Growing up in Scotland: health inequalities in the early years

This report investigates health inequalities in the early years in terms of risk factors and outcomes.


CHAPTER 4 AVOIDING NEGATIVE OUTCOMES

4.1 Key findings about the avoidance of negative health outcomes

The analysis in this chapter explored the factors associated with avoiding negative outcomes among disadvantaged children with a particular focus on the concept of resilience. Resilience has been defined as "the process of withstanding the negative effects of risk exposure, demonstrating positive adjustment in the face of adversity or trauma, and beating the odds associated with risks" (Bartley, 2006). The kinds of factors that have been thought to help children at high risk of negative outcomes to avoid them are wide ranging. This chapter explored a range of possible factors including: maternal, family and household characteristics and behaviours; neighbourhood characteristics; and social support networks.

The extent to which these measures were associated with negative outcomes was explored for all children in the first instance.

The key findings were:

  • The findings in relation to all children reinforce the evidence that there are strong associations between child outcomes and maternal health and behaviours such as smoking, long-term health problems or disability as well as confidence in parenting abilities. It should be recognised, though, that the experience of having a child with negative health outcomes may in itself influence some of these maternal measures.
  • A number of factors within households also showed associations with the avoidance of negative outcomes, for example the consumption of fruit and vegetables and higher levels of physical activity. The findings also suggest possible associations with measures relating to tenure stability and major life events, parental feelings about household income and the home learning environment (the latter is likely to be related to the measures of cognitive and language development used in this stage of the analysis).
  • It has also been suggested that neighbourhoods provide an important source of resilience for families. Based on two measures of satisfaction with local services and judgments of the child friendliness of local areas, positive assessments of these aspects were associated with fewer negative outcomes.
  • The extent of social support appeared to be associated with avoiding negative outcomes. Regular attendance at parent and toddler groups throughout the child's life and the ability to draw on support at short notice were both more common among children with low negative outcomes.

To identify resilience it is necessary to show what factors are associated with avoiding negative outcomes among children who are at an increased risk of them. It was clear from the analysis of health inequalities that for most of the negative outcomes of interest, children living in the most deprived areas, in the lowest income households and in semi-routine and routine households were most likely to experience them. Therefore the next stage of the analysis focused on children from disadvantaged backgrounds - those from any of the three socio-economic groups at most risk of negative outcomes. This approach disentangles the association between resilience and socio-economic background which might have explained the findings outlined above.

This showed that:

  • Only a few of the resilience measures were independently associated with avoiding negative outcomes. Therefore, factors such as area deprivation, income or socio-economic classification clearly have a major influence. In other words, this emphasizes the difficulty of countering very powerful economic and structural influences on early life.
  • The significant resilience measures were quite different in nature to each other. For example, children were less likely to have negative outcomes if their mother had not experienced long-term health problems, or if they lived in a household with at least one adult in full-time work, or if they had a more enriching home learning environment. These different kinds of factors would have very wide ranging policy implications.
  • Some of the significant associations that remain are surprising - for example, even within disadvantaged groups, older maternal age is a predictor of avoiding negative outcomes.
  • It is clear that most of the resilience measures that are significantly associated with avoiding negative outcomes do not sit entirely within the health domain and that effective action to promote resilience and address child health inequalities requires action at many different levels and from a wide range of agencies and bodies.

4.2 Introduction

This chapter starts by briefly mapping out what is meant by resilience in the wider health and child development literature. It then presents the measures from GUS that will be used to explore the concept. The final part of the chapter discusses the analysis conducted to attempt to answer the question of whether any factors appear to be associated with the avoidance of negative outcomes among children from disadvantaged background.

4.3 What is resilience?

Bartley (2006:4) suggests that: "The notion of resilience refers to the process of withstanding the negative effects of risk exposure, demonstrating positive adjustment in the face of adversity or trauma, and beating the odds associated with risks." The concept of resilience has a longer history in psychology and ecology than in the fields of health and social science (Tunstell et al., 2005). In psychology, resilience tends to focus on individual character traits and resources, and is a major concept in child development (see for example Goldstein and Brooks, 2005). In contrast, its adoption by researchers in the health field has seen resilience extended to mean something related to individuals, places and communities (Mitchell et al., 2009). Indeed, some place particular emphasis on the wider social context in which resilience is fostered, for example Gilligan (2004:94) states that "[t]he degree of resilience displayed by a person in a certain context may be said to be related to the extent to which that context has elements that nurture this resilience".

