 |
 |
|
|
|
 Estimated to occur in 9% of boys and 3% of girls aged 4 to 16 years (1), ADHD is the most common psychiatric syndrome in children. While psychostimulants are documented as effective in controlling the overactivity, inattention, and impulsiveness typical of ADHD (2), their use has been controversial (3). The most common stimulant used is MPH (4). ADHD and treatment with stimulant medications are more common among boys than among girls and among school-aged children than among preschoolers or teenagers (5–10). Other correlates of stimulant use in population-based studies appear to reflect the child’s context. For instance, urban residence has been associated with use in some (9) but not all (5) studies, and family income has been positively (9), negatively (11), and not (5) associated with stimulant use. Variations in levels of stimulant use have been correlated with geography (5,9,11,12) and prescriber characteristics (5,8, 9,11). In some jurisdictions, a few physicians account for large numbers of prescriptions (5,9). Identification and treatment of ADHD increased during the 1990s in the US (6,8,13) and in Canada (11). Stimulant use appears to have increased among girls (8), and the age of peak use has gone up (8), as have prevalence rates of use, suggesting persistence of use into the teenage years. The foregoing information regarding stimulant use has been obtained from large-scale administrative and prescription databases. These databases contain information recorded by health care providers and reflect physician diagnostic and prescribing practices. Databases such as these offer little information about parent and family characteristics that might influence use of medications for treatment of ADHD. Small-scale clinical studies identify psychosocial adversity such as family conflict, parent psychopathology, and low social class as common correlates of childhood ADHD (14–18). Negative parent–child interactions have also been implicated in the etiology of ADHD in some (15,19), but not all (20), studies. Clinical experience suggests that psychosocial adversity or poor parenting practices could interfere with obtaining medical treatment or adhering to treatment if it is prescribed. If this is so, then psychosocial adversity or poor parenting practices would be expected to lessen the likelihood that a child would use stimulants for ADHD. In addition, parents decide when to seek help from professionals, and only a minority of children who have ADHD receive specialized services from mental health agencies, social services, or special education programs (1). For these reasons, investigating the contribution of parent and family factors to stimulant use in children is important in understanding which children receive treatment for ADHD. The NLSCY is a large-scale, population-based database that collected information primarily from parents about their children. It offers an alternative data source to administrative and prescription databases for information regarding use of stimulant medications by Canadian children—one reflecting parent and family sources, not health care provider sources of information. An earlier report using the NLSCY noted that the prevalence of MPH use was less than 2% in 1994–1995 and increased by 36% in 1996–1997 (21). Use was noted to be greater among boys than among girls and among those of school-age than among preschoolers. While MPH use was presumed to occur among children with ADHD, no behavioural characteristics of children or information about parents was investigated. This study uses the NLSCY database to investigate psychosocial correlates of ADHD identified in clinical populations to see whether these characteristics are associated with MPH use in the nonclinical population. It tracks MPH use from 1994–1995 to 1998–1999 and investigates parent and child characteristics as correlates of stimulant use. It is hypothesized that boys of school age with high levels of hyperactivity and aggression will likely be the children using MPH, because these are common clinical correlates of children with ADHD (22,23). The secondary hypotheses are that parent and family characteristics associated clinically with ADHD, such as maternal depression, parental substance use, poor family functioning, and poor parenting skills, will also be associated with MPH use in children. It is anticipated that increases in stimulant use will occur over time among all age groups and both sexes. MethodUsing data from the NLSCY, we identified 3 cross-sectional cohorts of children aged 2 to 11 years in 1994–1995, 1995–1996, and 1998–1999. We measured associations between child, parent, and family characteristics and parent- reported MPH use. Information regarding correlates was collected by means of structured interviews in the child’s home with the person most knowledgeable about the child—in 89.9% of cases, the biological mother. The initial data collection in 1994–1995 used a random sampling frame of households in the Labour Force Survey, with clusters within age groups and large geographic areas, to be representative of the Canadian child