursors to oppositional defiant disorder and conduct disorder. In: Handbook of Disruptive Behavior Disorders, Quay HC, Hogan AE, eds. New York: Kluwer, pp 397Y418 Saudino KJ (2005), Behavioral genetics and child temperament. Dev Behav Pediatr 26:214Y223 Shaffer D, Fisher P, Piacentini J, Schwab-Stone M, Wicks J (1989), DISC-2: Parent Version. New York: Division of Child and Adolescent Psychiatry, New York State Psychiatric Institute Sherman DK, McGue M, Iacono W (1997), Twin concordance for attention deficit hyperactivity disorder: comparison of teachers_ and mothers_ reports. Am J Psychiatry 154:532Y535 Silberg JL, Parr T, Neale MC, Rutter M, Angold A, Eaves LJ (2003), Maternal smoking during pregnancy and risk to boys_ conduct disturbance: an examination of the causal hypothesis. Biol Psychiatry 53:130Y135 Thapar A, Fowler T, Rice F et al. (2003), Maternal smoking during pregnancy and attention deficit hyperactivity disorder symptoms in offspring. Am J Psychiatry 160:1985Y1989 Waldman ID, Rhee SH, Levy F, Hay D (2002), Causes of the overlap among symptoms of attention deficit hyperactivity disorder, oppositional defiant disorder, and conduct disorder. In: Attention, Genes, and ADHD, Levy F, Hay D, eds. Philadelphia: Taylor & Francis, pp 115Y138 Williams GM, O_Callaghan M, Najman JM et al. (1998), Maternal cigarette smoking and child psychiatric morbidity: a longitudinal study. Pediatrics 102:e11

369

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.