ARCHIVED - Early primary school outcomes associated with maternal use of alcohol and tobacco during pregnancy and with exposure to parent alcohol and tobacco use postnatally


Early primary school outcomes

3. Method

3.1 Data Source

This report used data from the BBBF Longitudinal Study database for analyses of the effects of prenatal and postnatal alcohol and tobacco on children’s health and developmental outcomes during early primary school.

The BBBF Longitudinal Study is one of the most ambitious research projects on the long-term impacts of early childhood prevention programming for disadvantaged children in Canada. The diversity of the participating communities (francophone, Aboriginal, recent immigrants, and multicultural) increases the likelihood that findings will be applicable to children across Canada.

The longitudinal study began in 1993 and is following two groups of children and their families who experienced up to four years of BBBF prevention programming. One group received Better Beginning programs from birth to age 4 (the younger group), and a second group received the programs from ages 4 to 8 (the older group). Also included in the longitudinal research is a comparison group of children and their families from several demographically matched communities that did not receive BBBF funding. (See for a complete description of the research design and analyses.)

Data from the younger children only are included in the present study, as these children were involved in the study from 3 months of age. From 1993 to 2003, data were collected on approximately 600 children when the children were 3 months, 18 months, 33 months and 48 months, and in Grades 1 (age 6 years) and 3 (age 8 years). Data were collected by trained researchers in each community via a parent interview, direct child measures and, beginning at 48 months, from the child’s teacher using a teacher report form. Over 100 outcome measures have been gathered at each data collection point, covering a wide range of child, parent/family and neighbourhood characteristics.

One of the unique features of the current study is the number of potentially confounding variables that were statistically controlled in all analyses. The measures used as covariates in all analyses were those that might bias the results due to factors other than smoking or drinking during pregnancy. By including these measures in the analyses, statistical controls were employed to remove any bias these variables may have had on the differences between groups. A complete list of the measures used as covariates in the analyses appears in Appendix 2, and includes measures of family income, maternal education, immigrant status, home language and single-parent status. Also included in this list of control variables is whether or not children resided in a BBBF or comparison community. Thus, any outcome differences resulting from Better Beginning program effects have been statistically eliminated from the following analyses.

3.2 Better Beginnings, Better Futures, Study Characteristics

The BBBF study has generated the most extensive and intensive longitudinal database involving disadvantaged children and families in Canada. The BBBF longitudinal study contains more information about early child development and parent behaviour in disadvantaged neighbourhoods than the National Longitudinal Survey of Children and Youth (NLSCY; Statistics Canada & Human Resources Development Canada, 1995), the Ontario Child Health Study (OCHS; Statistics Canada, 2004) and the Montreal Longitudinal Study (MLS; Tremblay, Mâsse, Kurtz & Vitaro, 1996). The NLSCY longitudinal samples are selected to match the general Canadian population in terms of socio-economic and other demographic variables. Hence in these longitudinal samples, there are relatively few children living in disadvantaged families. This is also true of the OCHS sample in Ontario. Further, since the OCHS and MLS began studying children longitudinally at ages 4 and 6, respectively, no data were collected in these two studies from mothers or children at or immediately following the children’s birth. Finally, neither the OCHS nor MLS collected as wide a variety of child outcome measures as the BBBF longitudinal study.

3.3 Measures of Maternal Alcohol and Tobacco Use

As part of the first parent interview, when their child was 3 months old, mothers in the BBBF study were asked a series of questions concerning, among other things, their use of alcohol and tobacco when they were pregnant with this child. These questions are similar to those used in the NLSCY and other population surveys and are presented in Table 1. The questions concerned mothers’ reports of alcohol use and cigarette smoking during their pregnancy, as well as indications of high-risk problem drinking using the four questions from the CAGE questionnaire (Ewing, 1984), and are described below.

Responses to the alcohol-use questions were categorized as “never drank,” “drank less than once per month,” “drank more than once per month.” Responses to the questions concerning cigarette smoking during pregnancy were categorized as “never smoked,” “smoked less than ½ pack per day,” “smoked more than ½ pack per day.”  