population. Statistical weighting procedures developed by Statistics Canada to be representative of cross-sectional population estimates were used. (For a full description of the NLSCY, see www.statcan.ca/english/freepub/89F0078XIE/free.htm.) At maximum, 4 children per household were identified. Those identified in Cycle 1 (1994–1995) were tracked for inclusion in Cycle 2 (1997–1998) and in Cycle 3 (1998–1999). Overall response rates to the survey were 87%, 91%, and 88%, respectively. Study Population For each data collection, the subset of children aged 2 to 11 years whose parent answered the question “Does he/she take any of the following prescribed medications on a regular basis: Ritalin?” were included. In 1994–1995, 17 814 observations were identified, representing 3 962 900 children. In 1996–1997, 13 575 observations were identified, representing 3 935 800 children; in 1998–1999, 18 980 observations were identified, representing 3 693 600 children. Owing to budgetary constraints in 1996–1997, families who had participated in a concurrent study were removed, shrinking the potential pool of subjects for that cycle. The families were recontacted in 1998–1999. Measures The dependent variable was MPH use (yes or no). During the 1990s, only 3 stimulants were available on the market in Canada: MPH, dextroamphetamine, and pemoline. MPH accounted for 88% of stimulants prescribed for ADHD in the US between 1990 and 1995 (8). Similarly, data from the Manitoba Centre for Health Policy documents that 79% of stimulant prescriptions between 1995 and 2004 were for MPH (M Brownell, unpublished data, 2005), providing Canadian data that most children receiving stimulants in Canada received MPH. The independent variables evaluated fell into 3 categories, sociodemographic, individual child, and parent or family factors. Sociodemographics included urban or rural household, low maternal education (did not complete high school), low income adequacy, and single-parent status. Income adequacy was determined according to Statistics Canada’s derived variable of household size and income. The 2 lowest categories were combined to indicate low income adequacy, a dichotomous grouping that corresponds closely to Canada’s poverty lines in 1995 (24). Child characteristics included sex, age, and child behaviour. Age was subgrouped into preschool, age 2 to 5 years, and school age (6 to 11 years) to investigate potential increased use over time in the 2 age categories separately. Child behaviour (hyperactivity and aggression) was measured as follows: Parent-reported descriptions of child hyperactivity and aggression were collected by means of the NLSCY Child Behaviour Scale (25). Items with known psychometric properties from other large community-based surveys of children were used to develop this scale. Separate scales were used for children aged 2 to 3 years and for those aged 4 to 11 years. For the younger age group, principal components analysis identified 5 underlying factors; the following 2 subscales were used for this study: Hyperactivity–Inattention, with Cronbach’s a = 0.80, and Physical Aggression–Opposition, with Cronbach’s a = 0.75 (25). For the children aged 4 to 11 years, principal components analysis identified 6 factors, of which 2 were used for this study: Hyperactivity–Inattention, with Cronbach’s a = 0.84, and Physical Aggression–Conduct Disorder, with Cronbach’s a = 0.77 (25). After stratification by age in 2-year increments, as well as by sex, those children scoring > 1.5 SD above the mean were classified as positive for hyperactivity or positive for aggression on the corresponding scales. Parent and family factors included family dysfunction, maternal depression, parenting style, parent smoking, and parent drinking. Family Dysfunction. The 12-item General Functioning subscale of the McMaster Family Assessment Device is a reliable and valid measure of family functioning (26) previously used in large-scale community surveys (27). Families scoring > 1.5 SD above the mean were identified as positive for dysfunction. Maternal Depression. An abbreviated version of the Centre for Epidemtiologic Studies Depression Scale Scale (28) was used to determine the frequency of maternal depressive symptoms. The full version has been previously used in large-scale community surveys (29). The scale estimates the presence and severity of 12 items describing mood symptoms over the past week. Mothers scoring > 1.5 SD above the mean were identified as positive for depression. Parenting Style. An 18-item Parenting Scale for children aged 2 to 11 years, based on Strayhorn and Weidman’s Parenting Practices Scale (30), was developed by Boyle (25). Principal components analysis (25) identified 3 underlying factors: positive parent–child interaction, with Cronbach’s a = 0.81; hostile-ineffective parenting, with Cronbach’s a = 0.71; and consistent parenting, with Cronbach’s a = 0.66. In addition, a scale on parent management techniques for children aged 2 to 11 years was developed (25). One factor, aversive parenting, was identified, with Cronbach’s a = 0.57. A score in the highest quartile was used to indicate