Long term impact of giving antibiotics before skin incision versus after cord clamping on children born by caesarean section: longitudinal study based on UK electronic health records

Text removed is struck through; text added is underlined. Unchanged paragraphs are summarised rather than repeated.

Objective for processing

The purpose of this government-funded carried out by the University of Birmingham, is to determine whether there is an association between the change in antibiotic prophylaxis in Caesarian section delivery and the incidence of subsequent admission of the child for a number of diseases. Before 2011, the National Institute for Health and Care Excellence (NICE) Clinical Guideline 132 on Caesarean Section (CS) advised administering intravenous prophylactic antibiotics for women undergoing CS after the baby’s cord had been clamped to prevent exposing the baby to antibiotics. In 2011, the guidance was changed to recommend giving antibiotics to women undergoing CS prior to skin incision. This was based on evidence that earlier administration reduces maternal infectious morbidity; although most such infections are mild and respond well to treatment. The current Cochrane review summarises data from 10 randomised trials (5,041 women) which showed a near halving of risk of all maternal infection (43%, 95% confidence interval (CI) 28-55%), endometritis (46%, CI 21-64%), and wound infection (41%, CI 19-56%) compared with giving antibiotics after clamping the baby’s umbilical cord. This Data Sharing Agreement permits the retention and processing of the data provided under previous iterations of this Agreement. Hospital Episode Statistics (HES) Admitted Patient Care (APC) data provided by NHS Digital will include pseudonymised records only for mothers/babies who have had a birth/delivery event between 1st April 2005 and 31st March 2018. The University will act as the sole controller and processor of the admissions data that are the subject of this application. The purpose of this government-funded research, carried out by the University of Birmingham, is to determine whether there is an association between the change in antibiotic prophylaxis in Caesarean section delivery and the incidence of subsequent admissions of the child for a number of diseases. Before 2011, the National Institute for Health and Care Excellence (NICE) Clinical Guideline 132 on Caesarean Section (CS) advised administering intravenous prophylactic antibiotics for women undergoing CS after the baby’s cord had been clamped to prevent exposing the baby to antibiotics. In 2011, the guidance was changed to recommend giving antibiotics to women undergoing CS prior to skin incision. This was based on evidence that earlier administration reduces maternal infectious morbidity; although most such infections are mild and respond well to treatment. The current Cochrane review summarises data from 10 randomised trials (5,041 women) which showed a near halving of risk of all maternal infection (43%, 95% confidence interval (CI) 28-55%), endometritis (46%, CI 21-64%), and wound infection (41%, CI 19-56%) compared with giving antibiotics after clamping the baby’s umbilical cord. Hospital Episode Statistics (HES) Admitted Patient Care (APC) data provided by NHS Digital will include pseudonymised records only for mothers/babies who have had a birth/delivery event between 1st April 2005 and 31st March 2018. The University of Birmingham are acting as the sole controller and processor of the data that are the subject of this Agreement. [4 paragraphs unchanged] Pseudonymised HES APC data provided by NHS Digital will be limited to women who have given birth in hospital between 1st April 2005 and 31st March 2018, and for those babies who are born in hospital in the same period. For the mothers, the HES data will be limited to episodes which start in the period from 30 days before birth to 56 days after birth. For the babies, HES data will be supplied for all episodes that begin before the child's 5th birthday. The period of data (2005-2018) covers a period of time either side of the change in antibiotic guidance, in order to understand the effect the change has had. Data will be supplied both for Caesarean Section deliveries (as the group of interest) as well as other delivery methods (to act as a comparator group). As the study is following babies born in this period forward in time, the study also need admissions data for all children under 5 (i.e. STARTAGE <5) in financial year 2018-2019. The key variable in the analysis is whether the birth occurred before or after a particular chance in practice. The change was introduced at different times in different providers, and in some cases different sites adopted changes at different times independent of Trust policy. The study has determined practice change dates for most of the maternity sites in England by individually contacting them and capturing the dates needed. The study now needs to attach these change dates to the birth records already held. The study team have involved the public throughout the development of the proposal. This has reconfirmed the importance of the research question, particularly: • the importance of assuring the baby’s health as a main priority when deciding on delivery options; • that uncertainty as to whether antibiotics given around the time of birth have an impact on children later in life should be resolved; • that a robust study design is required to ensure the validity of the findings; • a broad scope of important health and other outcomes which need to be considered; • that the project needs to clearly communicate findings in terms of risks and benefits; • that the findings regarding prophylactic antibiotics for CS should form part of the wider discussion regarding risks and benefits of medications in pregnancy. Two lay parent representatives are members of the Project Management Group and there is also an independent parent representative as a member of the Project Steering Group. Lay parent involvement will be particularly helpful to ensure that the study team research the outcomes that are important for parents and that the messages as the result of the study are helpful when making decisions about the timing of prophylactic antibiotic administration. Clear communication and publicising of key findings and messages are priorities of the study. The study team will achieve this by: • holding one PPI workshop to gain agreement on which outcomes to focus on early in the project; • holding another PPI workshop towards the end of the project to co-produce messages for dissemination via clinical networks, patient organisations and the media. The Project Management Group comprising of all co-applicants and the second PPI advisor will meet as required and formally at least bi-monthly throughout the project. The Project Steering Group will oversee, advise and monitor the project. It will be chaired by an independent chair ( a Professor of Perinatal and Paediatric Epidemiology, from the University of Leicester). The Project Steering Group will meet three times, at the beginning, middle and towards the end of the project. The Project Management Group and Project Steering Group offer advice regarding how the study should be managed but do not make any decisions regarding the collection or processing of the data under this Agreement. [3 paragraphs unchanged]

