Growing up in Scotland: health inequalities in the early years

CHAPTER 3 THE EXTENT AND CHARACTER OF HEALTH INEQUALITIES IN THE EARLY YEARS

3.1 Key findings about health inequalities in the first four years

• This analysis spanned the period from around the time of the children's birth to just before their fourth birthdays. A wide range of measures were used to illustrate inequalities in outcomes such as the children's birthweight, their experience of long-term health problems, accidents, poor psychosocial health and wider developmental problems. It also looked at a range of risk factors for poor health which included maternal smoking, maternal health, children's physical activity levels and their diet (including breastfeeding). All these outcomes and risk factors were explored in relation to area deprivation, household income, and household socio-economic classification.
• It showed that exposure to the kinds of risks that can impact on health and development in the early years, and have been shown in the wider literature to have implications for later life, are not uniformly or randomly distributed across the population at this very early point in life. Significant inequalities exist with those in the most deprived areas, the lowest income households or routine and semi-routine households found to have worse health outcomes, and higher exposures to risks for poor outcomes, than their more advantaged counterparts.
• Although overall levels of outcomes such as long-term health conditions and poor general health are relatively low in the early years, and appear not to change much each year, this analysis shows that there is in fact quite a high degree of individual-level change in health outcomes in this period. However, this would not necessarily be evident in an analysis that compared a different group of children over time without being able to explore individual pathways in the way that GUS permits.
• While the persistence of poor outcomes was quite variable, exposure to risks such as smoking and poor maternal health were somewhat more stable. For example, of those children whose mothers smoked at some point in their early years, most were exposed to this on a prolonged rather than temporary basis.
• Across all the outcomes and risk factors explored, inequalities in exposure to risk factors were generally larger than those evident for outcomes. However, within the outcomes explored, behavioural, psychosocial and linguistic problems showed much starker inequalities than physical ones such as poor general health.
• The more disadvantaged households can be said to face a double burden in their experience of health inequalities as both the children and adults within them are at greater risk of negative outcomes.

3.2 Introduction

This chapter maps out socio-demographic inequalities in the various health outcomes and risk factors for poor health outlined in the previous chapter over five stages of life. Although the first GUS interview took place when the children were 10 months old, retrospective information was collected during that first interview about the pregnancy and birth which has been utilised here. The five stages explored are:

• Pregnancy, birth and the first three months,
• 10 months,
• 22 months,
• 34 months, and
• 46 months.

It starts by looking at the measures relating to the pregnancy, birth and first three months. It then looks at three physical health outcomes captured across all four years between the interviews conducted at 10 and 46 months, as well as some specific behavioural and developmental outcomes at 46 months. It then looks at a number of risk factors for poor health measured at various points.

3.3 Pregnancy, birth and the first three months

The measures presented in this section focus on both the mother and the child. The remaining sections focus more heavily on factors associated with the child, with a smaller number of measures of maternal or parental factors likely to pose adverse risk for the child (such as smoking).

3.3.1 Risk factors and health outcomes in the early stages

Table 3.1 shows the total population prevalence for each of the factors of interest. The most common negative factor was three in ten (30%) mothers said that they had planned to bottle feed before their child was born. One in four (25%) mothers had smoked during their pregnancy and about one in eight (13%) described their health as not very, or not all good, during that time.

The best measure we have of the earliest point in the children's lives is whether their conception was planned or accidental. Around one in four (24%) pregnancies had not been planned at all and a further 17% were described as having not been planned but not actively prevented either. Fetal health can be linked to factors prior to conception as well as at the early stages of development. For example, folic acid supplementation is recommended prior to conception and for the first 12 weeks of pregnancy to help reduce the risk of neural tube defects. ⁶ Unplanned pregnancies are therefore less likely to involve these kinds of pre-conception steps that can promote fetal health.

Table 3.1 Risk factors for negative child outcomes, during pregnancy

Risk factor	%
Pregnancy planning
Planned by mother and father	58
Planned by mother but not father	1
Not planned but nothing done to prevent it	17
Not planned at all	24
Maternal smoking
Smoked frequently or occasionally	25
Did not smoke	75
Maternal health in pregnancy
Very/fairly good	87
Not very/not at all good	13
Breastfeeding plans before birth
Planned to breastfeed	64
Planned to bottle feed	30
No strong preference	6
Bases
Weighted	5109
Unweighted	5108

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

Table 3.2 looks at some of the circumstances of the children's births. Medical literature tends to make a distinction between children born at full term with a low birthweight, and the weights of premature babies or multiple births, as the reasons for their low weights can be rather different. In common with the approach taken by the Scottish Government long-term monitoring of inequality and low birthweight, the measure used here excludes multiple births (twins etc) but does not take account of pre-maturity which is itself linked to deprivation in singleton pregnancies (Scottish Government, 2009b).

Just 6% of singleton babies were low weight, while twice as many of all babies, 12%, spent time in an incubator or neonatal intensive care unit after birth. In common with Table 3.1, there was a much higher prevalence of exposure to a negative risk factor, never being breastfed (40%), than of a direct poor health outcome, such as low birthweight.

Table 3.2 Child health outcomes and risk factors for negative outcomes, at birth

Health outcome/risk factor	%
Birth weight (singleton births)
Low	6
Not low	94
Bases
Weighted	5074
Unweighted	5117
Time in SCBU/ NNU after birth
Yes	12
No	88
Was child ever breastfed
Yes	60
No	40
Bases
Weighted	5216
Unweighted	5216

Moving away from the very early stages, Table 3.3 looks at outcomes in terms of parents' reports of problems in the first three months. The most commonly reported problem was with sleeping (38% said this was a big or a bit of a problem), followed jointly by feeding or health problems other than allergies (17%), while allergies were reported by 7% of parents.

