Investigating the impact of the Health in Pregnancy Grant on birth outcomes in England, 2009-2011

Data is requested to assess the impact of the Health in Pregnancy Grant on birthweight and prematurity in England.

A resupply of data is requested due to problems with the data provided under the V0 Agreement. In particular, sex of the baby and the number of previous pregnancies was absent in the original data since data production focused on birth episodes of care rather than delivery episodes. Additionally, the researchers were not aware that selecting a particular field was necessary in order to obtain the selected IMD fields for all years.

In the previous data release, the researchers received a bespoke week of birth variable that is centred on the date of 6 April 2009 as requested. However, in the re-dissemination the researchers require a) a slightly different week of birth variable in addition to the one they currently have, which would be centred around 16 April 2011 (as this was the cut-off date for the abolition of the grant) and b) a binary ‘treatment’ variable equal to one if a baby’s expected date of delivery (EDD) was between 6 April 2009 and 16 April 2011 inclusive, and zero otherwise (as this captures the treatment eligibility rules for the Health in Pregnancy Grant). This is because the researchers need to be completely confident that the week of birth variable as currently provided is accurate as it is absolutely critical to their methodology, and given the confusion over delivery/birth episodes during data production so far they need to be confident that there are not any errors in the construction of the treatment variable.

This request to process HES data is made on the legal basis outlined in the GDPR, Article 6.1.(e). That is, this research establishes a task in the public interest as it is undertaken within the London School of Economics, and is necessary for the university to fulfil its function in the public interest (Article 6(1)(e)). The LSE considers this project to be a task in the public interest due to the potential societal implications of the policy being evaluated - the Health in Pregnancy Grant (2009-2011). It is also impossible to analyse the impact of the grant on prematurity without this data, as gestational age is unavailable elsewhere. In terms of processing special category data (such as ethnicity), the legal basis is Article 9(2)(j) - that is, the research is conducted for archival, research or statistical purposes that are in the public interest. The research is proportionate to the aim pursued (see below). The researchers have agreed to respect the right to data protection at all times, and will carry out suitable and specific measures to safeguard the fundamental rights and the interests of the data subject, including ensuring non-disclosive output and the maintenance of data protection measures when using the data.

This research establishes a public interest due to the societal implications of the policy under evaluation - the Health in Pregnancy Grant, 2009-2011. This grant was a lump sum of £190.00 given to all pregnant women in the UK from the third trimester of pregnancy, regardless of income or work status. In order to receive the grant, pregnant women had to make an appointment with their GP or midwife, when they would be invited to fill out a form. Participants would receive the money via a bank transfer within seven days. When the grant was introduced, policymakers made it clear that it had two main aims: first, to reduce low birthweight; second, to reduce prematurity. Both low birthweight and prematurity are associated with low-income and lack of funds during pregnancy for healthy nutrition and lifestyle.

The rationale of the grant, therefore, was that boosting women's incomes during pregnancy, would facilitate the purchase and consumption of healthier food, invest in healthy lifestyle choices and reduce any financial stress caused by having a baby (for example, the grant could help to cover the large one-off cost of buying a buggy). This would, the government argued, improve birthweight and reduce prematurity. The grant cost over £130 million pounds a year and is estimated to have affected over 700,000 pregnant women a year. It is therefore a matter of public significance both in terms of its human impact and its cost to the public purse. The grant was also abolished in 2011 specifically on the grounds that it was ineffective. No follow-up study of the grant was made, and take-up was not recorded.

A robust analysis of whether the grant was effective, therefore, is needed to investigate how public money is used, and whether it is being used efficiently and effectively. There is very strong evidence that health outcomes during pregnancy - birthweight and prematurity have a significant impact on future health, through childhood and adulthood. The grant was therefore seen as an early intervention policy which aimed to alleviate pressure on NHS services by reducing health problems throughout the lifecycle. By accessing the HES data, it is possible to identify whether the grant achieved this aim.

To assess the impact of the Health in Pregnancy Grant on birthweight and prematurity in England, the London School of Economics will use a methodology called a 'Regression Discontinuity Design' which makes use of the arbitrary eligibility rule that was attached to the grant: all babies with due dates on or after 6 April 2009-16 April 2011 were eligible.

This meant that if the mother had a due date of 5 April 2009, you were not eligible for the grant, but if you had a due date of 6 April 2009 you were. Since it is random whether you have a due date on 5 or 6 April, the LSE can make use of this to isolate the causal impact of the grant on birth outcomes. The methodology essentially mimics the setup of a randomised control trial, which is considered the gold standard in estimating the causal impact of policies. Expected week of birth (calculated using week of birth and gestational age) will be used as a proxy for whether an individual was likely to receive the grant or not.

The LSE researchers require data from 2006-2014 as this includes a 3-year window before and after the grant was introduced. This will enable the researchers to check whether there were any notable changes in birthweight or prematurity over the period and check that the methodology is valid.

The London School of Economics require NHS Digital to create a 'week of birth' variable. The week of birth variable is required in order to calculate women’s due dates, since due dates were eligibility criteria for the Health in Pregnancy Grant (all women with due dates of 6 April 2009 - 16 April 2011 inclusive were eligible for the grant).

Month of birth would be too large a category as it would not enable the researchers to precisely identify how close a baby’s due date was to the eligibility cut-off, and it would therefore significantly reduce the effectiveness of their methodology. Date of birth would be very fine-grained, by contrast, but it would be identifiable. Based on existing research using birth registrations data accessed through the ONS, the researchers have concluded that week of birth is a practicable alternative to using date of birth.’

The data subjects are women who gave birth in England from 2006-2012 and their babies. Since the grant had a cut-off eligibility date, all births before 6 April 2009 act as the control group.

The research builds on previous work conducted by the London School of Economics (LSE) using ONS Birth Registrations microdata (provided through the ONS's Secure Research Service). This work was awarded a Distinction and an academic prize as part of a Masters programme undertaken by one of the researchers at the LSE. This proposed follow up to the research has secured funding from the Social Policy Department at the LSE. It will be a standalone research project, unconnected to any wider programmes of work.

This is a small standalone project, which is funded and will be disseminated on its own. However, it does run alongside a similarly-themed, wider programme of work within the Centre for Analysis of Social Exclusion (CASE) at LSE on the impact of recent social policy changes on outcomes. The Funder will not place any restriction or limitations on any publications and there is no time restriction on the funding.

There is a separate NIHR funded research project of a broadly similar scope by Leyland et al (2017) that investigated the impact of the Health in Pregnancy Grant in Scotland which found no significant effects on birthweight. The proposed project will improve on this existing evidence for reasons both of methodology and of coverage. First, Leyland et al (2017) use an ‘interrupted time series’ methodology, which does not allow them to account for contemporary events that may have biased their results (in particular the economic crisis of 2008-09, which may have led to downward bias in their estimation of birthweight effects). By contrast, this project uses a regression discontinuity design which focuses on births just before and just after the eligibility cut-off, enabling more credible causal estimates to be drawn. Second, to date no research has been conducted in England about the impact of the Health in Pregnancy Grant. This project would fill this gap.

The dataset requested and the variables selected within it are essential for this research. The LSE researchers need access to anonymised, record-level data including crucially: week of birth of the baby, birthweight and gestational age. The ONS's birth registrations data do not include gestational age, so it is impossible to assess the impact of the grant on prematurity without the HES data.

'Week of birth' will be used to avoid having to use identifiable (full date of birth) in calculating due dates and to minimise the amount of data required. Due dates will be used as a proxy to separate the sample into a treatment and control group. Gestational age is needed for two reasons: first, to calculate due dates (in conjunction with week of birth); second, to construct an outcome variable for prematurity. Other important outcome variables are birthweight, first antenatal assessment date (to see whether the grant incentivised women to seek medical advice earlier during pregnancy), healthy neonate indicator, postnatal stay, neonatal level of care, and the well baby flag. Other useful variables to serve as controls are sex of the baby, birth status, number of babies, number of previous pregnancies, ethnic category, and mother's age at birth. To see whether the impact of the grant differed depending on levels of deprivation, the researchers also need access to the Index of Multiple Deprivation and Lower Super Output Area. This will enable them to see how the effect of the grant varied across different groups (by IMD, ethnicity, etc), given that it may have mattered more to more disadvantaged groups.

