Social Policy Research Paper No. 36

Contents

• Executive summary
• 1. Introduction
• 2. Outcomes for children in differing circumstances
• 3. Children's use of non-parental care
• 4. Children's health
• 5. Mothers' health
• 6. Family learning environments and child outcomes
• 7. Discussion
• Appendix A: Summary of variables included in the Outcome Index
• Appendix B: LSAC Research Consortium
• Endnotes
• References
• List of tables and figures

5. Mothers' health

• 5.1 Introduction
• 5.2 Findings
• 5.3 Discussion

Section summary

• LSAC mothers were broadly representative of all Australian mothers on parameters such as smoking and alcohol in pregnancy, mental health status, and prevalence of overweight and obesity.
• Serious maternal psychological distress was associated with a very substantial worsening of child outcomes. Better maternal general health and enjoyment of physical activity predicted better child Outcome Index scores, driven mainly by their impacts on the physical and social–emotional (rather than learning) domains.
• Prenatal health, smoking and alcohol use, and meeting nutritional and physical activity guidelines contributed relatively little to children's outcomes.

5.1 Introduction

This section addresses the following questions:

• What prenatal, postnatal, and current maternal health issues face Australian infants and 4 to 5 year-old children, and how strongly are these associated with child outcomes?
• What associations do maternal choices regarding healthful levels of physical activity, nutritional habits and substance use (smoking, alcohol) have with their children's overall functioning?
• When considered relative to each other, which factors appear to make the strongest contribution to children's overall wellbeing?

Like the child health data, maternal health data collected in Wave 1 were mainly cross-sectional. However, retrospective data regarding maternal health during pregnancy and the postnatal period allow us to start to appreciate the longitudinal nature of the development of good and poor child outcomes in the early years.

5.2 Findings

Maternal health variables

The maternal health variables (Table 20) could broadly be classified as prenatal, postnatal, current health, and current lifestyle. In the bivariate and multivariable analyses, we examined whether maternal health variables predicted Outcome Index scores, while noting the caveats regarding causal inferences above. All maternal health measures were self-reported.

The child outcomes examined for this section were the overall Outcome Index and the three domain scores. We examined the association between each maternal health variable with the continuous Outcome Index score and the proportion of study children with an Outcome Index score in the top 15 per cent and bottom 15 per cent of their cohort.

Figures 11 to 12 show the proportions of the child cohort according to the maternal health variables scoring in the top 15 per cent (that is, above the positive cut-off ) and bottom 15 per cent (that is, below the negative cut-off ) of total Outcome Index scores. Independent associations of each variable with the overall Outcome Index score for both the infant and child cohorts were examined using multivariable regression analyses controlling for all other prenatal, postnatal, current health and current lifestyle variables as well as for the set of socioeconomic variables referred to in Section 2. These results are presented in Tables 21 and 22.

The proportions of the child cohort according to each maternal health variable scoring in the top and bottom 15 per cent of overall Outcome Index scores show that some of the relationships are striking, while others are marginal or absent (see Figures 9 and 10). These analyses are bivariate, that is, not adjusted for sociodemographic or other variables. Comparable graphs for the infant cohort are not shown; relationships between health variables and high or low Outcome Index scores were generally either absent or weaker than (but in the same direction as) the child cohort.

