Growing up in Scotland: parenting and children's health

CHAPTER 4 IS PARENTING ASSOCIATED WITH CHILD HEALTH AND HEALTH BEHAVIOURS?

4.1 Introduction

This chapter investigates whether parenting is associated with child health outcomes and health behaviours. In the first part of the chapter, each dimension of parenting is examined individually. The last section of the chapter examines associations between the composite parenting index (as described in Chapter 3) and health. In both cases, associations are investigated before and after controlling for other family factors that may influence health ⁸ .

4.2 Key findings

• The prevalence of children in poor health and with poor health behaviours increased as the level of parenting skill decreased. After taking account of the child's social background and family circumstances, low overall parenting skills were associated with poorer general health, greater longstanding illness, poorer mental health, worse dental health, lower physical activity, higher screen time, lower fruit and vegetable consumption, and more snacking amongst children.
• The association between low parenting skills and children's social, behavioural or emotional difficulties was particularly strong.
• There were no associations between overall parenting skills and number of health problems in the past year, or accidents and injuries over the first five years, after allowing for other family factors.
• A wide range of different parenting skills appeared to be important for health, although the following aspects of parenting appeared particularly relevant for specific outcomes:
  • High levels of parent-child conflict were strongly associated with behavioural and emotional difficulties.
  • Low parental supervision was associated with poorer general health, greater long-standing illness and social, behavioural and emotional difficulties.
  • Joint mother-child activities and parental rules appeared important for health behaviours.

4.3 Associations between individual dimensions of parenting and child health

Table 4.1 presents associations between each parenting measure and child health outcomes. This analysis does not take account of the fact that many parenting practices may be related, and also ignores other important family influences on health (subsequent analysis will control for these relationships and other influences).

The table indicates that many of the associations were statistically significant after controlling for several basic demographic factors (the child's gender, the child's age in months at sweep 5, whether the child was first born or had older siblings and the number of children in the household at sweep 5 ⁹ ) but not other parenting measures. In summary:

• All aspects of parenting were associated with having moderate or severe difficulties with most also associated with general health and dental health (only mother-infant attachment and smacking were not).
• Health problems and accident/injuries were the two outcomes least likely to be associated with parenting.
• Home chaos was associated with each health outcome.
• Parental supervision was associated with all health outcomes except having three or more accidents/injuries.
• Both warmth and conflict in the mother-child relationship were associated with most health outcomes but neither were significantly related with having accidents/injuries or health problems.

Table 4.1 Summary of child health behaviour outcom

		General health fair, bad or very bad		Limiting long-standing illness		Total difficulties score moderate or severe		Three or more health problems last 12 months		Three or more accidents/ injuries sweeps 1-5		Dental decay		Row bases
Row percentages		%		%		%		%		%		%		Weighted	Unweighted
All		5		4		12		17		11		16		3478	3486
Mother-infant attachment	Low	6	NS	5	**	17	***	17	NS	11	NS	16	NS	1219	1238
	Medium	5		4		11		19		10		17		1098	1117
	High	4		3		9		17		11		17		1149	1121
Warmth of mother-child relationship	Low	7	***	7	***	21	***	19	NS	12	NS	21	***	1253	1203
	High	4		2		7		17		10		14		2198	2261
Mother and child activities	Low	6	*	4	NS	17	***	18	NS	13	NS	24	***	1299	1214
	Medium	5		4		11		16		10		14		1260	1271
	High	3		3		7		18		10		9		919	1001
Conflict in mother-child relationship	Low	4	**	2	***	3	***	16	NS	9	NS	15	*	1126	1169
	Medium	4		3		6		18		12		16		1272	1283
	High	7		7		29		19		12		19		1052	1011
Smacking	None	5	NS	4	NS	10	***	16	NS	9	*	15	NS	1783	1778
	Some	5		4		15		19		13		18		1695	1708
Parental supervision	Low	6	*	6	***	14	***	21	***	10	NS	20	***	1419	1333
	Medium	5		2		13		16		12		16		1235	1252
	High	3		3		7		14		10		13		823	900
Number of rules	Low	6	*	5	NS	18	***	19	NS	12	NS	24	***	1245	1187
	Medium	4		4		10		16		10		13		1897	1962
	High	3		3		5		17		12		9		336	337
Home chaos	Low	3	**	2	***	6	***	16	**	9	***	13	***	1704	1764
	Medium	6		5		10		15		10		16		563	570
	High	7		6		22		20		14		21		1211	1152

Note: Significance levels: *p<0.05, **p<0.01, ***p< 0.001, NS = not significant. Adjusting for child gender, age in months at sweep 5, birth order and number of children in household at sweep 5. Analyses do not adjust for relationships between parenting measures.

4.3.1 Associations after controlling for family influences and relationships between parenting measures

As discussed in section 3.4, the different parenting styles are related to each other. The analysis in the previous section does not control for these relationships. In this section, the first stage of analysis considers the association between parenting and health whilst controlling for the relationships between the parenting measures. At the second stage, the analysis further controlled for other family influences on health, namely: mother's ethnic group, age at birth of the survey child, educational qualifications and mental health; and family composition from sweeps 1 to 5, housing, household equivalised income and area deprivation. Full details of these analyses are provided in section 2.3 of the Technical Appendix. In this chapter, we focus on important features of the fully adjusted (stage 2) models.

Table 4.2 provides a summary of statistically significant associations between individual parenting measures and health outcomes after controlling for the relationships between the parenting measures and some family characteristics.

Table 4.2 Associations between individual parenting measures and child health after controlling for family influences and relationships between parenting measures

	Connection			Negativity		Control
	Mother-infant attachment	Warmth of mother-child relationship	Mother and child activities	Conflict in mother-child relationship	Smacking	Parental supervision	Rules	Home chaos
General health fair, bad or very bad						v
Limiting long-standing illness	v	v		v		v		v
Total difficulties score moderate or severe		v		v	v	v	v	v
Three or more health problems last 12 months
Three or more accidents/ injuries sweeps 1-5					v
Dental decay			v				v

n=3343 (unweighted). Ticks indicate significant relationships between variables. Models controlled for child's gender, age at sweep 5, birth order; number of children in household; mother's ethnic group, age at birth of the survey child, educational qualifications and mental health; family composition from sweeps 1 to 5, housing, household equivalised income and area deprivation; other parenting measures.

It can be seen that all parenting measures were associated with at least one health outcome, and that many different dimensions of parenting were associated with limiting longstanding illness or a high level of social, behavioural and emotional difficulties. However, no parenting measure was associated with having a high number of health problems in the last 12 months.

Detailed information about the magnitude of effects can be found in section 2.3 of the Technical Appendix. In summary, this information shows that:

• In the connection domain, the odds of children who experienced low levels of attachment, warmth or joint activities having poor health were 1.8 to 2.1 times higher than for children with more optimal parenting.
• In the negativity domain, the association between conflict and poor child mental health was particularly strong. The odds of children with a highly conflictual mother-child relationship having mild or severe social, emotional and behavioural difficulties were
  7 times higher than for children experiencing low levels of conflict.
• For other health outcomes, associations were such that the odds of children experiencing highly negative parenting (conflict and smacking) having poor health were 1.3 to 2.2 times higher than those for children with low levels of negative parenting.
• In the control domain, the odds of children experiencing low levels of supervision, rules and control having poor health were 1.8 to 2.2 times higher than for children with less optimal parenting.

Note that whilst these associations are statistically significant, it is not possible to determine causal direction from the findings.

We compared the findings in this section with a recent study using data from the Millennium Cohort Study ( MCS), referred to in Chapter 1 (Hobcraft and Kiernan 2010). The MCS, using the sample of children surveyed in England, examined associations between parenting behaviours at age 3 and two health outcomes also used in this report: general health (fair/poor/very poor) and total difficulties score (moderate/severe) at age 5.

Parenting behaviours in the MCS study covered similar 'domains' to GUS, with some similar or identical measures, although it did not include the parental supervision or home chaos measures used here. The MCS study included mother-reported Pianta measures of warmth and conflict (similar to GUS), interviewers' observations of positive and negative parenting, mother reports of reading with the child, disciplinary practices (frequency of smacking and shouting) and family organisation (regular bedtimes and mealtimes).

In multivariate analysis that took account of other family and maternal characteristics, the MCS study found that two measures of family organisation (regular bed and mealtimes) were the only parenting behaviours predicting poor general health. Pianta warmth and conflict, shouting and irregular mealtimes predicted a moderate/severe total difficulties score.

In order to see whether parenting was associated with a change in health from age 3 to 5, a second stage of analysis in the MCS study added health outcomes at age 3 to the models. At this stage some of the parenting measures dropped out of the analysis, leaving only irregular mealtimes as a predictor of poor general health and Pianta conflict and irregular mealtimes as a predictor of total difficulties.

Both MCS and GUS results suggest that parenting is more strongly associated with social, behavioural and emotional difficulties than with general health, in terms of the greater number of significant associations between parenting measures and difficulties. The MCS findings for Pianta conflict and routines in relation to total difficulties echo the findings of this study. In the GUS data set, because conflict and difficulties were measured at the same interview, it is possible that some of the strong association is due to reverse causation, with difficult child behaviour leading to conflict in the mother-child relationship. The MCS study had the advantage that associations between parenting and health outcomes were longitudinal in nature, and this temporal relationship adds strength to the likelihood that findings reported for total difficulties in GUS could also reflect earlier negative parenting. After the completion of current and future data collection, such analysis will be possible using GUS data.