Table 1
Questions in Parent Interview Concerning Maternal Alcohol Use and Smoking
 1. Did you drink alcohol during your pregnancy?
a. If yes, did you change the amount you drank while you were pregnant?
b. How often did you drink alcohol during this pregnancy?
c. When you drank alcohol during this pregnancy, how many drinks would you have, on average, each time?
CAGE 1. Did you ever feel that you ought to cut down on your drinking?
CAGE 2. Did people annoy you by criticizing your drinking?
CAGE 3. Did you ever feel bad or guilty about your drinking?
CAGE 4. Did you ever have a drink first thing in the morning to steady your nerves or get rid of a hangover?
1. Did you smoke cigarettes during your pregnancy?
a. If yes, how many cigarettes did you smoke on a typical day during this pregnancy?
b. Did you change your smoking pattern during this pregnancy?
c. If you stopped smoking, in which month of pregnancy did you stop?
2. Did any of the other people living in your household smoke cigarettes during your pregnancy?
a. If yes, how many?


3.4 Sample Size

The size of the longitudinal sample at the various data collection points (i.e. child ages 33 months, 48 months, Grade 1 and Grade 3) for which prenatal alcohol and tobacco use responses were available appear in Table 2 for alcohol use and Table 3 for tobacco use. The attrition in the longitudinal sample was approximately 19% from 48 months to Grade 3. Analyses of differences between families that were maintained in the dataset compared with those that dropped out yielded no indication of bias resulting from sample attrition (see Peters et al., 2000 for a thorough discussion of these attrition analyses). More specifically, with regard to the data analyzed for the present study, there were no differences between the retained sample and those lost in terms of mothers’ reports of smoking or drinking patterns during pregnancy.

As shown in Tables 2 and 3, the sample size when the children were 48 months of age is approximately 500; the sample was reduced to 407 by Grade 3. Over 20% of the sample at each point in time consisted of mothers who reported some prenatal alcohol consumption; over 30% reported some prenatal tobacco exposure. Approximately 6% of these mothers reported using alcohol more than once per month during pregnancy, and 9% reported smoking more than ½ pack of cigarettes per day. These rates are higher than those reported in the Canadian Community Health Survey (see Section 2).

Approximately 99% of the children were residing with their biological mother at age 3 months, and this decreased slightly over time to 97% at age 33 months and 96% at Grade 1 and Grade 3. Due to the small number of children living with a foster parent or guardian at Grade 1 (N = 12) and Grade 3 (N = 9), it was not possible to analyze the data to see if those living with non-biological parents differed in exposure to prenatal alcohol or tobacco when compared with those living with a biological parent. Thus, the results of analyses reported here apply almost exclusively to children who were living with at least one biological parent from birth to Grade 3.

Table 2
Longitudinal Sample Sizes: Alcohol Use During Pregnancy
Mother's reported drinking during pregnancy Child age at time of longitudinal data collection
33 Mos 48 Mos Gr. 1 Gr. 3
Never N/em> (%) 414 (77.4) 399 (76.9) 343 (77.1) 330 (78.4)
< once/mo. N/em> (%) 89 (16.6) 88 (17.0) 75 (16.9) 67 (15.9)
> once/mo. N/em> (%) 32 (6.0) 32 (6.2) 27 (6.1) 24 (5.7)
Total N/em> (%) 535 (100) 519 (100) 445 (100) 421 (100)

Table 3
Longitudinal Sample Sizes: Tobacco Use During Pregnancy
Mother's reported smoking during pregnancy
(# cig./day)
Child age at time of longitudinal data collection
33 Mos 48 Mos Gr. 1 Gr. 3
None N/em> (%) 352 (68.2) 345 (68.9) 290 (66.4) 275 (67.6)
< ½ pack N/em> (%) 118 (22.9) 110 (22.0) 101 (23.1) 88 (21.6)
> ½ pack N/em> (%) 46 (8.9) 46 (9.2) 46 (10.5) 44 (10.8)
Total N/em> (%) 516 (100) 501 (100) 437 (100) 407 (100)


3.5 Sample Definition

3.5.1 Tobacco Use

Due to the relatively imprecise data on the number of cigarettes smoked daily reported by the mothers, two categories of prenatal smoking were formed: those mothers that reported any smoking during pregnancy and those that reported no smoking. Thus, the smoking sample includes women who reported smoking less than a half pack per day (about two-thirds of the mothers) as well as heavier smokers.