A series of linked projects between 2003 and 2007 explored resilience and health in relation to a wide range of factors including the presence of strong personal relationships (within families and between individuals within communities), positive relationships between parents and children, enriching environments with opportunities for children to play and learn, neighbourhood support networks and social capital, and educational attainment (Bartley, 2006).

The ability to identify possible factors that promote resilience among individuals and communities which might act as a buttress to the kinds of socio-economic disadvantage that so often result in inequalities in outcomes - as presented in Chapter 3 - is of obvious interest to policy makers. However, Wilkinson (cited in Bartley, 2006) argues that while it is clearly right for societies to provide ways of protecting people from the negative consequences of adversity, and to continuously seek better means of doing so, these kinds of policy interventions are not necessarily any less expensive or less difficult to deliver than interventions that might diminish the root causes of disadvantage.

4.4 The definition of resilience in this report

In common with Bartley's (2006) definition above, to explore children's ability to withstand the negative effects of risk exposure we need to distinguish between those who have generally avoided adverse outcomes in their early years and those who have not. The summary measure of negative outcomes outlined at the end of the previous chapter is therefore the starting point for the analysis.

Resilience is arguably a rather meaningless concept if it cannot be demonstrated that negative outcomes have been avoided despite there being a high likelihood of them occurring. It was clear in Chapter 3 that for most of the negative outcomes of interest, children living in the most deprived areas, in the lowest income households and in semi-routine and routine households were the most likely to experience them.

One way of establishing whether children have avoided negative outcomes despite being in a high risk group (i.e. living in a deprived area, low income household, semi-routine and routine household) would be to look at the association between negative outcomes and various resilience measures in each of these three most disadvantaged groups separately. However, the sample is not big enough to restrict the analysis in this way. Instead a pragmatic choice was made to classify children as disadvantaged if they were from any of the three socio-economic groups at most risk of negative outcomes, but not in any of the least deprived categories for these measures (the highest income, the least deprived areas, or professional/managerial households). This additional exclusion was important because children living in the most deprived areas can come from families in the highest income or professional/managerial households. This resulted in an unweighted sub-sample of around 1,000 children.

The inclusion of a wide range of measures in the negative outcomes scale means that the proportion of children in this disadvantaged sub-group of around 1,000 children who avoided all ten of the negative outcomes (and therefore have scores of zero) is quite small. For this reason the analysis of resilience presented here focuses on children with scores of one or zero.

4.5 Potential measures of resilience

As outlined above, resilience can be operationalised in many different ways and in relation to numerous aspects of a child's early experiences. It is never easy to investigate a complex topic such as this using a study that did not have that specific aim as one of its central objectives. However, GUS is fairly broad in its reach and much of what it has covered relates either directly or indirectly to the theory of resilience so while it does not have as comprehensive a set of measures as might be ideal for this purpose, it certainly has enough to allow at least a preliminary scoping of the topic. This should therefore be treated as exploratory and intended to signal future possibilities for analysis of existing data or further questions in GUS.

The kinds of resilience measures identified from within the study's four sweeps can be grouped into four broad themes linked to the characteristics of parents, the family and household setting, the neighbourhood context, and the degree of social support available to the main carer, child and wider family unit. Although these four headings are all linked in some way to the wider literature on resilience, they are not meant to be definitive nor are they fixed. Indeed, many of the measures discussed below could be assigned to more than one of the headings used. What matters is not the allocation of individual measures to specific groups, but rather whether the measures themselves tell us anything meaningful or useful in terms of the central question being addressed:

  • What factors, if any, correlate with the avoidance of negative early health outcomes, among families from disadvantaged backgrounds?

The measures presented in the following tables include some of the risk factors that were explored in Chapter 3, such as maternal health and smoking, and children's diet and activity levels, as well as some new ones. As noted in the previous chapter, the negative outcomes scale excluded the risk factors for poor health explored in this report. This is because these risk factors are not only of interest in terms of the inequality in children's exposure to them, but also in terms of their potential to moderate negative outcomes.

To help introduce the resilience measures, and to illustrate the extent of their overall prevalence, the following tables compare the scale of negative outcomes for all children in relation to resilience. We will return to the question of whether any of these factors are associated with the avoidance of negative outcomes among children from disadvantaged backgrounds after these measures have been mapped out.