Table 3.3 Problems in first three months

	Big or bit of a problem	Not a problem
	%	%
Extent of problems with:
Sleeping	38	62
Feeding	17	83
Allergies	7	93
Other health problems	17	83
Bases
Weighted: 5205
Unweighted: 5205

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

The key point to note from these first three tables is that direct health outcome measures such as low birthweight, time in an incubator after birth, or early health problems are relatively less common than exposures to risk factors for poor health such as smoking in pregnancy or bottle feeding.

3.3.2 Inequalities in the early stages

Table 3.4 presents the above figures about the pregnancy, birth and first three months by quintiles of the Scottish Index of Multiple Deprivation (this splits all areas of Scotland into five roughly equally sized groups from the least to most deprived fifth of areas). Three key points stand out from this table. Firstly, with the exception of the problems reported at three months, all the factors display some social patterning indicative of health inequality, and in many cases these patterns show a steady linear trend in line with increasing deprivation (all associations presented in the table are statistically significant apart from the reported problems with allergies or other health problems).

The relative risk figures (see section 2.1.2 for explanation) presented in the penultimate column show the magnitude of the difference between the most and least disadvantaged areas. The biggest relative risks were for smoking during pregnancy (9% versus 43%, respectively) and for pregnancies that were not planned at all (8% versus 39%). The largest absolute differences in risk (that is, the size of the gap between the most and least deprived areas) were in relation to the two feeding measures (planning to bottle feed and never breastfeeding). The smallest inequalities were in relation to post-natal incubation (11% versus 15%) and low birthweight (5% versus 8%). The latest official statistics on low singleton birthweight show that the relative risk of low birthweight between the most and least deprived 10% of areas was 2.2 (Scottish Government, 2009b), therefore the GUS figures (which compare the most and least 20% of areas) are broadly in line with this.

The higher likelihood of reporting problems with sleep and, to an extent, feeding, among families in less deprived areas is interesting. The finding about feeding might be a result of problems with breastfeeding which is more common among less deprived areas, hence the reverse gradient. However, as this report is focusing on inequalities that confer a disadvantage for less affluent children extensive exploration of these patterns is outwith our scope.

Table 3.4 Health outcomes and risk factors in pregnancy, at birth and in first three months by Scottish Index of Multiple Deprivation quintile

	SIMD quintile
	5th - least deprived	4th	3rd	2nd	1st - most deprived	Relative risk	Risk difference
	%	%	%	%	%
Risk factors
Pregnancy not planned at all	8	17	21	29	39	4.8	31
Smoked in pregnancy	9	17	21	31	43	4.9	34
Health in pregnancy not very/not at all good	9	12	13	14	17	1.8	8
Planned to bottle feed	15	21	27	36	46	3.1	31
Never breastfed	21	29	35	47	60	2.8	38
Health outcomes
Low birth weight (singletons)	5	4	5	6	8	1.6	3
SCBU/ NNU after birth	11	11	11	12	15	1.4	4
Problems with:
Sleep	45	41	37	34	34	0.8	-11
Feeding	19	18	15	19	15	0.8	-4
Allergies	7	6	8	8	8	1.1	0
Other health problems	19	15	16	19	19	1.0	0
Bases
Weighted	919	989	995	945	1261
Unweighted	996	1042	1025	907	1138

Notes:
The relative risk is the prevalence in the most deprived areas divided by the prevalence in the least deprived areas, the risk difference is the difference in the prevalence in the two areas. These two measures were calculated using the raw data whereas the percentages in each column have been rounded to the nearest whole number; the relative risks and risk differences cannot therefore be calculated from the percentages presented.
Bases vary for each measure, those shown are the lowest of the range.

Table 3.5 looks at the factors shown above that had unequal distributions in favour of the least deprived in relation to household income only and reveals some interesting patterns. For example, the relative risk of both unplanned pregnancy and smoking in pregnancy is higher here than found in relation to area deprivation. This is accounted for by both lower levels of these factors among those in the highest income households than in the least deprived areas, as well as higher rates in the lowest income households compared with the most deprived quintiles.

The birthweight and incubator measures show smaller relative risks than the other measures, as seen above with area deprivation. However, it should be noted that relative risk is a measure subject to imprecision, in the same way that any percentage reported from a survey is, and will therefore have an associated margin of error. Therefore, tests of the significance of any differences between relative risks for the same measures across different socio-demographic measures would need to be conducted before definitive conclusions could be drawn.

Table 3.5 Health outcomes and risk factors in pregnancy, at birth and in first three months by equivalised household income quintile

	Equivalised household income quintile
	1st - highest	2nd	3rd	4th	5th - lowest	Relative risk	Risk difference
	%	%	%	%	%
Risk factors
Pregnancy not planned at all	6	12	19	28	49	7.5	42
Smoked in pregnancy	8	13	18	33	51	6.3	43
Health in pregnancy not very/not at all good	9	11	13	17	16	1.7	7
Planned to bottle feed	13	18	29	38	50	4.0	38
Never breastfed	20	27	40	48	64	3.2	44
Health outcomes
Low birthweight	6	5	6	8	9	1.5	3
SCBU/ NNU after birth	10	10	11	14	14	1.4	4
Bases
Weighted	851	967	834	950	979
Unweighted	903	999	845	935	910

Notes:
The relative risk is the prevalence in the lowest income quintiles divided by the prevalence in the highest income quintiles, the risk difference is the difference in the prevalence in the two income categories. These two measures were calculated using the raw data whereas the percentages in each column have been rounded to the nearest whole number; the relative risks and risk differences cannot therefore be calculated from the percentages presented.
Bases vary for each measure, those shown are the lowest of the range

As noted in Chapter 2, analysis was also carried out looking at household socio-economic classification. For all but one of the measures explored so far - maternal health - the difference between households headed by someone in a professional or managerial job and those headed by someone in a semi-routine or routine job showed a similar pattern to those presented above for area deprivation and income.