3.5.2 Alcohol Use

For prenatal alcohol consumption, several ways of categorizing the mothers’ reports of alcohol use were explored in conjunction with several of the child outcome measures. The most sensitive measure was whether the mother answered “Yes” to one or more of the four CAGE questions (see Table 1). If she did, she was considered a higher-risk drinker (MHRD) during pregnancy. If not, she was considered lower risk. The decision to use this method of identifying children who were exposed to higher versus lower risk of prenatal alcohol was based on several studies that indicated the use of scores of 1 or greater on the CAGE as being the most sensitive to problem drinking in women while scores of 2 or greater on the four CAGE questions have been found to be most sensitive to higher-risk drinking in men (Bradley, Boyd-Wickizer, Powell & Burman, 1998; Midanik, Zahnd & Klein, 1998; Moraes, Viellas & Reichenheim, 2005).

Note that this is a behavioural definition of higher risk based on reported feelings of guilt, annoyance, sober second thought (“I ought to cut back”) and hangover avoidance by morning drinking. The guilt, annoyance and second-thought criteria are likely to have captured a substantial number of cases where the amount of drinking was moderate as well as the behaviour of very heavy drinkers. We did not try to quantify the amount of alcohol consumed in any of the analyses reported here.

There is emerging evidence that the most severe harmful effects of prenatal alcohol exposure result from mothers’ binge drinking rather than from more regular or more frequent light or moderate consumption. Although more research is needed on more subtle outcomes resulting from prenatal exposure to lower concentrations of alcohol, the higher-risk versus lower-risk dichotomy of mothers’ prenatal alcohol consumption based on a CAGE score of 1 or more was adopted for analyses of children’s prenatal exposure to alcohol in this study.

Table 4
Sample Sizes for the Four Groups of Mothers Regarding Drinking and Smoking During Pregnancy
Data cycles Smoking during pregnancy Drinking during pregnancy (CAGE)
Lower-risk drinker Higher-risk drinker Total
33 Months Didn't smoke 302 22 324
Smoked 122 24 146
Total 424 46 470
Chi-sq test (Exact test): β2 (1) = 10.6, p<0.001
48 Months Didn't smoke 298 21 319
Smoked 115 24 139
Total 413 45 458
Chi-sq test (Exact test): β2 (1) = 12.5, p<0.001
Grade 1 Didn't smoke 249 16 265
Smoked 107 23 130
Total 356 39 395
Chi-sq test (Exact test): β2 (1) = 13.3, p<0.001
Grade 3 Didn't smoke 241 13 254
Smoked 97 21 118
Total 338 34 372
Chi-sq test (Exact test): β2 (1) = 15.6, p<0.001


3.5.3 Prevalence of Alcohol and Tobacco Use During Pregnancy

This strategy divided mothers into four groups regarding alcohol use and smoking during pregnancy:

  1. higher-risk drinking, smoking;
  2. higher-risk drinking, non-smoking;
  3. lower-risk drinking, smoking;
  4. lower-risk drinking, non-smoking (see Table 4 for sample sizes).

The chi-square statistical test results reported at each age reflect a highly statistically positive relationship between mothers’ reports of prenatal smoking and their reports of high-risk drinking.

We decided that the BBBF longitudinal dataset contained enough detailed information on mothers’ alcohol and tobacco use during pregnancy to warrant further analyses concerning relationships with children’s development, school readiness and functioning during early primary school. The sample sizes were considered to be adequate to allow analyses of the independent and combined association between prenatal exposure to alcohol and to tobacco, with a wide range of measures of child development.

3.6 Measures of Child Development

These domains correspond closely with the five domains of school readiness currently employed in Canada (Janus & Offord, 2000). A total of 79 child outcome measures were selected for preliminary analysis. Most of these measures were collected when the children were 48 months, 6 years (Grade1) and 8 years (Grade 3) of age. Three of the measures had been collected when the children were 33 months old. The specific child outcome measures selected for analysis are listed in Appendix 1 for each of the five domains of child development.

The BBBF dataset has measures of five major domains of children’s development.
  1. Children’s general development
  2. Cognitive development/academic performance,
  3. Social/emotional functioning
  4. Behaviour problems
  5. Child health


Page details

Date modified: