Reading the Evidence: Registers, Cohorts and Surveys on Bereaved Children

Only a few countries, mainly in the Nordic region and the Netherlands, can count how many children have a parent die, because their population registers link a death to the children left behind. Almost everywhere else, including the UK and the US, the figures are modelled estimates. The strongest evidence on what happens next also comes from registers, and even they can adjust only for what they record.

Behind every figure in this library is a child whose mum or dad has died. The numbers shape what gets funded and which services are planned, but only if people know what kind of number they are looking at. This study explains the main kinds of evidence on bereaved children and what each can and cannot tell us, using worked examples from our research library on bereaved children.

A few countries count bereaved children; most can only estimate

To count how many children have a parent die, a country needs to know, for each death, whether the person who died had children. Many death registration systems record facts about the person who died and nothing about who they leave behind. Population registers give every resident a personal number and record who is the parent of whom, so a parent’s death can be linked to their children.

Sweden publishes the result every year. In 2025, 2,717 children aged 0–17 had one or both parents die, 1.3 in every 1,000 children; in 2021 the figure was 3,198, or 1.5 per 1,000^[1]. Statistics Netherlands reports that just over 12,000 under-25s have had a parent die each year on average since 2000^[3]. Denmark’s Civil Registration System, set up in 1968, gives every resident a ten-digit personal number that allows “virtually complete long-term follow-up on emigration and death”^[4].

The UK is typical of the many countries without such links. “Data are not officially collected on the number of children bereaved each year in the UK”, so its 46,300 figure is a Childhood Bereavement Network estimate based on “mortality statistics, census data, and other sources”^[2]. In 2023 the UK government turned down a petition’s request to record children in the household at death registration, saying “The purpose of a death registration is to record information about the deceased person”^[5]. Worldwide, “Nearly 40% of the world’s deaths are not registered”^[6]. Our study of the missing official data on bereaved children looks at these gaps country by country.

Modelled estimates are careful, but each answers its own question

Where children cannot be counted, researchers estimate them, usually starting from death registrations by age and sex and applying fertility or census data to work out how many children adults of that age have. This careful work is the only way most countries can put any figure on childhood bereavement. But each model rests on assumptions and answers its own question.

The United States has two well-known models with different definitions. The Childhood Bereavement Estimation Model (CBEM), from Judi’s House/JAG Institute, uses CDC vital statistics. Its 2026 figures, based on 2020–2024 data, estimate that 9.1% of US children, about 6.6 million or 1 in 11, will have a parent or a sibling die by age 18^[7]. Judi’s House created the model in partnership with, and with funding from, the New York Life Foundation, the charitable arm of a US life insurer. A peer-reviewed model published in Nature Medicine in 2025 estimated that 2.91 million US children (4.2%) had, by 2021, experienced the death of a parent or a grandparent caregiver^[8]. One counts siblings, the other grandparent caregivers; one looks ahead to age 18, the other takes a snapshot. Neither is wrong, but they should never be compared as though they measured the same thing.

Global estimates add another layer. The first worldwide estimate of children whose parent or other caregiver died in COVID-19-associated deaths (the study’s term) used “mortality and fertility data to model minimum estimates” in 21 countries, then extrapolated. It estimated that 1,134,000 children had a parent or custodial grandparent die from March 2020 to April 2021, with a 95% credible interval of 884,000 to 1,185,000^[9]. Our study of children bereaved during COVID-19 sets out these estimates and their later updates.

Registers follow whole populations, but adjust only for what they record

For questions about what happens to children after a parent dies, the Nordic registers are the strongest evidence anywhere. Because they include everyone, nobody opts in or out, and because they are so large they can study rare outcomes precisely. One study combined everyone born in Denmark from 1968 to 2008 and in Sweden from 1973 to 2006 with a large sample from Finland; 189,094 of them had a parent die before they were 18^[10]. No survey could follow that many bereaved children.

But registers record administrative facts such as deaths, diagnoses, grades and income. They do not record how a family talked about a death or what support a child received. The three-country study named its main limitation as “the lack of data on post-bereavement information on the quality of the parent-child relationship, lifestyles, and common physical environment”^[10].