a high degree of the factor of interest. Parental Smoking. Parental smoking for either parent was recorded as present or absent in response to the question, “At the present time do you/does . . . smoke cigarettes?” This cut-off identified approximately one-half (52%) of the children in 1994–1995 as having a parent who smoked. Parental Alcohol Use. Parental alcohol use was recorded as positive if the frequency estimate of alcohol consumption for either parent was more than once monthly. This cut-off identified approximately one-half (56%) of the children in 1994–1995 as having a parent who drank alcohol more than once monthly. Statistical AnalysisStatistics Canada’s data publication guides were followed throughout all analyses (25). Data were weighted up to the population level according to weights derived by Statistics Canada for the purpose of cross-sectional analysis for each data collection. Corrections for the effect of sampling design on parameter estimates were produced with coefficients of variation derived by Statistics Canada. The prevalence of MPH use was estimated for preschoolers (aged 2 to 5 years) and school-aged children (aged 6 to 11 years) in each data collection. Chi-square was used to compare sample characteristics and the proportion of the sample with positive responses to question regarding MPH use at each data collection. The associations between potential correlates and MPH use for each data collection were also evaluated with chi-square. The following variables were included in the logistic regression models to be investigated as potential correlates of MPH use for each time point: child age, sex, low income adequacy, low maternal education, single-parent status, high level of child hyperactivity, high level of child aggression, presence of maternal depression, presence of family dysfunction, parental smoking, parental alcohol use, and 4 parenting style variables. Collinearity among the variables was investigated with correlation matrices. Backward selection procedure (with P value to stay = 0.05 and P value to remove = 0.10) was used to select variables to be included in the final model for each data collection. The Hosmer and Lemeshow test was used to evaluate goodness-of-fit for the logistic model. Statistical analyses were performed with SAS Version 8.2 (31). ResultsTable 1 illustrates the characteristics of the samples in 1994–1995, 1996–1997, and 1998–1999. Compared with the 1996–1997 and 1998–1999 samples, the 1994–1995 sample included fewer children from urban households (81.7%, compared with 86.8% and 86.7%; P < 0.001) and more mothers who did not complete high school education (16.3%, compared with 12.1% and 11.7%; P < 0.001). Compared with the 1994–1995 and 1996–1997 samples, the 1998–1999 sample included fewer children from families with low income adequacy (11.7%, compared with 17.9% and 17.7%; P < 0.001).
The overall prevalence of MPH use increased from 1.33% (95%CI, 1.17% to 1.50%) in 1994–1995 to 1.71% (95%CI, 1.49% to 1.93%) in 1996–1997 and remained higher at 1.60% (95%CI, 1.42% to 1.78%) in 1998–1999. A significant increase occurred among girls, where 0.52% (95%CI, 0.32% to 0.67%) used MPH in 1994–1995, 0.59% (95%CI, 0.41% to 0.78%) used MPH in 1996–1997, and 0.97% (95%CI, 0.77% to 1.17%) used MPH in 1998–1999. Use among school-age children (aged 6 to 11 years) increased from 1.94% (95%CI, 1.68% to 2.20%) in 1994–1995 to 2.62% (95%CI, 2.27% to 2.97%) in 1996–1997 and remained higher at 2.42% (95%CI, 2.14% to 2.71%) in 1998–1999. Table 2 illustrates child characteristics, sociodemographic factors, and parent or family risk factors associated with MPH use. In all data collections, the following variables were associated with MPH use: parent-reported hyperactivity and aggression, single-parent household, maternal depression, hostile–ineffective parenting, parental smoking, and absence of parental alcohol use.
Table 3 shows the results of the logistic regression models for 1994–1995, 1996–1997, and 1998–1999. In all data collections, parent-reported hyperactivity, age 6 to 11 years, and male sex were strong correlates of the child’s MPH use. The OR for children with parent-reported hyperactivity using MPH increased from 6.63 (95%CI, 4.94 to 8.90) in 1994–1995 to 12.54 (95%CI, 9.74 to 16.16) in 1998–1999. The OR for school-aged children (aged 6 to 11 years) using MPH ranged from 4.44 (95%CI, 2.98 to 6.62) in 1994–1995 to 8.15 (95%CI, 4.87 to 13.64) in 1996–1997, with no significant differences between samples. The OR for boys using MPH decreased from 4.86 (95%CI, 3.39 to 6.96) in 1994–1995 to 2.52 (95%CI, 1.92 to 3.30) in 1998–1999, consistent with the increased use among girls noted above. Parental alcohol use was negatively correlated with MPH use in all 3 samples: OR 0.49 (95%CI, 0.37 to 0.65) in 1994–1995, OR 0.74 (95%CI, 0.56 to 0.99) in 1996–1997, and OR 0.67 (95%CI, 0.52 to 0.87) in 1998–1999. The regression models using child and parent or family variables as correlates of MPH use fit the data better in the 1994–1995 sample than in the subsequent 2 samples.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||