 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
a School of Psychology, University of Ottawa, Ottawa, Ontario, Canada
b Research Unit on Children’s Psychosocial Maladjustment, University of Montreal, Montreal, Quebec, Canada
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONIMPLICATIONS AND LIMITATIONSCONCLUSIONSREFERENCES |
|---|
PARTICIPANTS. Four data-collection cycles of a nationwide survey of Canadian children were used to track the early development of hyperactivity. Children were 0 to 23 months at the first cycle in 1994 and 6 to 7 years at the fourth cycle in 2000.
OUTCOME MEASURES. Hyperactivity data were gathered from mothers on a biennial basis beginning when children were 24 months old. Information on potential prenatal and postnatal predictors was gathered from mothers at the first cycle.
DESIGN. Group-based semiparametric mixture modeling was used to estimate developmental trajectories, and logistic-regression analysis identified predictors of hyperactivity.
RESULTS. Four trajectories of hyperactive symptoms were identified: very low, low, moderate, and high. Statistically significant predictors for high and persistent hyperactivity, after controlling for all other factors, were maternal prenatal smoking, child male gender, maternal depression, and hostile parenting.
CONCLUSIONS. For the majority of children, the frequency of hyperactive symptoms decreased or remained low from 2 to 7 years. However, 7 children in 100 were classified as having high initial levels of hyperactive symptoms that persisted over time. Several prenatal and early postnatal risk factors identified these children, although additional variables will need to be identified to accurately predict high and persistent hyperactivity. Findings suggest that preventive interventions could target high-risk families during pregnancy and early childhood.
Key Words: developmental • early childhood • hyperactivity • predictors
Abbreviations: ADHD—attention-deficit/hyperactivity disorder • NLSCY—National Longitudinal Survey of Children and Youth • OR—odds ratio
Children who exhibit hyperactive symptoms are often referred for mental health services because of problems functioning at school, at home, and with peers. Because hyperactivity is a frequent childhood behavior problem of chronic duration, it is important to better understand its developmental course. A longitudinal study of 1000 at-risk boys found that 6% scored high on teacher-reported hyperactivity at 6 years old, and these boys continued to score high over time until midadolescence. The remaining boys exhibited either very low levels of hyperactivity or moderate to high levels that decreased over time.1 This type of information is important for distinguishing normative from problematic levels of hyperactivity and, more specifically, persistent problematic hyperactivity levels. Unfortunately, longitudinal studies on hyperactive symptoms during early childhood are limited. A recent review identified only 6 studies on hyperactivity during the preschool period, with all of the studies using a cross-sectional design and only 1 of the studies using data from a national sample.2 The first objective of the study, therefore, was to model the early development of hyperactive symptoms in a national population sample of children from 2 to 7 years of age. Studying early development means that effective interventions for high and chronic hyperactivity can be introduced at a young age and that they can be better tailored to meet developmental changes over time.3
The second study objective was to examine early predictors of high-level hyperactive symptoms as a way to better target prevention and intervention efforts. Past studies have identified a number of predictors of hyperactivity and, more specifically, attention-deficit/hyperactivity disorder (ADHD), including family history of ADHD, negative parenting behaviors, maternal depression, childhood oppositional/aggressive symptoms, and childhood difficult temperament.4–6 These studies, however, have tended to focus on boys or older children or have relied on nonrepresentative samples of children. Moreover, few studies have focused on early predictors that occur during pregnancy or early infancy. A review of pregnancy lifestyle habits indicated that maternal prenatal smoking and drinking generally increased the risk of hyperactivity in children and adolescents.7 However, only a few of the reviewed studies were based on national samples. A more recent study found that maternal prenatal smoking remained a significant predictor of childhood hyperactivity after controlling for genetic influences, social adversity, child low birth weight, and child comorbid conduct disorder.8 The study, however, was limited by its cross-sectional design. Our study tested child (eg, temperament) and maternal (eg, depression) characteristics present at birth, as well as postnatal factors (eg, parenting behaviors) before the child's second birthday. We included a number of variables of practical importance that could be modified or prevented.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONIMPLICATIONS AND LIMITATIONSCONCLUSIONSREFERENCES |
|---|
59000 Canadian households that are representative of the population of Canada. The NLSCY excluded children living in institutional facilities (0.5% of 0- to 11-year-old children), on Aboriginal reserves, and in the 2 Canadian territories, because they are not targeted by the Labor Force Survey. NLSCY budgetary constraints and attrition have led to loss of data over time. For example, cycle 2 eliminated certain cycle 1 responding households and reduced the maximum number of children surveyed per household from 4 to 2. In the end, there was a high overall response rate (91.6%), which resulted in a responding sample of 15468 children ages 2 to 13 years from 10261 households.10 The NLSCY includes both cross-sectional and longitudinal weights to adjust for possible bias resulting from attrition and to maintain sample representation.