The bigger issue is confounding. Families in which a parent dies young often differed from other families before the death. In the 1970 British birth cohort, 12% of children in bereaved families had a father in a professional or managerial job when they were born, against 18% in what the study calls intact families^[11]. Such circumstances shape children’s lives whether or not a parent dies. A register study can adjust for what it records, such as income and education, but not for everything, so its figures are carefully adjusted associations, not proof that the death alone caused a difference.

A worked example: watching a gap shrink

A Swedish study followed all 908,064 children born from 1991 to 2000; 22,634 of them, 2.5% of the cohort, had a parent die before age 17^[12]. Before any adjustment, 15.8% of the bereaved children left compulsory school without the grades for upper secondary education, against 9.1% of other children: 1.74 times the risk. Adjusting for parents’ education, income, health and other family factors reduced that to 1.36. The researchers then compared bereaved children with their own brothers and sisters bereaved only after finishing compulsory school, which removes everything siblings share, measured or not. The difference fell to 1.07 and was no longer statistically significant. A smaller gap in average grades remained^[12].

An earlier Swedish study of 772,117 people born from 1973 to 1981 found the same shape: a father’s death went with 2.04 times the odds of school failure before adjustment, and 1.40 times after^[13]. None of this means grief does not matter. It means the raw gap is partly a gap in circumstances, so support may need to take in the family’s wider situation. Our study of school attainment after a parent dies works through this evidence in full.

Cohorts and representative surveys see more of a child’s life, with smaller numbers

Cohort studies follow a chosen group over time, asking questions no register holds. The 1970 British Cohort Study draws on “11,000+ adults who were born in 1970” and interviewed at 30^[11]. In the US, the Add Health study began with “a nationally representative sample of U.S. adolescents in grades 7-12” in 1994–1995, and 14,796 participants later reported their education. Those who had a parent or sibling die had 0.60 times the odds of gaining a bachelor’s degree, in adjusted models^[14].

These studies can measure family life and relationships, which registers cannot. But even a large cohort holds far fewer bereaved children than a register, so estimates are less precise; people drop out over the years; and the British cohort used here was born more than 50 years ago. Definitions vary too: Add Health combines the death of a parent with that of a sibling^[14].

National surveys can estimate how common bereavement is where no register exists. Australia’s second national child mental health survey recruited “a nationally representative sample of households” in 2013–2014; 6,310 parents or carers took part, a 55% response rate. By age 17, 2.5% of young people had had a biological parent die^[15]. That is a sound estimate, but not a count.

In parts of sub-Saharan Africa, population cohorts have filled the gap left by incomplete death registration. In Rakai, Uganda, a community cohort of 92,441 people surveyed from 1995 to 2022 found that orphanhood prevalence (the study’s term for the share of children with one or both parents dead) fell from 21.5% in 2003–04 to 6.3% in 2020–22, with the steepest falls after HIV treatment became available. These figures cover 28 communities, not the whole country^[16].

Self-selected surveys tell us about experience, not about how common it is

Much of what is known about families’ experiences of support comes from surveys open to anyone who wants to take part. They put families’ own words on the record and show where support is missing. What they cannot do is say how common an experience is.

A UK survey of adults bereaved during the COVID-19 pandemic is an openly reported example. Participants “were recruited via media, social media, national associations and community/charitable organisations”, and its analysis of children’s experiences drew on 104 parents and guardians^[17]. Its findings on valued support and unmet need are real. But people who hear about a survey through a charity and choose to answer may differ from bereaved families in general, so any percentage from such a survey describes the people who answered, not everyone.

Of any survey figure, it helps to ask: were people sampled at random or self-selected, what share answered, and which country and year does it describe?

Odds, risks, rates and hazards are not the same number

Studies of bereaved children report their findings in several different measures, and they are easy to mix up:

• Risk ratio (or relative risk): how many times as likely an outcome is in one group as in another. The Swedish school study reported risk ratios^[12].
• Odds ratio: a comparison of odds rather than risks. It is close to the risk ratio only when an outcome is rare; for common outcomes it pushes a difference further from 1 than the risk ratio would. Add Health reported odds ratios^[14].
• Rate ratio and hazard ratio: comparisons of how quickly events happen over time, used by many register studies. The three-country Nordic study reported a mortality rate ratio of 1.50 for people whose parent died in childhood^[10].