4.5.1 Maternal factors

Table 4.1 looks at the association between the number of negative outcomes experienced and a range of possible resilience measures relating to mothers or main carers. The measures highlighted here focus on behaviours that might potentially protect children from experiencing negative outcomes. The main point of the table is to compare the prevalence of potential resilience measures across each of the four groups of children, from those on the left hand side with scores of one or less on the negative outcomes scale, through to those with scores of four to nine. 9 If children with 0-1 negative outcomes have greater exposure to the resilience measure than children with two or more this indicates that it might be associated with avoiding negative outcomes. The total "stock" of each form of resilience in the population is presented in the final column.

For example, 69% of all children had mothers who did not smoke when they were 10 or 34 months old. The prevalence of this was higher (76%) among children with one or less negative outcomes, and much lower (51%) among those with four or more. So having a mother who does not smoke is associated with avoiding negative outcomes. The patterns were similar for the three other measures explored previously in Chapter 3, i.e. long-term health problems, breastfeeding and maternal mental ill-health as measured by the SF-12.

Maternal education is often considered to be an important asset and many studies have shown that it promotes, or is at least associated with, positive outcomes for children (Bradshaw and Martin, 2008). Children with scores of one or less were more likely than those with scores of four to nine to have mothers educated to degree level or above (34% versus 18%) and less likely to have mothers with no qualifications (6% and 16%), or with standard grades (15% and 26%). Some of this pattern is accounted for by mothers' ages when their child is born, and it is clear that children with a low negative outcome score are more likely than those with higher scores to have a mother aged 25 or over.

The table also explores two attitudinal measures based on questions asked when the children were 10 months old. Mothers were asked to assess their parenting ability, having a mother who thought she was a very good parent was more common for children with scores of one or less. Conversely, having a mother who said she was an average or worse parent was much more common among children scoring four to nine.

A composite measure of mothers' attitudes towards asking for help about parenting was created by combining the answers to three statements: "If you ask for help or advice on parenting from professionals like doctors or social workers, they start interfering or trying to take over"; "it's difficult to ask people for help or advice about parenting unless you know them really well"; and "It's hard to know who to ask for help or advice about being a parent". Disagreement with the three statements was considered indicative of having more positive views about help-seeking. Positive views about help-seeking were more common among children with one or less negative outcomes.

The direction of the association between outcomes and the three final resilience measures in the table could well flow in the opposite direction to that suggested by proponents of resilience as a protective factor. For example, negative outcomes for children might impact on maternal well-being and on confidence levels in relation to their parenting skills, rather than the other way round. It is worth noting at this point that none of the patterns presented in the tables should be interpreted as implying anything concrete about the relationship between factors beyond the fact of their association. These results cannot be used to draw conclusions about causation or the direction these associations take.

It is clear from the outset that the association between outcomes and resilience displays a very similar pattern to that seen in Chapter 3 in relation to outcomes and socio-economic factors. The final stage of the analysis in this chapter addresses the fact that many of the associations outlined here are very likely to be partly explained by differences in socio-economic circumstances. The relationships apparent in Table 4.1 to Table 4.3 might well disappear (or at least reduce) when factors such as deprivation, income or NS- SEC are considered as well (see section 4.6).

Table 4.1 Number of negative health outcomes by maternal resilience measures

Resilience measures

Number of negative health outcomes

One or less

Two

Three

Four or more

Total

%

%

%

%

%

Mother did not smoke when child was 10 & 34 months old

76

69

59

51

69

No long-term health problem/ disability since child's birth

72

66

53

50

65

Child was breastfed

64

58

55

50

60

Maternal education

Degree

34

25

19

18

28

HE below degree

38

42

39

36

39

Higher grades

9

7

7

4

7

Standard grades

15

17

21

26

17

No qualifications

6

9

14

16

9

Maternal age

35+

24

17

18

14

21

25-34

57

56

49

42

54

15-24

19

27

33

44

26

Assessment of parenting ability

A very good parent

37

34

32

27

34

A better than average parent

28

26

23

22

26

Average or worse parent

35

39

44

50

39

Mean scores

Attitude to seeking help about parenting (higher mean score = more positive)

10.4

10.1

9.7

9.6

10.1

Standard error of mean

0.04

0.07

0.09

0.10

0.03

SF12 mental health scale (higher mean score = better mental health)

51.3

49.8

47.8

47.0

50.0

Standard error of mean

0.18

0.34

0.48

0.51

0.15

Bases

Weighted

2005

775

499

491

3770

Unweighted

2131

762

465

424

3782

Note:
Bases vary for each measure, those shown are the lowest of the range.

