Written evidence from the Maternal and Child Health Network (MatCHNet), MRC/CSO Social and Public Health Sciences Unit (CPM0009)

This response is submitted on behalf of the MRC/CSO Social and Public Health Sciences Unit, University of Glasgow and was prepared by members of the cross-UK Maternal and Child Health Network (Ruth Dundas, Emma Stewart, Anna Pearce, Alastair Leyland – University of Glasgow, Sinead Brophy – Swansea University, Richard Cookson – University of York, Ruth Gilbert & Pia Hardelid – NIHR Children and Families Policy Research Unit, UCL Great Ormond Street Institute of Child Health, Katie Harron – UCL Great Ormond Street Institute of Child Health, Joanne Given – Ulster University, Rachael Wood – Public Health Scotland).

About us

The MRC/CSO Social and Public Health Sciences Unit, University of Glasgow is an interdisciplinary group of sociologists, anthropologists, psychologists, epidemiologists, geographers, political scientists, public health physicians, statisticians, information scientists, trial managers and others. The Unit receives core-funding from the Medical Research Council and the Scottish Government Chief Scientist Office, as well as grant funding for specific projects from a range of sources. We conduct research to understand the determinants of population health and health inequalities, and to develop and test interventions to improve health and reduce inequalities, using a wide variety of methods including qualitative research, the collection, linkage, and analysis of social survey and routinely collected data, evidence synthesis, randomised controlled trials and natural experimental studies. Further information about the Unit is available at http://www.sphsu.mrc.ac.uk/.

The Maternal and Child Health Network (MatCHNet) is a UK Prevention Research Partnership (UKPRP) funded network that aims to evaluate policy impacts on child and maternal health by harnessing administrative data across the 4 UK nations. MatCHNet is focused on interventions to tackle the social determinants of health that can improve pregnancy and early childhood outcomes. More information can be found at: www.gla.ac.uk/matchnet.

Executive summary

1. Giving every child the best start in life is a key policy goal for a healthier society. Child poverty is bad for children’s health; it is associated with negative educational outcomes and adverse long-term social and psychological outcomes. This constrains children’s development and reduces adulthood life chances.
2. Administrative data that is routinely collected by welfare, health, education, and social care services should be deployed more effectively to evaluate the impact of a wide range of policies upon child poverty outcomes. Linked data from the tax and benefits system to data about children in households should be used to assess the impact of varying welfare provision across the 4 UK nations (such as Universal Credit, the Sure Start Maternity Grant, and the Scottish Child Payment). Additionally, de-identified data linked to health and education outcomes would provide insight into what policies can help tackle child poverty and improve children’s health.
3. A closer relationship across all Government Departments will establish such cross-sectoral linkage to realise the benefits of administrative data for society and children in poverty. A key initial data-sharing relationship to establish would be one between the Department for Work and Pensions and the Ministry for Housing Communities and Local Government. An additional, initial relationship would be between the Department for Work and Pensions and the Department for Health and Social Care.
4. In summary, our response specifically addresses Questions a1, a5 and b3, and covers the following:

i.         We provide evidence about the impact of child poverty upon health outcomes.

ii.       We discuss how child poverty can be measured using existing data.

iii.     We recommend the linkage of tax and benefits data to children in households.

iv.     We demonstrate how data linkage across Government Departments could answer important questions about child poverty.

a. Measurements and targets

Question 1: How should child poverty be measured and defined?

1. As well as the existing measures of absolute and relative child poverty, consideration should be given to the thresholds of child poverty e.g., poverty (<60% median income) and severe poverty (<50% median income).
2. In addition to a broad measure of child poverty, the Department for Work and Pensions (DWP) should release finer details of specific benefits and income levels of children/households to allow for thorough inspection of particular features of the current benefit system.
3. Child poverty should be measured at the individual and household level. DWP collects and maintains data for all benefit claimants and households across the UK. These data are therefore comparable and available at the household level. There are two reasons for this:

i.         Having data at this level will allow exposure measures to child poverty to be linked to datasets at other Government Departments at the individual/household level. This would facilitate detailed and robust research studies evaluating key policies that may impact child poverty and other outcomes.

ii.       Using administrative data that is routinely collected to measure and monitor child poverty provides maximum benefit and minimum cost of data collection.