All of these are relative. A ratio of 1.50 means 50% higher than the comparison group, which can still be a small absolute risk. Good studies give absolute figures too, as the Swedish school study did with its 15.8% and 9.1%^[12].

The choice of measure can change a conclusion. A 2021 meta-analysis of 15 cohort studies of parental death in childhood and later anxiety, mood and psychotic disorders found pooled hazard ratios of 1.48 and incidence rate ratios of 1.37, both significant, but a pooled odds ratio of 0.87, which was not. The authors judged this “likely an artefact of study design”, and noted the data “were clustered in four countries”^[18].

Only trials can show what helps, and that evidence is thinner

Registers, cohorts and surveys describe what happens to bereaved children, but cannot show whether a kind of support changes it, because families who seek support differ from those who do not. The strongest test is a randomised trial, where chance decides who receives a programme.

A well-known example is the US Family Bereavement Program. Its trial randomly assigned 244 children and young people aged 8 to 16, from 156 families, to a 12-session programme for caregivers and children or to a comparison group given written materials. Fifteen years later, the programme group had 0.332 times the odds of major depression^[19]. It is one trial, of a few hundred children, in one country: a strong result and a thin evidence base at once. Our study of what helps bereaved children looks at the wider evidence.

What the evidence asks of us

• Count bereaved children where it can be done. Sweden and the Netherlands show counting is possible when records link parents and children^[1][3]. In the UK, bereavement organisations are calling on the government “to record official data on the number of children affected by bereavement”^[2], in ways that protect families’ privacy.
• Strengthen the foundations. With nearly 40% of the world’s deaths unregistered^[6], better civil registration is the first step to knowing how many children are bereaved in much of the world.
• Label every number. Charities, services and journalists can help by saying, each time, whether a figure is counted or modelled, which country and year it covers, and who it includes: parents only, or siblings and caregivers too.
• Fund trials of support. Descriptive evidence on bereaved children is strong in a few countries; evidence on what helps is thinner, and even a strong result such as the Family Bereavement Program’s comes from a single trial^[19]. Funders can help close that gap.
• Keep listening to families. Self-selected surveys cannot measure prevalence, but they surface needs that no register records^[17]. Reported clearly as what they are, they belong alongside the numbers.

Language: this piece uses plain words such as “died” rather than euphemisms, because they are clearer.

Why we're publishing this

Bereaved children, their families and the organisations that support them often have to make their case with numbers. We wrote this guide so those numbers can be used with confidence, and so a charity can tell a commissioner exactly what kind of figure it is quoting. The case is theirs; our part is to explain the evidence plainly. This study is part of our research library on children bereaved of a parent.

How to read this data

The figures here are worked examples chosen to illustrate methods, not a full review. The Swedish and Dutch figures are register counts; the UK, US and global COVID-19 figures are modelled estimates, and the two US models use different definitions. The Australian figure comes from a household survey with a 55% response rate, the Ugandan figures from one community cohort, and the UK pandemic survey was self-selected. Effect sizes are given in each study’s own measure and should not be compared across studies. Most register evidence comes from the Nordic countries; we found far less from Asia, Africa and Latin America.

Use this data

Free to cite with attribution to Understanding Bereavement. Charities, schools, services and journalists are welcome to reuse these figures and charts. For the underlying figures as a spreadsheet, get in touch.

Important: This study summarises published research about groups of people for general information. It is not medical advice or bereavement support, and it cannot describe or predict anything about an individual child or family. If you are worried about a child or about yourself, please speak to a GP or another qualified professional.