4.4 Associations between individual dimensions of parenting and child health behaviours

Table 4.3 presents associations between each parenting measure and child health behaviours. As in Table 4.1, this analysis does not take account of the fact that many parenting practices may be related, and also ignores other important family influences on health, although some basic demographic factors (the child's gender, the child's age in months at sweep 5, whether the child was first born or had older siblings and the number of children in the household at sweep 5 - see earlier footnote in this chapter) are controlled for.

As for the child health outcomes, many of the associations between individual parenting measures and health behaviours were statistically significant after taking account of basic demographic factors. In particular, this analysis showed that:

• Mother-child activities and rules were significantly associated with all health behaviours.
• Home chaos was associated with all but one (physical activity) of the health behaviours.
• Screen time and fruit and vegetable consumption were associated with most parenting measures.
• Early mother-infant attachment was not associated with any of the behaviours selected for this report.

Table 4.3 Associations between parenting and child health behaviours

		Physical activity - low		Screen time 2 hours plus on termtime weekday		Fruit and vegetable consumption - low		Sweets/crisps/sugary soft drinks more than once daily		Row bases
Row percentages		%		%		%		%		Weighted	Unweighted
All		38		40		69		35		3478	3486
Mother-infant attachment	Low	40	NS	40	NS	71	NS	35	NS	1219	1238
	Medium	39		37		68		33		1098	1117
	High	35		41		67		36		1149	1121
Warmth of mother-child relationship	Low	42	***	43	*	71	NS	39	***	1253	1203
	High	35		38		68		33		2198	2261
Mother and child activities	Low	45	***	47	***	79	***	43	***	1299	1214
	Medium	35		38		68		34		1260	1271
	High	31		32		56		26		919	1001
Conflict in mother-child relationship	Low	35	NS	34	***	65	***	33	NS	1126	1169
	Medium	38		40		68		34		1272	1283
	High	39		46		73		38		1052	1011
Smacking	None	38	NS	37	**	66	***	34	NS	1783	1778
	Some	37		42		72		36		1695	1708
Parental supervision	Low	39	NS	44	***	72	**	36	NS	1419	1333
	Medium	38		37		68		35		1235	1252
	High	36		36		65		33		823	900
Number of rules	Low	42	***	49	***	77	***	42	***	1245	1187
	Medium	36		35		65		32		1897	1962
	High	31		29		60		28		336	337
Home chaos	Low	37	NS	34	***	65	***	31	***	1704	1764
	Medium	36		40		72		33		563	570
	High	40		47		72		42		1211	1152

Note: Significance levels: *p<0.05, **p<0.01, ***p< 0.001, NS = not significant. Adjusting for child gender, age in months at sweep 5, birth order and number of children in household at sweep 5. Analyses do not adjust for relationships between parenting measures.

4.4.1 Associations after controlling for family influences and the relationship between parenting measures

Table 4.4 provides a summary of statistically significant associations between individual parenting measures and health behaviours after controlling for family influences and the relationship between parenting measures. Whilst some of the associations shown in Table 4.3 have dropped out, with the exception of early mother-infant attachment, all parenting measures were associated with at least one health behaviour. Joint activities and parental rules were both associated with three out of four health behaviours.

Table 4.4 Associations between individual parenting measures and child health behaviours after controlling for family influences and relationships between parenting measures

	Connection			Negativity		Control
	Mother-infant attachment	Warmth of mother-child relationship	Mother and child activities	Conflict in mother-child relationship	Smacking	Parental supervision	Rules	Home chaos
Physical activity - not meeting guidelines		v	v
Screen time 2 hours plus on term-time week day				v	v		v
Fruit and vegetable consumption - low			v		v	v	v
Sweets/crisps/sugary soft drinks more than once daily			v				v	v

n=3343 (unweighted). Ticks indicate significant associations between variables. Models controlled for child's gender, age at sweep 5, birth order; number of children in household; mother's ethnic group, age at birth of the survey child, educational qualifications and mental health; family composition from sweeps 1 to 5, housing, household equivalised income and area deprivation; other parenting measures.

Detail on the magnitude of associations between parenting and health behaviours has been provided in section 2.4 of the Technical Appendix. Below we provide a summary of those data:

• In the connection domain, the odds of children experiencing low levels of warmth and joint activities with their mother having poor health behaviours were 1.6 to 2.6 times higher than those of children with high levels of connection.
• In the negativity domain, the odds of children experiencing high levels of conflict and smacking having poor health behaviours were 1.2 to 1.3 times higher than for children with low levels of negativity.
• In the control domain the odds of children experiencing low levels of supervision and rules or a high level of home chaos having poorer health behaviours were 1.2 to 2.0 higher than those of children with high parental control.