There is another grant available to qualifying mothers called the Sure Start Maternity Grant. The Sure Start Maternity Grant was introduced in 2000 as a means-tested cash transfer for women in receipt of benefits (by contrast, the Health in Pregnancy Grant was universal). The SSMG was expanded in April 2000, December 2000, and April 2002, before being restricted to first-born children only in April 2011. This separate grant has been considered as a potential contributor to the research outcomes. This project's methodology will enable LSE to disentangle the marginal effect of the Health in Pregnancy Grant and the Sure Start Maternity Grant due to the unique EDD cut-off that applied to the HPG (6 April 2009): no changes to SSMG were made at this cut-off. LSE will also be able to isolate the effect of the abolition of the HPG on 16 April 2011 from the restriction of the SSMG to the first-born child on 11 April 2011 by looking at effects separately for first-borns.

However, with the data requested under this Agreement, LSE will not be able to identify the effect of the introduction of the SSMG in 2000 (and its expansion later in 2000 and in 2002) on birthweight and compare with the effect of the HPG. This is something that is currently being considered by LSE.

The purpose of the research cannot be achieved in a less intrusive way, as record-level data is necessary to implement this analysis of the impact of the grant on birthweight and prematurity. No identifiable or sensitive variables are being requested.

The data has not been narrowed by demographics (e.g. age), clinical factors (e.g. diagnosis/procedure) or live/singleton birth status, since the aims of the Health in Pregnancy Grant were broad-based and universal, and as such were intended to improve birth outcomes for all women and babies, regardless of these characteristics. The researchers would like to investigate and compare the impact of the grant on stillbirths, live births, and multiple births, as it may be that the grant affected these groups differently.

In terms of episodes, only maternity episodes are required - and specifically those variables that have been previously specified. The unborn child and neonatal records are fundamental to the purpose of this research as they enable the measurement of the health and status of the baby, which was the main target of the Health in Pregnancy Grant.

In selecting the necessary variables, the researchers have been careful to minimise the amount of fields necessary to achieve the purpose of the research. Indeed, the request does not include any identifiable/high risk field.

The data controller is the London School of Economics. Data is processed by the LSE Secure Research Computing Governance Group who are substantive employees of LSE within the secure environment. The VIRTUS datacentre provides physical storage of the data only. VIRTUS Holdco Ltd do not access data held under this agreement as they only supply the building. Therefore, any access to the data held under this agreement would be considered a breach of the agreement. This includes granting of access to the database(s) containing the data.

The research is funded by the Social Policy Department at the LSE, and their role will be to help publish the findings of the research. This will include a public seminar, and the publication of a free working paper and/or copy of a peer-reviewed journal article via LSE Research Online.

The main output from the data analysis will be a submission to peer-reviewed academic journals, specifically the Journal of Health Economics alternatively the Journal of Public Health

All figures produced will be aggregate figures (e.g. coefficients, means) and the researchers will ensure that each figure contains a sufficient cell count (e.g. the ONS's requirement of 10) to avoid disclosure of patient data.

In addition to the peer-reviewed journal submission, the researchers intend to hold a public seminar at the Centre for Analysis of Social Exclusion (CASE) at the LSE at which the findings of the research will be presented to a range of stakeholders, policymakers and members of the public. The peer-reviewed journal is likely to require a fee from readers for access, but the full content will be made available for free via LSE Research Online, as part of LSE’s commitment to the Open Access model.

The researchers may also write blogs or other short articles summarising the findings to ensure they communicate the findings of the research to wider audiences, including interested groups and civil society. The researchers' preliminary research from ONS Birth Registrations data suggests the Health in Pregnancy Grant had a positive impact on birth weight in England and Wales. Findings from this research will inform whether the Health in Pregnancy Grant was effective.

The target date for the submission of the peer-reviewed journal and the CASE Seminar is 2022.

LSE are also looking to disseminate outputs to relevant patient-focussed charities, including (but not limited to) BLISS, who work with families who have low birth weight babies. A list of patient-focussed charities to include in dissemination is soon to be drawn up.

All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide.

Data was disseminated in late December 2020. The applicant has therefore not had sufficient time with the data to deliver any public benefits so far. However, data cleaning and analysis have begun and it is intended that extending this agreement will enable this work to yield the public benefits outlined above.