Processing activities

Pseudonymised HES APC data provided by NHS Digital will be limited to women who have given birth in hospital between 1st April 2005 and 31st March 2018, and for those babies who are born in hospital in the same period. For the mothers, the HES data will be limited to episodes which start in the period from 30 days before birth to 56 days after birth. For the babies, HES data will be supplied for all episodes that begin before the child's 5th birthday. The period of data (2005-2018) covers a period of time either side of the change in antibiotic guidance, in order to understand the effect the change has had. Data will be supplied both for Caesarian Section deliveries (as the group of interest) as well as other delivery methods (to act as a comparator group). As the study is following babies born in this period forward in time, the study also need admissions data for all children under 5 (i.e. STARTAGE <5) in financial year 2018-2019. The University of Birmingham will provide NHS Digital with an EPIKEY to allow NHS Digital to provide UoB with a bridge file consisting of EPIKEY and SITETRET. PROCODE, if used as intended should identify a provider using the full five character code of the treatment provider that should be able to identify the individual site. This was the field that the study requested to perform this function. However, it has been found that usual coding practice is variable and frequently the two character suffix in the code, used to identify a site of treatment are often not used. In fact this would give the study sufficient specificity in less than 25% of cases and badly compromise the sample size. SITETRET, a more specific code, is also imperfectly applied, but would offer much greater accuracy and allow the study to attribute a change date to approximately 75% of the sample which would be methodologically acceptable. This application requests one additional field in addition to the data the study is already using to complete this work, that of SITETRET. The key variable in the analysis is whether the birth occurred before or after a particular chance in practice. The change was introduced at different times in different providers, and in some cases different sites adopted changes at different times independent of Trust policy. The study has determined practice change dates for most of the maternity sites in England by individually contacting them and capturing the dates needed. The study now needs to attach these change dates to the birth records already held. The University of Birmingham will provide NHS Digital with an EPIKEY to allow NHS Digital to provide UoB with a bridge file consisting of EPIKEY and SITETRET. PROCODE, if used as intended should identify a provider using the full five character code of the treatment provider that should be able to identify the individual site. This was the field that the study requested to perform this function. However, it has been found that usual coding practice is variable and frequently the two character suffix in the code, used to identify a site of treatment are often not used. In fact this would give the study sufficient specificity in less than 25% of cases and badly compromise the sample size. SITETRET, a more specific code, is also imperfectly applied, but would offer much greater accuracy and allow the study to attribute a change date to approximately 75% of the sample which would be methodologically acceptable. [4 paragraphs unchanged] An extensive set of data cleaning algorithms will be run to identify duplicate episodes and exclude spells where the degree of data missingness missing-ness would make even probabilistic linkage difficult to undertake. Researchers will then link [126 words unchanged] place (intended), status of person conducting delivery, postcode district and ethnic group). [7 paragraphs unchanged]