4.5.2 Home and family resilience measures

The measures in Table 4.2 include the two diet questions and the physical activity scale explored in Chapter 3. These are included here as potential indicators of behaviours that might help to build resilience among disadvantaged children if they are encouraged in the home. The differences between the groups are not as stark as was the case with some of the factors shown in Table 4.1. In contrast, how much experience parents had had with children before the study child was born does not appear to have any association with the number of negative outcomes experienced.

The measures of tenure stability and major life events are an attempt at capturing the extent to which children have experienced upheaval in their first four years of life. It is possible that these kinds of disruptions in early life could result in already disadvantaged children being at greater risk of negative outcomes when compared with similarly disadvantaged children who have had more stable lives. Neither are perfect measures and there is of course a direct correlation between attrition in a study like this and major upheavals of these kinds so it is possible that the families in the sample are not wholly typical of the wider population when it comes to measures such as these. Although the differences between the groups are relatively small at around seven percentage points, these measures are worth exploring further at the next stage of the analysis.

The income measure is based on responses to a question included every year that asks parents how they feel about their household income. The scale ranges from "living very comfortably on present income" to "finding it very difficult". Answers from all four years were combined and the sample was split into four roughly equal sized groups ranging from the most positive quarter to the least positive. Although we have direct measures of income this arguably taps a rather different aspect which is closer to capturing the extent to which families are free from the stresses associated with money worries. Children with scores of one or less are more likely to live in households with more positive feelings about their income and, conversely, those scoring four to nine were twice as likely to live in households in the least positive group.

The final measure presented in Table 4.2 is an index of the children's home learning environment. It was originally developed to assess the association between children's activities at 10, 22 and 34 months and their cognitive development at 34 months (Melhuish, 2010). The index covers aspects such as: how often the children have been read to; done activities such as painting, singing rhymes, or playing educational games; and the number of books in the home. Higher scores on the index indicate children who have experienced a higher number of these items. The negative outcomes scale includes the two cognitive ability measures at 34 months that have been shown to be associated with the home learning index so it is not surprising that higher scores on it are also associated with low scores on the negative outcomes scale.

Table 4.2 Number of negative health outcomes by home/family resilience measures

Resilience measures

Number of negative health outcomes

One or less

Two

Three

Four or more

Total

%

%

%

%

%

Eats 2+ different fruits a day

88

86

80

75

85

Eats 2+ different vegetables a day

73

68

66

63

70

High physical activity level

26

28

24

19

25

Experience with children (prior to child's birth)

A lot

9

10

9

13

10

Quite a lot

15

14

14

13

14

Not very much

14

16

10

9

13

None at all

12

11

12

9

12

Already had children

51

49

56

55

52

Lived at same address since 10 months old (high stability)

68

64

63

61

66

No major life events since child 10 months old*

40

38

35

33

38

Feelings about income over 4 years (quartiles)

1st - Most positive

29

19

13

13

23

2nd

22

22

21

18

21

3rd

29

31

30

30

29

4th - least positive

20

28

36

39

26

Mean scores

Home learning environment (higher mean score = more enriched environment)

46.7

44.2

43.1

38.9

44.7

Standard error of mean

0.21

0.39

0.47

0.55

0.17

Bases

Weighted

2039

790

509

503

3841

Unweighted

2159

775

473

435

3842

Notes:
Bases vary for each measure, those shown are the lowest of the range.
*The events covered each year from 22 months were: new parent/partner; parent not resident full time; parent married; new baby; another child moving into or out of house; death of sibling, parent or grandparent; illness of parent or sibling. It is of course possible that other kinds of major life events will have happened to these families, but they have not been captured by this set of questions.