This request to process HES data is made on the legal basis outlined in the GDPR, Article 6.1.(e). That is, this research establishes a task in the public interest as it is undertaken within the London School of Economics, and is necessary for the university to fulfil its function in the public interest (Article 6(e)). The LSE considers this project to be a task in the public interest due to the potential societal implications of the policy being evaluated - the Health in Pregnancy Grant (2009-2011). It is also impossible to analyse the impact of the grant on prematurity without this data, as gestational age is unavailable elsewhere. In terms of processing special category data (such as ethnicity), the legal basis is Article 9(2)(j) - that is, the research is conducted for archival, research or statistical purposes that are in the public interest. The research is proportionate to the aim pursued (see below). The researchers have agreed to respect the right to data protection at all times, and will carry out suitable and specific measures to safeguard the fundamental rights and the interests of the data subject, including ensuring non-disclosive output and the maintenance of data protection measures when using the data.

This research establishes a public interest due to the societal implications of the policy under evaluation - the Health in Pregnancy Grant, 2009-2011. This grant was a lump sum of £190.00 given to all pregnant women in the UK from the third trimester of pregnancy, regardless of income or work status. In order to receive the grant, pregnant women had to make an appointment with their GP or midwife, when they would be invited to fill out a form. Participants would receive the money via a bank transfer within seven days. When the grant was introduced, policymakers made it clear that it had two main aims: first, to reduce low birthweight; second, to reduce prematurity. Both low birthweight and prematurity are associated with low-income and lack of funds during pregnancy for healthy nutrition and lifestyle.

The rationale of the grant, therefore, was that boosting women's incomes during pregnancy, would facilitate the purchase and consumption of healthier food, invest in healthy lifestyle choices and reduce any financial stress caused by having a baby (for example, the grant could help to cover the large one-off cost of buying a buggy). This would, the government argued, improve birthweight and reduce prematurity. The grant cost over £130 million pounds a year and is estimated to have affected over 700,000 pregnant women a year. It is therefore a matter of public significance both in terms of its human impact and its cost to the public purse. The grant was also abolished in 2011 specifically on the grounds that it was ineffective. No follow-up study of the grant was made, and take-up was not recorded.

A robust analysis of whether the grant was effective, therefore, is needed to investigate how public money is used, and whether it is being used efficiently and effectively. There is very strong evidence that health outcomes during pregnancy - birthweight and prematurity have a significant impact on future health, through childhood and adulthood. The grant was therefore seen as an early intervention policy which aimed to alleviate pressure on NHS services by reducing health problems throughout the lifecycle. By accessing the HES data, it is possible to identify whether the grant achieved this aim.

To assess the impact of the Health in Pregnancy Grant on birthweight and prematurity in England, the London School of Economics will use a methodology called a 'Regression Discontinuity Design' which makes use of the arbitrary eligibility rule that was attached to the grant: all babies with due dates on or after 6 April 2009-16 April 2011 were eligible.

This meant that if the mother had a due date of 5 April 2009, you were not eligible for the grant, but if you had a due date of 6 April 2009 you were. Since it is random whether you have a due date on 5 or 6 April, the LSE can make use of this to isolate the causal impact of the grant on birth outcomes. The methodology essentially mimics the setup of a randomised control trial, which is considered the gold standard in estimating the causal impact of policies. Expected week of birth (calculated using week of birth and gestational age) will be used as a proxy for whether an individual was likely to receive the grant or not.

The LSE researchers require data from 2006-2014 as this includes a 3-year window before and after the grant was introduced. This will enable the researchers to check whether there were any notable changes in birthweight or prematurity over the period and check that the methodology is valid.

The London School of Economics require NHS Digital to create a 'week of birth' variable. The week of birth variable is required in order to calculate women’s due dates, since due dates were eligibility criteria for the Health in Pregnancy Grant (all women with due dates of 6 April 2009 - 16 April 2011 inclusive were eligible for the grant).

Month of birth would be too large a category as it would not enable the researchers to precisely identify how close a baby’s due date was to the eligibility cut-off, and it would therefore significantly reduce the effectiveness of their methodology. Date of birth would be very fine-grained, by contrast, but it would be identifiable. Based on existing research using birth registrations data accessed through the ONS, the researchers have concluded that week of birth is a practicable alternative to using date of birth.’

The data subjects are women who gave birth in England from 2006-2012 and their babies. Since the grant had a cut-off eligibility date, all births before 6 April 2009 act as the control group.

The research builds on previous work conducted by the London School of Economics (LSE) using ONS Birth Registrations microdata (provided through the ONS's Secure Research Service). This work was awarded a Distinction and an academic prize as part of a Masters programme undertaken by one of the researchers at the LSE. This proposed follow up to the research has secured funding from the Social Policy Department at the LSE. It will be a standalone research project, unconnected to any wider programmes of work.