Expected output

Due to interruptions in normal working during the COVID-19 pandemic, and staff absences within the study team, there has been significant disruption to planned timescales. The study team have secured a non-funded extension until November 2022 to complete the final report to funders and complete all of the planned outputs. [1 paragraph unchanged] The following outputs will be produced: have been produced or are planned: 1) Production of lay summary of the findings for wider dissemination, including a dissemination event at the end of the project with lay, clinical stakeholders and professional organisations. Target date 01/07/2020 1) A report draft was submitted to funders (NIHR) in February 2021 and has been peer reviewed (reference HTA 16/150/01). The findings have been well received and the project team are preparing their amended version in response to reviewers comments which were received in April 2021. The final report is still undergoing a review by the NIHR which is expected by the end of 2021. 2) A dissemination event to the clinical directors of maternity units. Target date May 2020 2) Production of lay summary of the findings for wider dissemination, including a dissemination event at the end of the project with lay, clinical stakeholders and professional organisations, was originally scheduled for July 2020. This has been deferred owing to the pandemic. A dissemination event to the clinical directors of maternity units has also been postponed. The study team now plan to run these events early in 2022. 3) At least one peer reviewed publication to be aimed at a journal of general medical interest. Target date for submission is July 2020 3) A paper reporting the finding to the team submission is nearing completion and will be submitted to the British Medical Journal by the end of 2021. Any statistical outputs from the research will be highly derived, and principally consist of correlation co-efficients and odds ratios. A correlation co-efficient is a value that measures the strength and a direction of a relationship between two variables in a data set. An odds ratio is a ratio of the odds of an event happening in one population divided by the odds of an event happening in a comparator population. In this case, researchers are interested in the odds of admission in childhood for acute asthma and other conditions in children exposed to different modes of delivery and c-section deliveries pre and post guideline changes. Another paper on detailed results of subgroup analyses regarding the impact of prophylactic pre‐incision antibiotics on child health and maternal infectious morbidity is planned and will be submitted to an appropriate specialty journal such as the British Journal of Obstetrics and Gynaecology. Publications: the team anticipate at least two peer‐reviewed publications in such journals as BMJ and BJOG, one reporting the overall results and their implications, and another paper on detailed results of subgroup analyses regarding the impact of prophylactic pre‐incision antibiotics on child health and maternal infectious morbidity. The statistical outputs from the research are highly derived, and principally consist of correlation co-efficients and odds ratios. A correlation co-efficient is a value that measures the strength and a direction of a relationship between two variables in a data set. An odds ratio is a ratio of the odds of an event happening in one population divided by the odds of an event happening in a comparator population. In this case, researchers are interested in the odds of admission in childhood for acute asthma and other conditions in children exposed to different modes of delivery and C-section deliveries pre and post guideline changes. Patient information resource: in collaboration with the PPI co‐applicant, second PPI advisor and during the PPI workshop, researchers will produce a lay are producing an information resource summarising the key findings with respect to the benefits and [21 words unchanged] the relevant stakeholders, including the patient organisations and clinical directors for maternity. [1 paragraph unchanged] The study team continue to require access to the data under this Agreement as the paper planned for submission to the British Medical Journal with the final results may require (depending on the reviewers’ comments) further or slightly different analysis.

Expected measurable benefits

The latest data (from 2016) suggest that 24.5% of all delivery events involve Caesarian Caesarean section, with over 120,000 babies being delivered this way. This study will [98 words unchanged] of evidence regarding the long term effects of such treatment on children. [1 paragraph unchanged]

Benefits reported

The study has discovered from the arm which does not use HES [22 words unchanged] slightly less likelihood of them developing auto-immune conditions that present in primary care care. This is both interesting and counter-intuitive as the study expected to find an increase. It is now imperative that the HES arm of the study is completed to provide confirmation of these findings, however this cannot be done with appropriate statistical rigor without the SITTRET field for the reasons stated in section 5b. The analysis of the data under this Agreement has demonstrated that there is no evidence that the change in anti-biotic protocols for C-section deliveries is associated with illnesses of the children delivered in early life course. It had been hypothesised that in-utero exposure to antibiotics immediately prior to birth may increase the likelihood of developing auto-immune illnesses as this was thought to be highly biologically plausible and limited evidence from some small observational studies had suggested that this may be the case. This has caused heightened anxiety on the part of expectant mothers who undergo C-section deliveries, as well as concern in the clinical community. The study team found no difference in this very large study between exposed and unexposed babies. The size of the study, involving some 8 million births was very highly powered, so the chances that there is actually an effect and, by chance, the team have missed the signal, is infinitesimally small. This is of particular interest to NICE whose guideline is affected by this work. The guideline will not be changed by the study, as it effectively validates the safety of the change in practice, but it will probably add to the information offered by NICE in support of it.

Text removed is struck through; text added is underlined. Unchanged paragraphs are summarised rather than repeated.