4. Access to these data would need to be controlled and access only allowed for bona fide people who have completed information governance training. Both the data and means to link it should be available to researchers and other non-governmental parties, as well as inter-governmental sharing. These data could be deposited and analysed in the ONS Safe Research Setting. This provides high quality security and access is only allowed for approved researchers and projects.

Question 5: What is the impact of child poverty?

1. Poverty is bad for children’s health, with negative effects in pregnancy, infancy, and the early years. Poverty is linked to increased infant mortality rates. Research in England documents a disproportionate rise in infant mortality in the poorest areas of the country and suggests that a third of the increase in infant mortality (between 2014 and 2017) can be directly attributed to child poverty [1]. Poverty and food insecurity during pregnancy is also linked to poor fetal growth [2].
2. Children living in disadvantaged socioeconomic circumstances (SECs) suffer from worse health than their more advantaged peers [3]. Exposure to poverty during childhood is associated with poor physical and mental health outcomes that persist into adolescence and adulthood [4]. For example, research has linked poverty in early childhood to a higher risk of obesity [5] and being overweight [5].
3. Furthermore, the impacts of poverty accumulate throughout childhood – the longer a child is exposed to poverty the worse their outcomes.
4. Poverty is largely responsible for many of the inequitable differences seen within the UK population, for example between lone parent and couple families [6].
5. Child poverty is associated with poor educational outcomes and adverse long-term social and psychological outcomes, which impacts upon children’s development and limits adulthood life chances [7]. Thus, poverty leads to the intergenerational transference of health inequalities.
6. Policies aimed at tackling child poverty can lead to improvements in child health and socio-emotional well-being through pathways to impact [3, 8].
7. International research evidence indicates that level of family income and increased welfare provisions can significantly improve child and maternal health outcomes. A review of 27 quantitative and qualitative studies found strong positive effects on birthweight and maternal mental health [9]. Likewise, a systematic review on cash transfers and health outcomes in low, middle and high income countries documented the health benefits of increasing the material circumstances of families with children [10]. For example, a Canadian scheme resulted in a reduced risk of low birth weight, preterm birth and small gestation for age [11]. Similarly, dividends received by families in Alaska are associated with increase in birth weight, decrease in low birth weight and improved Apgar scores at birth [12].
8. Macro-level change such as improved social security systems can reduce child poverty. Economic modelling suggests that increasing the main child element of Universal Credit by £40 per week as well as abolishing the benefit cap and the two-child limit could cut child poverty by 100,000 in Scotland [13]; reducing the number of Scottish children living in poverty (230,000) by nearly 50% [14].

Q5 (continued): And how can it best be measured?

1. We see the ‘measurement’ of poverty in two distinct ways – a definition of child poverty and a need to understand the drivers of child poverty. Both are necessary to inform action to reduce child poverty. Below we focus on how to measure the latter.
2. It is not purely the measurement of child poverty that is needed. While determining levels of child poverty is important by itself, it does not inform policy makers on what actions might reduce poverty or ameliorate its effects. Therefore, there is a need to understand the determinants of poverty and the mediators/modifiers of how poverty affects key health, social and economic outcomes.
3. During the past year, MatCHNet has conducted a rapid policy review of the UK policy landscape to extensively scope policies that potentially impact upon child and maternal health outcomes. Utilising a systematic search strategy, guided by the social determinants of health framework, 88 policies have been identified in the areas of welfare, employment, health, housing, education, and environmental policies. Specifically, this includes 32 different welfare policies.
4. It is vital to evaluate and prioritise policies to understand what works best at improving child poverty and associated health outcomes. This understanding will permit the evaluation and understanding of the effect of policies across government, both singly and interdependently, on poverty and associated outcomes.
5. Linked data is essential to conducting these policy evaluations. There is a need to have the ability to easily link child poverty exposure to other outcomes such as health (including behaviours), mortality and education.

In the section below, we cover the current limitations of existing data and what we see as the ideal situation - linkage of tax and benefits data to children in households - to allow for comparative analysis. The optimum situation may be difficult to establish quickly, therefore we also provide recommendations for immediate improvements and current best practice.