Prenatal maternal health variables

• Physical health problems during pregnancy: close to 6 per cent of mothers reported having diabetes in both cohorts. The prevalence of diabetes in pregnancy (mainly gestational diabetes) is uncertain because of its rapidly changing epidemiology and the multiple sets of criteria in use internationally for its diagnosis (Agarwal et al. 2005). However, the LSAC prevalence appears to be at the lower end of recently-quoted rates of ≥8 per cent, suggesting either that the LSAC cohorts had lower rates of gestational diabetes (that is, were healthier than average) or that not all cases had been detected. Of children whose mothers reported having gestational diabetes, 21 per cent fell into the lowest 15 per cent of Outcome Index scores, compared to less than 15 per cent of children whose mothers did not report this condition.
• About 8 per cent in both cohorts reported hypertension, comparable to the 9.8 per cent prevalence (0.6 per cent chronic hypertension, 4.2 per cent pre-eclampsia, 0.3 per cent chronic hypertension with pre-eclampsia, and 4.3 per cent gestational hypertension) reported in a recent large New South Wales study of women delivering babies in 2000–2002 (Roberts et al. 2005). Children whose mothers reported hypertension during pregnancy were slightly overrepresented in the lowest 15 per cent of Outcome Index scores compared to children whose mothers did not report this condition (18 per cent versus 15 per cent).
• Smoking: close to 20 per cent of mothers in both cohorts reported smoking during pregnancy, a figure which is comparable to the 17 per cent reported in five states and territories in 2003 (Laws & Sullivan 2005); heavy smoking was, however, rarely reported (see Table 20). Of infants whose mothers reported smoking more than two packs of cigarettes per day during pregnancy, 30 per cent fell in the lowest 15 per cent of Outcome Index scores, compared to 15 per cent of infants whose mothers did not smoke during pregnancy. The percentage of the child cohort scoring in the top 15 per cent of Outcome Index scores fell from no smoking (17 per cent), through occasional smoking (11 per cent), to daily smoking (5 per cent), while the percentage scoring in the bottom 15 per cent of Outcome Index scores rose in a stepwise fashion from 13 per cent for no smoking to over 27 per cent for smoking most days.
• Alcohol consumption during pregnancy was reported more often for the infant cohort (38 per cent) than the child cohort (28 per cent). Reported rates from other studies vary greatly (from 22 to 76 per cent) depending on the sampling strategy, but the rates for both LSAC cohorts appear somewhat higher than reported in the 2005 Alcohol Awareness Survey, in which 23 per cent of Australian women aged 25 to 49 years reported continuing to drink during pregnancy (Roy Morgan Research 2005). Less than 1 per cent of mothers in both cohorts reported consumption 'most days' during pregnancy. The percentage of children scoring in the lowest and highest 15 per cent of Outcome Index scores showed little variation according to the frequency of alcohol consumption during pregnancy (did not drink, drank occasionally, drank most days), with the perhaps surprising exception that a high percentage of children whose mothers reported drinking most days during pregnancy (29 per cent) scored above the positive cut-off for the Outcome Index.

Postnatal maternal health variables

• Postnatal depression: 15 per cent of child cohort mothers reported that they experienced postnatal depression during the first year after the study child was born, in keeping with Australian epidemiologic studies using community measures (such as the Edinburgh Postnatal Depression Scale) during the first postnatal year (Boyce & Stubbs 1994; Brown & Lumley 1998). This question was not asked of mothers of the infant cohort, since in most cases a full year had not elapsed since birth at that time so that, for most of the infants, the current maternal health and lifestyle variables could equally well be considered to be 'postnatal'. Of children whose mothers reported postnatal depression, 20 per cent fell below the negative cut-off for the Outcome Index compared to 13 per cent of children whose mothers reported not having this condition.