3.4 Health measures in the first four years of life

The previous section looked at early outcomes and risk factors for poor later outcomes in combination. As the number of measures that can be explored from 10 months onwards is much larger, a sharper distinction is made from this point onwards between children's direct health outcomes (physical, psychosocial and developmental/cognitive) and risk factors for poor outcomes.

The following sections therefore start by outlining the prevalence of a number of health outcomes and exposures to risks likely to have an adverse impact on health at 10, 22, 34 and 46 months. This should help to introduce readers to each measure as well as provide a baseline point of comparison for the detailed tables that follow looking at differences by socio-demographic group. For some measures the figures for specific years are explored, whereas for others the full potential of the longitudinal design of the study has been tapped and the measures focus on multiple outcomes across the years rather than just at specific time points.

3.5 Overview of health outcomes

3.5.1 Physical health

This section starts by presenting the overall figures for the main health outcome measures available at each sweep of data collection, as well as some composite measures that look at children's experiences spanning the period between 10 and 46 months.

The number of outcome measures increased notably at later sweeps with the introduction of standardised assessments of cognitive ability (at 34 months), child psychosocial health (at 46 months), and direct measurements of height and weight at 46 months.

Table 3.6 shows that the prevalence of long-term conditions was highest at 46 months (17%) and ranged between 11% and 14% in the three earlier years with no obvious pattern. Almost one in three (28%) children were described by their parents as having a long-term condition at some point in their first four years of life. However, as Figure 3.1 illustrates, these categories appear to have been very fluid as very few children were described as having a long-term condition on more than one occasion between the ages of 10 and 46 months, and just 4% were persistently in this category.

The proportion of parents or carers describing their child's health as fair, bad or very bad followed a similar pattern. The overall prevalence was low in most years, at around 6-7%, but a higher proportion - 18% of children - were described as having fair or worse health at some point in this period. Persistent fair or worse health was very uncommon; just 3% of children were described in these terms in at least three of the four years.

A high degree of movement into and out of poor states of health is not surprising as many childhood illnesses and conditions can be serious in the short to medium term but do not go on to become enduring problems. For example, some skin, respiratory or allergic conditions clear up within a couple of years, while many serious conditions present at birth or in the first year of life will have been treated successfully by the age of 3 or 4.

Figure 3.1 Persistent long-term health problems and fair/bad/very bad health

Accidents requiring medical attention were relatively uncommon at 10 months, but doubled in the following year (from 10% to 23%), probably due to increased independent (but faltering) mobility as the children began to walk. They show a slight decline after that to just under one in five children requiring medical attention for an accident in year preceding the interviews at 34 and 46 months. Overall, a fifth (20%) of children experienced two or more accidents requiring attention between their birth and 46 months.

It is also worth noting that the accident measure only includes incidents for which the child was taken to a doctor, dentist, health centre or hospital. This is designed to help parents differentiate between more and less serious accidents. However, it means that the accidents captured by this measure might vary depending on the parents' propensity to seek help or feel the need for their child to be checked. This could be related to parents' confidence, experience and attitude to risk, as well as on the accessibility of local health services. This is not to diminish the usefulness of the measure, but is simply highlighted to clarify what was measured so that can be taken into consideration when the results are presented.

Table 3.6 Health outcomes at 10, 22, 34 and 46 months

Health outcome	10 months	22 months	34 months	46 months
	%	%	%	%
Long-term health problems or disabilities
Yes	13	11	14	17
No	87	89	86	83
Long-term health problems at least once between 10 and 46 months	-	-	-	28
General health (parent assessed)
Very good/good	94	93	94	93
Fair/bad/very bad	6	7	6	7
Fair/bad/very bad health at least once between 10 and 46 months	-	-	-	18
Accidents requiring medical attention in past year
0	90	77	81	82
1 or more	10	23	19	18
2 or more accidents between 10 and 46 months	-	-	-	20
Bases
Weighted	5217	4512	4193	3994
Unweighted	5217	4512	4193	3994

Note:
Bases vary for each measure, those shown are for the whole sample

3.5.2 Problems reported by parents

Table 3.7 looks at some of the kinds of problems parents reported over the course of the first four years of their child's life. The nature of the problems asked about changed over this period to reflect the fact that different developmental stages often pose different challenges.

The questions about allergies and other health problems showed some fluctuations over the stages but with no notable patterns. In contrast, problems with sleeping became less common by 34 months while feeding became more of an issue as time progressed and was in fact the most commonly reported of the problems by the age of 46 months (41% said this was a bit of or a big problem). Two behavioural measures were introduced at 34 months relating to interactions with other children and general behaviour. At 46 months twice as many parents reported general behavioural problems than said their child's behaviour to other children was a problem (31% versus 16%).

Table 3.7 Problems reported by parents in the last three months, at 0-3, 10, 22, 34 and 46 months

Big or bit of a problem:	0-3 months	10 months	22 months	34 months	46 months
		%	%	%	%
Sleeping	38	33	n/a	29	n/a
Feeding	17	14	n/a	35	41
Allergies	7	10	n/a	11	12
Other health problems	17	16	n/a	14	14
Behaviour to other children	n/a	n/a	n/a	19	16
General behaviour	n/a	n/a	n/a	26	31
Bases
Weighted	5217	5217	4512	4193	3994
Unweighted	5217	5217	4512	4193	3994

Note:
Bases vary for each measure, those shown are for the whole sample
n/a= not asked or not asked in a directly comparable way

The direct cognitive assessments carried out at age 34 months measured language development and problem solving ability. These were reported in detail in Bromley (2009) which showed that below average attainment was significantly associated with socio-economic disadvantage. In addition to these assessments, parents were asked if they had any concerns about their child's language development at 46 months. ⁷ Around a fifth (19%) of children were judged to have at least one problem with their language development. The correspondence between these two separate measurements was high; children who scored below average on the vocabulary assessment at 34 months were more likely than those with average scores or above to be described as having language problems by their parents a year later. The analysis below in Section 3.6 focuses on the question about language concerns at 46 months.