This is a small standalone project, which is funded and will be disseminated on its own. However, it does run alongside a similarly-themed, wider programme of work within the Centre for Analysis of Social Exclusion (CASE) at LSE on the impact of recent social policy changes on outcomes. The Funder will not place any restriction or limitations on any publications and there is no time restriction on the funding given that the work is due to be completed in July 2020.

The dataset requested and the variables selected within it are essential for this research. The LSE researchers need access to anonymised, record-level data including crucially: week of birth of the baby, birthweight and gestational age. The ONS's birth registrations data do not include gestational age, so it is impossible to assess the impact of the grant on prematurity without the HES data.

'Week of birth' will be used to avoid having to use identifiable (full date of birth) in calculating due dates and to minimise the amount of data required. Due dates will be used as a proxy to separate the sample into a treatment and control group. Gestational age is needed for two reasons: first, to calculate due dates (in conjunction with week of birth); second, to construct an outcome variable for prematurity. Other important outcome variables are birthweight, first antenatal assessment date (to see whether the grant incentivised women to seek medical advice earlier during pregnancy), healthy neonate indicator, postnatal stay, neonatal level of care, and the well baby flag. Other useful variables to serve as controls are sex of the baby, birth status, number of babies, number of previous pregnancies, ethnic category, and mother's age at birth. To see whether the impact of the grant differed depending on levels of deprivation, the researchers also need access to the Index of Multiple Deprivation and Lower Super Output Area. This will enable them to see how the effect of the grant varied across different groups (by IMD, ethnicity, etc), given that it may have mattered more to more disadvantaged groups.

The purpose of the research cannot be achieved in a less intrusive way, as record-level data is necessary to implement this analysis of the impact of the grant on birthweight and prematurity. No identifiable or sensitive variables are being requested.

The data has not been narrowed by demographics (e.g. age), clinical factors (e.g. diagnosis/procedure) or live/singleton birth status, since the aims of the Health in Pregnancy Grant were broad-based and universal, and as such were intended to improve birth outcomes for all women and babies, regardless of these characteristics. The researchers would like to investigate and compare the impact of the grant on stillbirths, live births, and multiple births, as it may be that the grant affected these groups differently.

In terms of episodes, only maternity episodes are required - and specifically those variables that have been previously specified. The unborn child and neonatal records are fundamental to the purpose of this research as they enable the measurement of the health and status of the baby, which was the main target of the Health in Pregnancy Grant.

In selecting the necessary variables, the researchers have been careful to minimise the amount of fields necessary to achieve the purpose of the research. Indeed, the request does not include any identifiable/high risk field.

The data controller is the London School of Economics. The Data Processor will be the LSE specifically the LSE Secure Research Computing Governance Group, whose secure environment will be used for the data controllers to access the data. The VIRTUS datacentre provides physical storage of the data only - the processing is conducted by LSE only. The research is funded by the Social Policy Department at the LSE, and their role will be to help publish the findings of the research. This will include a public seminar, and the publication of a free working paper and/or copy of a peer-reviewed journal article via LSE Research Online.

The main output from the data analysis will be a submission to peer-reviewed academic journals, specifically the Journal of Health Economics alternatively the Journal of Public Health

All figures produced will be aggregate figures (e.g. coefficients, means) and the researchers will ensure that each figure contains a sufficient cell count (e.g. the ONS's requirement of 10) to avoid disclosure of patient data.

In addition to the peer-reviewed journal submission, the researchers intend to hold a public seminar at the Centre for Analysis of Social Exclusion (CASE) at the LSE at which the findings of the research will be presented to a range of stakeholders, policymakers and members of the public. The peer-reviewed journal is likely to require a fee from readers for access, but the full content will be made available for free via LSE Research Online, as part of LSE’s commitment to the Open Access model.

The researchers may also write blogs or other short articles summarising the findings to ensure they communicate the findings of the research to wider audiences, including interested groups and civil society. The researchers' preliminary research from ONS Birth Registrations data suggests the Health in Pregnancy Grant had a positive impact on birth weight in England and Wales. Findings from this research will inform whether the Health in Pregnancy Grant was effective.

The target date for the submission of the peer-reviewed journal and the CASE Seminar is 2021.

All outputs will contain only data that is aggregated with small numbers suppressed in line with the HES Analysis Guide.

Yielded Benefits is not a requirement for new applications.