At cycle 1, information was collected from the person most knowledgeable about the child (ie, biological mother in 90% of cases). Mothers participated in a personal home interview with a trained interviewer. Data also were collected from older children and from teachers of school-aged children.9 We selected children 0 to 23 months of age at cycle 1 because of our interest in the early to middle childhood developmental period. By cycle 4, children were 7 years old (ie, 
95 months of age). Children were also required to have complete hyperactivity data for 
2 of the data cycles to ensure reliable estimation of developmental trajectories (only 1.5% were eliminated because of incomplete data), resulting in a sample of 2946 children. Table 1 shows that children were evenly distributed across gender and age group, and the majority of children (88.5%) were of white ethnicity. Slightly >1 in 10 mothers (11.2%) had their first child at <20 years of age, and 2.7% were <20 years at the birth of the study child. Approximately 1 in 5 mothers (20.7%) and fathers (22.5%) did not complete high school, and 1 (21.1%) in 4 mothers and 1 (8.3%) in 10 fathers were unemployed. Approximately 1 (11.2%) in 10 children were living in a single-parent household where other siblings were present (58.7%). Close to 17% of households in our sample were substantially worse off than the average Canadian household, as defined by the low-income cutoff, which estimates the minimum income required by a family to meet its basic needs for food, shelter, and clothing. Different cutoffs are established according to family size and the size of the community in which the family lives. Given the stratified, multistage probability sampling strategy of the NLSCY, it should be noted that these sample demographics are representative of Canadian families with children aged 0 to 23 months in 1994.
|
Maternal Depression
At cycle 1, mothers provided information on current symptoms of depression (ie, over the past week) by rating 12 items from the Center for Epidemiologic Studies Depression Scale11 along a 4-point scale from 1 (rarely or none of the time; <1 day) to 4 (most or all of the time; 5–7 days). Higher scores indicated greater depressive symptoms. The Cronbach 
reliability was .82 for 0- to 11-year-old children at cycle 1.
Perinatal Information
At cycle 1, mothers provided information on tobacco and alcohol use during pregnancy. They also answered questions about the infant's status (eg, birth weight). We focused exclusively on the presence/absence of prenatal smoking and drinking (and not on frequency and timing during pregnancy), because findings suggest, for example, that the majority of pregnant smokers continue to smoke throughout their pregnancy.12, 13
Parenting Practices
At cycle 1, mothers rated 7 items from the Parent Practices Scale14 along a 5-point scale from 0 (never) to 4 (many times each day). The measure yields 2 types of parenting practices, namely positive interaction (5 items with a Cronbach reliability of .73 for 0- to 23-month-old children at cycle 1) and hostility (2 items with a Cronbach reliability of .40 for 0- to 23-month-old children at cycle 1). Higher scores indicated greater use of the parenting practice. Sample positive interaction items include, "How often do you do something special with the child that he/she enjoys?" and "How often do you and your child talk or play with each other, focusing attention on each other for five minutes or more, just for fun?" Sample hostility items include, "How often do you get annoyed with your child for saying or doing something that he/she is not supposed to do?" and "How often do you tell your child that he/she is bad or not as good as others?"
Family Dysfunction
At cycle 1, mothers rated 12 items from the McMaster Family Assessment Device15 along a 4-point scale from 0 (strongly agree) to 3 (strongly disagree). Higher scores indicated greater family dysfunction. The Cronbach reliability was .88 for 0- to 11-year-old children at cycle 1. Sample items include, "We express feelings to each other" and "We are able to make decisions about how to solve problems."
Child Temperament
At cycle 1, mothers rated 7 items from the Infant Characteristics Questionnaire16 along a 7-point scale from 1 (very little; very easy) to 7 (a lot; very difficult). Higher scores indicated a more difficult temperament. The Cronbach reliability was .77 for 0- to 11-month-old children and .79 for 12- to 23-month-old children at cycle 1. Sample items include, "How much does your child cry and fuss in general?" and "How easy or difficult is it for you to calm or soothe your child when he or she is upset?"
Child Outcome Variable
Hyperactive Symptoms
The items on hyperactive symptoms were adapted from the widely used Child Behavior Checklist17 and have been used in 2 Canadian longitudinal studies.18, 19 At each data-collection cycle, mothers rated hyperactive symptoms along a 3-point scale from 0 (never or not true) to 2 (often or very true). Note that mothers did not rate the amount of impairment arising from these symptoms. Child behavior assessments began at 24 months of age, so data in our study were only available for cycles 2, 3, and 4 (because children were 0–23 months at cycle 1). The 3 hyperactivity items were, "Can't sit still, is restless, or hyperactive"; "Has difficulty awaiting turn in games or groups"; and "Cannot settle to anything for more than a few moments." Higher scores indicated greater hyperactive symptoms. The Cronbach reliability coefficient was .80 for 24- to 35-month-old children and .84 for 4- to 11-year-old children at cycle 1.