3.5.3 Psychosocial health

Some notable additions to the questionnaire were made when the children were 46 months old. Firstly, there was a standardised assessment of the children's social, emotional and psychological development using the Strengths and Difficulties Questionnaire (Goodman, 1997). Parents answered 25 questions about a range of aspects of their child's behaviour from which normal, borderline and abnormal scores for various domains, as well as an aggregate assessment, can be derived. The detailed analysis of this presented in section 3.6 focuses on children with borderline and abnormal scores for the total difficulties measure and for the conduct and hyperactivity domains.

By way of introduction, Table 3.8 presents the overall figures for all children and shows that children were more likely to be assessed as having problems in relation to conduct (14%) and hyperactivity (12%) and that abnormal scores for the total difficulties measure were much less common at just 5%. Boys were more likely than girls to have borderline or abnormal scores in relation to total difficulties, conduct, hyperactivity, and pro-social behaviour, whereas differences were less pronounced for emotional symptoms and peer problems.

Table 3.8 Strengths and difficulties at age 46 months, by sex

	Boys	Girls	All
	%	%	%
Emotional symptoms
- Normal	92	93	92
- Borderline	5	4	4
- Abnormal	3	3	3
Conduct problems
- Normal	65	72	68
- Borderline	19	15	17
- Abnormal	16	13	14
Hyperactivity
- Normal	76	85	80
- Borderline	9	6	8
- Abnormal	14	9	12
Peer problems
- Normal	82	85	84
- Borderline	9	8	9
- Abnormal	8	7	8
Pro-social behaviour
- Normal	86	93	89
- Borderline	10	4	7
- Abnormal	5	3	4
Total difficulties score
- Normal	86	91	88
- Borderline	8	6	7
- Abnormal	7	4	5
Bases
Weighted	2026	1909	3935
Unweighted	2015	1926	3941

Note: Bases vary for each measure, those shown are for the total difficulties score (the lowest of the range)

3.5.4 Body mass index

Obesity is a growing problem in most of the developed world and much of the developing world too (Foresight, 2008). Childhood obesity is a particular concern and is the subject of specific policy interventions in Scotland, including a national outcome to reduce the rate of increase in unhealthy weight among children between 2008 and 2011 (Scottish Government, 2007) and a new guideline for clinical practice ( SIGN, 2010).

Direct measures of the children's height and weight were taken at 46 months. These were used to derive their body mass index ( BMI) which was compared with standard growth charts for children of this age to assess whether they were underweight, normal weight, overweight or obese. The most recent Scottish Health Survey report contains full details of the methodology used in Scotland to calculate children's BMI, which GUS also follows (Gray and Leyland, 2009). Although BMI is not a perfect measure of body fat, because it cannot take account of skeletal or muscle density, it is nevertheless a good enough approximation to be useful in studies such as GUS.

The Scottish Government's preferred indicator of BMI in children is the proportion outwith the healthy weight range (underweight, overweight and obese combined); 28% of the children were in this category at age 46 months (10% were obese, 16% overweight and 2% underweight).

3.6 Inequalities in health outcomes

3.6.1 Area deprivation

Table 3.9 compares the proportion of children in each deprivation category whose parents described them as having: a long-term health problem or disability at least once between their birth and 46 months; fair, bad or very bad health on at least one occasion; and children who have had more than 2 accidents requiring medical attention since their birth.

One in ten (11%) children in the least deprived areas were described as having poor health at least once since their birth compared with one in four (24%) in the most deprived areas. The prevalence of this appeared to increase in a linear fashion across the groups as deprivation increased. Both the relative risk and the absolute difference between the least and most deprived groups were similar for long-term health problems and accident rates. However, the overall patterns were somewhat different. The risk of having a long-term health problem at least once since birth increased in line with increasing deprivation, from 24% of children in the least deprived areas to 33% in the most. In contrast, the risk of having had two or more accidents since birth was largely similar across the first four groups (ranging between 17% and 20% with no obvious pattern) but was higher among children in the most deprived areas at 26% (the difference between children in the most and least deprived areas was statistically significant).

This suggests that reducing inequalities in the overall rate of poor health and long-term conditions in children in the early years might require action targeted across the whole population. In contrast, accident rates might benefit from a more targeted approach focusing on children in the most deprived areas. Further analysis of the nature of the accidents might assist this. The overview of health outcomes presented in section 3.5.1 showed that accidents were at their most common between the ages of one and two. Further analysis of the association between deprivation and accidents in each individual year of life suggests that this is also the point at which inequalities in accident rates are at their most pronounced (18% of children in the least deprived areas had an accident between the age of 10 and 22 months compared with 28% in the most deprived areas). This might therefore be the age at which interventions to reduce accident rates could have most impact.

Turning now to focus on some of the health and developmental problems asked about at 46 months, Table 3.9 shows some small, but statistically significant, differences between parental reports of problems with allergies, asthma and other health problems by area deprivation. Further analysis of this found that problems with allergies and asthma were also significantly associated with deprivation at 34 months, but not at 10 months. The overall difference between groups was quite small at 46 months (10% in the least deprived areas versus 14% in the most) so this emergence of a pattern over time might simply be caused by the findings at 10 months being anomalous, or it could possibly reflect a real change in the burden of these kinds of conditions among more deprived children over time. Similarly, parental reports of problems associated with other health issues were associated with deprivation at 46 months but not at any of the earlier stages (12% in the least deprived areas mentioned this at 46 months versus 16% in the most). These two patterns might be an interesting area for further investigation.