Limitations of current approaches and available data

1. A unique feature of MatCHNet is the ambition to conduct outcome evaluations of policies using natural experiment designs across the 4 UK nations. This is constrained with the currently available data. In order to conduct such studies, data is required for the whole population therefore survey data will not provide large enough sample sizes to conduct robust evaluations, and certainly will not adequately cover Wales, Scotland and Northern Ireland [15]. Current population level data that has the potential to be used (Census, Index of Multiple Deprivation, including Income Deprivation Affecting Children Index) have limitations.
2. The Census covers the whole population of the UK, but in terms of measuring child poverty, census data is incomplete. The Census only collects information on occupation but not income or benefits data. It is not easy to link individual records in the Census to data from other sectors. In Northern Ireland, health data can only be linked to the Census in context of NI Longitudinal study (NILS), which limits analysis.
3. Across the UK, there are four indices of multiple deprivation: Index of Multiple Deprivation (IMD), Scottish Index of Multiple Deprivation (SIMD), Welsh Index of Multiple Deprivation (WIMD) and Northern Ireland Multiple Deprivation Measure (NIMDM).
4. Each one calculates an Income Domain, which is a population weighted count of means tested benefits claimants. These Income Domains have the potential to be used to measure child poverty, and the Ministry for Housing Communities and Local Government (MHCLG) do produce estimates for the Income Deprivation Affecting Children Index (IDACI), which measures the proportion of all children aged 0 to 15 living in income deprived families [16]. NB. these indices are intended to measure neighbourhood deprivation, but researchers are often forced to use these as a poor proxy for household level socio-economic circumstances because no other data are available.
5. This measure is inadequate because it is only available for analysis at the Lower Super Output Area in England and Wales (average population size 1500), Datazone level in Scotland (average population size 700), and Super Output Area in Northern Ireland (average population size 2,000). This restricts the utility of evaluating policies using this measure of child poverty as it does not identify individual children or households. As this is an aggregate measure, not all children in poverty will be living in a LSOA defined as high on the IDACI and there will be children in poverty who live in areas defined as low on the IDACI. For example, a report by MHLGC showed that the most deprived decile of IMD includes between 26% and 60% of households that are income deprived [16]. Such a blunt measure of poverty does not adequately reflect economic disparities, resulting in no or weak relationships between health and deprivation in some areas.
6. As data about benefits and benefits claimants at the population level is lacking, researchers have used panel surveys and inferred recipient status based upon the month Universal Credit was introduced in a respondent’s residential area [17]. Variability in the roll out across the UK and a lack of Universal Credit data for Northern Ireland respondents limits this type of invaluable research.
7. Existing measures such as the Indices of Multiple Deprivation do not capture the changeable nature of income, which fluctuates frequently and sometime dramatically in response to life events such as relationship breakdown or job loss. As mentioned earlier, the negative impacts of income are cumulative. The only way we can truly capture and understand the relationship between poverty and health, and the potential for policies to alter short and long-term outcomes, is using administrative data on income and benefits.
8. Inaccurate measures of income at household level substantially underestimates the steepness of the gradient between poverty and health, educational outcomes, and wellbeing. In turn, policies based on such inaccurate measures could result in unintended harm and unmet need. 

Outline of ideal situation

1. The optimum solution would be to use the data DWP collects and maintains for all benefit claimants and households across the UK. These data are comparable and available at the household level.
2. These exposure measures to child poverty could be linked to datasets at other Government Departments at the individual/household level to allow for detailed research studies.
3. Access to these data would need to be controlled and only allowed for bona fide researchers who have completed information governance training. These data could be deposited and analysed in the ONS Safe Research Setting. This provides high quality security and access is only allowed for approved researchers and projects. There are equivalent Trusted Research Environments in the devolved nations e.g., SAIL in Wales, the National Safe Haven in Scotland, and the Northern Ireland Statistics and Research Agency. These have been used extensively to allow researchers to access individual health data.
4. The costs of data access should not be prohibitive [18].