Current maternal health variables

• General health: this was rated using the same widely used five point global health item (derived from the SF-6, see Ware et al. 1992) as for the children, with response options of excellent, very good, good, fair and poor. Only 61 per cent of mothers of the infant cohort and 58 per cent of mothers of the child cohort rated their own health as excellent/very good, with the remainder (39 per cent and 42 per cent respectively) reporting their own health to be good, fair or poor. This contrasts with how mothers see the general health of their children (Section 4), with 87 per cent of infants and children rated as being in excellent/very good health. Children whose mothers reported good/fair/poor health were overrepresented in the lowest 15 per cent of Outcome Index scores (23 per cent compared to 10 per cent of children whose mothers reported excellent or very good health).
• Serious psychological distress: the Kessler six-item screen of adult mental health (K-6) (Kessler et al. 2003) probes symptoms of anxiety and depression, is well validated and is widely used in large Australian surveys. Possible scores range from 6 to 30; these scores were dichotomised using the standard recommended cut points for the K-6. Serious maternal psychological distress (a score ≤18) was reported by 4 per cent of mothers of the infant cohort and 6 per cent of the child cohort.
• The percentage of children scoring in the lowest and 15 per cent of Outcome Index scores differed strikingly according to maternal mental health status. Only 3 per cent of children of distressed mothers had Outcome Index scores above the positive cut-off, while more than 40 per cent had scores below the negative cut-off.
• The BMI status of mothers in both cohorts was nearly identical, with 55 per cent classified as not overweight, 26 per cent overweight (BMI›25 up to 30 kg/m2), and 18 to 19 per cent obese (BMI›30kgm2), in keeping with other recent surveys. Only 10 per cent of children of obese mothers were above the Outcome Index positive cut-off, while 22 per cent were below the negative cut-off (compared to 17 per cent and 13 per cent of children of non-overweight mothers, respectively).

Current maternal lifestyle variables

• Current alcohol consumption: reported daily intake was low, with the median number of alcoholic drinks per day being 0.14 for the infant cohort and 0.20 for the child cohort, and only 2 per cent and 3 per cent of mothers respectively averaging more than two standard drinks daily—the current NHMRC definition of 'risky drinking' (NHMRC 2001). Forty-five per cent of mothers in the child cohort, and 48 per cent in the infant cohort, reported either never drinking or drinking less than one drink per month. Surprisingly, children whose mothers averaged more than two standard drinks daily were underrepresented in the bottom 15 per cent of Outcome Index scores (8 per cent compared to 16 per cent of children whose mothers who drank less than two standard drinks daily).
• Regarding binge drinking, although the NHMRC considers more than five standard drinks in a session to be 'a binge' for women, it does not stipulate how often this must occur to constitute 'binge drinking' (NHMRC 2001). If defined as five or more standard drinks in a sitting twice a month or more, 7 per cent of mothers in the infant cohort and 12 per cent of mothers in the child cohort could be classified as binge drinkers. The percentage of children scoring in the lowest 15 per cent of Outcome Index scores showed little variation according to maternal binge drinking.
• Current smoking: 20 per cent of mothers in both cohorts were smokers, nearly identical to 2001 Australian national figures of 20.8 per cent of women over 14 years of age reported to smoke at least occasionally (AIHW 2001). Of those who smoked, 60 per cent reported smoking up to half a packet per day, 6 to 7 per cent up to a full pack per day, and 1 to 2 per cent more than a pack a day. The majority of parents (86 to 88 per cent) reported that no one smoked inside the house; 8 to 10 per cent of households had one person and 4 per cent had two or more people who smoked inside.
• The percentage of children scoring in the highest and lowest 15 per cent varied considerably by current maternal smoking and the number of people smoking inside the study child's household (Figure 12). Only 3 per cent of children whose mothers smoked more than two packets of cigarettes daily scored above the positive Outcome Index cut-off (compared to 17 per cent of children whose mothers did not smoke), while 35 per cent scored below the negative cut-off (compared to 14 per cent of children whose mothers did not smoke). Similarly, 28 per cent of children living with two or more reported indoor smokers scored below the negative cut-off (compared to 14 per cent of children not living with indoor smokers), and just 3 per cent scored above the positive cut-off (compared to 16 per cent of children not living with indoor smokers).
• Mothers reported on their own nutrition, by estimating the number of serves of fruit and of vegetables per day using standardised questions developed for Australia. Only 3.5 per cent of mothers across both cohorts reported meeting current recommendations (five or more serves of vegetables and two or more serves of fruit per day) (NHMRC 2003a). Children whose mothers met these recommendations were overrepresented in the top 15 per cent of Outcome Index scores (20 per cent compared to 15 per cent of children whose mothers did not meet the recommendations).
• Mothers also reported on their own physical activity levels. Close to 30 per cent in both cohorts reported that they found physical activity to be not enjoyable/a little enjoyable, around 30 per cent reported it as very enjoyable, with the remaining 40 per cent finding it moderately enjoyable. The percentage of the child cohort scoring in the top 15 per cent of Outcome Index scores rose in a stepwise fashion from 'not enjoyable/a little enjoyable' (12 per cent) to 'very enjoyable' (20 per cent), while the percentage scoring in the bottom 15 per cent of Outcome Index scores fell from 'not enjoyable/a little enjoyable' (21 per cent) through to 'very enjoyable' (10 per cent). According to their reports, 17 to 18 per cent of mothers meet the current recommendation of at least 30 minutes of moderate to vigorous physical activity (MVPA) five or more times per week, with the median frequency of 30 minutes daily MVPA being two days per week for both cohorts (Department of Health and Ageing 1999). The percentage of children scoring in the highest or lowest 15 per cent of Outcome Index scores showed little variation according to MVPA status.