Table 3.9 shows that parents in the most deprived areas were twice as likely as those in the least deprived areas to have concerns about their child's language, 26% compared with 12%. The increase from the least to most deprived areas appeared to follow a fairly linear pattern.

The BMI measure introduced at 46 months was not significantly associated with area deprivation. This was true for both the proportion of children whose weight was outwith the healthy range, as well as for the subset of children classified as obese (both measures were also explored separately for boys and girls and the same lack of association was found). Previous analyses of children's BMI and area deprivation in Scotland was inconclusive (Hirani and Stamatakis, 2005). On the whole, few associations were significant and those that were did not follow any obvious pattern. However, analysis of these trends in England, where the sample size for the analysis is much larger, has tended to show higher levels of obesity among children from more deprived areas, in semi-routine and routine households and in low income households (Scholes and Heeks, 2008; Jotanga, et al., 2005). However, it is worth noting that these analyses were based on a much wider age range of children whereas the GUS sample children are all the same age. It is therefore possible that an association between socio-demographic factors and unhealthy weight emerges when children are older.

As outlined above, the range of available measures of behavioural problems was much greater at 46 months; all showed very similar patterns of increasing reports of problems in line with increasing deprivation. For example, the proportion of children described by a parent as having problematic behaviour towards other children rose from 10% to 24%, while reported general behavioural problems rose from 28% to 37%, between the least and most deprived areas.

These patterns were supported by the standardised assessment of psychosocial health, which includes behaviour, conducted using the Strengths and Difficulties Questionnaire ( SDQ). Figure 3.2 presents the abnormal and borderline scores across deprivation categories for each of the SDQ's domains, as well as the total difficulties score (derived from all the domains apart from pro-social behaviour). This clearly demonstrates a strong association between deprivation and poor psychosocial health at this very young age; the proportion of children with borderline or abnormal scores increased in line with increasing deprivation. The difference between children in the least and most deprived areas was the most extreme in relation to conduct problems (23% versus 41% had borderline or abnormal scores for this), hyperactivity (13% versus 27%), and total difficulties (7% versus 20%). Table 3.9 also presents the proportions for these three particular domains.

The SDQ conduct domain covers behaviours such as lying, cheating, stealing, fighting, having tantrums as well as general obedience. The hyperactivity domain captures aspects such as restlessness, fidgeting, poor concentration, compulsiveness and low attention span. All of these kinds of difficulties have significant consequences for children's well-being, their relationships with carers and other children, and their ability to settle into the formal school environment. The fact that 31% of children were assessed as having conduct problems before their fourth birthday is somewhat worrying. More concerning is the fact that the prevalence of this almost doubles between the least and most deprived areas. For many children these kinds of problems will not be severe enough to warrant intervention, but those who would benefit are spatially concentrated in more deprived areas which has obvious resource implications for service providers.

Table 3.9 Health outcomes by Scottish Index of Multiple Deprivation quintile

Health outcomes	SIMD quintile
	5th - least deprived	4th	3rd	2nd	1st - most deprived	Relative risk	Risk difference
	%	%	%	%	%
Longitudinal measures:
Long-term health problems at least once between 10 and 46 months	24	25	27	30	33	1.4	9
Fair/bad/very bad health at least once since birth	11	14	16	22	24	2.2	13
2 or more accidents since birth	17	18	20	18	26	1.5	9
Measures at 46 months:
A bit of a/big problem:
Allergies/asthma	10	12	10	13	14	1.3	3
Other health problems	12	11	14	16	16	1.3	3
Behaviour to other children	10	13	15	16	24	2.3	14
Behaviour in general	28	26	33	31	37	1.4	10
Any language development concerns	12	13	18	23	26	2.2	14
BMI outside healthy range (ns)	26	28	28	30	29	1.1	3
Borderline/abnormal SDQ scores:
Total difficulties	7	6	11	13	20	2.7	13
Conduct	23	25	32	34	41	1.8	18
Hyperactivity	13	14	19	24	27	2.1	14
Bases
Weighted	746	774	761	731	923
Unweighted	867	855	821	663	735

Notes:
The relative risk is the prevalence in the most deprived areas divided by the prevalence in the least deprived areas, the risk difference is the difference in the prevalence in the two areas. These two measures were calculated using the raw data whereas the percentages in each column have been rounded to the nearest whole number; the relative risks and risk differences cannot therefore be calculated from the percentages presented.
Bases vary for each measure, those shown are the lowest of the range.

Figure 3.2 Strengths and Difficulties Questionnaire individual domain scores, by SIMD quintile (46 months)

3.6.2 Household income

Table 3.10 presents the patterns by household income for a selection of the outcome measures, suggesting a very strong link between outcomes and income. The patterns for household income are broadly similar to those for deprivation with the relative risks greatest for language development and behavioural problems, and poor general health since birth.

The relative risks associated with the SDQ scores appear even starker for household income than with area deprivation, but it should be noted that the overall proportions in each income quintile were actually very similar to those in the deprivation quintiles. Very small differences in the underlying prevalence figures can result in disproportionately bigger differences in the relative risks.