Statistical Analyses
The first step was to identify distinctive groups of developmental trajectories for hyperactive symptoms in children from ages 24 to 95 months (ie, 2–7 years, corresponding with NLSCY cycles 2–4). To estimate the trajectories, we used an accelerated longitudinal design with 2 age cohorts, specifically children who, at cycle 1, were between 0 and 11 months and 12 and 23 months. We also grouped the ages in 3-month intervals (eg, 24–26, 27–29, 30–32, etc). The Appendix illustrates the accelerated longitudinal design by cohort over time. By combining the data from the 2 different age cohorts (each of which covers only a segment of the total time period of interest), we are able to study the progression of hyperactive symptoms over the entire developmental period from 24 to 95 months. Note that this design does not impact on the predictor variables, which were collected at cycle 1 for the entire sample of 2946 children aged 0 to 23 months.
|
The second step was to use logistic regressions to identify predictors of children in the high hyperactivity group compared with those who did not exhibit high levels of hyperactive symptoms. Note again that data on the predictors were derived from cycle 1 of the NLSCY. The first regression model focused on variables present before or at the birth of the child, and the second added postnatal parenting and family functioning variables. In addition to the NLSCY longitudinal weights, we also incorporated additional weighting procedures using the posterior probabilities of group membership from the trajectory analyses. This was done to take into account the ambiguity that can exist in assigning children to the different trajectory groups. All of the predictor variables were categorized as follows: maternal age was coded as 14 to 19 years, 20 to 25 years, and 26 years; maternal depression was scored using established clinical cutoffs11, 26; and parenting practices, family dysfunction, and child temperament were scored using cutoffs of 1 SD above or below the mean score for our sample. Table 2 indicates that 14.6% of mothers in the sample reported experiencing depression. Approximately 1 (22.7%) in 5 mothers reported smoking while pregnant, and close to 1 (17.7%) in 5 reported prenatal drinking. Low mother-child positive interaction was reported by 13.9% of mothers, and 14.3% reported high hostile interactions with their child. Close to 1 (17.7%) in 5 mothers indicated high family dysfunction. Low birth weight characterized 6.2% of our sample, and 12.8% of mothers reported a difficult temperament in their child.
|
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONIMPLICATIONS AND LIMITATIONSCONCLUSIONSREFERENCES |
|---|
|
|
|
1 of these 3 risk factors were in the high and persistent hyperactivity trajectory group. This sensitivity rating suggests that high-level persistent hyperactivity in children would be accurately identified 2 of 3 times with a scale composed of prenatal smoking, maternal depression, and hostile parenting. The positive predictive value of such a scale, however, was low (11%). This finding suggests that a scale composed only of prenatal smoking, maternal depression, and hostile parenting would not have much predictive use, because it would result in a large number of false positives (ie, children being wrongly labeled as exhibiting high and persistent hyperactive symptoms). |
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONIMPLICATIONS AND LIMITATIONSCONCLUSIONSREFERENCES |
|---|
7 children in 100 were classified as having high initial levels of hyperactive symptoms that persisted over time. Children with these atypical levels of hyperactivity could be those that meet diagnostic criteria for ADHD given previous research showing that behavior problems, in particular at high levels, are stable from early to late childhood.27–29 In any case, the findings suggest that as early as 2 years of age, there is a small proportion of children that could be targeted for interventions that help them (and their parents) better manage problematic symptoms that could escalate over time. The study also tested a number of potential predictors of hyperactivity present early in the child's life. There were 3 significant predictors: maternal prenatal smoking, maternal depression, and maternal hostile parenting. Smoking during pregnancy increased children's risk for high-level and persistent hyperactive symptoms by >2.5 times, after having controlled for a number of other possible risk factors. These results confirm previous studies7 but also show that the impact of prenatal smoking can be observed in early childhood and increases up to 7 years of age. Several authors have proposed that compounds found in tobacco smoke (especially nicotine) may damage brain structure and/or functioning at critical times during the development of the fetus or that tobacco smoke may alter maternal physiology in such a way that there is a decrease in the flow of oxygen and nutrients and/or an increase in the flow of harmful tobacco smoke components to the fetus.30, 31
A common finding is the association between maternal depression and children's symptoms. However, only a few studies have focused specifically on hyperactivity.4–6 Children with a depressed mother were >2 times more likely to have high and persistent levels of hyperactive symptoms than children with a nondepressed mother, after having controlled for a number of other potential risk factors. Although there is concern that depressed mothers are biased toward overreporting symptoms in their children, several studies have found evidence against this hypothesis and have shown mothers (regardless of depression status) to be reliable and accurate reporters of their children's symptoms.32–34 In addition, we did not assess maternal depression and child hyperactivity during the same time period to help guard against this possible confound (ie, maternal depression was assessed 2 years before the maternal assessments of hyperactivity).
