Missingness as a marker of health inequality: implications of data gaps for maternal and child outcomes

“Behind missing information in maternity care may be barriers to sharing, hearing, recording or carrying information across services.”
— Reflecting themes from My Data Story, a public and community engagement project.

This study is being led by Dr Talia Hubble and Prof Katie Harron (UCL) and was funded by NIHR. The study started in 2026 and aims to finish in 2027.

‍This project was reviewed and approved through ECHILD's data access process. Learn how ECHILD data is kept safe.‍

  • We want to know whether missing information in routinely collected records might tell us something important about the people those records belong to. In particular, we want to understand whether incomplete records are more common among women experiencing social disadvantage or fragmented care, and whether missing information is associated with poorer outcomes for mothers and babies.

  • Missing information is usually treated as a problem to be overcome when analysing data. But it may also contain important information in its own right. For example, information may be more likely to be missing when women face barriers to accessing care, move between services, book late for maternity care or have more fragmented contact with the health system.

    By understanding who is not fully represented in large datasets - and why - we aim to improve how health inequalities are identified and addressed. Our goal is to increase visibility in the dataset amongst underserved groups.

  • We will use linked records of births in ECHILD to look at patterns of missingness and investigate their relationship with maternal and neonatal outcomes.

Find out more

  • Why does this matter?

    Information that is routinely collected when women receive maternity care is used for research and to help plan and improve health services. However, these records are not always complete. Information such as ethnicity, social circumstances or details about someone’s pregnancy and care may be missing.

    Missing information is usually considered a problem with the data. But it may also tell us something important about access and experience of healthcare.

    If some groups are more likely to have incomplete records than others, they may then be less well represented in the research and evidence used to make decisions about maternity care. This could mean that existing health inequalities are underestimated or overlooked.

    What do we want to find out?

    We want to understand whether missing information is more common among some groups of women, and identify whether it is associated with social disadvantage, difficulties accessing care or other indicators of inequality.

    We also want to understand if women with incomplete records are more likely to experience poorer outcomes for themselves or their babies.

    What we will do

    We will use birth and delivery records in ECHILD from 2015-2023 to assess the extent of missing data, and whether missing data is more common in particular groups (for example, those facing social disadvantage).

    We will then explore whether missingness is related to maternal or neonatal outcomes, such as severe maternal complications, preterm birth, low birthweight, or stillbirth.

  • Background/aims: 

    This study aims to investigate whether there are non-random patterns to missingness in the ECHILD dataset and if these patterns may act as markers of social vulnerability. We will further investigate whether the extent and pattern of missingness are associated with adverse maternal and neonatal outcomes.

    The overarching goal is to shift the perception of missingness from being seen as a statistical inconvenience to being understood as an informative signal of exclusion, disadvantage, and risk.

    Methods and analysis:

    This will be a retrospective cohort study using population-level data from the ECHILD database. We will:

    (1)    Quantify the extent and distribution of missing data in key maternal and child health variables across linked health datasets.

    (2)    Explore associations between missingness and known sociodemographic factors, healthcare access indicators, and other markers of social vulnerability including maternal education and social care experience

    (3)    Evaluate whether missingness is independently associated with adverse maternal and neonatal outcomes

    (4)    Evaluate whether the magnitude of missingness for each case predicts worse outcomes

    Descriptive and multivariable analyses will be used to characterise patterns of missingness and their association with maternal and neonatal outcomes (such as severe maternal morbidity, preterm birth, low birthweight, stillbirth, and admission to the neonatal unit).

    Stratified analyses will assess variation by sociodemographic subgroups. Sensitivity analyses will also be used to compare results with and without inclusion of records with missing data and with missing data re-coded to the extremes of variables, to highlight the consequence of exclusion. These analyses will explore the extent to which conventional approaches to missing data could alter estimates of health inequalities.

    ECHILD data used:

    HES: Admitted Patient Care 2015-2023; mother-baby link

    NPD: Key Stage 4, CiN, CLA.

  • The study will generate a peer-reviewed research publication describing the social patterning and prognostic significance of missingness in linked maternal and child health data.

    PPIE activities:

    My Data Story was a public involvement and community engagement project developed with three community organisations supporting women who had given birth in the UK and were from underserved backgrounds. Through creative workshops, women explored why information may be missing from maternity records and co-produced a framework describing personal, relational and system-level missingness.

    The project highlighted how missing information may reflect issues including trust and safety, communication barriers, fragmented care, mobility and disconnected health systems. These findings directly informed the aims of this ECHILD study, which will test at population level whether missingness clusters among women experiencing social or structural disadvantage and whether it is associated with poorer maternal and neonatal outcomes.