Immediate improvements

1. It may take time for data sharing agreements to be in place to allow for the cross-sectoral linkage described above. Currently, there are immediate improvements that can be done.
2. A number of UK Government committees and other bodies have strongly recommended improvements in data linkage and the UK Government have committed to cross-government data linkage [19]. DWP should make progress in data linkage with other Government Departments.
3. The Department for Work and Pensions (and the Ministry for Housing Communities and Local Government who produce IMD figures) should produce their outputs currently used for IMD for households (e.g., on benefits/<60% of median income) by household (identified by a unique property reference number UPRN – encrypted for research) instead of LSOA.
4. Linkage of household level information to an encrypted UPRN is increasingly being used in administrative data analyses of health and education data in Wales and Scotland, and in England (ONS, NHS England). Linkage of an encrypted UPRN with categorical information on household benefits and income, could be non-identifying.
5. Onward linkage of the encrypted UPRN linked-household income information to de-identified administrative data, from health and/or education data, would allow timely evaluations at scale of the impact of benefits or income policies on children’s health and education.

b. Joint working

Question 3: What would be the merits of having a cross-government child poverty strategy?

1. It is key to have a cross-government child poverty strategy as the determinants of child poverty cut across all sectors of government. The social determinants of health are key drivers of child poverty and many policies that may affect child poverty, will come from departments other than DWP, for example employment, housing, health, education, and environment.
2. This is evidenced by MatCHNet’s rapid policy review; we identified 88 different policies in the areas of welfare, employment, health, housing, education, and environment that potentially impact upon maternal and child health outcomes across the UK.
3. As one specific example, both adverse childhood experiences (ACEs) and disadvantaged socioeconomic conditions (SECs) in early life are linked with poor outcomes across the life course [20]. While health and social care services can detect ACEs and target resources, these are individual level, after-the event markers. This does not tackle the underlying causes of social and economic disadvantage.
4. To determine which policies successfully reduce child poverty (and associated health outcomes), data are required to assess the efficacy and cost effectiveness of different policies.
5. Some research has been conducted in this area [21-24], but there are many more policies which need to be evaluated. For example, what is the difference in outcomes from the Scottish Best Start Grant Scheme versus the Sure Start Maternity Grant? How does this impact upon child poverty rates and related health outcomes? Also, what has been the impact of Scotland moving to Best Start Foods from Healthy Start?
6. Political devolution in the UK provides a natural experiment to explore the effect of different poverty strategies [25] alongside variations in policies across different Government Departments. Linked data across different policy domains (e.g., welfare and health; welfare and education) is essential to tackling the following urgent questions:

i.         What is the effect of different poverty strategies in 4 UK nations?

ii.       What are the impacts of the new Scottish Child Payment upon child poverty?

iii.     How has the £20 uplift in Universal Credit affected child poverty and child health?

iv.     What is the impact of the two-child limit (UC) on child poverty and health outcomes? Has there been a disproportionate effect in Northern Ireland?

Recommendations

1. The Department for Work and Pensions (and the Ministry for Housing Communities and Local Government who produce IMD figures) should produce their outputs currently used for IMD for households (e.g., on benefits/<60% of median income) by household (identified by a unique property reference number UPRN – encrypted for research) instead of LSOA. UPRN is increasingly being used in Wales and Scotland, and is starting in England (with ONS and in NHS England).
2. DWP should provide a more detailed range of level of income and benefits, to allow evaluation of policy changes over time.
3. DWP and MHCLG should work together to produce categorical metrics of household income/benefits (and other components of IMD) at different geographical output levels (e.g., UPRN, postcode, Output Area and LSOA) to distribute to government data providers and local authorities. Dissemination of these as encrypted or identified level will depend on the purpose and authorisation.
4. DWP to deposit household level data in Trusted Research Environments to allow for robust policy evaluation.
5. DWP to make data available to researchers and policy evaluators without prohibitive costs.

References

1.              Taylor-Robinson, D., et al., Assessing the impact of rising child poverty on the unprecedented rise in infant mortality in England, 2000–2017: time trend analysis. BMJ Open, 2019. 9(10): e029424.

2.              Goin, D.E., et al., Maternal experience of multiple hardships and fetal growth: extending environmental mixtures methodology to social exposures. Epidemiology, 2021. 32(1): 18-26.

3.              Pearce, A., et al., Pathways to inequalities in child health. Archives of Diseases in Childhood, 2019. 104(10): 998-1003.

4.              Nelson, C.A., et al., Adversity in childhood is linked to mental and physical health throughout life. BMJ, 2020. 371: m3048.

5.              Lai, E.T.C., et al., Poverty dynamics and health in late childhood in the UK: evidence from the Millennium Cohort Study. Archives of Disease in Childhood, 2019. 104(11): 1049.