Table 3.10 Health outcomes by equivalised household income quintile

Health outcomes	Equivalised household income quintile
	1st - highest	2nd	3rd	4th	5th - lowest	Relative risk	Risk difference
	%	%	%	%	%
Longitudinal measures:
Long-term health problems at least once between 10 and 46 months	26	24	27	28	34	1.3	8
Fair/bad/very bad health at least once since birth	12	12	15	21	26	2.2	14
2 or more accidents since birth	16	19	18	22	24	1.5	8
Measures at 46 months:
Any language development concerns	10	10	16	23	28	2.8	18
Borderline/abnormal SDQ scores:
Total difficulties	4	7	8	14	21	4.8	17
Conduct	23	25	26	35	44	1.9	21
Hyperactivity	12	15	19	20	28	2.3	16
Bases
Weighted	625	684	732	748	946
Unweighted	726	763	774	719	765

Notes:
The relative risk is the prevalence in the lowest income quintile divided by the prevalence in the highest income quintile, the risk difference is the difference in the prevalence in the two income categories. These two measures were calculated using the raw data whereas the percentages in each column have been rounded to the nearest whole number; the relative risks and risk differences cannot therefore be calculated from the percentages presented.
Bases vary for each measure, those shown are the lowest of the range.

3.6.3 Socio-economic classification ( NS- SEC)

The analysis of the same outcomes by NS- SEC showed that in all cases the prevalence of poor outcomes was largely similar for children in lower supervisory and technical households and semi-routine and routine households. In most cases children in managerial and professional households stood out as the least likely of all the NS- SEC groups to be at risk of poor outcomes.

3.6.4 Conclusion

The Scottish Government and COSLA's Early Years Framework, and Equally Well, consider health inequalities in the early years to be important explanations for differences in children's readiness to learn and adapt to the formal school environment. Policy interventions to try and narrow the gap between more and less advantaged children in terms of their health, wellbeing and wider development before they enter the education system need to be alert to the extent to which some children are behind their peers at this key stage. This analysis has attempted to reflect some of the problems children in Scotland face before they have reached the age of four.

The prevalence at such a young age of the kinds of psychosocial and language development problems outlined in the preceding sections illustrates the kinds of challenges schools and parents face at that crucial transition stage. A sizeable minority of children in the least advantaged social groups have experienced poor health, or a long-term condition beyond the usual array of acute illness children commonly experience in the early years, or multiple accidents requiring medical attention before they are four. More notably, problems with language development and with behaviour are clearly evident. The fact that these kinds of negative outcome are very unequally distributed among children, with those in the most disadvantaged groups at greatest risk, highlights the imperative for direct early interventions to remedy the immediate consequences of these outcomes, as well as the need for policy to address the broader social and economic influences that foster inequalities.

Having looked at health outcomes for children, the next section now explores exposures to risk factors for negative outcomes in the early years.

3.7 Exposure to risk factors likely to have an adverse impact on health

The first table in this section, Table 3.11, looks at some risk factors for poor health outcomes over the four years. As described in Chapter 2, these risks can have both immediate consequences, or long-term implications, or both. Although these tend to have been measured at less frequent intervals than most of the outcome measures discussed above, these risk factor measures provide important information about children's exposure to them in the early years.

Table 3.11 Risk factors for poor health, at 10, 22, 34 and 46 months

Risk factors	10 months	22 months	34 months	46 months
	%	%	%	%
Maternal factors:
Current smoker	28	n/a	28	n/a
Smoker when child was 10 months and 34 months			23
Long-term health problem or disability	16	18	17	20
Long-term health problem or disability at least once since child's birth				35
Child factors:
Low fruit consumption (0-1 different types/ day)	n/a	15	n/a	n/a
Low vegetable consumption (0-1 different types/day)	n/a	30	n/a	n/a
Consumes sweets/chocolate at least once a day	n/a	43	n/a	n/a
Consumes crisps/savoury snacks at least once a day	n/a	46	n/a	n/a
Has non-diet soft drinks at least once a day	n/a	12	n/a	n/a
Bases
Weighted	5187	4475	4071	3978
Unweighted	5188	4481	4150	3981

Notes:
Bases vary for each measure, those shown are for the lowest of the range
n/a= not asked or not asked in a directly comparable way

A quarter of children (24%) had a mother who both smoked when they were 10 and 34 months old. A further 8% of mothers either smoked when their child was 10 months but gave up by the time they were 34 months, or were non-smokers at 10 months but smokers at 34 months. Table 3.12 and Table 3.13 below focus on the quarter of children that can be said to have been exposed to maternal smoke on a prolonged basis, at least when they were 10 and 34 months old (some will have been exposed for longer, but this information was not collected when the children were 46 months old).

The proportion of children whose mothers reported having a long-term health problem or disability increased a little between 10 and 46 months (from 16% to 20%). As with children's long-term conditions, there was a high degree of movement into and out of this category. Just 6% of children had a mother with a long-term condition throughout their first four years of life, but 35% did so for at least one period. The tables below focus on the prevalence of this latter measure. Maternal mental health was also explored when the children were 10 months old, using a scale from a standardized measure, the SF12 questionnaire mental health component. ⁸ This scale is designed to yield an average score for the whole population of 50; differences in mean SF12 maternal mental health component scores when the children were 10 months are presented in section 3.8.

Two aspects of the children's lifestyles were measured at 22 and 34 months. At 22 months, parents were asked how many different types of fruit and vegetable their children eat on a typical day. While this data cannot be used to judge whether children met the "5 a day" recommendation, it is a useful proxy measure of how much fruit and vegetables they consume. Low consumption of either of these items was defined as eating none, or just one, type per day.

Low fruit consumption was less common than low vegetable consumption (15% versus 30%), which might suggest that parents find it easier to feed their children fruit than vegetables. Parents were also asked how often their children consume sweets, crisps, savoury snacks, and non-diet drinks. Over four in ten children were reportedly eating confectionery or crisps/savoury snacks every day at 22 months, though far fewer, just 12%, were drinking non-diet drinks as often as this.