4.5.3 Neighbourhood resilience measures

Table 4.3 looks at the extent to which neighbourhoods can confer resilience. As the introduction to this chapter outlined, community resilience has featured prominently in the literature around resilience and health in recen t years. GUS includes a fairly large number of questions about services and parents' views of them, as well as more general items to measure satisfaction with neighbourhoods. The two measures in the table are composite scales based on two sets of questions asked when the children were 34 months old. The child friendliness scale was based on five questions originally developed for use in the 'Starting Well Demonstration Project' evaluation in Glasgow (Mackenzie et al., 2004) and was explored in full in a previous GUS report (Bradshaw et al., 2009). The questions covered aspects such as whether the area is a good place to bring up children, whether people in the local neighbourhood can be trusted with children.

As shown in the table, children who scored one or less on the negative outcomes scale were more likely than those with scores of four-nine to live in an area rated as being highly child friendly (21% versus 12%), and were less likely to live in areas with a low rating (15% versus 26%). This is likely to be explained in part by area deprivation which is related to both negative outcomes and perceptions of child friendliness. The next stage of the analysis in Section 4.6 addresses this.

The second scale measured parents' satisfaction with services and facilities in their local area. The aspects covered were local health and childcare services, educational establishments, and facilities for adults, teenagers and young children (see Bradshaw et al., 2009). Once again, children with scores of one or less on the negative outcomes scale were more likely than those with higher scores to live in areas rated highly in terms of their local services, and less likely to live in areas with low ratings.

Table 4.3 Number of negative health outcomes by neighbourhood resilience measures

Resilience measures

Number of negative health outcomes

One or less

Two

Three

Four or more

Total

%

%

%

%

%

Child friendliness of local area

High

21

15

13

12

18

Medium

65

62

62

62

63

Low

15

23

25

26

19

Satisfaction with local facilities

High

34

29

26

22

30

Medium

29

24

21

23

26

Low

37

47

53

56

44

Base (child friendliness)

Weighted

1884

721

483

460

3548

Unweighted

2029

714

454

400

3596

Base (satisfaction with facilities)

Weighted

1473

593

401

373

2840

Unweighted

1581

583

370

326

2860

Note:
Bases vary for each measure, those shown are the lowest of the range.

4.5.4 Social support networks

Finally, in Table 4.4, we look at measures of social support and networks. In many ways this table overlaps with the previous tables as it encompasses aspects related to parental behaviour and neighbourhood networks (in the form of attending parent and toddler groups) as well as support for families from grandparents, friends or other family members.

The first measure in Table 4.4 uses the question asked each year about attendance at parent/toddler groups to assess how many years mothers reported doing this. Although the differences between children with low and high scores on the negative outcome scale were not large, they were statistically significant.

The next measure used a similar approach and calculated the total number of years in which parents reported they would have difficulties finding someone at short notice to look after their child for a day. About half of mothers said this would not be difficult for them every year it was asked, but the proportion was higher among children with low scores on the negative outcomes scale and lower among those with scores of four to nine. The pattern was similar in relation to whether parents said at 10 months that they had family or friends with medical knowledge or training who they could call on for advice about their child's health. In contrast, a similar proportion across all groups had a high level of support from the child's grandparents at the age of 10 months.

Table 4.4 Number of negative health outcomes by social support resilience measures

Resilience measures

Number of negative health outcomes

One or less

Two

Three

Four or more

Total

Years parent has attended a parent/toddler group

Four

10

10

6

5

9

Three

21

18

17

12

18

Two

20

21

19

22

20

One

19

21

22

23

20

Never

31

31

36

39

32

Years parent has said it would be difficult to find help at short notice to look after child

Never difficult

55

49

47

38

50

One

17

17

19

19

17

Two

11

13

15

15

13

Three

9

11

12

16

11

Four

8

10

7

12

9

High level of grandparental support when child 10 months

22

24

24

24

23

Friend/family member with medical knowledge when child 10 months

45

41

42

35

43

Bases

Weighted

2039

789

509

504

3841

Unweighted

2159

775

473

435

3842

Note:
Bases vary for each measure, those shown are the lowest of the range.

Most of the figures presented in the above tables suggest an association between the avoidance of negative outcomes and individual, family, neighbourhood and support network related resilience measures. However, we know from Chapter 3 that children from the least disadvantaged socio-economic groups are the most likely to have few negative outcomes in their first four years. The association between low negative outcomes and parents having friends or family members with medical knowledge or training is likely to result from people from more advantaged groups being more likely to know people like this, rather than from any direct benefits that this access to medical knowledge might confer.