Impact of maternal health variables on child outcomes: multivariable analyses

Tables 21 (infant cohort) and 22 (child cohort) show multivariable linear regression relationships between the maternal health variables and the continuous Outcome Index and domain scores, all adjusted for the nine key sociodemographic variables identified in Section 2. Table 21 shows that the final multivariable models for the infant cohort accounted for only small proportions of the total variance in scores (5.5 per cent of the overall Outcome Index and 5 per cent or less for each domain score). Over and above the contribution of the sociodemographic variables, the maternal health variables as a group accounted for 3.6 per cent of the variability in overall Outcome Index score.

There was no evidence that diabetes, hypertension, or cigarette smoking during pregnancy were associated with overall Outcome Index or domain scores (all p›0.10), suggesting that the observed correlation between heavy prenatal smoking and infant Outcome Index scores may be largely explicable by sociodemographic gradients. There was strong evidence (p‹0.001) that maternal general health (but not serious psychological distress) affected infant overall and physical and social–emotional scores, all of which were on average 2 to 3 points lower if the mother reported only good, fair or poor (as opposed to excellent or very good) general health. There was no evidence that daily alcohol consumption, maternal smoking, number of residents smoking indoors, or maternal fruit and vegetable intake were associated with Outcome Index scores (all p›0.30). There was strong evidence (p‹0.001) that increasing maternal enjoyment of physical activity was associated with increasing overall Outcome Index, social–emotional and learning domain scores, but not physical domain scores. Regarding maternal BMI status, the single finding that overweight and obesity were associated with slightly higher social–emotional domain scores is of uncertain importance.

Table 22 shows that the final multivariable models for the child cohort accounted for considerably larger proportions of the total variance in scores (20 per cent for the overall Outcome Index, 7 per cent for the physical domain, 17 per cent for the social–emotional domain, and 14 per cent for the learning domain respectively). Over and above the contribution of the nine sociodemographic variables, the maternal health variables as a group accounted for 6.4 per cent of variability in Outcome Index scores. Regarding maternal health impacts on the LSAC children, Table 22 shows the following strong associations:

• Poorer maternal general health was associated with lowered overall Outcome Index, physical and social–emotional domain scores, even more than in the infant cohort (with scores falling by 3 to 4 points on average, p‹0.001).
• Serious psychological distress was associated with lower scores on all Outcome Index measures, with the largest effect size being for overall Outcome Index and social–emotional domain scores (with scores falling by 5 and 6 points on average—more than half a standard deviation—respectively, both p‹0.001).
• Maternal enjoyment of physical activity was associated with higher overall Outcome Index (p=0.002) and social–emotional domain (p‹0.001) scores (with scores falling by 2 points on average).

For all other maternal health variables, there was either no evidence of association with Outcome Index or domain scores (diabetes and hypertension in pregnancy; current alcohol and smoking; and fruit and vegetable intake) or the associations were weak and/or inconsistent in direction (cigarette smoking during pregnancy, number of residents who smoke indoors, BMI status).