Physical activity was measured at 34 months by asking how much time children spent doing various physical activities in the previous week (such as running, cycling, swimming). A total estimate of time was derived from their answers and this measure was used to split the children into four equal sized groups ranging from the least active quarter to the most active (see Marryat et al., (2009) for a full discussion of children's activity at 34 months). The tables in section 3.8 focus on the 26% of children in least active group and explore whether low activity levels vary according to social groups.

3.8 Inequalities in exposure to risk factors for poor health outcomes

3.8.1 Area deprivation

Table 3.12 explores differences in exposure to the risk factors discussed in section 3.7 by area deprivation. The first point to note is that some of the risk factors for poor outcomes show much higher prevalences, and often greater inequality in their distribution, than was the case for the direct health outcomes explored in the corresponding Table 3.9. For example, with maternal smoking, the factor with the largest absolute and relative difference between the levels of deprivation, there was a fivefold increase in children's prolonged exposure to this between those in the least and most deprived areas (from 8% to 41%). The proportion of children whose mothers had experienced a long-term health problem or disability at least once since their birth also increased in line with deprivation, from 27% in the least deprived areas to 42% in the most. Similarly, the variation in mean scores on the mental health component of the SF12 questionnaire was significant; mothers in the most deprived areas had the lowest scores indicating higher levels of mental health problems.

All three indicators highlight the extent of the double burden of health inequalities experienced in households with young children. Not only are the children in more deprived areas at greater risk of poor outcomes, so too are their mothers (and wider family), which in turn acts as a negative risk factor for the children. Improving the health and outcomes for children therefore requires interventions targeted at improving outcomes for their close carers as well.

Children's eating habits all show large absolute and relative differences between the most and least deprived areas. Under one in ten (8%) of children in the least deprived areas consumed 0-1 different types of fruit a day compared with almost one in four (23%) of children in the most deprived areas. The corresponding proportions who typically eat 0-1 different vegetables a day were 22% and 37%. The reverse was true for daily consumption of sweets, crisps or non-diet drinks. Around a third of children in the least deprived areas consumed sweets or crisps every day compared with half of those in the most deprived areas. It is worth noting that the absolute differences between the most and least deprived areas were the same for both fruit and vegetables (15 percentage points). However, the relative risk is larger for fruit consumption because the underlying prevalence of low fruit consumption was lower overall. This is a good illustration of the need to consider large relative risks in the context of the underlying prevalence figures. In this instance it would be incorrect to conclude that the larger relative risk means a greater problem exists in relation to fruit than vegetable consumption.

Rather than looking at all five of the diet measures, the remaining tables in this section focus on fruit and vegetable consumption. The health benefits of a diet with a wide range of fruit and vegetables are well established so children who consume a limited variety of them can therefore be considered to be at an objective disadvantage relative to those who do not. In contrast, regular consumption of sweets, crisps and sugary drinks is not necessarily directly harmful unless the rest of the diet is unbalanced, dental hygiene is poor and activity levels are low.

As outlined in section 3.7, the questions about physical activity at 34 months were used to group the children according to their overall level across a number of activities. Table 3.12 looks at the least active children, defined as being in the lowest 25% of the distribution. Children in the least deprived areas were around half as likely as those in the most to be in this low activity group (18% versus 34%). The association between area deprivation and activity levels could be due to a lack of resources at the household level (such as access to a garden) or it could be related to the quality and provision of open spaces and play facilities in the local area. There is certainly scope for much further exploration of this than there is space in this report, and a much more detailed assessment of children's activity has been introduced at age 6 which would help with this.

Table 3.12 Risk factors for poor health outcomes by Scottish Index of Multiple Deprivation quintile

Risk factors	SIMD quintile
	5th - least deprived	4th	3rd	2nd	1st - most deprived	Relative risk	Risk difference
	%	%	%	%	%
Maternal factors:
Mother smoked when child was 10 & 34 months old	8	15	21	28	41	5.1	33
Long-term health problem/ disability at least once since child's birth	27	32	35	37	42	1.6	15
Mean score on SF12 mental health scale (10 months)	51.4	50.9	49.9	49.3	48.7	-	-2.7
Standard error of mean	0.21	0.26	0.33	0.37	0.31	-	-
Child factors:
Eating habits (22 months)
Eats 0-1 different fruits a day	8	11	13	20	23	2.9	15
Eats 0-1 different vegetables a day	22	29	28	34	37	1.7	15
Eats sweets at least daily	30	35	42	48	56	1.9	26
Eats crisps at least daily	35	42	46	49	54	1.5	19
Has sugary drinks at least daily	6	10	11	15	17	2.6	10
Low physical activity level (34 months)	18	22	25	31	34	1.9	16
Weighted bases	769	789	804	730	980
Unweighted bases	895	867	854	671	792

Notes:
The relative risk is the prevalence in the most deprived areas divided by the prevalence in the least deprived areas, the risk difference is the difference in the prevalence in the two areas. These two measures were calculated using the raw data whereas the percentages in each column have been rounded to the nearest whole number; the relative risks and risk differences cannot therefore be calculated from the percentages presented.
Bases vary for each measure, those shown are the lowest of the range.

3.8.2 Houshold income

Table 3.13 reveals some very similar patterns to Table 3.12. The absolute and relative risks of prolonged exposure to maternal smoking were greater for income than with area deprivation which is due to a marginally higher smoking rate among those in the lowest income households, and a marginally lower rate among those in the highest, than was the case in relation to the corresponding deprivation quintiles.