Sources

1. Statistics Sweden. Children and young persons aged 0–21 whose parents have died during the year by sex, age and which parent who died, 2015–2025 (table LE0102T36). Statistics Sweden, updated June 2026. statistikdatabasen.scb.se/…/LE0102T36
2. Child Bereavement UK. UK death and bereavement statistics (citing Childhood Bereavement Network estimates based on 2019–2021 data). Child Bereavement UK, 2026. childbereavementuk.org/death-bereavement-statistics
3. Statistics Netherlands (CBS). Bijna 100 duizend jongeren halfwees [Almost 100,000 young people have had a parent die]. CBS, 11 October 2022. cbs.nl/nl-nl/nieuws/2022/41/bijna-100-duizend-jongeren-halfwees
4. Schmidt M, Pedersen L, Sørensen HT. The Danish Civil Registration System as a tool in epidemiology. European Journal of Epidemiology, 2014. europepmc.org/article/MED/24965263
5. UK Parliament petitions. Record the number of bereaved children to ensure they are supported (petition 636718), with government response of 12 October 2023. UK Parliament, 2023. petition.parliament.uk/petitions/636718
6. World Health Organization. Civil registration and vital statistics (fact sheet). WHO, 2024. who.int/news-room/fact-sheets/detail/civil-registration-and-vital-statistics
7. Judi’s House/JAG Institute. Childhood Bereavement Estimation Model (CBEM): 2026 national highlights. Judi’s House/JAG Institute, 2026. judishouse.org/research-tools/cbem
8. Villaveces A, Chen Y, Tucker S, et al. Orphanhood and caregiver death among children in the United States by all-cause mortality, 2000–2021. Nature Medicine, 2025. europepmc.org/article/MED/39794613
9. Hillis SD, Unwin HJT, Chen Y, et al. Global minimum estimates of children affected by COVID-19-associated orphanhood and deaths of caregivers: a modelling study. The Lancet, 2021. europepmc.org/article/MED/34298000
10. Li J, Li J, Vestergaard M, et al. Mortality after parental death in childhood: a nationwide cohort study from three Nordic countries. PLoS Medicine, 2014. europepmc.org/article/MED/25051501
11. Parsons S. Long-term impact of childhood bereavement: preliminary analysis of the 1970 British Cohort Study (BCS70). Childhood Wellbeing Research Centre / Department for Education, 2011. assets.publishing.service.gov.uk/…/CWRC-00081-2011.pdf
12. Liu C, Grotta A, Hiyoshi A, Berg L, Rostila M. School outcomes among children following death of a parent. JAMA Network Open, 2022. europepmc.org/article/MED/35394516
13. Berg L, Rostila M, Saarela J, Hjern A. Parental death during childhood and subsequent school performance. Pediatrics, 2014. europepmc.org/article/MED/24616354
14. Thyden NH, Slaughter-Acey J, Widome R, Warren JR, Osypuk TL. Family deaths in the early life course and their association with later educational attainment in a longitudinal cohort study. Social Science & Medicine, 2023. europepmc.org/article/MED/37595424
15. Breen LJ, Chih H, Nastoska S, Skinner S, Lawrence D. Parental loss: prevalence estimates and mental health impacts from the second Australian Child and Adolescent Survey of Mental Health and Wellbeing. Australian Journal of Psychology, 2026. europepmc.org/article/MED/42614886
16. Khalifa A, Malden DE, Ratmann O, et al. Incidence and prevalence of orphanhood in Rakai, Uganda: a population-based cohort study, 1995–2022. The Lancet Global Health, 2026. europepmc.org/article/MED/41519154
17. Harrop E, Goss S, Longo M, et al. Parental perspectives on the grief and support needs of children and young people bereaved during the COVID-19 pandemic: qualitative findings from a national survey. BMC Palliative Care, 2022. europepmc.org/article/MED/36210432
18. McKay MT, Cannon M, Healy C, Syer S, O’Donnell L, Clarke MC. A meta-analysis of the relationship between parental death in childhood and subsequent psychiatric disorder. Acta Psychiatrica Scandinavica, 2021. europepmc.org/article/MED/33604893
19. Sandler I, Tein JY, Zhang N, Wolchik SA. Developmental pathways of the Family Bereavement Program to prevent major depression 15 years later. Journal of the American Academy of Child and Adolescent Psychiatry, 2023. europepmc.org/article/MED/36898606