None of the patterns highlighted above should be considered in any way to indicate possible causal links between resilience and outcomes. So, while the above analysis has been useful in setting the scene for the next stage of the analysis, and in illustrating the extent of certain resilience measures in the population, it has little to offer by way of useful recommendations for policy to help children from disadvantaged backgrounds avoid negative outcomes. The next section turns its focus to some analysis that might prove more useful in this respect.

4.6 What factors appear to protect disadvantaged children from negative outcomes?

4.6.1 Analysis method

This stage of the analysis focused only on children from more disadvantaged backgrounds. To recap, to identify resilience it is necessary to show what factors are associated with avoiding negative outcomes among children who are at an increased risk of them. As set out in Section 4.4, this was done by restricting this final stage of the analysis to children who live in either the most deprived areas, in the lowest income households, or in semi-routine and routine households - and do not live the least deprived area deprivation quintile, the highest household income quintile or in professional and managerial households.

The resilience measures set out in the preceding tables were explored using logistic regression. This technique assessed the extent to which each of the resilience factors had an independent association with avoiding negative outcomes (having a score of one or less on the negative outcomes scale), when all other factors were taken into account. This whole approach helped to overcome the problem discussed above of how to disentangle the association between resilience and socio-economic background.

There are many ways to approach analysis such as this and certainly there are more sophisticated statistical techniques that could be applied to control for the association between affluence and the resilience measures outlined above. However, this approach was chosen because it was thought to be clear and easily interpretable. 10 This analysis hopefully highlights the fact that this is an area worthy of more detailed exploration, and illustrates the increasing analytic potential that GUS is offering as the study progresses.

4.6.2 Regression results

Table A1 in the Appendix presents the full results of the logistic regression. Table 4.5 highlights the key statistically significant findings. In addition to the resilience measures explored above in Section 4.5, some other factors were included in the model, either because they are known to be critical in terms of explaining outcomes (such as sex), or because they capture additional important aspects of resilience that have specific resonance among more disadvantaged families (such as adult employment status). Although the largest differences between the most and least disadvantaged children will have been accounted for by removing those from the least deprived areas, the highest income households, and professional and managerial households, the remaining categories for these three measures were still included in the analysis.

The number of factors explored in the regression was relatively high for this type of analysis. One of the dangers associated with this kind of approach is that using a standard threshold of 5% for statistical significance will result in one in twenty findings being significant by chance. This therefore needs to be borne in mind when interpreting the results. One option is to raise the threshold to 1% so the risk of chance findings reduces. However, with this analysis it is also possible that real differences in the population will not be detected as significant because the sample size in the disadvantaged sub-population of around 1,000 cases is too small. Further restricting the interpretation by setting a stricter significance level therefore increases the risk of missing genuine results. The key point is that the findings of all types of analysis should be interpreted with reference to the prior hypotheses that led to the analysis being conducted in the first place, and with regard to the existing evidence in the field. It should not be considered in isolation nor treated as definitive.

Odds ratios, and their confidence intervals, are a useful indication of the size of the effect estimated in this kind of analysis. 11 While statistical significance is an important indicator of whether a finding has relevance in the wider population, effect sizes are arguably more important when it comes to determining the policy significance of findings. A factor that is highly statistically significant but has a small effect might not warrant much action, especially if the costs associated with it would be hard to justify in terms of its likely overall impact.

The factors that were found to have an independent association with avoiding negative outcomes (after taking into account all the measures shown in the table in the appendix) are set out below. The analysis compared the odds of the groups listed in the left hand column of avoiding negative outcomes relative to the odds of the comparison group shown in brackets. The second column presents the odds ratio associated with each factor and the third shows the range of values for the odds ratio that we can be 95% confident includes the true population value for that factor. In all cases the odds ratio is higher than one which shows that the odds of avoiding negative outcomes increased relative to the comparison group. 12

The wider the confidence interval, the less precise the estimate, for example, the confidence interval for maternal age suggests that compared with children whose mothers were aged 15-24 when they were born, the odds of children whose mothers were aged 35 and over avoiding negative outcomes ranged between 1.21 and 3.14. This means that the effect of having a mother aged 35 or over could be as small as a 21% increase in odds, or it could be as large as a 214% increase. This wide interval will be a result of the relatively small sample size for older mothers and all that can be concluded is that there is a positive association but its magnitude cannot be precisely estimated.