The proportion of children whose mothers have had a long-term health problem at least once since their birth was also higher in the lowest income households than in the most deprived areas, which helps to illustrate the importance of being able to look at inequality from more than one perspective to include measures of direct household resources as well as those pertaining to the areas in which people live. The patterns in relation to diet and physical activity also confirm the fact that household level resources are a strong determinant of exposure to risk factors for poor health in the early years and that further exploration of these patterns to disaggregate the household and area level influences of these risks could be enlightening.

Table 3.13 Risk factors for poor health outcomes by equivalised household income quintile

Risk factors	Equivalised household income quintile
	1st - highest	2nd	3rd	4th	5th - lowest	Relative risk	Risk difference
	%	%	%	%	%
Maternal factors:
Mother smoked when child was 10 & 34 months old	6	11	18	27	45	7.5	39
Long-term health problem/ disability at least once since child's birth	26	28	34	34	47	1.8	21
Mean score on SF12 mental health scale (10 months)	51.4	50.6	50.8	49.3	48.0	-	-3.4
Standard error of mean	0.24	0.29	0.36	0.33	0.41	-	-
Child factors:
Eating habits (22 months)
Eats 0-1 different fruits a day	7	10	13	18	25	3.6	18
Eats 0-1 different vegetables a day	22	24	28	35	39	1.7	16
Low physical activity level (34 months)	15	21	26	28	33	2.2	18
Weighted bases	617	763	709	816	911
Unweighted bases	710	846	743	787	742

Notes:
The relative risk is the prevalence in the lowest income quintile divided by the prevalence in the highest income quintile the risk difference is the difference in the prevalence in the two income categories. These two measures were calculated using the raw data whereas the percentages in each column have been rounded to the nearest whole number; the relative risks and risk differences cannot therefore be calculated from the percentages presented.
Bases vary for each measure, those shown are the lowest of the range.

3.8.3 Socio-economic classification ( NS- SEC)

Prolonged maternal smoking increased progressively across the five NS- SEC categories with the gap at its most extreme between professional and managerial households and semi-routine and routine households. The pattern for low fruit consumption was similar. In contrast, levels of low vegetable consumption and low activity were broadly similar across all groups except for professional and managerial households where these were the least common. Both the relative risk and absolute difference between the highest and lowest NS- SEC groups for maternal long-term health problems was lower than was evident for deprivation or household income. The pattern of association was not linear, 31% of mothers in professional and managerial households experienced long-term health problems at least once since their child was born compared with 34%-40% in all other groups. It is possible that poor maternal health could impact on household income if mothers' earning potential is affected as a consequence. However, as this NS- SEC measure reflects the household member with the highest status, there is probably less of a possibility that the relationship could work in that direction for this factor.

3.8.4 Conclusion

Some of the patterns discussed in the three preceding sections (3.8.1 to 3.8.3) looking at inequalities in exposure to risk factors are notable for the way in which the same measure can sometimes display a different pattern of association depending on the socio-demographic through which it is viewed. In addition, different variables revealed variations in relation to the same demographic factor, as shown in the above discussion of maternal ill-health and NS- SEC.

This reinforces the points made in Chapter 1 about the multifaceted nature of health inequality and its manifestations. However, the broad picture mapped out in this section does confirm to the overall message that drew the preceding section about outcomes to a conclusion.

Exposure to the kinds of risks that can affect health and development in the early years, and can have implications for decades to come, are not uniformly or randomly distributed across the population. Children from less affluent backgrounds are at a significant disadvantage in terms of their exposure to factors as seemingly diverse as physical activity, maternal mental health or smoking rates. The fact that the diets of children in less affluent circumstances appear to be higher in energy dense foods and lower in fruit and vegetables, and that low activity levels are highest among this group, suggests that the absence of health inequalities in unhealthy weight or obesity could well be temporary.

3.9 Summary measure of negative outcomes

To aid the analysis in the next chapter a summary measure of negative outcomes was created by summing a number of the negative outcomes experienced by children in their early years. This scale was not meant to have any substantive meaning in terms of estimating the overall prevalence of negative outcomes among children in the population; that would be a highly reductive approach and would be meaningless as it could never include all potential outcomes that are important to capture. However, for analytic purposes a scale such as this can be more helpful to explore associations with other factors than one or two single measures, which risks missing key children if the outcomes selected did not happen to be the ones that best reflect outcomes in general. In line with the definition of health set out in Chapter 1, the scale included a mixture of physical health outcomes as well as developmental, behavioural and psychosocial ones and spanned the period from birth to 46 months. For reasons that will explained further in the next chapter, it did not include any of the risk factors presented in the section above as this would limit the potential to use these in the exploration of resilience. The items in the scale were as follows, children were given one point for each outcome they had experienced:

• Low birthweight
• Time in an incubator or neo-natal unit following birth
• Fair/bad/very bad health at least once between 10 and 46 months
• Two or more accidents between 10 and 46 months
• Long-term health problems at least once between 10 and 46 months
• Any language development difficulties (reported by parent at 46 months)
• General behaviour problems (reported by parent at 46 months)
• Borderline/abnormal total difficulties SDQ scores (parental assessment at 46 months)
• Below average verbal ability (direct assessment at 34 months)
• Below average problem solving ability (direct assessment at 34 months)

Although the maximum score possible was 10, no child scored more than nine and just 0.5% had a score as high as this. A quarter (24%) of children had a score of zero and a further three in ten (29%) scored just one. The figure below shows the distribution of the scores for all children (in the first bar presented for each score). In the second set of bars the scores of children living in the most deprived area quintile are shown. Doing this illustrates the fact that children in more deprived areas are less likely than average to have low scores and more likely to have higher scores.

Figure 3.3 Negative health outcome scale in the 0 to 4 years period for all children and for those in the most deprived quintile of the Scottish Index of Multiple Deprivation