Objective for processing

The purpose of this government-funded research (carried carried out by the University of Birmingham) Birmingham, is to determine whether there is an association between the change in [139 words unchanged] CI 19-56%) compared with giving antibiotics after clamping the baby’s umbilical cord. Data Hospital Episode Statistics (HES) Admitted Patient Care (APC) data provided by NHS Digital will include pseudonymised records only for mothers/babies who have had a birth/delivery event between 1st April 2005 and 31st March 2018. The University will act as the sole controller and processor of the admissions data that are the subject of this application. [4 paragraphs unchanged] The GDPR Legal basis for this application: Article 6(1)(e): processing is necessary for the performance of a task carried out in the public interest or in the exercise of official authority vested in the controller; Article 9(2)(j): processing is necessary for archiving purposes in the public interest, scientific or historical research purposes or statistical purposes in accordance with Article 89(1) based on Union or Member State law which shall be proportionate to the aim pursued, respect the essence of the right to data protection and provide for suitable and specific measures to safeguard the fundamental rights and the interests of the data subject.

Processing activities

Data Pseudonymised HES APC data provided by NHS Digital will be limited to women who have given [110 words unchanged] as well as other delivery methods (to act as a comparator group). As the study is following babies born in this period forward in time, the study also need admissions data for all children under 5 (i.e. STARTAGE <5) in financial year 2018-2019. Researchers will use this data to create three distinct HES data sets. Firstly to identify birth events of babies born in English hospitals between April 2005 and March 2018 (hereafter called the baby data set). Researchers will build continuous inpatient spells from these data, after having removed duplicate records or records with critical data items missing. The team will go on to build a second set of continuous inpatient spells containing delivery events (hereinafter called the delivery data set). This application requests one additional field in addition to the data the study is already using to complete this work, that of SITETRET. The key variable in the analysis is whether the birth occurred before or after a particular chance in practice. The change was introduced at different times in different providers, and in some cases different sites adopted changes at different times independent of Trust policy. The study has determined practice change dates for most of the maternity sites in England by individually contacting them and capturing the dates needed. The study now needs to attach these change dates to the birth records already held. The University of Birmingham will provide NHS Digital with an EPIKEY to allow NHS Digital to provide UoB with a bridge file consisting of EPIKEY and SITETRET. PROCODE, if used as intended should identify a provider using the full five character code of the treatment provider that should be able to identify the individual site. This was the field that the study requested to perform this function. However, it has been found that usual coding practice is variable and frequently the two character suffix in the code, used to identify a site of treatment are often not used. In fact this would give the study sufficient specificity in less than 25% of cases and badly compromise the sample size. SITETRET, a more specific code, is also imperfectly applied, but would offer much greater accuracy and allow the study to attribute a change date to approximately 75% of the sample which would be methodologically acceptable. After receipt of data from NHS Digital, there will be no further flows of data. Researchers will use these data to create three distinct HES data sets. Firstly to identify birth events of babies born in English hospitals between April 2005 and March 2018 (hereafter called the baby data set). Researchers will build continuous inpatient spells from these data, after having removed duplicate records or records with critical data items missing. The team will go on to build a second set of continuous inpatient spells containing delivery events (hereinafter called the delivery data set). [7 paragraphs unchanged] The data will not be used for commercial purposes, not provided in record level form to any third party and not used for direct marketing. marketing and there will be no attempt to identify individuals. Processing will be carried out by substantive employees from the University of Birmingham only. [1 paragraph unchanged]

Expected output

[2 paragraphs unchanged] 1) Production of lay summary of the findings for wider dissemination, including [6 words unchanged] of the project with lay, clinical stakeholders and professional organisations. Target date 01/04/2020 01/07/2020 2) A dissemination event to the clinical directors of maternity units. Target date 20/02/2020 May 2020 3) At least one peer reviewed publication to be aimed at a journal of general medical interest. Target date: 01/04/2020 date for submission is July 2020 [4 paragraphs unchanged]

Benefits reported

Yielded Benefits is not a requirement for new applications. The study has discovered from the arm which does not use HES data that there is no extra risk to babies from the change in cord clamping procedure, in fact there appears to be slightly less likelihood of them developing auto-immune conditions that present in primary care This is both interesting and counter-intuitive as the study expected to find an increase. It is now imperative that the HES arm of the study is completed to provide confirmation of these findings, however this cannot be done with appropriate statistical rigor without the SITTRET field for the reasons stated in section 5b.

Unchanged: Expected measurable benefits.