6.              Pearce, A., H. Lewis, and C. Law, The role of poverty in explaining health variations in 7-year-old children from different family structures: findings from the UK Millennium Cohort Study. J Epidemiol Community Health, 2013. 67(2): 181-9.

7.              Wickham, S., et al., Poverty and child health in the UK: using evidence for action. Archives of disease in childhood, 2016. 101(8): 759-766.

8.              Pearce, A., et al., Family structure and socio-emotional wellbeing in the early years: a life course approach. Longitudinal and Life Course Studies, 2014. 5(3): 20.

9.              Gibson, M., W. Hearty, and P. Craig, The public health effects of interventions similar to basic income: a scoping review. The Lancet Public Health, 2020. 5(3): e165-e176.

10.              Siddiqi, A., A. Rajaram, and S.P. Miller, Do cash transfer programmes yield better health in the first year of life? A systematic review linking low-income/middle-income and high-income contexts. Archives of Disease in Childhood, 2018. 103(10): 920.

11.              Brownell, M.D., et al., Unconditional prenatal income supplement and birth outcomes. Pediatrics, 2016. 137(6): e20152992.

12.              Chung, W., H. Ha, and B. Kim, Money transfer and birth weight: Evidence from the Alaska Permanent Fund Dividend. Economic Inquiry, 2016. 54(1): 576-590.

13.              Fraser of Allander Institute, Modelling the Economic Impact of a Citizen’s Basic Income in Scotland: Final Report. 2020, Fraser of Allander Institute, University of Strathclyde: https://www.sbs.strath.ac.uk/download/Fraser/202004/Modelling-Economic-Impact.pdf.

14.              Scottish Government, Poverty and Income Inequality in Scotland 2016-19. 2020, Scottish Government: https://www.gov.scot/publications/poverty-income-inequality-scotland-2016-19/pages/3/.

15.              Zylbersztejn, A., R. Gilbert, and P. Hardelid, Preventing child deaths: what do administrative data tell us? Archives of Disease in Childhood, 2020. 105(1): 15.

16.              Noble, S., et al., The English Indices of Deprivation 2019: Research Report. 2019, Ministry of Housing, Communities and Local Government: London.

17.              Wickham, S., et al., Effects on mental health of a UK welfare reform, Universal Credit: a longitudinal controlled study. The Lancet Public Health, 2020. 5(3): e157-e164.

18.              Cavallaro, F., et al., Reducing barriers to data access for research in the public interest—lessons from Covid-19. The BMJ Opinion, 2020: https://blogs.bmj.com/bmj/2020/07/06/reducing-barriers-to-data-access-for-research-in-the-public-interest-lessons-from-covid-19/.

19.              Government Analysis Function, Joined Up Data in Government: The Future of Data Linking Methods. 2020, Office for National Statistics: https://www.gov.uk/government/publications/joined-up-data-in-government-the-future-of-data-linking-methods/joined-up-data-in-government-the-future-of-data-linkage-methods#acknowledgements.

20.              Straatmann, V.S., et al., How do early-life adverse childhood experiences mediate the relationship between childhood socioeconomic conditions and adolescent health outcomes in the UK? Journal of Epidemiology and Community Health, 2020. 74(11): 969.

21.              Harron, K., et al., Using Linked Administrative Data for Monitoring and Evaluating the Family Nurse Partnership in England: A Scoping Report. 2016, Family Nurse Partnership: London, http://repository.tavistockandportman.ac.uk/1448/1/FNP%20Report.pdf.

22.              Leyland, A.H., et al., Evaluation of health in pregnancy grants in Scotland: a natural experiment using routine data. Public Health Research, 2017. 5(6): doi: 10.3310/phr05060.

23.              Cattan, S., et al., The Health Effects of Sure Start. 2019, The Institute of Fiscal Studies: London.

24.              Dundas, R., et al., Evaluation of the Healthy Start Voucher Scheme in UK: a natural experiment using the Growing Up in Scotland record linkage study and the Infant Feeding Survey. NIHR Project Protocol, 2016: Available at: https://www.journalslibrary.nihr.ac.uk/programmes/phr/1316410/#/.

25.              Rogers, M., Devolution and Child Poverty Policies: A Four Nations Perspective. 2019, British Academy: London.

February 21