Contrary to a number of previous studies that place maternal depression within the context of impaired parenting practices,5, 35–40 we found the impact of maternal depression on children's hyperactivity to be unaffected by negative parenting practices. In other words, negative mother-child interactions and hostile parenting did not seem to mediate the impact of maternal depression on childhood hyperactivity. There is a need for additional longitudinal studies that examine maternal depression specifically in relation to children's hyperactivity. Finally, results showed that mothers' hostile parenting significantly increased children's risk for hyperactivity. Although previous research has linked negative parenting practices to childhood behavior problems, there are virtually no studies that focused specifically on parenting practices and children's hyperactivity.6 Again, more research is needed on parenting practices specifically in relation to hyperactivity in children.
Contrary to past studies,4–8 we did not find low birth weight, difficult temperament, or prenatal drinking to be statistically significant predictors of childhood hyperactive symptoms. The prevalence of low birth weight was only 6.2% in our sample, which may have limited the statistical power to detect significant findings. Low statistical power is a plausible explanation, given that low birth weight was not significantly correlated with maternal prenatal smoking, maternal depression, and maternal hostile parenting. Child temperament showed strong statistical associations with child gender, maternal depression, and maternal hostile parenting (all of which were significant predictors of hyperactivity), and so the effects of temperament may have been attenuated by these other variables. A similar explanation might account for the lack of statistical significance for maternal prenatal drinking, which was significantly correlated with child gender and maternal smoking during pregnancy. These findings also suggest that adding the variables of low birth weight, difficult temperament, and maternal prenatal drinking to a scale composed of prenatal smoking, maternal depression, and hostile parenting might increase the predictive use of identifying children at risk for high and persistent hyperactivity.
|
IMPLICATIONS AND LIMITATIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONIMPLICATIONS AND LIMITATIONSCONCLUSIONSREFERENCES |
|---|
1 of these 3 easily identifiable and amenable factors 2 of 3 times. Therefore, family planning services are needed to inform women about the dangers of smoking during pregnancy and to assist pregnant women to stop smoking. Intervention programs that address mental health problems, such as depression, and that improve parenting practices should be available and readily accessible by members of a community. Although it seems important to introduce such interventions, we also found that these 3 factors alone did not accurately predict hyperactivity, suggesting that they are necessary but not sufficient. Therefore, more research is needed to identify additional factors that, in conjunction with prenatal smoking, maternal depression, and hostile parenting, could be used to develop a scale that predicts early childhood hyperactivity. In terms of interventions for young children who exhibit hyperactive symptoms, a recent study41 offers pharmacologic and psychosocial guidelines. |
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONIMPLICATIONS AND LIMITATIONSCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
We thank Danielle Forest and Alain Girard for statistical assistance and Mark Zoccolillo for comments on previous drafts.
|
FOOTNOTES |
|---|
Address correspondence to Elisa Romano, PhD, University of Ottawa, School of Psychology, 11 Marie Curie St, Ottawa, Ontario, Canada K1N 6N5. E-mail: eromano{at}uottawa.ca
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONIMPLICATIONS AND LIMITATIONSCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  L. M. Furman Attention-Deficit Hyperactivity Disorder (ADHD): Does New Research Support Old Concepts? J Child Neurol, July 1, 2008; 23(7): 775 - 784. [Abstract] [PDF]
|
|
|
|
 |
|
 L. K. Jacobsen, M. R. Picciotto, C. J. Heath, S. J. Frost, K. A. Tsou, R. A. Dwan, M. P. Jackowski, R. T. Constable, and W. E. Mencl Prenatal and Adolescent Exposure to Tobacco Smoke Modulates the Development of White Matter Microstructure J. Neurosci., December 5, 2007; 27(49): 13491 - 13498. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Kutzbach, C. G. Summers, A. M. Holleschau, R. A. King, and J. T. MacDonald The Prevalence of Attention-Deficit/ Hyperactivity Disorder Among Persons With Albinism J Child Neurol, December 1, 2007; 22(12): 1342 - 1347. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2006 American Academy of Pediatrics. All rights reserved. |
||
|
||
indicates very low (4.5%), accompanying line, very low predicted; 
, low (42.0%), accompanying line, low predicted; 
, moderate (46.3%), accompanying line, moderate predicted; •, high (7.2%), accompanying line, high predicted.