Table 4.5 Factors significantly associated with avoiding negative outcomes among more disadvantaged children

Odds ratio

95% confidence interval

Child factors

**Being a girl (being a boy)

1.49

1.12-1.97

Maternal factors

**More positive attitudes to seeking help (scale)

1.10

1.02-1.19

*Mother did not have any long-term health conditions or disabilities in child's first four years (one or more conditions)

1.51

1.08-2.12

*Age at child's birth 35+ (age 15-24)

1.95

1.21-3.14

Household factors

**A more highly enriched home learning environment (scale)

1.03

1.02-1.04

*At least one adult in the household in full-time employment (no adult in full-time employment)

2.03

1.19-3.45

Neighbourhood factors

*Medium level of satisfaction with the facilities in the local area (low level)

1.59

1.15-2.21

*Overall measure is significant at 5% level, **Overall measure is significant at 1% level or below.

The first point to note is that actually very few of the resilience measures explored in Tables 4.1-4.4 showed a significant association. This in part confirms the suspicions noted above that the relationships between outcomes and resilience were a product of their underlying socio-economic distribution.

However, it is also true that a number of the significant findings are perhaps surprising. For example, maternal age at birth is highly socially patterned so to find a significant association even when the most advantaged children have been removed from the analysis suggests that the current policy focus on younger mothers is well placed. In addition to this, the Scottish Government's sexual health strategy includes policies to widen young women's contraceptive choices with the aim of raising the age of women's first conception.

As the scale included factors such as below average cognitive development, language and behavioural problems, all of which are much more common in boys, it is unsurprising that that this analysis found a difference between boys and girls. However, it should be borne in mind that the scale might not have been very good at detecting the kinds of negative outcomes that girls experience, so this finding should not be used to conclude that only boys have additional support needs in the early years.

Melhuish (2010) concluded that the particularly strong association between the home learning environment index and the verbal aspect of the cognitive development assessment was related to the fact that the measure includes a high number of language-related activities, and the fact that language development changes markedly around the time that the assessments were carried out. The negative health outcomes scale included two negative language development measures (lower than average verbal ability at 34 months and parental reports of problems with language development at 46 months) so the association between the home learning index and avoiding negative outcomes found here is also likely to be a result of this. However, as this is a particularly critical stage of life for language and wider cognitive development it is appropriate that the outcomes being captured reflect this. These findings add to the evidence that appears to be mounting in favour of these kinds of activities being of intrinsic value in relation to children's outcomes. Further analysis of this index in relation to other outcomes captured in GUS might prove useful.

The finding that higher satisfaction levels with services were associated with avoiding negative outcomes needs some careful reflection. 13 It is possible that parents of children who experience multiple negative outcomes have greater need for the kinds of services covered in this measure and that their lower satisfaction levels reflect a greater awareness of local service provision relative to those with children with better outcomes. Although it could reflect something about the kinds of services available within communities that is independent of the area characteristics measured by the deprivation index.

Chapter 3 described the fact that children from disadvantaged backgrounds face a double burden of health inequality in terms of their own increased risks of negative outcomes as well as those of their immediate family. The fact that maternal long-term health was associated with avoiding negative outcomes reinforces this message. Action to prevent children experiencing negative outcomes in their early years therefore seems to require attention to their main carer's health as well as to their own.

The fact that the factors associated with resilience range from ones specific to the child through to aspects of the local neighbourhood suggest that it was correct to operationalise resilience as something with many levels. However, the significant factors are quite different in nature which could also suggest that resilience was either too broadly specified in the analysis, or is too wide ranging to be considered as a single concept. Either way, there are likely to be very different policy responses required to promote factors such as having at least one adult in a household in full-time work, having a positive attitude towards seeking help, or living in an area where services are highly rated.

One very obvious implication does stand out - many of the factors that appear to be associated with avoiding negative health outcomes fall outside the traditional remit of the health service. This certainly chimes with the cross-portfolio approach to setting and delivering outcomes set out in Equally Well, and in the Early Years Framework. It therefore reinforces the message that attempts to reduce health inequality and to promote the best start in life will only succeed if they are acknowledged as having policy implications across the board and not just within one or two limited domains.

Back to top