Substantial amounts of data were missing for both the postnatal depression and maternal BMI status variables. Because postnatal depression had the highest number of missing cases and showed little relationship to overall Outcome Index or domain scores, it was excluded from the multivariable analyses reported above (which increased the available child sample by around 250 individuals). Maternal BMI status was missing for a further 224 infants and 217 children, and made a borderline contribution to the learning domain score (p=0.02) in the child cohort. Data on BMI status were more likely to be missing when a language other then English was spoken at home, with lower maternal education and low family income (both infants and children); lower occupational class (infants only); and being an Aboriginal or Torres Strait Islander (child cohort only) (all p‹0.05). Therefore, the multivariable analysis was repeated without the inclusion of maternal BMI status as a potential predictor. Total variance explained was nearly identical with and without maternal BMI status and no conclusion changed as a result.

Figure 11: Low Outcome Index: percentage of child cohort

Figure 12: High Outcome Index: percentage of child cohort

5.3 Discussion

These findings confirm the major role of mothers' own health in their children's outcomes. LSAC mothers were broadly representative of all Australian mothers on parameters such as smoking and alcohol in pregnancy, mental health status, and prevalence of overweight and obesity. This gives confidence that the associations between maternal health and children's outcomes revealed in the multivariable analyses are likely to be generalisable to the broader population of Australian children.

Of all the child and maternal health variables studied, serious maternal psychological distress predicted the greatest fall in children's outcomes (by half a standard deviation or more). Rates of serious maternal psychological distress were 1.5 times higher in mothers of preschoolers (6 per cent) than mothers of infants (4 per cent). This is not an isolated finding—Queensland's Longitudinal Mater Study (Najman et al. 2000) also reported that maternal depression was more common prenatally and at age 4 than in the postnatal year. There appears to be a need for services and supports directed at improving maternal mental health throughout the entire preschool period, and perhaps beyond.

Maternal general health and enjoyment of physical activity were other powerful predictors of child Outcome Index scores, driven mainly by their impacts on the physical and social–emotional (rather than learning) domains, though their impacts were less sizeable than that of maternal mental health. However, prenatal health, smoking and alcohol use, and meeting nutritional and physical activity guidelines contributed relatively little to children's outcomes in these analyses. These are interesting findings, given that children's own nutritional habits appeared to make a significant and substantial contribution to their outcomes, and that the prenatal maternal health variables studied (hypertension, diabetes, smoking, and alcohol intake) would be expected to contribute to conditions such as intrauterine growth restriction and preterm birth—which did show strong relationships with children's outcomes. It is recognised that separating 'child' and 'maternal' variables is artificial, since they are so closely intertwined particularly during pregnancy. Therefore, analyses combining child and maternal health variables might alter somewhat the relationships depicted in Sections 4 and 5, but were beyond the scope of this report. It is also likely that sociodemographic gradients 'explain' why some of the associations seen in the bivariate were not evident in the multivariable analyses.

The high prevalence of maternal overweight and obesity is of great concern, since maternal BMI status typically exceeds all other risk factors for current and future obesity in their offspring. Though beyond the scope of this report, this relationship has been demonstrated to hold for the LSAC children (Wake et al. 2007). However, it is not surprising that maternal BMI contributed little to children's outcomes, since child overweight/obesity itself had little impact on outcomes. Over time, it is likely that relationships between maternal BMI status, child BMI status, and child outcomes will emerge and intensify.

Taken together, the data in Sections 4 and 5 strongly suggest that children's physical health, mothers' psychological health, and the healthful behaviours of both children and mothers make major contributions to Australian children's outcomes throughout the early years.

Return to top

Next: 6. Family learning environments and child outcomes

Previous: 4. Children's health

Copyright & Disclaimer | Privacy policy | Contact us | FaHCSIA home

© Commonwealth of Australia, 2008
Last modified 24 April, 2009