4.5 Which dimensions of parenting are most important?

As sections 4.3 and 4.4 show, each of the parenting domains (connection, negativity and control) contained parenting measures with significant associations with both child health and health behaviours. The magnitude of these significant associations did not differ greatly between child health and health behaviours, although high conflict was unique in its particularly strong association with social, emotional and behavioural difficulties. Within each 'domain', each dimension of parenting was associated with several health/health behaviour outcomes (the one exception being mother-infant attachment, which was only associated with limiting long-term illness).

These findings of a complex network of associations between all domains and dimensions of parenting suggest that a wide, rather than a narrow, range of parenting skills is important to benefit both children's health and their health behaviours. Although it is difficult to single out particular dimensions of parenting, it is worth noting that low parental supervision was associated with three out of six health outcomes, and bordered on statistical significance for a fourth outcome. For child health behaviours, joint mother-child activities and parental rules appeared more important, as each measure was associated with three out of the four health behaviours studied.

4.6 Associations between index of parenting skills and child health and health behaviours

In this section, we consider the association between classification on the index of parenting skills (low, average or high) and child health and health behaviours.

Figures 4-A and 4-B show the proportion of children in poor health and with poor health behaviours according to their grouping on the parenting skills index. The graphs show that the prevalence of children in poor health and with poor health behaviours increased as the level of parenting skill decreased. The difference between the low parenting skill group and the other two groups was most pronounced for social, behavioural and emotional difficulties.

Figure 4-A Percentage of children in poor health according to parenting skill index group

Figure 4-B Percentage of children with poor health behaviours according to parenting skill index group

4.6.1 Associations between parenting index and health and health behaviours after controlling for family influences

Two-stage analysis of associations between the parenting index and both health outcomes and health behaviours were performed in a similar fashion to that described for the previous sections. Full results are presented in section 2.5 of the Technical Appendix. Table 4.5 summarises statistically significant associations between the parenting index and health outcomes/health behaviours after controlling for family influences and the relationships between the parenting measures. There were significant associations between the parenting index and all health outcomes and health behaviours, with two exceptions: health problems and accidents/injuries.

Table 4.5 Associations between parenting index and child health and health behaviours after controlling for family influences

		Parenting index
Child health	General health fair, bad or very bad	v
	Limiting long-standing illness	v
	Total difficulties score borderline or abnormal	v
	Three or more health problems last 12 months
	Three or more accidents/injuries sweeps 1-5
	Dental decay	v
Child health behaviours	Physical activity - low	v
	Screen time 2 hours plus on term-time week day	v
	Fruit and vegetable consumption - under 5 portions	v
	Sweets/crisps/sugary soft drinks more than once daily	v

N=3343 (unweighted). Ticks indicate significant associations between variables. Models adjusted for child's gender, age at sweep 5, birth order; number of children in household; mother's ethnic group, age at birth of the survey child, educational qualifications and mental health; family composition from sweeps 1 to 5, housing, household equivalised income and area deprivation.

Full details of the strength of the associations are included in the technical appendix. In summary, we found that, when children in the low parenting skills group were compared with those in the high parenting skills group:

• Low parenting skill showed the strongest association with social, behavioural and emotional difficulties. The odds of children of parents in the low skill group having mild or severe difficulties were more than eleven times higher than those of children with parents in the high skill group.
• For the other health outcomes, the odds of having poor health for children with low skill parents ranged from being two to over four times higher than for children with high skill parents.
• The odds of children of with low-skilled parents having poor health behaviours were 1.5 times to over 2 times higher than those for children with high skill parents.

For two health outcomes (limiting long-term illness and social, emotional and behavioural difficulties) and two health behaviours (screen time and fruit and vegetable consumption), children of parents with average skills were also more likely to have poor health and poor health behaviour than children in the high parenting skills group. Elsewhere, differences between the average and high skills groups were not statistically significant (p<0.05).

This analysis further confirms that highly skilled parenting is associated with more positive health outcomes and health behaviours in children.

4.7 Summary

Parenting skills are associated with a range of child health outcomes and behaviours. Even after taking account of the child's social background and family circumstances, low overall parenting skills were associated with poorer general health, greater longstanding illness, poorer mental health, worse dental health, lower physical activity, higher screen time, lower fruit and vegetable consumption, and more snacking amongst children.

A wide range of different parenting skills were important for health, although certain aspects of parenting appeared particularly relevant for specific outcomes. For example, high levels of parent-child conflict were strongly associated with behavioural and emotional difficulties, whereas joint mother-child activities and parental rules appeared more important for health behaviours.

At an overall level, parenting skill was more strongly related to certain health outcomes and behaviours than others. The association between low parenting skills and children's social, behavioural or emotional difficulties was particularly strong. In contrast, there were no associations between overall parenting skills and number of health problems in the past year, or accidents and injuries over the first five years after allowing for other family factors.